From 140756ad30361a0a44d5f9d7ec8db9e8684863fc Mon Sep 17 00:00:00 2001 From: Blagodarenko Date: Sun, 20 Aug 2017 14:48:59 +0300 Subject: [PATCH] miner-burst-1.170820 * Optimization of the algorithm - Removed support for burstdev pool (poolV2) - "ShowMsg" parameter was deleted - "ShowUpdates" parameter was deleted - "SendBestOnly" parameter was deleted - "UseFastRcv" parameter was deleted + Added statistics file stat.csv (height; baseTarget; deadline) * bugs fixed --- miner.cpp | 484 ++++---- miner.h | 17 +- miner.vcxproj | 2 +- mshabal256_avx2.c | 2 + mshabal_avx1.c | 50 +- rapidjson/allocators.h | 42 +- rapidjson/document.h | 1081 +++++++++++++---- rapidjson/encodedstream.h | 58 +- rapidjson/encodings.h | 233 ++-- rapidjson/error/en.h | 21 +- rapidjson/error/error.h | 27 +- rapidjson/filereadstream.h | 13 +- rapidjson/filewritestream.h | 17 +- rapidjson/fwd.h | 151 +++ rapidjson/internal/biginteger.h | 14 +- rapidjson/internal/diyfp.h | 23 +- rapidjson/internal/dtoa.h | 50 +- rapidjson/internal/ieee754.h | 3 +- rapidjson/internal/regex.h | 734 +++++++++++ rapidjson/internal/stack.h | 68 +- rapidjson/internal/strfunc.h | 32 +- rapidjson/internal/strtod.h | 39 +- rapidjson/internal/swap.h | 46 + rapidjson/istreamwrapper.h | 115 ++ rapidjson/memorybuffer.h | 2 +- rapidjson/memorystream.h | 16 +- rapidjson/msinttypes/stdint.h | 8 +- rapidjson/ostreamwrapper.h | 81 ++ rapidjson/pointer.h | 1358 +++++++++++++++++++++ rapidjson/prettywriter.h | 98 +- rapidjson/rapidjson.h | 314 +++-- rapidjson/reader.h | 1333 +++++++++++++++----- rapidjson/schema.h | 2016 +++++++++++++++++++++++++++++++ rapidjson/stream.h | 179 +++ rapidjson/stringbuffer.h | 30 +- rapidjson/writer.h | 468 +++++-- 36 files changed, 7999 insertions(+), 1226 deletions(-) create mode 100644 rapidjson/fwd.h create mode 100644 rapidjson/internal/regex.h create mode 100644 rapidjson/internal/swap.h create mode 100644 rapidjson/istreamwrapper.h create mode 100644 rapidjson/ostreamwrapper.h create mode 100644 rapidjson/pointer.h create mode 100644 rapidjson/schema.h create mode 100644 rapidjson/stream.h diff --git a/miner.cpp b/miner.cpp index cc24a53..c0360c6 100644 --- a/miner.cpp +++ b/miner.cpp @@ -25,19 +25,20 @@ void Log_init(void) ss << "Logs\\" << cur_time.wYear << "-" << cur_time.wMonth << "-" << cur_time.wDay << "_" << cur_time.wHour << "_" << cur_time.wMinute << "_" << cur_time.wSecond << ".log"; std::string filename = ss.str(); - if (fopen_s(&fp_Log, filename.c_str(), "wt") != 0) + if ((fp_Log = _fsopen(filename.c_str(), "wt", _SH_DENYNO)) == NULL) { wattron(win_main, COLOR_PAIR(12)); wprintw(win_main, "LOG: file openinig error\n", 0); wattroff(win_main, COLOR_PAIR(12)); use_log = false; } + Log(version); } } void Log(char const *const strLog) { - if (use_log && fp_Log) + if (use_log) { // если строка содержит интер, то добавить время if (strLog[0] == '\n') @@ -53,7 +54,7 @@ void Log(char const *const strLog) void Log_server(char const *const strLog) { size_t len_str = strlen(strLog); - if ((len_str> 0) && use_log && fp_Log) + if ((len_str> 0) && use_log) { char * Msg_log = (char*)HeapAlloc(hHeap, HEAP_ZERO_MEMORY, len_str * 2 + 1); if (Msg_log == nullptr) ShowMemErrorExit(); @@ -90,7 +91,7 @@ void Log_server(char const *const strLog) void Log_llu(unsigned long long const llu_num) { - if (use_log && fp_Log) + if (use_log) { fprintf_s(fp_Log, "%llu", llu_num); fflush(fp_Log); @@ -99,7 +100,7 @@ void Log_llu(unsigned long long const llu_num) void Log_u(size_t const u_num) { - if (use_log && fp_Log) + if (use_log) { fprintf_s(fp_Log, "%u", (unsigned)u_num); fflush(fp_Log); @@ -161,10 +162,7 @@ int load_config(char const *const filename) { Log("\nMode: "); if(strcmp(document["Mode"].GetString(), "solo") == 0) miner_mode = 0; - else - if(strcmp(document["Mode"].GetString(), "pool") == 0) miner_mode = 1; - else - if (strcmp(document["Mode"].GetString(), "poolV2") == 0) miner_mode = 2; + else miner_mode = 1; Log_u(miner_mode); } @@ -198,14 +196,6 @@ int load_config(char const *const filename) if(document.HasMember("UseHDDWakeUp") && (document["UseHDDWakeUp"].IsBool())) use_wakeup = document["UseHDDWakeUp"].GetBool(); Log_u(use_wakeup); - Log("\nShowMsg: "); - if(document.HasMember("ShowMsg") && (document["ShowMsg"].IsBool())) show_msg = document["ShowMsg"].GetBool(); - Log_u(show_msg); - - Log("\nShowUpdates: "); - if(document.HasMember("ShowUpdates") && (document["ShowUpdates"].IsBool())) show_updates = document["ShowUpdates"].GetBool(); - Log_u(show_updates); - Log("\nSendInterval: "); if(document.HasMember("SendInterval") && (document["SendInterval"].IsUint())) send_interval = (size_t)document["SendInterval"].GetUint(); Log_u(send_interval); @@ -214,10 +204,6 @@ int load_config(char const *const filename) if(document.HasMember("UpdateInterval") && (document["UpdateInterval"].IsUint())) update_interval = (size_t)document["UpdateInterval"].GetUint(); Log_u(update_interval); - Log("\nUseFastRcv: "); - if(document.HasMember("UseFastRcv") && (document["UseFastRcv"].IsBool())) use_fast_rcv = document["UseFastRcv"].GetBool(); - Log_u(use_fast_rcv); - Log("\nDebug: "); if(document.HasMember("Debug") && (document["Debug"].IsBool())) use_debug = document["Debug"].GetBool(); Log_u(use_debug); @@ -264,10 +250,6 @@ int load_config(char const *const filename) if (document.HasMember("ShowWinner") && (document["ShowWinner"].IsBool())) show_winner = document["ShowWinner"].GetBool(); Log_u(show_winner); - Log("\nSendBestOnly: "); - if (document.HasMember("SendBestOnly") && (document["SendBestOnly"].IsBool())) send_best_only = document["SendBestOnly"].GetBool(); - Log_u(send_best_only); - Log("\nTargetDeadline: "); if (document.HasMember("TargetDeadline") && (document["TargetDeadline"].IsInt64())) my_target_deadline = document["TargetDeadline"].GetUint64(); Log_llu(my_target_deadline); @@ -487,49 +469,113 @@ size_t Get_index_acc(unsigned long long const key) LeaveCriticalSection(&bestsLock); return bests.size() - 1; } + +//////////////////////// /* void gen_nonce(unsigned long long addr, unsigned long long start_nonce, unsigned long long count) { #define PLOT_SIZE (4096 * 64) #define HASH_SIZE 32 #define HASH_CAP 4096 - char * final = (char *)calloc(32, 1); - char * gendata = (char *)calloc(16 + PLOT_SIZE, 1); + #define SCOOP_SIZE 64 + + unsigned long long nonce1; + unsigned long long nonce2; + unsigned long long nonce3; + unsigned long long nonce4; + //char * final = (char *)calloc(32, 1); + char *final1 = new char[32]; + char *final2 = new char[32]; + char *final3 = new char[32]; + char *final4 = new char[32]; + + char *gendata1 = new char[16 + PLOT_SIZE]; + char *gendata2 = new char[16 + PLOT_SIZE]; + char *gendata3 = new char[16 + PLOT_SIZE]; + char *gendata4 = new char[16 + PLOT_SIZE]; + char * cache = (char *)calloc(64 * count, 1); char *xv = (char*)&addr; - sph_shabal_context x; - size_t i; + mshabal_context *mx = new mshabal_context[sizeof(mshabal_context)]; + size_t len; - //gendata[PLOT_SIZE] = xv[7]; gendata[PLOT_SIZE+1] = xv[6]; gendata[PLOT_SIZE+2] = xv[5]; gendata[PLOT_SIZE+3] = xv[4]; - //gendata[PLOT_SIZE+4] = xv[3]; gendata[PLOT_SIZE+5] = xv[2]; gendata[PLOT_SIZE+6] = xv[1]; gendata[PLOT_SIZE+7] = xv[0]; - for (size_t i = 0; i < 8; i++) gendata[PLOT_SIZE+i] = xv[7-i]; - - for (unsigned long long z = start_nonce; z < (start_nonce + count); z++) { - xv = (char*)&z; - //gendata[PLOT_SIZE + 8] = xv[7]; gendata[PLOT_SIZE + 9] = xv[6]; gendata[PLOT_SIZE + 10] = xv[5]; gendata[PLOT_SIZE + 11] = xv[4]; - //gendata[PLOT_SIZE + 12] = xv[3]; gendata[PLOT_SIZE + 13] = xv[2]; gendata[PLOT_SIZE + 14] = xv[1]; gendata[PLOT_SIZE + 15] = xv[0]; - for (i = 8; i < 16; i++) gendata[PLOT_SIZE + i] = xv[15 - i]; - - for (i = PLOT_SIZE; i > 0; i -= HASH_SIZE) { - sph_shabal256_init(&x); + for (size_t i = 0; i < 8; i++) + { + gendata1[PLOT_SIZE + i] = xv[7 - i]; + gendata2[PLOT_SIZE + i] = xv[7 - i]; + gendata3[PLOT_SIZE + i] = xv[7 - i]; + gendata4[PLOT_SIZE + i] = xv[7 - i]; + } + + for (unsigned long long z = start_nonce; z < (start_nonce + count); z+=4) { +// xv = (char*)&z; +// for (i = 8; i < 16; i++) gendata[PLOT_SIZE + i] = xv[15 - i]; + + nonce1 = z + 0; + nonce2 = z + 1; + nonce3 = z + 2; + nonce4 = z + 3; + char *xv1 = (char*)&nonce1; + char *xv2 = (char*)&nonce2; + char *xv3 = (char*)&nonce3; + char *xv4 = (char*)&nonce4; + for (size_t i = 8; i < 16; i++) + { + gendata1[PLOT_SIZE + i] = xv1[15 - i]; + gendata2[PLOT_SIZE + i] = xv2[15 - i]; + gendata3[PLOT_SIZE + i] = xv3[15 - i]; + gendata4[PLOT_SIZE + i] = xv4[15 - i]; + } + + for (size_t i = PLOT_SIZE; i > 0; i -= HASH_SIZE) + { + avx1_mshabal_init(mx, 256); len = PLOT_SIZE + 16 - i; - if (len > HASH_CAP) len = HASH_CAP; - sph_shabal256(&x, (const unsigned char*)&gendata[i], len); - sph_shabal256_close(&x, &gendata[i - HASH_SIZE]); + if (len > HASH_CAP) len = HASH_CAP; + avx1_mshabal(mx, &gendata1[i], &gendata2[i], &gendata3[i], &gendata4[i], len); + avx1_mshabal_close(mx, 0, 0, 0, 0, 0, &gendata1[i - HASH_SIZE], &gendata2[i - HASH_SIZE], &gendata3[i - HASH_SIZE], &gendata4[i - HASH_SIZE]); + + // ceil because (4096*32*2-32)/64 == 4095,5 but scoop 4095 is not complete yet + //int scoopNumberReady = i / SCOOP_SIZE; + //if (i % SCOOP_SIZE) ++scoopNumberReady; + //if (scoopNumberReady == scoop) break; + if ((i / SCOOP_SIZE == scoop) && (i % SCOOP_SIZE == 0)) break; + } - sph_shabal256_init(&x); - sph_shabal256(&x, (const unsigned char*)gendata, 16 + PLOT_SIZE); - sph_shabal256_close(&x, final); + + //avx1_mshabal_init(mx, 256); + //avx1_mshabal(mx, gendata1, gendata2, gendata3, gendata4, 16 + PLOT_SIZE); + //avx1_mshabal_close(mx, 0, 0, 0, 0, 0, final1, final2, final3, final4); // XOR with final - //for(i = 0; i < PLOT_SIZE; i ++) gendata[i] ^= (final[i % HASH_SIZE]); - //for (i = scoop * 64; i < ((scoop + 1) * 64); i++) gendata[i] ^= (final[i % HASH_SIZE]); - gendata[scoop * 64] ^= (final[(scoop * 64) % HASH_SIZE]); - memmove(&cache[(z - start_nonce) * 64], &gendata[scoop * 64], 64); + for (size_t i = scoop * SCOOP_SIZE; i < scoop * SCOOP_SIZE + SCOOP_SIZE; i++) + { + gendata1[i] ^= (final1[i % 32]); + gendata2[i] ^= (final2[i % 32]); + gendata3[i] ^= (final3[i % 32]); + gendata4[i] ^= (final4[i % 32]); + } + gendata1[scoop * 64] ^= (final1[(scoop * 64) % HASH_SIZE]); + gendata2[scoop * 64] ^= (final2[(scoop * 64) % HASH_SIZE]); + gendata3[scoop * 64] ^= (final3[(scoop * 64) % HASH_SIZE]); + gendata4[scoop * 64] ^= (final4[(scoop * 64) % HASH_SIZE]); + + memmove(&cache[(z - start_nonce) * 64], &gendata1[scoop * 64], 64); + memmove(&cache[(z - start_nonce + 1) * 64], &gendata2[scoop * 64], 64); + memmove(&cache[(z - start_nonce + 2) * 64], &gendata3[scoop * 64], 64); + memmove(&cache[(z - start_nonce + 3) * 64], &gendata4[scoop * 64], 64); } - free(final); - free(gendata); - + //free(final); + //free(gendata); + delete[] gendata1; + delete[] gendata2; + delete[] gendata3; + delete[] gendata4; + delete[] final1; + delete[] final2; + delete[] final3; + delete[] final4; + delete[] mx; //acc = Get_index_acc(key); //procscoop_sph(n, size, cache, 0, std::string("generator")); @@ -539,24 +585,25 @@ void gen_nonce(unsigned long long addr, unsigned long long start_nonce, unsigned void generator_i(size_t number) { - unsigned long long start_nonce = 100000000 * (number + 1);// * (local_num + 1); - unsigned long long size = 400; + unsigned long long start_nonce = 10000000000 * (number + 1);// * (local_num + 1); + unsigned long long size = 1000; clock_t start_work_time; - //wprintw(win_main, "\ngenerator RUNING !"); + wprintw(win_main, "\ngenerator RUNING !"); while (!stopThreads) { start_work_time = clock(); gen_nonce(bests[0].account_id, start_nonce, size); - wprintw(win_main, "\n%llu\tnoces/min: %f", start_nonce, (float)((float)size * CLOCKS_PER_SEC * 60 / (float)(clock() - start_work_time))); + wprintw(win_main, "\n%llu\tnoces/min: %f (scoop %llu)", start_nonce, (float)((float)size * CLOCKS_PER_SEC * 60 / (float)(clock() - start_work_time)), scoop); //wrefresh(win_main); start_nonce = start_nonce + size; }; return; } -*/ +*/ +//////////////////////////// void proxy_i(void) { @@ -647,7 +694,7 @@ void proxy_i(void) RtlSecureZeroMemory(tmp_buffer, buffer_size); iResult = recv(ClientSocket, tmp_buffer, (int)(buffer_size - 1), 0); strcat_s(buffer, buffer_size, tmp_buffer); - } while ((iResult > 0) && !use_fast_rcv); + } while (iResult > 0); Log("\nProxy get info: "); Log_server(buffer); unsigned long long get_accountId = 0; @@ -758,7 +805,7 @@ void proxy_i(void) { find[0] = 0; wattron(win_main, COLOR_PAIR(15)); - wprintw(win_main, "PROXY: %s\n", buffer, 0); + wprintw(win_main, "PROXY: %s\n", buffer, 0);//You can crash the miner when the proxy is enabled and you open the address in a browser. wprintw(win_main, "PROXY: %s\n", "Error", 0); wattroff(win_main, COLOR_PAIR(15)); } } @@ -799,8 +846,7 @@ void send_i(void) for (auto iter = shares.begin(); iter != shares.end();) { - if (send_best_only) //Гасим шару если она больше текущего targetDeadline, актуально для режима Proxy - { + //Гасим шару если она больше текущего targetDeadline, актуально для режима Proxy if ((iter->best / baseTarget) > bests[Get_index_acc(iter->account_id)].targetDeadline) { if (use_debug) @@ -815,7 +861,6 @@ void send_i(void) LeaveCriticalSection(&sharesLock); continue; } - } RtlSecureZeroMemory(&hints, sizeof(hints)); hints.ai_family = AF_INET; @@ -867,24 +912,10 @@ void send_i(void) { unsigned long long total = total_size / 1024 / 1024 / 1024; for (auto It = satellite_size.begin(); It != satellite_size.end(); ++It) total = total + It->second; - bytes = sprintf_s(buffer, buffer_size, "POST /burst?requestType=submitNonce&accountId=%llu&nonce=%llu&deadline=%llu HTTP/1.0\r\nHost: %s:%s\r\nX-Miner: Blago %s\r\nX-Capacity: %llu\r\nContent-Length: 0\r\nConnection: close\r\n\r\n", iter->account_id, iter->nonce, iter->best, nodeaddr.c_str(), nodeport.c_str(), version, total, 0); - } - if (miner_mode == 2) - { - char* f1 = (char*)HeapAlloc(hHeap, HEAP_ZERO_MEMORY, MAX_PATH); - char* str_len = (char*)HeapAlloc(hHeap, HEAP_ZERO_MEMORY, MAX_PATH); - if ((f1 == nullptr) || (str_len == nullptr)) ShowMemErrorExit(); - - int len = sprintf_s(f1, MAX_PATH, "%llu:%llu:%llu", iter->account_id, iter->nonce, height); - _itoa_s(len, str_len, MAX_PATH - 1, 10); - - bytes = sprintf_s(buffer, buffer_size, "POST /pool/submitWork HTTP/1.0\r\nHost: %s:%s\r\nContent-Type: text/plain;charset=UTF-8\r\nContent-Length: %i\r\n\r\n%s", nodeaddr.c_str(), nodeport.c_str(), len, f1); - HeapFree(hHeap, 0, f1); - HeapFree(hHeap, 0, str_len); + bytes = sprintf_s(buffer, buffer_size, "POST /burst?requestType=submitNonce&accountId=%llu&nonce=%llu&deadline=%llu HTTP/1.0\r\nHost: %s:%s\r\nX-Miner: Blago %s\r\nX-Capacity: %llu\r\nContent-Length: 0\r\nConnection: close\r\n\r\n", iter->account_id, iter->nonce, iter->best, nodeaddr.c_str(), nodeport.c_str(), version, total); } // Sending to server - iResult = send(ConnectSocket, buffer, bytes, 0); if (iResult == SOCKET_ERROR) { @@ -904,15 +935,12 @@ void send_i(void) wprintw(win_main, "%s [%20llu] sent DL: %15llu %5llud %02llu:%02llu:%02llu\n", tbuffer, iter->account_id, dl, (dl) / (24 * 60 * 60), (dl % (24 * 60 * 60)) / (60 * 60), (dl % (60 * 60)) / 60, dl % 60, 0); wattroff(win_main, COLOR_PAIR(9)); - if (show_msg) wprintw(win_main, "send: %s\n", buffer, 0); // показываем послание - Log("\nSender: Sent: "); Log_server(buffer); - EnterCriticalSection(&sessionsLock); //sessions.push_back({ ConnectSocket, iter->account_id, dl, iter->best, iter->nonce }); sessions.push_back({ ConnectSocket, dl, *iter }); LeaveCriticalSection(&sessionsLock); - if (send_best_only) bests[Get_index_acc(iter->account_id)].targetDeadline = dl; + bests[Get_index_acc(iter->account_id)].targetDeadline = dl; EnterCriticalSection(&sharesLock); iter = shares.erase(iter); LeaveCriticalSection(&sharesLock); @@ -944,7 +972,7 @@ void send_i(void) do{ iResult = recv(ConnectSocket, &buffer[pos], (int)(buffer_size - pos - 1), 0); if (iResult > 0) pos += (size_t)iResult; - } while ((iResult > 0) && !use_fast_rcv); + } while (iResult > 0); if (iResult == SOCKET_ERROR) { @@ -961,11 +989,8 @@ void send_i(void) } else //что-то получили от сервера { - if (show_msg) wprintw(win_main, "\nReceived: %s\n", buffer, 0); - Log("\nSender: Received: "); Log_server(buffer); if (network_quality < 100) network_quality++; - //получили пустую строку, переотправляем дедлайн if (buffer[0] == '\0') { @@ -1125,7 +1150,8 @@ void procscoop_m_4(unsigned long long const nonce, unsigned long long const n, c char res2[32]; char res3[32]; unsigned posn; - mshabal_context x; + mshabal_context x, init_x; + avx1_mshabal_init(&init_x, 256); for (unsigned long long v = 0; v < n; v += 4) { @@ -1133,9 +1159,8 @@ void procscoop_m_4(unsigned long long const nonce, unsigned long long const n, c memcpy(&sig1[32], &cache[(v + 1) * 64], 64); memcpy(&sig2[32], &cache[(v + 2) * 64], 64); memcpy(&sig3[32], &cache[(v + 3) * 64], 64); - - avx1_mshabal_init(&x, 256); + memcpy(&x, &init_x, sizeof(init_x)); // optimization: avx1_mshabal_init(&x, 256); avx1_mshabal(&x, (const unsigned char*)sig0, (const unsigned char*)sig1, (const unsigned char*)sig2, (const unsigned char*)sig3, 64 + 32); avx1_mshabal_close(&x, 0, 0, 0, 0, 0, res0, res1, res2, res3); @@ -1163,8 +1188,6 @@ void procscoop_m_4(unsigned long long const nonce, unsigned long long const n, c if ((*wertung / baseTarget) <= bests[acc].targetDeadline) { - if (send_best_only) - { if (bests[acc].nonce == 0 || *wertung < bests[acc].best) { Log("\nfound deadline="); Log_llu(*wertung / baseTarget); Log(" nonce="); Log_llu(nonce + v + posn); Log(" for account: "); Log_llu(bests[acc].account_id); Log(" file: "); Log((char*)file_name.c_str()); @@ -1185,30 +1208,6 @@ void procscoop_m_4(unsigned long long const nonce, unsigned long long const n, c wattroff(win_main, COLOR_PAIR(2)); } } - } - else - { - if (bests[acc].nonce == 0 || *wertung < bests[acc].best) - { - Log("\nfound deadline="); Log_llu(*wertung / baseTarget); Log(" nonce="); Log_llu(nonce + v); Log(" for account: "); Log_llu(bests[acc].account_id); Log(" file: "); Log((char*)file_name.c_str()); - EnterCriticalSection(&bestsLock); - bests[acc].best = *wertung; - bests[acc].nonce = nonce + v + posn; - bests[acc].DL = *wertung / baseTarget; - LeaveCriticalSection(&bestsLock); - } - EnterCriticalSection(&sharesLock); - shares.push_back({ file_name, bests[acc].account_id, *wertung, nonce + v + posn }); - LeaveCriticalSection(&sharesLock); - if (use_debug) - { - char tbuffer[9]; - _strtime_s(tbuffer); - wattron(win_main, COLOR_PAIR(2)); - wprintw(win_main, "%s [%20llu] found DL: %9llu\n", tbuffer, bests[acc].account_id, *wertung / baseTarget, 0); - wattroff(win_main, COLOR_PAIR(2)); - } - } } } } @@ -1223,10 +1222,17 @@ void procscoop_m256_8(unsigned long long const nonce, unsigned long long const n char sig5[32 + 64]; char sig6[32 + 64]; char sig7[32 + 64]; + char res0[32]; + char res1[32]; + char res2[32]; + char res3[32]; + char res4[32]; + char res5[32]; + char res6[32]; + char res7[32]; cache = data; unsigned long long v; - char tbuffer[9]; - + memmove(sig0, signature, 32); memmove(sig1, signature, 32); memmove(sig2, signature, 32); @@ -1236,6 +1242,9 @@ void procscoop_m256_8(unsigned long long const nonce, unsigned long long const n memmove(sig6, signature, 32); memmove(sig7, signature, 32); + mshabal256_context x, init_x; + mshabal256_init(&init_x, 256); + for (v = 0; v stagger) { cache_size_local = stagger - i; // остаток - //wprintw(win_main, "%llu\n", cache_size_local); + #ifdef __AVX2__ + if (cache_size_local < 8) + { + wattron(win_main, COLOR_PAIR(12)); + wprintw(win_main, "WARNING: %llu\n", cache_size_local); + wattroff(win_main, COLOR_PAIR(12)); + } + #else + #ifdef __AVX__ + if (cache_size_local < 4) + { + wattron(win_main, COLOR_PAIR(12)); + wprintw(win_main, "WARNING: %llu\n", cache_size_local); + wattroff(win_main, COLOR_PAIR(12)); + } + #endif + #endif } bytes = 0; b = 0; @@ -2030,9 +1971,7 @@ void pollLocal(void) { Log("\n*! GMI: connect function failed with error: "); Log_u(WSAGetLastError()); } else { - int bytes; - if (miner_mode == 2) bytes = sprintf_s(buffer, buffer_size, "GET /pool/getMiningInfo HTTP/1.0\r\nHost: %s:%s\r\nContent-Type: text/plain;charset=UTF-8\r\n\r\n", updateraddr.c_str(), updaterport.c_str()); - else bytes = sprintf_s(buffer, buffer_size, "POST /burst?requestType=getMiningInfo HTTP/1.0\r\nHost: %s:%s\r\nContent-Length: 0\r\nConnection: close\r\n\r\n", nodeaddr.c_str(), nodeport.c_str()); + int bytes = sprintf_s(buffer, buffer_size, "POST /burst?requestType=getMiningInfo HTTP/1.0\r\nHost: %s:%s\r\nContent-Length: 0\r\nConnection: close\r\n\r\n", nodeaddr.c_str(), nodeport.c_str()); iResult = send(UpdaterSocket, buffer, bytes, 0); if (iResult == SOCKET_ERROR) { @@ -2040,17 +1979,13 @@ void pollLocal(void) { Log("\n*! GMI: send request failed: "); Log_u(WSAGetLastError()); } else{ - if (show_updates) wprintw(win_main, "Sent: \n%s\n", buffer, 0); - Log("\n* GMI: Sent: "); Log_server(buffer); - - RtlSecureZeroMemory(buffer, buffer_size); size_t pos = 0; iResult = 0; do{ iResult = recv(UpdaterSocket, &buffer[pos], (int)(buffer_size - pos - 1), 0); if (iResult > 0) pos += (size_t)iResult; - } while ((iResult > 0) && !use_fast_rcv); + } while (iResult > 0); if (iResult == SOCKET_ERROR) { if (network_quality > 0) network_quality--; @@ -2059,8 +1994,7 @@ void pollLocal(void) { else { if (network_quality < 100) network_quality++; Log("\n* GMI: Received: "); Log_server(buffer); - if (show_updates) wprintw(win_main, "Received: %s\n", buffer, 0); - + // locate HTTP header char *find = strstr(buffer, "\r\n\r\n"); if (find == nullptr) Log("\n*! GMI: error message from pool"); @@ -2104,6 +2038,100 @@ void pollLocal(void) { HeapFree(hHeap, 0, buffer); } + +void pollLocal2(void) { + size_t const buffer_size = 1000; + char *buffer = (char*)HeapAlloc(hHeap, HEAP_ZERO_MEMORY, buffer_size); + if (buffer == nullptr) ShowMemErrorExit(); + + int iResult = 0; + SOCKET UpdaterSocket = INVALID_SOCKET; + SOCKADDR_STORAGE LocalAddr = { 0 }; + SOCKADDR_STORAGE RemoteAddr = { 0 }; + DWORD dwLocalAddr = sizeof(LocalAddr); + DWORD dwRemoteAddr = sizeof(RemoteAddr); + BOOL bSuccess; + + UpdaterSocket = socket(AF_INET, SOCK_STREAM, 0); + timeval timeout; + timeout.tv_sec = 1; + timeout.tv_usec = 0; + + bSuccess = WSAConnectByNameA(UpdaterSocket, (LPCSTR)updateraddr.c_str(), (LPCSTR)updaterport.c_str(), &dwLocalAddr, (SOCKADDR*)&LocalAddr, &dwRemoteAddr, (SOCKADDR*)&RemoteAddr, &timeout, NULL); + if (!bSuccess) { + if (network_quality > 0) network_quality--; + Log("\n*! GMI: WsaConnectByName failed with error: "); Log_u(WSAGetLastError()); + Log(updateraddr.c_str()); + } + else { + setsockopt(UpdaterSocket, SOL_SOCKET, SO_UPDATE_CONNECT_CONTEXT, NULL, 0); + + int bytes = sprintf_s(buffer, buffer_size, "POST /burst?requestType=getMiningInfo HTTP/1.0\r\nHost: %s:%s\r\nContent-Length: 0\r\nConnection: close\r\n\r\n", nodeaddr.c_str(), nodeport.c_str()); + iResult = send(UpdaterSocket, buffer, bytes, 0); + if (iResult == SOCKET_ERROR) + { + if (network_quality > 0) network_quality--; + Log("\n*! GMI: send request failed: "); Log_u(WSAGetLastError()); + } + else{ + RtlSecureZeroMemory(buffer, buffer_size); + size_t pos = 0; + iResult = 0; + do{ + iResult = recv(UpdaterSocket, &buffer[pos], (int)(buffer_size - pos - 1), 0); + if (iResult > 0) pos += (size_t)iResult; + } while (iResult > 0); + if (iResult == SOCKET_ERROR) + { + if (network_quality > 0) network_quality--; + Log("\n*! GMI: get mining info failed:: "); Log_u(WSAGetLastError()); + } + else { + if (network_quality < 100) network_quality++; + Log("\n* GMI: Received: "); Log_server(buffer); + + // locate HTTP header + char *find = strstr(buffer, "\r\n\r\n"); + if (find == nullptr) Log("\n*! GMI: error message from pool"); + else { + rapidjson::Document gmi; + if (gmi.Parse<0>(find).HasParseError()) Log("\n*! GMI: error parsing JSON message from pool"); + else { + if (gmi.IsObject()) + { + if (gmi.HasMember("baseTarget")) { + if (gmi["baseTarget"].IsString()) baseTarget = _strtoui64(gmi["baseTarget"].GetString(), 0, 10); + else + if (gmi["baseTarget"].IsInt64()) baseTarget = gmi["baseTarget"].GetInt64(); + } + + if (gmi.HasMember("height")) { + if (gmi["height"].IsString()) height = _strtoui64(gmi["height"].GetString(), 0, 10); + else + if (gmi["height"].IsInt64()) height = gmi["height"].GetInt64(); + } + + if (gmi.HasMember("generationSignature")) { + strcpy_s(str_signature, gmi["generationSignature"].GetString()); + if (xstr2strr(signature, 33, gmi["generationSignature"].GetString()) == 0) Log("\n*! GMI: Node response: Error decoding generationsignature\n"); + } + if (gmi.HasMember("targetDeadline")) { + if (gmi["targetDeadline"].IsString()) targetDeadlineInfo = _strtoui64(gmi["targetDeadline"].GetString(), 0, 10); + else + if (gmi["targetDeadline"].IsInt64()) targetDeadlineInfo = gmi["targetDeadline"].GetInt64(); + } + } + } + } + } + } + } + closesocket(UpdaterSocket); + HeapFree(hHeap, 0, buffer); +} + + + void updater_i(void) { if (updateraddr.length() <= 3) { Log("\nGMI: ERROR in UpdaterAddr"); @@ -2648,7 +2676,7 @@ int main(int argc, char **argv) { double pcFreq = double(li.QuadPart); std::thread proxy; - //std::vector generator; + std::vector generator; InitializeCriticalSection(&sessionsLock); InitializeCriticalSection(&bestsLock); @@ -2656,6 +2684,7 @@ int main(int argc, char **argv) { char tbuffer[9]; unsigned long long bytesRead = 0; + FILE * pFileStat; shares.reserve(20); bests.reserve(4); @@ -3013,9 +3042,9 @@ int main(int argc, char **argv) { wmove(win_progress, 1, 1); wattron(win_progress, COLOR_PAIR(14)); if (deadline == 0) - wprintw(win_progress, "%3llu%% %6llu GB (%.2f MB/s). no deadline Network quality: %3u%%", (bytesRead * 4096 * 100 / total_size), (bytesRead / (256 * 1024)), threads_speed, network_quality, 0); + wprintw(win_progress, "%3llu%% %6llu GB (%.2f MB/s). no deadline Connection: %3u%%", (bytesRead * 4096 * 100 / total_size), (bytesRead / (256 * 1024)), threads_speed, network_quality, 0); else - wprintw(win_progress, "%3llu%% %6llu GB (%.2f MB/s). Deadline =%10llu Network quality: %3u%%", (bytesRead * 4096 * 100 / total_size), (bytesRead / (256 * 1024)), threads_speed, deadline, network_quality, 0); + wprintw(win_progress, "%3llu%% %6llu GB (%.2f MB/s). Deadline =%10llu Connection: %3u%%", (bytesRead * 4096 * 100 / total_size), (bytesRead / (256 * 1024)), threads_speed, deadline, network_quality, 0); wattroff(win_progress, COLOR_PAIR(14)); wrefresh(win_main); @@ -3037,6 +3066,7 @@ int main(int argc, char **argv) { Log("\nInterrupt Sender. "); if (sender.joinable()) sender.join(); + /* if (can_generate != 0){ Log("\nInterrupt Generator. "); @@ -3046,6 +3076,16 @@ int main(int argc, char **argv) { can_generate = 1; } */ + + + fopen_s(&pFileStat, "stat.csv", "a+t"); + if (pFileStat != nullptr) + { + fprintf(pFileStat, "%llu;%llu;%llu\n", old_height, old_baseTarget, deadline); + fclose(pFileStat); + } + + } if (pass != nullptr) HeapFree(hHeap, 0, pass); diff --git a/miner.h b/miner.h index 9869b6c..1af8480 100644 --- a/miner.h +++ b/miner.h @@ -19,6 +19,7 @@ using namespace rapidjson; #pragma comment(lib,"Ws2_32.lib") #include +#include // Need for SO_UPDATE_CONNECT_CONTEXT #include "curses.h" //#include "panel.h" @@ -52,12 +53,12 @@ HANDLE hHeap; bool exit_flag = false; #ifdef __AVX2__ - char const *const version = "v1.170603_AVX2"; + char const *const version = "v1.170820_AVX2"; #else #ifdef __AVX__ - char const *const version = "v1.170603_AVX"; + char const *const version = "v1.170820_AVX"; #else - char const *const version = "v1.170603"; + char const *const version = "v1.170820"; #endif #endif @@ -91,22 +92,22 @@ volatile int stopThreads = 0; size_t miner_mode = 0; // . 0=, 1= size_t cache_size = 100000; // std::vector paths_dir; // -bool show_msg = false; // -bool show_updates = false; // +//bool show_msg = false; // +//bool show_updates = false; // FILE * fp_Log = nullptr; // - size_t send_interval = 100; // size_t update_interval = 1000; // short win_size_x = 80; short win_size_y = 60; -bool use_fast_rcv = false; +//bool use_fast_rcv = false; bool use_debug = false; bool enable_proxy = false; -bool send_best_only = true; +//bool send_best_only = true; bool use_wakeup = false; bool use_log = true; // bool use_boost = false; // bool show_winner = false; // -//short can_generate = 1; // 0 - disable; 1 - can start generate; 2 - already run generate +//short can_generate = 0; // 0 - disable; 1 - can start generate; 2 - already run generator SYSTEMTIME cur_time; // diff --git a/miner.vcxproj b/miner.vcxproj index 674d944..56e15b8 100644 --- a/miner.vcxproj +++ b/miner.vcxproj @@ -175,7 +175,7 @@ Speed true true - AdvancedVectorExtensions + StreamingSIMDExtensions true AssemblyAndSourceCode true diff --git a/mshabal256_avx2.c b/mshabal256_avx2.c index 82c66cf..27581ae 100644 --- a/mshabal256_avx2.c +++ b/mshabal256_avx2.c @@ -44,6 +44,7 @@ extern "C" { const unsigned char *buf6, const unsigned char *buf7, size_t num) { + _mm256_zeroupper(); union { u32 words[64 * MSHABAL256_FACTOR]; __m256i data[16]; @@ -235,6 +236,7 @@ extern "C" { _mm256_storeu_si256((__m256i *)sc->state + j + 28, C[j]); } + _mm256_zeroupper(); #undef M } diff --git a/mshabal_avx1.c b/mshabal_avx1.c index f1d3b37..aafd899 100644 --- a/mshabal_avx1.c +++ b/mshabal_avx1.c @@ -222,7 +222,7 @@ extern "C" { _mm_storeu_si128((__m128i *)sc->state + j + 12, B[j]); _mm_storeu_si128((__m128i *)sc->state + j + 28, C[j]); } - + // _mm256_zeroupper(); #undef M } @@ -265,53 +265,6 @@ extern "C" { sc->out_size = out_size; } - - - static const mshabal_u32 A_init_256[] = { - C32(0x52F84552), C32(0xE54B7999), C32(0x2D8EE3EC), C32(0xB9645191), - C32(0xE0078B86), C32(0xBB7C44C9), C32(0xD2B5C1CA), C32(0xB0D2EB8C), - C32(0x14CE5A45), C32(0x22AF50DC), C32(0xEFFDBC6B), C32(0xEB21B74A) - }; - - static const mshabal_u32 B_init_256[] = { - C32(0xB555C6EE), C32(0x3E710596), C32(0xA72A652F), C32(0x9301515F), - C32(0xDA28C1FA), C32(0x696FD868), C32(0x9CB6BF72), C32(0x0AFE4002), - C32(0xA6E03615), C32(0x5138C1D4), C32(0xBE216306), C32(0xB38B8890), - C32(0x3EA8B96B), C32(0x3299ACE4), C32(0x30924DD4), C32(0x55CB34A5) - }; - - static const mshabal_u32 C_init_256[] = { - C32(0xB405F031), C32(0xC4233EBA), C32(0xB3733979), C32(0xC0DD9D55), - C32(0xC51C28AE), C32(0xA327B8E1), C32(0x56C56167), C32(0xED614433), - C32(0x88B59D60), C32(0x60E2CEBA), C32(0x758B4B8B), C32(0x83E82A7F), - C32(0xBC968828), C32(0xE6E00BF7), C32(0xBA839E55), C32(0x9B491C60) - }; - - - void avx1_mshabal_init2(mshabal_context *cc, unsigned out_size) - { - //for (unsigned u = 0; u < 176; u++) cc->state[u] = 0; - - memset(cc->buf0, 0, sizeof cc->buf0); - memset(cc->buf1, 0, sizeof cc->buf1); - memset(cc->buf2, 0, sizeof cc->buf2); - memset(cc->buf3, 0, sizeof cc->buf3); - - for (unsigned u = 0; u < 176; u = u + 12 + 16 + 16) - { - memcpy(cc->state + u, A_init_256, sizeof(u32) * 12); - memcpy(cc->state + u + 12, B_init_256, sizeof(u32) * 16); - memcpy(cc->state + u + 12 + 16, C_init_256, sizeof(u32) * 16); - } - - cc->Wlow = 1; - cc->Whigh = 0; - cc->ptr = 0; - //cc->Whigh = cc->Wlow = C32(0xFFFFFFFF); - cc->out_size = out_size; - } - - /* see shabal_small.h */ void avx1_mshabal(mshabal_context *sc, const void *data0, const void *data1, @@ -435,7 +388,6 @@ extern "C" { out = (u32*)dst3; for (z = 0; z < out_size_w32; z++) out[(size_t)z] = sc->state[off + (size_t)(z << 2) + 3]; } - _mm256_zeroupper(); } //#pragma optimize("", on) diff --git a/rapidjson/allocators.h b/rapidjson/allocators.h index d74a671..655f4a3 100644 --- a/rapidjson/allocators.h +++ b/rapidjson/allocators.h @@ -179,9 +179,10 @@ class MemoryPoolAllocator { size = RAPIDJSON_ALIGN(size); if (chunkHead_ == 0 || chunkHead_->size + size > chunkHead_->capacity) - AddChunk(chunk_capacity_ > size ? chunk_capacity_ : size); + if (!AddChunk(chunk_capacity_ > size ? chunk_capacity_ : size)) + return NULL; - void *buffer = reinterpret_cast(chunkHead_ + 1) + chunkHead_->size; + void *buffer = reinterpret_cast(chunkHead_) + RAPIDJSON_ALIGN(sizeof(ChunkHeader)) + chunkHead_->size; chunkHead_->size += size; return buffer; } @@ -194,14 +195,16 @@ class MemoryPoolAllocator { if (newSize == 0) return NULL; + originalSize = RAPIDJSON_ALIGN(originalSize); + newSize = RAPIDJSON_ALIGN(newSize); + // Do not shrink if new size is smaller than original if (originalSize >= newSize) return originalPtr; // Simply expand it if it is the last allocation and there is sufficient space - if (originalPtr == (char *)(chunkHead_ + 1) + chunkHead_->size - originalSize) { + if (originalPtr == reinterpret_cast(chunkHead_) + RAPIDJSON_ALIGN(sizeof(ChunkHeader)) + chunkHead_->size - originalSize) { size_t increment = static_cast(newSize - originalSize); - increment = RAPIDJSON_ALIGN(increment); if (chunkHead_->size + increment <= chunkHead_->capacity) { chunkHead_->size += increment; return originalPtr; @@ -209,11 +212,13 @@ class MemoryPoolAllocator { } // Realloc process: allocate and copy memory, do not free original buffer. - void* newBuffer = Malloc(newSize); - RAPIDJSON_ASSERT(newBuffer != 0); // Do not handle out-of-memory explicitly. - if (originalSize) - std::memcpy(newBuffer, originalPtr, originalSize); - return newBuffer; + if (void* newBuffer = Malloc(newSize)) { + if (originalSize) + std::memcpy(newBuffer, originalPtr, originalSize); + return newBuffer; + } + else + return NULL; } //! Frees a memory block (concept Allocator) @@ -227,15 +232,20 @@ class MemoryPoolAllocator { //! Creates a new chunk. /*! \param capacity Capacity of the chunk in bytes. + \return true if success. */ - void AddChunk(size_t capacity) { + bool AddChunk(size_t capacity) { if (!baseAllocator_) - ownBaseAllocator_ = baseAllocator_ = RAPIDJSON_NEW(BaseAllocator()); - ChunkHeader* chunk = reinterpret_cast(baseAllocator_->Malloc(sizeof(ChunkHeader) + capacity)); - chunk->capacity = capacity; - chunk->size = 0; - chunk->next = chunkHead_; - chunkHead_ = chunk; + ownBaseAllocator_ = baseAllocator_ = RAPIDJSON_NEW(BaseAllocator)(); + if (ChunkHeader* chunk = reinterpret_cast(baseAllocator_->Malloc(RAPIDJSON_ALIGN(sizeof(ChunkHeader)) + capacity))) { + chunk->capacity = capacity; + chunk->size = 0; + chunk->next = chunkHead_; + chunkHead_ = chunk; + return true; + } + else + return false; } static const int kDefaultChunkCapacity = 64 * 1024; //!< Default chunk capacity. diff --git a/rapidjson/document.h b/rapidjson/document.h index c6acbd9..3133a2f 100644 --- a/rapidjson/document.h +++ b/rapidjson/document.h @@ -20,40 +20,37 @@ #include "reader.h" #include "internal/meta.h" #include "internal/strfunc.h" +#include "memorystream.h" +#include "encodedstream.h" #include // placement new +#include -#ifdef _MSC_VER RAPIDJSON_DIAG_PUSH +#ifdef _MSC_VER RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant -#elif defined(__GNUC__) -RAPIDJSON_DIAG_PUSH -RAPIDJSON_DIAG_OFF(effc++) +RAPIDJSON_DIAG_OFF(4244) // conversion from kXxxFlags to 'uint16_t', possible loss of data +#ifdef _MINWINDEF_ // see: http://stackoverflow.com/questions/22744262/cant-call-stdmax-because-minwindef-h-defines-max +#ifndef NOMINMAX +#pragma push_macro("min") +#pragma push_macro("max") +#undef min +#undef max +#endif #endif - -/////////////////////////////////////////////////////////////////////////////// -// RAPIDJSON_HAS_STDSTRING - -#ifndef RAPIDJSON_HAS_STDSTRING -#ifdef RAPIDJSON_DOXYGEN_RUNNING -#define RAPIDJSON_HAS_STDSTRING 1 // force generation of documentation -#else -#define RAPIDJSON_HAS_STDSTRING 0 // no std::string support by default #endif -/*! \def RAPIDJSON_HAS_STDSTRING - \ingroup RAPIDJSON_CONFIG - \brief Enable RapidJSON support for \c std::string - - By defining this preprocessor symbol to \c 1, several convenience functions for using - \ref rapidjson::GenericValue with \c std::string are enabled, especially - for construction and comparison. - \hideinitializer -*/ -#endif // !defined(RAPIDJSON_HAS_STDSTRING) +#ifdef __clang__ +RAPIDJSON_DIAG_OFF(padded) +RAPIDJSON_DIAG_OFF(switch-enum) +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif -#if RAPIDJSON_HAS_STDSTRING -#include -#endif // RAPIDJSON_HAS_STDSTRING +#ifdef __GNUC__ +RAPIDJSON_DIAG_OFF(effc++) +#if __GNUC__ >= 6 +RAPIDJSON_DIAG_OFF(terminate) // ignore throwing RAPIDJSON_ASSERT in RAPIDJSON_NOEXCEPT functions +#endif +#endif // __GNUC__ #ifndef RAPIDJSON_NOMEMBERITERATORCLASS #include // std::iterator, std::random_access_iterator_tag @@ -69,6 +66,9 @@ RAPIDJSON_NAMESPACE_BEGIN template class GenericValue; +template +class GenericDocument; + //! Name-value pair in a JSON object value. /*! This class was internal to GenericValue. It used to be a inner struct. @@ -155,6 +155,7 @@ class GenericMemberIterator Otherwise, the copy constructor is implicitly defined. */ GenericMemberIterator(const NonConstIterator & it) : ptr_(it.ptr_) {} + Iterator& operator=(const NonConstIterator & it) { ptr_ = it.ptr_; return *this; } //! @name stepping //@{ @@ -257,6 +258,7 @@ struct GenericStringRef { typedef CharType Ch; //!< character type of the string //! Create string reference from \c const character array +#ifndef __clang__ // -Wdocumentation /*! This constructor implicitly creates a constant string reference from a \c const character array. It has better performance than @@ -279,11 +281,13 @@ struct GenericStringRef { In such cases, the referenced string should be \b copied to the GenericValue instead. */ +#endif template GenericStringRef(const CharType (&str)[N]) RAPIDJSON_NOEXCEPT : s(str), length(N-1) {} //! Explicitly create string reference from \c const character pointer +#ifndef __clang__ // -Wdocumentation /*! This constructor can be used to \b explicitly create a reference to a constant string pointer. @@ -302,18 +306,23 @@ struct GenericStringRef { In such cases, the referenced string should be \b copied to the GenericValue instead. */ +#endif explicit GenericStringRef(const CharType* str) - : s(str), length(internal::StrLen(str)){ RAPIDJSON_ASSERT(s != NULL); } + : s(str), length(NotNullStrLen(str)) {} //! Create constant string reference from pointer and length +#ifndef __clang__ // -Wdocumentation /*! \param str constant string, lifetime assumed to be longer than the use of the string in e.g. a GenericValue \param len length of the string, excluding the trailing NULL terminator \post \ref s == str && \ref length == len \note Constant complexity. */ +#endif GenericStringRef(const CharType* str, SizeType len) - : s(str), length(len) { RAPIDJSON_ASSERT(s != NULL); } + : s(RAPIDJSON_LIKELY(str) ? str : emptyString), length(len) { RAPIDJSON_ASSERT(str != 0 || len == 0u); } + + GenericStringRef(const GenericStringRef& rhs) : s(rhs.s), length(rhs.length) {} //! implicit conversion to plain CharType pointer operator const Ch *() const { return s; } @@ -322,13 +331,24 @@ struct GenericStringRef { const SizeType length; //!< length of the string (excluding the trailing NULL terminator) private: - //! Disallow copy-assignment - GenericStringRef operator=(const GenericStringRef&); + SizeType NotNullStrLen(const CharType* str) { + RAPIDJSON_ASSERT(str != 0); + return internal::StrLen(str); + } + + /// Empty string - used when passing in a NULL pointer + static const Ch emptyString[]; + //! Disallow construction from non-const array template GenericStringRef(CharType (&str)[N]) /* = delete */; + //! Copy assignment operator not permitted - immutable type + GenericStringRef& operator=(const GenericStringRef& rhs) /* = delete */; }; +template +const CharType GenericStringRef::emptyString[] = { CharType() }; + //! Mark a character pointer as constant string /*! Mark a plain character pointer as a "string literal". This function can be used to avoid copying a character string to be referenced as a @@ -343,7 +363,7 @@ struct GenericStringRef { */ template inline GenericStringRef StringRef(const CharType* str) { - return GenericStringRef(str, internal::StrLen(str)); + return GenericStringRef(str); } //! Mark a character pointer as constant string @@ -401,6 +421,127 @@ template struct IsGenericValue : IsGenericValueImpl::Type {}; } // namespace internal +/////////////////////////////////////////////////////////////////////////////// +// TypeHelper + +namespace internal { + +template +struct TypeHelper {}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsBool(); } + static bool Get(const ValueType& v) { return v.GetBool(); } + static ValueType& Set(ValueType& v, bool data) { return v.SetBool(data); } + static ValueType& Set(ValueType& v, bool data, typename ValueType::AllocatorType&) { return v.SetBool(data); } +}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsInt(); } + static int Get(const ValueType& v) { return v.GetInt(); } + static ValueType& Set(ValueType& v, int data) { return v.SetInt(data); } + static ValueType& Set(ValueType& v, int data, typename ValueType::AllocatorType&) { return v.SetInt(data); } +}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsUint(); } + static unsigned Get(const ValueType& v) { return v.GetUint(); } + static ValueType& Set(ValueType& v, unsigned data) { return v.SetUint(data); } + static ValueType& Set(ValueType& v, unsigned data, typename ValueType::AllocatorType&) { return v.SetUint(data); } +}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsInt64(); } + static int64_t Get(const ValueType& v) { return v.GetInt64(); } + static ValueType& Set(ValueType& v, int64_t data) { return v.SetInt64(data); } + static ValueType& Set(ValueType& v, int64_t data, typename ValueType::AllocatorType&) { return v.SetInt64(data); } +}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsUint64(); } + static uint64_t Get(const ValueType& v) { return v.GetUint64(); } + static ValueType& Set(ValueType& v, uint64_t data) { return v.SetUint64(data); } + static ValueType& Set(ValueType& v, uint64_t data, typename ValueType::AllocatorType&) { return v.SetUint64(data); } +}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsDouble(); } + static double Get(const ValueType& v) { return v.GetDouble(); } + static ValueType& Set(ValueType& v, double data) { return v.SetDouble(data); } + static ValueType& Set(ValueType& v, double data, typename ValueType::AllocatorType&) { return v.SetDouble(data); } +}; + +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsFloat(); } + static float Get(const ValueType& v) { return v.GetFloat(); } + static ValueType& Set(ValueType& v, float data) { return v.SetFloat(data); } + static ValueType& Set(ValueType& v, float data, typename ValueType::AllocatorType&) { return v.SetFloat(data); } +}; + +template +struct TypeHelper { + typedef const typename ValueType::Ch* StringType; + static bool Is(const ValueType& v) { return v.IsString(); } + static StringType Get(const ValueType& v) { return v.GetString(); } + static ValueType& Set(ValueType& v, const StringType data) { return v.SetString(typename ValueType::StringRefType(data)); } + static ValueType& Set(ValueType& v, const StringType data, typename ValueType::AllocatorType& a) { return v.SetString(data, a); } +}; + +#if RAPIDJSON_HAS_STDSTRING +template +struct TypeHelper > { + typedef std::basic_string StringType; + static bool Is(const ValueType& v) { return v.IsString(); } + static StringType Get(const ValueType& v) { return StringType(v.GetString(), v.GetStringLength()); } + static ValueType& Set(ValueType& v, const StringType& data, typename ValueType::AllocatorType& a) { return v.SetString(data, a); } +}; +#endif + +template +struct TypeHelper { + typedef typename ValueType::Array ArrayType; + static bool Is(const ValueType& v) { return v.IsArray(); } + static ArrayType Get(ValueType& v) { return v.GetArray(); } + static ValueType& Set(ValueType& v, ArrayType data) { return v = data; } + static ValueType& Set(ValueType& v, ArrayType data, typename ValueType::AllocatorType&) { return v = data; } +}; + +template +struct TypeHelper { + typedef typename ValueType::ConstArray ArrayType; + static bool Is(const ValueType& v) { return v.IsArray(); } + static ArrayType Get(const ValueType& v) { return v.GetArray(); } +}; + +template +struct TypeHelper { + typedef typename ValueType::Object ObjectType; + static bool Is(const ValueType& v) { return v.IsObject(); } + static ObjectType Get(ValueType& v) { return v.GetObject(); } + static ValueType& Set(ValueType& v, ObjectType data) { return v = data; } + static ValueType& Set(ValueType& v, ObjectType data, typename ValueType::AllocatorType&) { return v = data; } +}; + +template +struct TypeHelper { + typedef typename ValueType::ConstObject ObjectType; + static bool Is(const ValueType& v) { return v.IsObject(); } + static ObjectType Get(const ValueType& v) { return v.GetObject(); } +}; + +} // namespace internal + +// Forward declarations +template class GenericArray; +template class GenericObject; + /////////////////////////////////////////////////////////////////////////////// // GenericValue @@ -428,17 +569,21 @@ class GenericValue { typedef GenericValue* ValueIterator; //!< Value iterator for iterating in array. typedef const GenericValue* ConstValueIterator; //!< Constant value iterator for iterating in array. typedef GenericValue ValueType; //!< Value type of itself. + typedef GenericArray Array; + typedef GenericArray ConstArray; + typedef GenericObject Object; + typedef GenericObject ConstObject; //!@name Constructors and destructor. //@{ //! Default constructor creates a null value. - GenericValue() RAPIDJSON_NOEXCEPT : data_(), flags_(kNullFlag) {} + GenericValue() RAPIDJSON_NOEXCEPT : data_() { data_.f.flags = kNullFlag; } #if RAPIDJSON_HAS_CXX11_RVALUE_REFS //! Move constructor in C++11 - GenericValue(GenericValue&& rhs) RAPIDJSON_NOEXCEPT : data_(rhs.data_), flags_(rhs.flags_) { - rhs.flags_ = kNullFlag; // give up contents + GenericValue(GenericValue&& rhs) RAPIDJSON_NOEXCEPT : data_(rhs.data_) { + rhs.data_.f.flags = kNullFlag; // give up contents } #endif @@ -446,6 +591,16 @@ class GenericValue { //! Copy constructor is not permitted. GenericValue(const GenericValue& rhs); +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + //! Moving from a GenericDocument is not permitted. + template + GenericValue(GenericDocument&& rhs); + + //! Move assignment from a GenericDocument is not permitted. + template + GenericValue& operator=(GenericDocument&& rhs); +#endif + public: //! Constructor with JSON value type. @@ -453,13 +608,13 @@ class GenericValue { \param type Type of the value. \note Default content for number is zero. */ - explicit GenericValue(Type type) RAPIDJSON_NOEXCEPT : data_(), flags_() { - static const unsigned defaultFlags[7] = { + explicit GenericValue(Type type) RAPIDJSON_NOEXCEPT : data_() { + static const uint16_t defaultFlags[7] = { kNullFlag, kFalseFlag, kTrueFlag, kObjectFlag, kArrayFlag, kShortStringFlag, kNumberAnyFlag }; RAPIDJSON_ASSERT(type <= kNumberType); - flags_ = defaultFlags[type]; + data_.f.flags = defaultFlags[type]; // Use ShortString to store empty string. if (type == kStringType) @@ -471,10 +626,50 @@ class GenericValue { \tparam SourceAllocator allocator of \c rhs \param rhs Value to copy from (read-only) \param allocator Allocator for allocating copied elements and buffers. Commonly use GenericDocument::GetAllocator(). + \param copyConstStrings Force copying of constant strings (e.g. referencing an in-situ buffer) \see CopyFrom() */ - template< typename SourceAllocator > - GenericValue(const GenericValue& rhs, Allocator & allocator); + template + GenericValue(const GenericValue& rhs, Allocator& allocator, bool copyConstStrings = false) { + switch (rhs.GetType()) { + case kObjectType: { + SizeType count = rhs.data_.o.size; + Member* lm = reinterpret_cast(allocator.Malloc(count * sizeof(Member))); + const typename GenericValue::Member* rm = rhs.GetMembersPointer(); + for (SizeType i = 0; i < count; i++) { + new (&lm[i].name) GenericValue(rm[i].name, allocator, copyConstStrings); + new (&lm[i].value) GenericValue(rm[i].value, allocator, copyConstStrings); + } + data_.f.flags = kObjectFlag; + data_.o.size = data_.o.capacity = count; + SetMembersPointer(lm); + } + break; + case kArrayType: { + SizeType count = rhs.data_.a.size; + GenericValue* le = reinterpret_cast(allocator.Malloc(count * sizeof(GenericValue))); + const GenericValue* re = rhs.GetElementsPointer(); + for (SizeType i = 0; i < count; i++) + new (&le[i]) GenericValue(re[i], allocator, copyConstStrings); + data_.f.flags = kArrayFlag; + data_.a.size = data_.a.capacity = count; + SetElementsPointer(le); + } + break; + case kStringType: + if (rhs.data_.f.flags == kConstStringFlag && !copyConstStrings) { + data_.f.flags = rhs.data_.f.flags; + data_ = *reinterpret_cast(&rhs.data_); + } + else + SetStringRaw(StringRef(rhs.GetString(), rhs.GetStringLength()), allocator); + break; + default: + data_.f.flags = rhs.data_.f.flags; + data_ = *reinterpret_cast(&rhs.data_); + break; + } + } //! Constructor for boolean value. /*! \param b Boolean value @@ -484,96 +679,125 @@ class GenericValue { */ #ifndef RAPIDJSON_DOXYGEN_RUNNING // hide SFINAE from Doxygen template - explicit GenericValue(T b, RAPIDJSON_ENABLEIF((internal::IsSame))) RAPIDJSON_NOEXCEPT + explicit GenericValue(T b, RAPIDJSON_ENABLEIF((internal::IsSame))) RAPIDJSON_NOEXCEPT // See #472 #else explicit GenericValue(bool b) RAPIDJSON_NOEXCEPT #endif - : data_(), flags_(b ? kTrueFlag : kFalseFlag) { + : data_() { // safe-guard against failing SFINAE RAPIDJSON_STATIC_ASSERT((internal::IsSame::Value)); + data_.f.flags = b ? kTrueFlag : kFalseFlag; } //! Constructor for int value. - explicit GenericValue(int i) RAPIDJSON_NOEXCEPT : data_(), flags_(kNumberIntFlag) { + explicit GenericValue(int i) RAPIDJSON_NOEXCEPT : data_() { data_.n.i64 = i; - if (i >= 0) - flags_ |= kUintFlag | kUint64Flag; + data_.f.flags = (i >= 0) ? (kNumberIntFlag | kUintFlag | kUint64Flag) : kNumberIntFlag; } //! Constructor for unsigned value. - explicit GenericValue(unsigned u) RAPIDJSON_NOEXCEPT : data_(), flags_(kNumberUintFlag) { + explicit GenericValue(unsigned u) RAPIDJSON_NOEXCEPT : data_() { data_.n.u64 = u; - if (!(u & 0x80000000)) - flags_ |= kIntFlag | kInt64Flag; + data_.f.flags = (u & 0x80000000) ? kNumberUintFlag : (kNumberUintFlag | kIntFlag | kInt64Flag); } //! Constructor for int64_t value. - explicit GenericValue(int64_t i64) RAPIDJSON_NOEXCEPT : data_(), flags_(kNumberInt64Flag) { + explicit GenericValue(int64_t i64) RAPIDJSON_NOEXCEPT : data_() { data_.n.i64 = i64; + data_.f.flags = kNumberInt64Flag; if (i64 >= 0) { - flags_ |= kNumberUint64Flag; + data_.f.flags |= kNumberUint64Flag; if (!(static_cast(i64) & RAPIDJSON_UINT64_C2(0xFFFFFFFF, 0x00000000))) - flags_ |= kUintFlag; + data_.f.flags |= kUintFlag; if (!(static_cast(i64) & RAPIDJSON_UINT64_C2(0xFFFFFFFF, 0x80000000))) - flags_ |= kIntFlag; + data_.f.flags |= kIntFlag; } else if (i64 >= static_cast(RAPIDJSON_UINT64_C2(0xFFFFFFFF, 0x80000000))) - flags_ |= kIntFlag; + data_.f.flags |= kIntFlag; } //! Constructor for uint64_t value. - explicit GenericValue(uint64_t u64) RAPIDJSON_NOEXCEPT : data_(), flags_(kNumberUint64Flag) { + explicit GenericValue(uint64_t u64) RAPIDJSON_NOEXCEPT : data_() { data_.n.u64 = u64; + data_.f.flags = kNumberUint64Flag; if (!(u64 & RAPIDJSON_UINT64_C2(0x80000000, 0x00000000))) - flags_ |= kInt64Flag; + data_.f.flags |= kInt64Flag; if (!(u64 & RAPIDJSON_UINT64_C2(0xFFFFFFFF, 0x00000000))) - flags_ |= kUintFlag; + data_.f.flags |= kUintFlag; if (!(u64 & RAPIDJSON_UINT64_C2(0xFFFFFFFF, 0x80000000))) - flags_ |= kIntFlag; + data_.f.flags |= kIntFlag; } //! Constructor for double value. - explicit GenericValue(double d) RAPIDJSON_NOEXCEPT : data_(), flags_(kNumberDoubleFlag) { data_.n.d = d; } + explicit GenericValue(double d) RAPIDJSON_NOEXCEPT : data_() { data_.n.d = d; data_.f.flags = kNumberDoubleFlag; } + + //! Constructor for float value. + explicit GenericValue(float f) RAPIDJSON_NOEXCEPT : data_() { data_.n.d = static_cast(f); data_.f.flags = kNumberDoubleFlag; } //! Constructor for constant string (i.e. do not make a copy of string) - GenericValue(const Ch* s, SizeType length) RAPIDJSON_NOEXCEPT : data_(), flags_() { SetStringRaw(StringRef(s, length)); } + GenericValue(const Ch* s, SizeType length) RAPIDJSON_NOEXCEPT : data_() { SetStringRaw(StringRef(s, length)); } //! Constructor for constant string (i.e. do not make a copy of string) - explicit GenericValue(StringRefType s) RAPIDJSON_NOEXCEPT : data_(), flags_() { SetStringRaw(s); } + explicit GenericValue(StringRefType s) RAPIDJSON_NOEXCEPT : data_() { SetStringRaw(s); } //! Constructor for copy-string (i.e. do make a copy of string) - GenericValue(const Ch* s, SizeType length, Allocator& allocator) : data_(), flags_() { SetStringRaw(StringRef(s, length), allocator); } + GenericValue(const Ch* s, SizeType length, Allocator& allocator) : data_() { SetStringRaw(StringRef(s, length), allocator); } //! Constructor for copy-string (i.e. do make a copy of string) - GenericValue(const Ch*s, Allocator& allocator) : data_(), flags_() { SetStringRaw(StringRef(s), allocator); } + GenericValue(const Ch*s, Allocator& allocator) : data_() { SetStringRaw(StringRef(s), allocator); } #if RAPIDJSON_HAS_STDSTRING //! Constructor for copy-string from a string object (i.e. do make a copy of string) /*! \note Requires the definition of the preprocessor symbol \ref RAPIDJSON_HAS_STDSTRING. */ - GenericValue(const std::basic_string& s, Allocator& allocator) : data_(), flags_() { SetStringRaw(StringRef(s), allocator); } + GenericValue(const std::basic_string& s, Allocator& allocator) : data_() { SetStringRaw(StringRef(s), allocator); } #endif + //! Constructor for Array. + /*! + \param a An array obtained by \c GetArray(). + \note \c Array is always pass-by-value. + \note the source array is moved into this value and the sourec array becomes empty. + */ + GenericValue(Array a) RAPIDJSON_NOEXCEPT : data_(a.value_.data_) { + a.value_.data_ = Data(); + a.value_.data_.f.flags = kArrayFlag; + } + + //! Constructor for Object. + /*! + \param o An object obtained by \c GetObject(). + \note \c Object is always pass-by-value. + \note the source object is moved into this value and the sourec object becomes empty. + */ + GenericValue(Object o) RAPIDJSON_NOEXCEPT : data_(o.value_.data_) { + o.value_.data_ = Data(); + o.value_.data_.f.flags = kObjectFlag; + } + //! Destructor. /*! Need to destruct elements of array, members of object, or copy-string. */ ~GenericValue() { if (Allocator::kNeedFree) { // Shortcut by Allocator's trait - switch(flags_) { + switch(data_.f.flags) { case kArrayFlag: - for (GenericValue* v = data_.a.elements; v != data_.a.elements + data_.a.size; ++v) - v->~GenericValue(); - Allocator::Free(data_.a.elements); + { + GenericValue* e = GetElementsPointer(); + for (GenericValue* v = e; v != e + data_.a.size; ++v) + v->~GenericValue(); + Allocator::Free(e); + } break; case kObjectFlag: for (MemberIterator m = MemberBegin(); m != MemberEnd(); ++m) m->~Member(); - Allocator::Free(data_.o.members); + Allocator::Free(GetMembersPointer()); break; case kCopyStringFlag: - Allocator::Free(const_cast(data_.s.str)); + Allocator::Free(const_cast(GetStringPointer())); break; default: @@ -638,12 +862,13 @@ class GenericValue { \tparam SourceAllocator Allocator type of \c rhs \param rhs Value to copy from (read-only) \param allocator Allocator to use for copying + \param copyConstStrings Force copying of constant strings (e.g. referencing an in-situ buffer) */ template - GenericValue& CopyFrom(const GenericValue& rhs, Allocator& allocator) { - RAPIDJSON_ASSERT((void*)this != (void const*)&rhs); + GenericValue& CopyFrom(const GenericValue& rhs, Allocator& allocator, bool copyConstStrings = false) { + RAPIDJSON_ASSERT(static_cast(this) != static_cast(&rhs)); this->~GenericValue(); - new (this) GenericValue(rhs, allocator); + new (this) GenericValue(rhs, allocator, copyConstStrings); return *this; } @@ -660,6 +885,20 @@ class GenericValue { return *this; } + //! free-standing swap function helper + /*! + Helper function to enable support for common swap implementation pattern based on \c std::swap: + \code + void swap(MyClass& a, MyClass& b) { + using std::swap; + swap(a.value, b.value); + // ... + } + \endcode + \see Swap() + */ + friend inline void swap(GenericValue& a, GenericValue& b) RAPIDJSON_NOEXCEPT { a.Swap(b); } + //! Prepare Value for move semantics /*! \return *this */ GenericValue& Move() RAPIDJSON_NOEXCEPT { return *this; } @@ -709,7 +948,7 @@ class GenericValue { else return data_.n.u64 == rhs.data_.n.u64; - default: // kTrueType, kFalseType, kNullType + default: return true; } } @@ -757,20 +996,58 @@ class GenericValue { //!@name Type //@{ - Type GetType() const { return static_cast(flags_ & kTypeMask); } - bool IsNull() const { return flags_ == kNullFlag; } - bool IsFalse() const { return flags_ == kFalseFlag; } - bool IsTrue() const { return flags_ == kTrueFlag; } - bool IsBool() const { return (flags_ & kBoolFlag) != 0; } - bool IsObject() const { return flags_ == kObjectFlag; } - bool IsArray() const { return flags_ == kArrayFlag; } - bool IsNumber() const { return (flags_ & kNumberFlag) != 0; } - bool IsInt() const { return (flags_ & kIntFlag) != 0; } - bool IsUint() const { return (flags_ & kUintFlag) != 0; } - bool IsInt64() const { return (flags_ & kInt64Flag) != 0; } - bool IsUint64() const { return (flags_ & kUint64Flag) != 0; } - bool IsDouble() const { return (flags_ & kDoubleFlag) != 0; } - bool IsString() const { return (flags_ & kStringFlag) != 0; } + Type GetType() const { return static_cast(data_.f.flags & kTypeMask); } + bool IsNull() const { return data_.f.flags == kNullFlag; } + bool IsFalse() const { return data_.f.flags == kFalseFlag; } + bool IsTrue() const { return data_.f.flags == kTrueFlag; } + bool IsBool() const { return (data_.f.flags & kBoolFlag) != 0; } + bool IsObject() const { return data_.f.flags == kObjectFlag; } + bool IsArray() const { return data_.f.flags == kArrayFlag; } + bool IsNumber() const { return (data_.f.flags & kNumberFlag) != 0; } + bool IsInt() const { return (data_.f.flags & kIntFlag) != 0; } + bool IsUint() const { return (data_.f.flags & kUintFlag) != 0; } + bool IsInt64() const { return (data_.f.flags & kInt64Flag) != 0; } + bool IsUint64() const { return (data_.f.flags & kUint64Flag) != 0; } + bool IsDouble() const { return (data_.f.flags & kDoubleFlag) != 0; } + bool IsString() const { return (data_.f.flags & kStringFlag) != 0; } + + // Checks whether a number can be losslessly converted to a double. + bool IsLosslessDouble() const { + if (!IsNumber()) return false; + if (IsUint64()) { + uint64_t u = GetUint64(); + volatile double d = static_cast(u); + return (d >= 0.0) + && (d < static_cast(std::numeric_limits::max())) + && (u == static_cast(d)); + } + if (IsInt64()) { + int64_t i = GetInt64(); + volatile double d = static_cast(i); + return (d >= static_cast(std::numeric_limits::min())) + && (d < static_cast(std::numeric_limits::max())) + && (i == static_cast(d)); + } + return true; // double, int, uint are always lossless + } + + // Checks whether a number is a float (possible lossy). + bool IsFloat() const { + if ((data_.f.flags & kDoubleFlag) == 0) + return false; + double d = GetDouble(); + return d >= -3.4028234e38 && d <= 3.4028234e38; + } + // Checks whether a number can be losslessly converted to a float. + bool IsLosslessFloat() const { + if (!IsNumber()) return false; + double a = GetDouble(); + if (a < static_cast(-std::numeric_limits::max()) + || a > static_cast(std::numeric_limits::max())) + return false; + double b = static_cast(static_cast(a)); + return a >= b && a <= b; // Prevent -Wfloat-equal + } //@} @@ -784,7 +1061,7 @@ class GenericValue { //!@name Bool //@{ - bool GetBool() const { RAPIDJSON_ASSERT(IsBool()); return flags_ == kTrueFlag; } + bool GetBool() const { RAPIDJSON_ASSERT(IsBool()); return data_.f.flags == kTrueFlag; } //!< Set boolean value /*! \post IsBool() == true */ GenericValue& SetBool(bool b) { this->~GenericValue(); new (this) GenericValue(b); return *this; } @@ -837,8 +1114,14 @@ class GenericValue { return member->value; else { RAPIDJSON_ASSERT(false); // see above note - static GenericValue NullValue; - return NullValue; + + // This will generate -Wexit-time-destructors in clang + // static GenericValue NullValue; + // return NullValue; + + // Use static buffer and placement-new to prevent destruction + static char buffer[sizeof(GenericValue)]; + return *new (buffer) GenericValue(); } } template @@ -852,16 +1135,16 @@ class GenericValue { //! Const member iterator /*! \pre IsObject() == true */ - ConstMemberIterator MemberBegin() const { RAPIDJSON_ASSERT(IsObject()); return ConstMemberIterator(data_.o.members); } + ConstMemberIterator MemberBegin() const { RAPIDJSON_ASSERT(IsObject()); return ConstMemberIterator(GetMembersPointer()); } //! Const \em past-the-end member iterator /*! \pre IsObject() == true */ - ConstMemberIterator MemberEnd() const { RAPIDJSON_ASSERT(IsObject()); return ConstMemberIterator(data_.o.members + data_.o.size); } + ConstMemberIterator MemberEnd() const { RAPIDJSON_ASSERT(IsObject()); return ConstMemberIterator(GetMembersPointer() + data_.o.size); } //! Member iterator /*! \pre IsObject() == true */ - MemberIterator MemberBegin() { RAPIDJSON_ASSERT(IsObject()); return MemberIterator(data_.o.members); } + MemberIterator MemberBegin() { RAPIDJSON_ASSERT(IsObject()); return MemberIterator(GetMembersPointer()); } //! \em Past-the-end member iterator /*! \pre IsObject() == true */ - MemberIterator MemberEnd() { RAPIDJSON_ASSERT(IsObject()); return MemberIterator(data_.o.members + data_.o.size); } + MemberIterator MemberEnd() { RAPIDJSON_ASSERT(IsObject()); return MemberIterator(GetMembersPointer() + data_.o.size); } //! Check whether a member exists in the object. /*! @@ -949,8 +1232,8 @@ class GenericValue { \return Iterator to member, if it exists. Otherwise returns \ref MemberEnd(). */ - MemberIterator FindMember(const std::basic_string& name) { return FindMember(StringRef(name)); } - ConstMemberIterator FindMember(const std::basic_string& name) const { return FindMember(StringRef(name)); } + MemberIterator FindMember(const std::basic_string& name) { return FindMember(GenericValue(StringRef(name))); } + ConstMemberIterator FindMember(const std::basic_string& name) const { return FindMember(GenericValue(StringRef(name))); } #endif //! Add a member (name-value pair) to the object. @@ -967,20 +1250,21 @@ class GenericValue { RAPIDJSON_ASSERT(IsObject()); RAPIDJSON_ASSERT(name.IsString()); - Object& o = data_.o; + ObjectData& o = data_.o; if (o.size >= o.capacity) { if (o.capacity == 0) { o.capacity = kDefaultObjectCapacity; - o.members = reinterpret_cast(allocator.Malloc(o.capacity * sizeof(Member))); + SetMembersPointer(reinterpret_cast(allocator.Malloc(o.capacity * sizeof(Member)))); } else { SizeType oldCapacity = o.capacity; o.capacity += (oldCapacity + 1) / 2; // grow by factor 1.5 - o.members = reinterpret_cast(allocator.Realloc(o.members, oldCapacity * sizeof(Member), o.capacity * sizeof(Member))); + SetMembersPointer(reinterpret_cast(allocator.Realloc(GetMembersPointer(), oldCapacity * sizeof(Member), o.capacity * sizeof(Member)))); } } - o.members[o.size].name.RawAssign(name); - o.members[o.size].value.RawAssign(value); + Member* members = GetMembersPointer(); + members[o.size].name.RawAssign(name); + members[o.size].value.RawAssign(value); o.size++; return *this; } @@ -1159,18 +1443,14 @@ class GenericValue { MemberIterator RemoveMember(MemberIterator m) { RAPIDJSON_ASSERT(IsObject()); RAPIDJSON_ASSERT(data_.o.size > 0); - RAPIDJSON_ASSERT(data_.o.members != 0); + RAPIDJSON_ASSERT(GetMembersPointer() != 0); RAPIDJSON_ASSERT(m >= MemberBegin() && m < MemberEnd()); - MemberIterator last(data_.o.members + (data_.o.size - 1)); - if (data_.o.size > 1 && m != last) { - // Move the last one to this place - *m = *last; - } - else { - // Only one left, just destroy - m->~Member(); - } + MemberIterator last(GetMembersPointer() + (data_.o.size - 1)); + if (data_.o.size > 1 && m != last) + *m = *last; // Move the last one to this place + else + m->~Member(); // Only one left, just destroy --data_.o.size; return m; } @@ -1200,7 +1480,7 @@ class GenericValue { MemberIterator EraseMember(ConstMemberIterator first, ConstMemberIterator last) { RAPIDJSON_ASSERT(IsObject()); RAPIDJSON_ASSERT(data_.o.size > 0); - RAPIDJSON_ASSERT(data_.o.members != 0); + RAPIDJSON_ASSERT(GetMembersPointer() != 0); RAPIDJSON_ASSERT(first >= MemberBegin()); RAPIDJSON_ASSERT(first <= last); RAPIDJSON_ASSERT(last <= MemberEnd()); @@ -1208,11 +1488,39 @@ class GenericValue { MemberIterator pos = MemberBegin() + (first - MemberBegin()); for (MemberIterator itr = pos; itr != last; ++itr) itr->~Member(); - std::memmove(&*pos, &*last, (MemberEnd() - last) * sizeof(Member)); - data_.o.size -= (last - first); + std::memmove(&*pos, &*last, static_cast(MemberEnd() - last) * sizeof(Member)); + data_.o.size -= static_cast(last - first); return pos; } + //! Erase a member in object by its name. + /*! \param name Name of member to be removed. + \return Whether the member existed. + \note Linear time complexity. + */ + bool EraseMember(const Ch* name) { + GenericValue n(StringRef(name)); + return EraseMember(n); + } + +#if RAPIDJSON_HAS_STDSTRING + bool EraseMember(const std::basic_string& name) { return EraseMember(GenericValue(StringRef(name))); } +#endif + + template + bool EraseMember(const GenericValue& name) { + MemberIterator m = FindMember(name); + if (m != MemberEnd()) { + EraseMember(m); + return true; + } + else + return false; + } + + Object GetObject() { RAPIDJSON_ASSERT(IsObject()); return Object(*this); } + ConstObject GetObject() const { RAPIDJSON_ASSERT(IsObject()); return ConstObject(*this); } + //@} //!@name Array @@ -1220,7 +1528,7 @@ class GenericValue { //! Set this value as an empty array. /*! \post IsArray == true */ - GenericValue& SetArray() { this->~GenericValue(); new (this) GenericValue(kArrayType); return *this; } + GenericValue& SetArray() { this->~GenericValue(); new (this) GenericValue(kArrayType); return *this; } //! Get the number of elements in array. SizeType Size() const { RAPIDJSON_ASSERT(IsArray()); return data_.a.size; } @@ -1237,8 +1545,9 @@ class GenericValue { */ void Clear() { RAPIDJSON_ASSERT(IsArray()); - for (SizeType i = 0; i < data_.a.size; ++i) - data_.a.elements[i].~GenericValue(); + GenericValue* e = GetElementsPointer(); + for (GenericValue* v = e; v != e + data_.a.size; ++v) + v->~GenericValue(); data_.a.size = 0; } @@ -1250,16 +1559,16 @@ class GenericValue { GenericValue& operator[](SizeType index) { RAPIDJSON_ASSERT(IsArray()); RAPIDJSON_ASSERT(index < data_.a.size); - return data_.a.elements[index]; + return GetElementsPointer()[index]; } const GenericValue& operator[](SizeType index) const { return const_cast(*this)[index]; } //! Element iterator /*! \pre IsArray() == true */ - ValueIterator Begin() { RAPIDJSON_ASSERT(IsArray()); return data_.a.elements; } + ValueIterator Begin() { RAPIDJSON_ASSERT(IsArray()); return GetElementsPointer(); } //! \em Past-the-end element iterator /*! \pre IsArray() == true */ - ValueIterator End() { RAPIDJSON_ASSERT(IsArray()); return data_.a.elements + data_.a.size; } + ValueIterator End() { RAPIDJSON_ASSERT(IsArray()); return GetElementsPointer() + data_.a.size; } //! Constant element iterator /*! \pre IsArray() == true */ ConstValueIterator Begin() const { return const_cast(*this).Begin(); } @@ -1276,7 +1585,7 @@ class GenericValue { GenericValue& Reserve(SizeType newCapacity, Allocator &allocator) { RAPIDJSON_ASSERT(IsArray()); if (newCapacity > data_.a.capacity) { - data_.a.elements = (GenericValue*)allocator.Realloc(data_.a.elements, data_.a.capacity * sizeof(GenericValue), newCapacity * sizeof(GenericValue)); + SetElementsPointer(reinterpret_cast(allocator.Realloc(GetElementsPointer(), data_.a.capacity * sizeof(GenericValue), newCapacity * sizeof(GenericValue)))); data_.a.capacity = newCapacity; } return *this; @@ -1296,7 +1605,7 @@ class GenericValue { RAPIDJSON_ASSERT(IsArray()); if (data_.a.size >= data_.a.capacity) Reserve(data_.a.capacity == 0 ? kDefaultArrayCapacity : (data_.a.capacity + (data_.a.capacity + 1) / 2), allocator); - data_.a.elements[data_.a.size++].RawAssign(value); + GetElementsPointer()[data_.a.size++].RawAssign(value); return *this; } @@ -1350,7 +1659,7 @@ class GenericValue { GenericValue& PopBack() { RAPIDJSON_ASSERT(IsArray()); RAPIDJSON_ASSERT(!Empty()); - data_.a.elements[--data_.a.size].~GenericValue(); + GetElementsPointer()[--data_.a.size].~GenericValue(); return *this; } @@ -1376,35 +1685,48 @@ class GenericValue { ValueIterator Erase(ConstValueIterator first, ConstValueIterator last) { RAPIDJSON_ASSERT(IsArray()); RAPIDJSON_ASSERT(data_.a.size > 0); - RAPIDJSON_ASSERT(data_.a.elements != 0); + RAPIDJSON_ASSERT(GetElementsPointer() != 0); RAPIDJSON_ASSERT(first >= Begin()); RAPIDJSON_ASSERT(first <= last); RAPIDJSON_ASSERT(last <= End()); ValueIterator pos = Begin() + (first - Begin()); for (ValueIterator itr = pos; itr != last; ++itr) itr->~GenericValue(); - std::memmove(pos, last, (End() - last) * sizeof(GenericValue)); - data_.a.size -= (last - first); + std::memmove(pos, last, static_cast(End() - last) * sizeof(GenericValue)); + data_.a.size -= static_cast(last - first); return pos; } + Array GetArray() { RAPIDJSON_ASSERT(IsArray()); return Array(*this); } + ConstArray GetArray() const { RAPIDJSON_ASSERT(IsArray()); return ConstArray(*this); } + //@} //!@name Number //@{ - int GetInt() const { RAPIDJSON_ASSERT(flags_ & kIntFlag); return data_.n.i.i; } - unsigned GetUint() const { RAPIDJSON_ASSERT(flags_ & kUintFlag); return data_.n.u.u; } - int64_t GetInt64() const { RAPIDJSON_ASSERT(flags_ & kInt64Flag); return data_.n.i64; } - uint64_t GetUint64() const { RAPIDJSON_ASSERT(flags_ & kUint64Flag); return data_.n.u64; } + int GetInt() const { RAPIDJSON_ASSERT(data_.f.flags & kIntFlag); return data_.n.i.i; } + unsigned GetUint() const { RAPIDJSON_ASSERT(data_.f.flags & kUintFlag); return data_.n.u.u; } + int64_t GetInt64() const { RAPIDJSON_ASSERT(data_.f.flags & kInt64Flag); return data_.n.i64; } + uint64_t GetUint64() const { RAPIDJSON_ASSERT(data_.f.flags & kUint64Flag); return data_.n.u64; } + //! Get the value as double type. + /*! \note If the value is 64-bit integer type, it may lose precision. Use \c IsLosslessDouble() to check whether the converison is lossless. + */ double GetDouble() const { RAPIDJSON_ASSERT(IsNumber()); - if ((flags_ & kDoubleFlag) != 0) return data_.n.d; // exact type, no conversion. - if ((flags_ & kIntFlag) != 0) return data_.n.i.i; // int -> double - if ((flags_ & kUintFlag) != 0) return data_.n.u.u; // unsigned -> double - if ((flags_ & kInt64Flag) != 0) return (double)data_.n.i64; // int64_t -> double (may lose precision) - RAPIDJSON_ASSERT((flags_ & kUint64Flag) != 0); return (double)data_.n.u64; // uint64_t -> double (may lose precision) + if ((data_.f.flags & kDoubleFlag) != 0) return data_.n.d; // exact type, no conversion. + if ((data_.f.flags & kIntFlag) != 0) return data_.n.i.i; // int -> double + if ((data_.f.flags & kUintFlag) != 0) return data_.n.u.u; // unsigned -> double + if ((data_.f.flags & kInt64Flag) != 0) return static_cast(data_.n.i64); // int64_t -> double (may lose precision) + RAPIDJSON_ASSERT((data_.f.flags & kUint64Flag) != 0); return static_cast(data_.n.u64); // uint64_t -> double (may lose precision) + } + + //! Get the value as float type. + /*! \note If the value is 64-bit integer type, it may lose precision. Use \c IsLosslessFloat() to check whether the converison is lossless. + */ + float GetFloat() const { + return static_cast(GetDouble()); } GenericValue& SetInt(int i) { this->~GenericValue(); new (this) GenericValue(i); return *this; } @@ -1412,18 +1734,19 @@ class GenericValue { GenericValue& SetInt64(int64_t i64) { this->~GenericValue(); new (this) GenericValue(i64); return *this; } GenericValue& SetUint64(uint64_t u64) { this->~GenericValue(); new (this) GenericValue(u64); return *this; } GenericValue& SetDouble(double d) { this->~GenericValue(); new (this) GenericValue(d); return *this; } + GenericValue& SetFloat(float f) { this->~GenericValue(); new (this) GenericValue(static_cast(f)); return *this; } //@} //!@name String //@{ - const Ch* GetString() const { RAPIDJSON_ASSERT(IsString()); return ((flags_ & kInlineStrFlag) ? data_.ss.str : data_.s.str); } + const Ch* GetString() const { RAPIDJSON_ASSERT(IsString()); return (data_.f.flags & kInlineStrFlag) ? data_.ss.str : GetStringPointer(); } //! Get the length of string. /*! Since rapidjson permits "\\u0000" in the json string, strlen(v.GetString()) may not equal to v.GetStringLength(). */ - SizeType GetStringLength() const { RAPIDJSON_ASSERT(IsString()); return ((flags_ & kInlineStrFlag) ? (data_.ss.GetLength()) : data_.s.length); } + SizeType GetStringLength() const { RAPIDJSON_ASSERT(IsString()); return ((data_.f.flags & kInlineStrFlag) ? (data_.ss.GetLength()) : data_.s.length); } //! Set this value as a string without copying source string. /*! This version has better performance with supplied length, and also support string containing null character. @@ -1450,7 +1773,7 @@ class GenericValue { \return The value itself for fluent API. \post IsString() == true && GetString() != s && strcmp(GetString(),s) == 0 && GetStringLength() == length */ - GenericValue& SetString(const Ch* s, SizeType length, Allocator& allocator) { this->~GenericValue(); SetStringRaw(StringRef(s, length), allocator); return *this; } + GenericValue& SetString(const Ch* s, SizeType length, Allocator& allocator) { return SetString(StringRef(s, length), allocator); } //! Set this value as a string by copying from source string. /*! \param s source string. @@ -1458,7 +1781,15 @@ class GenericValue { \return The value itself for fluent API. \post IsString() == true && GetString() != s && strcmp(GetString(),s) == 0 && GetStringLength() == length */ - GenericValue& SetString(const Ch* s, Allocator& allocator) { return SetString(s, internal::StrLen(s), allocator); } + GenericValue& SetString(const Ch* s, Allocator& allocator) { return SetString(StringRef(s), allocator); } + + //! Set this value as a string by copying from source string. + /*! \param s source string reference + \param allocator Allocator for allocating copied buffer. Commonly use GenericDocument::GetAllocator(). + \return The value itself for fluent API. + \post IsString() == true && GetString() != s.s && strcmp(GetString(),s) == 0 && GetStringLength() == length + */ + GenericValue& SetString(StringRefType s, Allocator& allocator) { this->~GenericValue(); SetStringRaw(s, allocator); return *this; } #if RAPIDJSON_HAS_STDSTRING //! Set this value as a string by copying from source string. @@ -1468,11 +1799,35 @@ class GenericValue { \post IsString() == true && GetString() != s.data() && strcmp(GetString(),s.data() == 0 && GetStringLength() == s.size() \note Requires the definition of the preprocessor symbol \ref RAPIDJSON_HAS_STDSTRING. */ - GenericValue& SetString(const std::basic_string& s, Allocator& allocator) { return SetString(s.data(), SizeType(s.size()), allocator); } + GenericValue& SetString(const std::basic_string& s, Allocator& allocator) { return SetString(StringRef(s), allocator); } #endif //@} + //!@name Array + //@{ + + //! Templated version for checking whether this value is type T. + /*! + \tparam T Either \c bool, \c int, \c unsigned, \c int64_t, \c uint64_t, \c double, \c float, \c const \c char*, \c std::basic_string + */ + template + bool Is() const { return internal::TypeHelper::Is(*this); } + + template + T Get() const { return internal::TypeHelper::Get(*this); } + + template + T Get() { return internal::TypeHelper::Get(*this); } + + template + ValueType& Set(const T& data) { return internal::TypeHelper::Set(*this, data); } + + template + ValueType& Set(const T& data, AllocatorType& allocator) { return internal::TypeHelper::Set(*this, data, allocator); } + + //@} + //! Generate events of this value to a Handler. /*! This function adopts the GoF visitor pattern. Typical usage is to output this JSON value as JSON text via Writer, which is a Handler. @@ -1488,35 +1843,35 @@ class GenericValue { case kTrueType: return handler.Bool(true); case kObjectType: - if (!handler.StartObject()) + if (RAPIDJSON_UNLIKELY(!handler.StartObject())) return false; for (ConstMemberIterator m = MemberBegin(); m != MemberEnd(); ++m) { RAPIDJSON_ASSERT(m->name.IsString()); // User may change the type of name by MemberIterator. - if (!handler.Key(m->name.GetString(), m->name.GetStringLength(), (m->name.flags_ & kCopyFlag) != 0)) + if (RAPIDJSON_UNLIKELY(!handler.Key(m->name.GetString(), m->name.GetStringLength(), (m->name.data_.f.flags & kCopyFlag) != 0))) return false; - if (!m->value.Accept(handler)) + if (RAPIDJSON_UNLIKELY(!m->value.Accept(handler))) return false; } return handler.EndObject(data_.o.size); case kArrayType: - if (!handler.StartArray()) + if (RAPIDJSON_UNLIKELY(!handler.StartArray())) return false; - for (GenericValue* v = data_.a.elements; v != data_.a.elements + data_.a.size; ++v) - if (!v->Accept(handler)) + for (const GenericValue* v = Begin(); v != End(); ++v) + if (RAPIDJSON_UNLIKELY(!v->Accept(handler))) return false; return handler.EndArray(data_.a.size); case kStringType: - return handler.String(GetString(), GetStringLength(), (flags_ & kCopyFlag) != 0); + return handler.String(GetString(), GetStringLength(), (data_.f.flags & kCopyFlag) != 0); default: RAPIDJSON_ASSERT(GetType() == kNumberType); - if (IsInt()) return handler.Int(data_.n.i.i); + if (IsDouble()) return handler.Double(data_.n.d); + else if (IsInt()) return handler.Int(data_.n.i.i); else if (IsUint()) return handler.Uint(data_.n.u.u); else if (IsInt64()) return handler.Int64(data_.n.i64); - else if (IsUint64()) return handler.Uint64(data_.n.u64); - else return handler.Double(data_.n.d); + else return handler.Uint64(data_.n.u64); } } @@ -1525,16 +1880,16 @@ class GenericValue { template friend class GenericDocument; enum { - kBoolFlag = 0x100, - kNumberFlag = 0x200, - kIntFlag = 0x400, - kUintFlag = 0x800, - kInt64Flag = 0x1000, - kUint64Flag = 0x2000, - kDoubleFlag = 0x4000, - kStringFlag = 0x100000, - kCopyFlag = 0x200000, - kInlineStrFlag = 0x400000, + kBoolFlag = 0x0008, + kNumberFlag = 0x0010, + kIntFlag = 0x0020, + kUintFlag = 0x0040, + kInt64Flag = 0x0080, + kUint64Flag = 0x0100, + kDoubleFlag = 0x0200, + kStringFlag = 0x0400, + kCopyFlag = 0x0800, + kInlineStrFlag = 0x1000, // Initial flags of different types. kNullFlag = kNullType, @@ -1552,16 +1907,27 @@ class GenericValue { kObjectFlag = kObjectType, kArrayFlag = kArrayType, - kTypeMask = 0xFF // bitwise-and with mask of 0xFF can be optimized by compiler + kTypeMask = 0x07 }; static const SizeType kDefaultArrayCapacity = 16; static const SizeType kDefaultObjectCapacity = 16; + struct Flag { +#if RAPIDJSON_48BITPOINTER_OPTIMIZATION + char payload[sizeof(SizeType) * 2 + 6]; // 2 x SizeType + lower 48-bit pointer +#elif RAPIDJSON_64BIT + char payload[sizeof(SizeType) * 2 + sizeof(void*) + 6]; // 6 padding bytes +#else + char payload[sizeof(SizeType) * 2 + sizeof(void*) + 2]; // 2 padding bytes +#endif + uint16_t flags; + }; + struct String { - const Ch* str; SizeType length; - unsigned hashcode; //!< reserved + SizeType hashcode; //!< reserved + const Ch* str; }; // 12 bytes in 32-bit mode, 16 bytes in 64-bit mode // implementation detail: ShortString can represent zero-terminated strings up to MaxSize chars @@ -1570,15 +1936,15 @@ class GenericValue { // to store has the maximal length of MaxSize then str[LenPos] will be 0 and therefore act as // the string terminator as well. For getting the string length back from that value just use // "MaxSize - str[LenPos]". - // This allows to store 11-chars strings in 32-bit mode and 15-chars strings in 64-bit mode - // inline (for `UTF8`-encoded strings). + // This allows to store 13-chars strings in 32-bit mode, 21-chars strings in 64-bit mode, + // 13-chars strings for RAPIDJSON_48BITPOINTER_OPTIMIZATION=1 inline (for `UTF8`-encoded strings). struct ShortString { - enum { MaxChars = sizeof(String) / sizeof(Ch), MaxSize = MaxChars - 1, LenPos = MaxSize }; + enum { MaxChars = sizeof(static_cast(0)->payload) / sizeof(Ch), MaxSize = MaxChars - 1, LenPos = MaxSize }; Ch str[MaxChars]; - inline static bool Usable(SizeType len) { return (MaxSize >= len); } - inline void SetLength(SizeType len) { str[LenPos] = (Ch)(MaxSize - len); } - inline SizeType GetLength() const { return (SizeType)(MaxSize - str[LenPos]); } + inline static bool Usable(SizeType len) { return (MaxSize >= len); } + inline void SetLength(SizeType len) { str[LenPos] = static_cast(MaxSize - len); } + inline SizeType GetLength() const { return static_cast(MaxSize - str[LenPos]); } }; // at most as many bytes as "String" above => 12 bytes in 32-bit mode, 16 bytes in 64-bit mode // By using proper binary layout, retrieval of different integer types do not need conversions. @@ -1607,69 +1973,79 @@ class GenericValue { double d; }; // 8 bytes - struct Object { - Member* members; + struct ObjectData { SizeType size; SizeType capacity; + Member* members; }; // 12 bytes in 32-bit mode, 16 bytes in 64-bit mode - struct Array { - GenericValue* elements; + struct ArrayData { SizeType size; SizeType capacity; + GenericValue* elements; }; // 12 bytes in 32-bit mode, 16 bytes in 64-bit mode union Data { String s; ShortString ss; Number n; - Object o; - Array a; - }; // 12 bytes in 32-bit mode, 16 bytes in 64-bit mode + ObjectData o; + ArrayData a; + Flag f; + }; // 16 bytes in 32-bit mode, 24 bytes in 64-bit mode, 16 bytes in 64-bit with RAPIDJSON_48BITPOINTER_OPTIMIZATION + + RAPIDJSON_FORCEINLINE const Ch* GetStringPointer() const { return RAPIDJSON_GETPOINTER(Ch, data_.s.str); } + RAPIDJSON_FORCEINLINE const Ch* SetStringPointer(const Ch* str) { return RAPIDJSON_SETPOINTER(Ch, data_.s.str, str); } + RAPIDJSON_FORCEINLINE GenericValue* GetElementsPointer() const { return RAPIDJSON_GETPOINTER(GenericValue, data_.a.elements); } + RAPIDJSON_FORCEINLINE GenericValue* SetElementsPointer(GenericValue* elements) { return RAPIDJSON_SETPOINTER(GenericValue, data_.a.elements, elements); } + RAPIDJSON_FORCEINLINE Member* GetMembersPointer() const { return RAPIDJSON_GETPOINTER(Member, data_.o.members); } + RAPIDJSON_FORCEINLINE Member* SetMembersPointer(Member* members) { return RAPIDJSON_SETPOINTER(Member, data_.o.members, members); } // Initialize this value as array with initial data, without calling destructor. void SetArrayRaw(GenericValue* values, SizeType count, Allocator& allocator) { - flags_ = kArrayFlag; + data_.f.flags = kArrayFlag; if (count) { - data_.a.elements = (GenericValue*)allocator.Malloc(count * sizeof(GenericValue)); - std::memcpy(data_.a.elements, values, count * sizeof(GenericValue)); + GenericValue* e = static_cast(allocator.Malloc(count * sizeof(GenericValue))); + SetElementsPointer(e); + std::memcpy(e, values, count * sizeof(GenericValue)); } else - data_.a.elements = NULL; + SetElementsPointer(0); data_.a.size = data_.a.capacity = count; } //! Initialize this value as object with initial data, without calling destructor. void SetObjectRaw(Member* members, SizeType count, Allocator& allocator) { - flags_ = kObjectFlag; + data_.f.flags = kObjectFlag; if (count) { - data_.o.members = (Member*)allocator.Malloc(count * sizeof(Member)); - std::memcpy(data_.o.members, members, count * sizeof(Member)); + Member* m = static_cast(allocator.Malloc(count * sizeof(Member))); + SetMembersPointer(m); + std::memcpy(m, members, count * sizeof(Member)); } else - data_.o.members = NULL; + SetMembersPointer(0); data_.o.size = data_.o.capacity = count; } //! Initialize this value as constant string, without calling destructor. void SetStringRaw(StringRefType s) RAPIDJSON_NOEXCEPT { - flags_ = kConstStringFlag; - data_.s.str = s; + data_.f.flags = kConstStringFlag; + SetStringPointer(s); data_.s.length = s.length; } //! Initialize this value as copy string with initial data, without calling destructor. void SetStringRaw(StringRefType s, Allocator& allocator) { - Ch* str = NULL; - if(ShortString::Usable(s.length)) { - flags_ = kShortStringFlag; + Ch* str = 0; + if (ShortString::Usable(s.length)) { + data_.f.flags = kShortStringFlag; data_.ss.SetLength(s.length); str = data_.ss.str; } else { - flags_ = kCopyStringFlag; + data_.f.flags = kCopyStringFlag; data_.s.length = s.length; - str = (Ch *)allocator.Malloc((s.length + 1) * sizeof(Ch)); - data_.s.str = str; + str = static_cast(allocator.Malloc((s.length + 1) * sizeof(Ch))); + SetStringPointer(str); } std::memcpy(str, s, s.length * sizeof(Ch)); str[s.length] = '\0'; @@ -1678,8 +2054,8 @@ class GenericValue { //! Assignment without calling destructor void RawAssign(GenericValue& rhs) RAPIDJSON_NOEXCEPT { data_ = rhs.data_; - flags_ = rhs.flags_; - rhs.flags_ = kNullFlag; + // data_.f.flags = rhs.data_.f.flags; + rhs.data_.f.flags = kNullFlag; } template @@ -1699,7 +2075,6 @@ class GenericValue { } Data data_; - unsigned flags_; }; //! GenericValue with UTF8 encoding @@ -1724,7 +2099,22 @@ class GenericDocument : public GenericValue { typedef Allocator AllocatorType; //!< Allocator type from template parameter. //! Constructor - /*! \param allocator Optional allocator for allocating memory. + /*! Creates an empty document of specified type. + \param type Mandatory type of object to create. + \param allocator Optional allocator for allocating memory. + \param stackCapacity Optional initial capacity of stack in bytes. + \param stackAllocator Optional allocator for allocating memory for stack. + */ + explicit GenericDocument(Type type, Allocator* allocator = 0, size_t stackCapacity = kDefaultStackCapacity, StackAllocator* stackAllocator = 0) : + GenericValue(type), allocator_(allocator), ownAllocator_(0), stack_(stackAllocator, stackCapacity), parseResult_() + { + if (!allocator_) + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + } + + //! Constructor + /*! Creates an empty document which type is Null. + \param allocator Optional allocator for allocating memory. \param stackCapacity Optional initial capacity of stack in bytes. \param stackAllocator Optional allocator for allocating memory for stack. */ @@ -1732,13 +2122,13 @@ class GenericDocument : public GenericValue { allocator_(allocator), ownAllocator_(0), stack_(stackAllocator, stackCapacity), parseResult_() { if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); } #if RAPIDJSON_HAS_CXX11_RVALUE_REFS //! Move constructor in C++11 GenericDocument(GenericDocument&& rhs) RAPIDJSON_NOEXCEPT - : ValueType(std::move(rhs)), + : ValueType(std::forward(rhs)), // explicit cast to avoid prohibited move from Document allocator_(rhs.allocator_), ownAllocator_(rhs.ownAllocator_), stack_(std::move(rhs.stack_)), @@ -1778,6 +2168,54 @@ class GenericDocument : public GenericValue { } #endif + //! Exchange the contents of this document with those of another. + /*! + \param rhs Another document. + \note Constant complexity. + \see GenericValue::Swap + */ + GenericDocument& Swap(GenericDocument& rhs) RAPIDJSON_NOEXCEPT { + ValueType::Swap(rhs); + stack_.Swap(rhs.stack_); + internal::Swap(allocator_, rhs.allocator_); + internal::Swap(ownAllocator_, rhs.ownAllocator_); + internal::Swap(parseResult_, rhs.parseResult_); + return *this; + } + + // Allow Swap with ValueType. + // Refer to Effective C++ 3rd Edition/Item 33: Avoid hiding inherited names. + using ValueType::Swap; + + //! free-standing swap function helper + /*! + Helper function to enable support for common swap implementation pattern based on \c std::swap: + \code + void swap(MyClass& a, MyClass& b) { + using std::swap; + swap(a.doc, b.doc); + // ... + } + \endcode + \see Swap() + */ + friend inline void swap(GenericDocument& a, GenericDocument& b) RAPIDJSON_NOEXCEPT { a.Swap(b); } + + //! Populate this document by a generator which produces SAX events. + /*! \tparam Generator A functor with bool f(Handler) prototype. + \param g Generator functor which sends SAX events to the parameter. + \return The document itself for fluent API. + */ + template + GenericDocument& Populate(Generator& g) { + ClearStackOnExit scope(*this); + if (g(*this)) { + RAPIDJSON_ASSERT(stack_.GetSize() == sizeof(ValueType)); // Got one and only one root object + ValueType::operator=(*stack_.template Pop(1));// Move value from stack to document + } + return *this; + } + //!@name Parse from stream //!@{ @@ -1790,13 +2228,13 @@ class GenericDocument : public GenericValue { */ template GenericDocument& ParseStream(InputStream& is) { - ValueType::SetNull(); // Remove existing root if exist - GenericReader reader(&stack_.GetAllocator()); + GenericReader reader( + stack_.HasAllocator() ? &stack_.GetAllocator() : 0); ClearStackOnExit scope(*this); parseResult_ = reader.template Parse(is, *this); if (parseResult_) { RAPIDJSON_ASSERT(stack_.GetSize() == sizeof(ValueType)); // Got one and only one root object - this->RawAssign(*stack_.template Pop(1)); // Add this-> to prevent issue 13. + ValueType::operator=(*stack_.template Pop(1));// Move value from stack to document } return *this; } @@ -1855,7 +2293,7 @@ class GenericDocument : public GenericValue { \param str Read-only zero-terminated string to be parsed. */ template - GenericDocument& Parse(const Ch* str) { + GenericDocument& Parse(const typename SourceEncoding::Ch* str) { RAPIDJSON_ASSERT(!(parseFlags & kParseInsituFlag)); GenericStringStream s(str); return ParseStream(s); @@ -1876,6 +2314,42 @@ class GenericDocument : public GenericValue { GenericDocument& Parse(const Ch* str) { return Parse(str); } + + template + GenericDocument& Parse(const typename SourceEncoding::Ch* str, size_t length) { + RAPIDJSON_ASSERT(!(parseFlags & kParseInsituFlag)); + MemoryStream ms(reinterpret_cast(str), length * sizeof(typename SourceEncoding::Ch)); + EncodedInputStream is(ms); + ParseStream(is); + return *this; + } + + template + GenericDocument& Parse(const Ch* str, size_t length) { + return Parse(str, length); + } + + GenericDocument& Parse(const Ch* str, size_t length) { + return Parse(str, length); + } + +#if RAPIDJSON_HAS_STDSTRING + template + GenericDocument& Parse(const std::basic_string& str) { + // c_str() is constant complexity according to standard. Should be faster than Parse(const char*, size_t) + return Parse(str.c_str()); + } + + template + GenericDocument& Parse(const std::basic_string& str) { + return Parse(str.c_str()); + } + + GenericDocument& Parse(const std::basic_string& str) { + return Parse(str); + } +#endif // RAPIDJSON_HAS_STDSTRING + //!@} //!@name Handling parse errors @@ -1890,10 +2364,26 @@ class GenericDocument : public GenericValue { //! Get the position of last parsing error in input, 0 otherwise. size_t GetErrorOffset() const { return parseResult_.Offset(); } + //! Implicit conversion to get the last parse result +#ifndef __clang // -Wdocumentation + /*! \return \ref ParseResult of the last parse operation + + \code + Document doc; + ParseResult ok = doc.Parse(json); + if (!ok) + printf( "JSON parse error: %s (%u)\n", GetParseError_En(ok.Code()), ok.Offset()); + \endcode + */ +#endif + operator ParseResult() const { return parseResult_; } //!@} //! Get the allocator of this document. - Allocator& GetAllocator() { return *allocator_; } + Allocator& GetAllocator() { + RAPIDJSON_ASSERT(allocator_); + return *allocator_; + } //! Get the capacity of stack in bytes. size_t GetStackCapacity() const { return stack_.GetCapacity(); } @@ -1910,9 +2400,10 @@ class GenericDocument : public GenericValue { }; // callers of the following private Handler functions - template friend class GenericReader; // for parsing + // template friend class GenericReader; // for parsing template friend class GenericValue; // for deep copying +public: // Implementation of Handler bool Null() { new (stack_.template Push()) ValueType(); return true; } bool Bool(bool b) { new (stack_.template Push()) ValueType(b); return true; } @@ -1922,6 +2413,14 @@ class GenericDocument : public GenericValue { bool Uint64(uint64_t i) { new (stack_.template Push()) ValueType(i); return true; } bool Double(double d) { new (stack_.template Push()) ValueType(d); return true; } + bool RawNumber(const Ch* str, SizeType length, bool copy) { + if (copy) + new (stack_.template Push()) ValueType(str, length, GetAllocator()); + else + new (stack_.template Push()) ValueType(str, length); + return true; + } + bool String(const Ch* str, SizeType length, bool copy) { if (copy) new (stack_.template Push()) ValueType(str, length, GetAllocator()); @@ -1936,7 +2435,7 @@ class GenericDocument : public GenericValue { bool EndObject(SizeType memberCount) { typename ValueType::Member* members = stack_.template Pop(memberCount); - stack_.template Top()->SetObjectRaw(members, (SizeType)memberCount, GetAllocator()); + stack_.template Top()->SetObjectRaw(members, memberCount, GetAllocator()); return true; } @@ -1977,38 +2476,152 @@ class GenericDocument : public GenericValue { //! GenericDocument with UTF8 encoding typedef GenericDocument > Document; -// defined here due to the dependency on GenericDocument -template -template -inline -GenericValue::GenericValue(const GenericValue& rhs, Allocator& allocator) -{ - switch (rhs.GetType()) { - case kObjectType: - case kArrayType: { // perform deep copy via SAX Handler - GenericDocument d(&allocator); - rhs.Accept(d); - RawAssign(*d.stack_.template Pop(1)); - } - break; - case kStringType: - if (rhs.flags_ == kConstStringFlag) { - flags_ = rhs.flags_; - data_ = *reinterpret_cast(&rhs.data_); - } else { - SetStringRaw(StringRef(rhs.GetString(), rhs.GetStringLength()), allocator); - } - break; - default: // kNumberType, kTrueType, kFalseType, kNullType - flags_ = rhs.flags_; - data_ = *reinterpret_cast(&rhs.data_); - } -} +//! Helper class for accessing Value of array type. +/*! + Instance of this helper class is obtained by \c GenericValue::GetArray(). + In addition to all APIs for array type, it provides range-based for loop if \c RAPIDJSON_HAS_CXX11_RANGE_FOR=1. +*/ +template +class GenericArray { +public: + typedef GenericArray ConstArray; + typedef GenericArray Array; + typedef ValueT PlainType; + typedef typename internal::MaybeAddConst::Type ValueType; + typedef ValueType* ValueIterator; // This may be const or non-const iterator + typedef const ValueT* ConstValueIterator; + typedef typename ValueType::AllocatorType AllocatorType; + typedef typename ValueType::StringRefType StringRefType; + + template + friend class GenericValue; + + GenericArray(const GenericArray& rhs) : value_(rhs.value_) {} + GenericArray& operator=(const GenericArray& rhs) { value_ = rhs.value_; return *this; } + ~GenericArray() {} + + SizeType Size() const { return value_.Size(); } + SizeType Capacity() const { return value_.Capacity(); } + bool Empty() const { return value_.Empty(); } + void Clear() const { value_.Clear(); } + ValueType& operator[](SizeType index) const { return value_[index]; } + ValueIterator Begin() const { return value_.Begin(); } + ValueIterator End() const { return value_.End(); } + GenericArray Reserve(SizeType newCapacity, AllocatorType &allocator) const { value_.Reserve(newCapacity, allocator); return *this; } + GenericArray PushBack(ValueType& value, AllocatorType& allocator) const { value_.PushBack(value, allocator); return *this; } +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + GenericArray PushBack(ValueType&& value, AllocatorType& allocator) const { value_.PushBack(value, allocator); return *this; } +#endif // RAPIDJSON_HAS_CXX11_RVALUE_REFS + GenericArray PushBack(StringRefType value, AllocatorType& allocator) const { value_.PushBack(value, allocator); return *this; } + template RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (const GenericArray&)) PushBack(T value, AllocatorType& allocator) const { value_.PushBack(value, allocator); return *this; } + GenericArray PopBack() const { value_.PopBack(); return *this; } + ValueIterator Erase(ConstValueIterator pos) const { return value_.Erase(pos); } + ValueIterator Erase(ConstValueIterator first, ConstValueIterator last) const { return value_.Erase(first, last); } + +#if RAPIDJSON_HAS_CXX11_RANGE_FOR + ValueIterator begin() const { return value_.Begin(); } + ValueIterator end() const { return value_.End(); } +#endif -RAPIDJSON_NAMESPACE_END +private: + GenericArray(); + GenericArray(ValueType& value) : value_(value) {} + ValueType& value_; +}; -#if defined(_MSC_VER) || defined(__GNUC__) -RAPIDJSON_DIAG_POP +//! Helper class for accessing Value of object type. +/*! + Instance of this helper class is obtained by \c GenericValue::GetObject(). + In addition to all APIs for array type, it provides range-based for loop if \c RAPIDJSON_HAS_CXX11_RANGE_FOR=1. +*/ +template +class GenericObject { +public: + typedef GenericObject ConstObject; + typedef GenericObject Object; + typedef ValueT PlainType; + typedef typename internal::MaybeAddConst::Type ValueType; + typedef GenericMemberIterator MemberIterator; // This may be const or non-const iterator + typedef GenericMemberIterator ConstMemberIterator; + typedef typename ValueType::AllocatorType AllocatorType; + typedef typename ValueType::StringRefType StringRefType; + typedef typename ValueType::EncodingType EncodingType; + typedef typename ValueType::Ch Ch; + + template + friend class GenericValue; + + GenericObject(const GenericObject& rhs) : value_(rhs.value_) {} + GenericObject& operator=(const GenericObject& rhs) { value_ = rhs.value_; return *this; } + ~GenericObject() {} + + SizeType MemberCount() const { return value_.MemberCount(); } + bool ObjectEmpty() const { return value_.ObjectEmpty(); } + template ValueType& operator[](T* name) const { return value_[name]; } + template ValueType& operator[](const GenericValue& name) const { return value_[name]; } +#if RAPIDJSON_HAS_STDSTRING + ValueType& operator[](const std::basic_string& name) const { return value_[name]; } +#endif + MemberIterator MemberBegin() const { return value_.MemberBegin(); } + MemberIterator MemberEnd() const { return value_.MemberEnd(); } + bool HasMember(const Ch* name) const { return value_.HasMember(name); } +#if RAPIDJSON_HAS_STDSTRING + bool HasMember(const std::basic_string& name) const { return value_.HasMember(name); } +#endif + template bool HasMember(const GenericValue& name) const { return value_.HasMember(name); } + MemberIterator FindMember(const Ch* name) const { return value_.FindMember(name); } + template MemberIterator FindMember(const GenericValue& name) const { return value_.FindMember(name); } +#if RAPIDJSON_HAS_STDSTRING + MemberIterator FindMember(const std::basic_string& name) const { return value_.FindMember(name); } +#endif + GenericObject AddMember(ValueType& name, ValueType& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + GenericObject AddMember(ValueType& name, StringRefType value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } +#if RAPIDJSON_HAS_STDSTRING + GenericObject AddMember(ValueType& name, std::basic_string& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } #endif + template RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (ValueType&)) AddMember(ValueType& name, T value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + GenericObject AddMember(ValueType&& name, ValueType&& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + GenericObject AddMember(ValueType&& name, ValueType& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + GenericObject AddMember(ValueType& name, ValueType&& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + GenericObject AddMember(StringRefType name, ValueType&& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } +#endif // RAPIDJSON_HAS_CXX11_RVALUE_REFS + GenericObject AddMember(StringRefType name, ValueType& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + GenericObject AddMember(StringRefType name, StringRefType value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + template RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (GenericObject)) AddMember(StringRefType name, T value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } + void RemoveAllMembers() { value_.RemoveAllMembers(); } + bool RemoveMember(const Ch* name) const { return value_.RemoveMember(name); } +#if RAPIDJSON_HAS_STDSTRING + bool RemoveMember(const std::basic_string& name) const { return value_.RemoveMember(name); } +#endif + template bool RemoveMember(const GenericValue& name) const { return value_.RemoveMember(name); } + MemberIterator RemoveMember(MemberIterator m) const { return value_.RemoveMember(m); } + MemberIterator EraseMember(ConstMemberIterator pos) const { return value_.EraseMember(pos); } + MemberIterator EraseMember(ConstMemberIterator first, ConstMemberIterator last) const { return value_.EraseMember(first, last); } + bool EraseMember(const Ch* name) const { return value_.EraseMember(name); } +#if RAPIDJSON_HAS_STDSTRING + bool EraseMember(const std::basic_string& name) const { return EraseMember(ValueType(StringRef(name))); } +#endif + template bool EraseMember(const GenericValue& name) const { return value_.EraseMember(name); } + +#if RAPIDJSON_HAS_CXX11_RANGE_FOR + MemberIterator begin() const { return value_.MemberBegin(); } + MemberIterator end() const { return value_.MemberEnd(); } +#endif + +private: + GenericObject(); + GenericObject(ValueType& value) : value_(value) {} + ValueType& value_; +}; + +RAPIDJSON_NAMESPACE_END +#ifdef _MINWINDEF_ // see: http://stackoverflow.com/questions/22744262/cant-call-stdmax-because-minwindef-h-defines-max +#ifndef NOMINMAX +#pragma pop_macro("min") +#pragma pop_macro("max") +#endif +#endif +RAPIDJSON_DIAG_POP #endif // RAPIDJSON_DOCUMENT_H_ diff --git a/rapidjson/encodedstream.h b/rapidjson/encodedstream.h index 7c8863f..223601c 100644 --- a/rapidjson/encodedstream.h +++ b/rapidjson/encodedstream.h @@ -15,13 +15,19 @@ #ifndef RAPIDJSON_ENCODEDSTREAM_H_ #define RAPIDJSON_ENCODEDSTREAM_H_ -#include "rapidjson.h" +#include "stream.h" +#include "memorystream.h" #ifdef __GNUC__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) #endif +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +#endif + RAPIDJSON_NAMESPACE_BEGIN //! Input byte stream wrapper with a statically bound encoding. @@ -57,10 +63,38 @@ class EncodedInputStream { Ch current_; }; +//! Specialized for UTF8 MemoryStream. +template <> +class EncodedInputStream, MemoryStream> { +public: + typedef UTF8<>::Ch Ch; + + EncodedInputStream(MemoryStream& is) : is_(is) { + if (static_cast(is_.Peek()) == 0xEFu) is_.Take(); + if (static_cast(is_.Peek()) == 0xBBu) is_.Take(); + if (static_cast(is_.Peek()) == 0xBFu) is_.Take(); + } + Ch Peek() const { return is_.Peek(); } + Ch Take() { return is_.Take(); } + size_t Tell() const { return is_.Tell(); } + + // Not implemented + void Put(Ch) {} + void Flush() {} + Ch* PutBegin() { return 0; } + size_t PutEnd(Ch*) { return 0; } + + MemoryStream& is_; + +private: + EncodedInputStream(const EncodedInputStream&); + EncodedInputStream& operator=(const EncodedInputStream&); +}; + //! Output byte stream wrapper with statically bound encoding. /*! \tparam Encoding The interpretation of encoding of the stream. Either UTF8, UTF16LE, UTF16BE, UTF32LE, UTF32BE. - \tparam InputByteStream Type of input byte stream. For example, FileWriteStream. + \tparam OutputByteStream Type of input byte stream. For example, FileWriteStream. */ template class EncodedOutputStream { @@ -77,8 +111,8 @@ class EncodedOutputStream { void Flush() { os_.Flush(); } // Not implemented - Ch Peek() const { RAPIDJSON_ASSERT(false); } - Ch Take() { RAPIDJSON_ASSERT(false); } + Ch Peek() const { RAPIDJSON_ASSERT(false); return 0;} + Ch Take() { RAPIDJSON_ASSERT(false); return 0;} size_t Tell() const { RAPIDJSON_ASSERT(false); return 0; } Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } size_t PutEnd(Ch*) { RAPIDJSON_ASSERT(false); return 0; } @@ -142,11 +176,11 @@ class AutoUTFInputStream { // FF FE UTF-16LE // EF BB BF UTF-8 - const unsigned char* c = (const unsigned char *)is_->Peek4(); + const unsigned char* c = reinterpret_cast(is_->Peek4()); if (!c) return; - unsigned bom = c[0] | (c[1] << 8) | (c[2] << 16) | (c[3] << 24); + unsigned bom = static_cast(c[0] | (c[1] << 8) | (c[2] << 16) | (c[3] << 24)); hasBOM_ = false; if (bom == 0xFFFE0000) { type_ = kUTF32BE; hasBOM_ = true; is_->Take(); is_->Take(); is_->Take(); is_->Take(); } else if (bom == 0x0000FEFF) { type_ = kUTF32LE; hasBOM_ = true; is_->Take(); is_->Take(); is_->Take(); is_->Take(); } @@ -166,7 +200,7 @@ class AutoUTFInputStream { // xx xx xx xx UTF-8 if (!hasBOM_) { - unsigned pattern = (c[0] ? 1 : 0) | (c[1] ? 2 : 0) | (c[2] ? 4 : 0) | (c[3] ? 8 : 0); + int pattern = (c[0] ? 1 : 0) | (c[1] ? 2 : 0) | (c[2] ? 4 : 0) | (c[3] ? 8 : 0); switch (pattern) { case 0x08: type_ = kUTF32BE; break; case 0x0A: type_ = kUTF16BE; break; @@ -193,7 +227,7 @@ class AutoUTFInputStream { //! Output stream wrapper with dynamically bound encoding and automatic encoding detection. /*! \tparam CharType Type of character for writing. - \tparam InputByteStream type of output byte stream to be wrapped. + \tparam OutputByteStream type of output byte stream to be wrapped. */ template class AutoUTFOutputStream { @@ -227,8 +261,8 @@ class AutoUTFOutputStream { void Flush() { os_->Flush(); } // Not implemented - Ch Peek() const { RAPIDJSON_ASSERT(false); } - Ch Take() { RAPIDJSON_ASSERT(false); } + Ch Peek() const { RAPIDJSON_ASSERT(false); return 0;} + Ch Take() { RAPIDJSON_ASSERT(false); return 0;} size_t Tell() const { RAPIDJSON_ASSERT(false); return 0; } Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } size_t PutEnd(Ch*) { RAPIDJSON_ASSERT(false); return 0; } @@ -254,6 +288,10 @@ class AutoUTFOutputStream { RAPIDJSON_NAMESPACE_END +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + #ifdef __GNUC__ RAPIDJSON_DIAG_POP #endif diff --git a/rapidjson/encodings.h b/rapidjson/encodings.h index 90b46ed..0df1c34 100644 --- a/rapidjson/encodings.h +++ b/rapidjson/encodings.h @@ -120,19 +120,45 @@ struct UTF8 { } } + template + static void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + if (codepoint <= 0x7F) + PutUnsafe(os, static_cast(codepoint & 0xFF)); + else if (codepoint <= 0x7FF) { + PutUnsafe(os, static_cast(0xC0 | ((codepoint >> 6) & 0xFF))); + PutUnsafe(os, static_cast(0x80 | ((codepoint & 0x3F)))); + } + else if (codepoint <= 0xFFFF) { + PutUnsafe(os, static_cast(0xE0 | ((codepoint >> 12) & 0xFF))); + PutUnsafe(os, static_cast(0x80 | ((codepoint >> 6) & 0x3F))); + PutUnsafe(os, static_cast(0x80 | (codepoint & 0x3F))); + } + else { + RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); + PutUnsafe(os, static_cast(0xF0 | ((codepoint >> 18) & 0xFF))); + PutUnsafe(os, static_cast(0x80 | ((codepoint >> 12) & 0x3F))); + PutUnsafe(os, static_cast(0x80 | ((codepoint >> 6) & 0x3F))); + PutUnsafe(os, static_cast(0x80 | (codepoint & 0x3F))); + } + } + template static bool Decode(InputStream& is, unsigned* codepoint) { -#define COPY() c = is.Take(); *codepoint = (*codepoint << 6) | ((unsigned char)c & 0x3Fu) -#define TRANS(mask) result &= ((GetRange((unsigned char)c) & mask) != 0) +#define COPY() c = is.Take(); *codepoint = (*codepoint << 6) | (static_cast(c) & 0x3Fu) +#define TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) #define TAIL() COPY(); TRANS(0x70) - Ch c = is.Take(); + typename InputStream::Ch c = is.Take(); if (!(c & 0x80)) { - *codepoint = (unsigned char)c; + *codepoint = static_cast(c); return true; } - unsigned char type = GetRange((unsigned char)c); - *codepoint = (0xFF >> type) & (unsigned char)c; + unsigned char type = GetRange(static_cast(c)); + if (type >= 32) { + *codepoint = 0; + } else { + *codepoint = (0xFFu >> type) & static_cast(c); + } bool result = true; switch (type) { case 2: TAIL(); return result; @@ -152,7 +178,7 @@ struct UTF8 { template static bool Validate(InputStream& is, OutputStream& os) { #define COPY() os.Put(c = is.Take()) -#define TRANS(mask) result &= ((GetRange((unsigned char)c) & mask) != 0) +#define TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) #define TAIL() COPY(); TRANS(0x70) Ch c; COPY(); @@ -160,7 +186,7 @@ struct UTF8 { return true; bool result = true; - switch (GetRange((unsigned char)c)) { + switch (GetRange(static_cast(c))) { case 2: TAIL(); return result; case 3: TAIL(); TAIL(); return result; case 4: COPY(); TRANS(0x50); TAIL(); return result; @@ -196,12 +222,12 @@ struct UTF8 { template static CharType TakeBOM(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - Ch c = Take(is); - if ((unsigned char)c != 0xEFu) return c; + typename InputByteStream::Ch c = Take(is); + if (static_cast(c) != 0xEFu) return c; c = is.Take(); - if ((unsigned char)c != 0xBBu) return c; + if (static_cast(c) != 0xBBu) return c; c = is.Take(); - if ((unsigned char)c != 0xBFu) return c; + if (static_cast(c) != 0xBFu) return c; c = is.Take(); return c; } @@ -209,13 +235,15 @@ struct UTF8 { template static Ch Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - return is.Take(); + return static_cast(is.Take()); } template static void PutBOM(OutputByteStream& os) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(0xEFu); os.Put(0xBBu); os.Put(0xBFu); + os.Put(static_cast(0xEFu)); + os.Put(static_cast(0xBBu)); + os.Put(static_cast(0xBFu)); } template @@ -255,22 +283,38 @@ struct UTF16 { RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); unsigned v = codepoint - 0x10000; os.Put(static_cast((v >> 10) | 0xD800)); - os.Put((v & 0x3FF) | 0xDC00); + os.Put(static_cast((v & 0x3FF) | 0xDC00)); + } + } + + + template + static void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputStream::Ch) >= 2); + if (codepoint <= 0xFFFF) { + RAPIDJSON_ASSERT(codepoint < 0xD800 || codepoint > 0xDFFF); // Code point itself cannot be surrogate pair + PutUnsafe(os, static_cast(codepoint)); + } + else { + RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); + unsigned v = codepoint - 0x10000; + PutUnsafe(os, static_cast((v >> 10) | 0xD800)); + PutUnsafe(os, static_cast((v & 0x3FF) | 0xDC00)); } } template static bool Decode(InputStream& is, unsigned* codepoint) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputStream::Ch) >= 2); - Ch c = is.Take(); + typename InputStream::Ch c = is.Take(); if (c < 0xD800 || c > 0xDFFF) { - *codepoint = c; + *codepoint = static_cast(c); return true; } else if (c <= 0xDBFF) { - *codepoint = (c & 0x3FF) << 10; + *codepoint = (static_cast(c) & 0x3FF) << 10; c = is.Take(); - *codepoint |= (c & 0x3FF); + *codepoint |= (static_cast(c) & 0x3FF); *codepoint += 0x10000; return c >= 0xDC00 && c <= 0xDFFF; } @@ -281,8 +325,8 @@ struct UTF16 { static bool Validate(InputStream& is, OutputStream& os) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputStream::Ch) >= 2); RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputStream::Ch) >= 2); - Ch c; - os.Put(c = is.Take()); + typename InputStream::Ch c; + os.Put(static_cast(c = is.Take())); if (c < 0xD800 || c > 0xDFFF) return true; else if (c <= 0xDBFF) { @@ -300,28 +344,29 @@ struct UTF16LE : UTF16 { static CharType TakeBOM(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); CharType c = Take(is); - return (unsigned short)c == 0xFEFFu ? Take(is) : c; + return static_cast(c) == 0xFEFFu ? Take(is) : c; } template static CharType Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - CharType c = (unsigned char)is.Take(); - c |= (unsigned char)is.Take() << 8; - return c; + unsigned c = static_cast(is.Take()); + c |= static_cast(static_cast(is.Take())) << 8; + return static_cast(c); } template static void PutBOM(OutputByteStream& os) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(0xFFu); os.Put(0xFEu); + os.Put(static_cast(0xFFu)); + os.Put(static_cast(0xFEu)); } template static void Put(OutputByteStream& os, CharType c) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(c & 0xFFu); - os.Put((c >> 8) & 0xFFu); + os.Put(static_cast(static_cast(c) & 0xFFu)); + os.Put(static_cast((static_cast(c) >> 8) & 0xFFu)); } }; @@ -332,28 +377,29 @@ struct UTF16BE : UTF16 { static CharType TakeBOM(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); CharType c = Take(is); - return (unsigned short)c == 0xFEFFu ? Take(is) : c; + return static_cast(c) == 0xFEFFu ? Take(is) : c; } template static CharType Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - CharType c = (unsigned char)is.Take() << 8; - c |= (unsigned char)is.Take(); - return c; + unsigned c = static_cast(static_cast(is.Take())) << 8; + c |= static_cast(is.Take()); + return static_cast(c); } template static void PutBOM(OutputByteStream& os) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(0xFEu); os.Put(0xFFu); + os.Put(static_cast(0xFEu)); + os.Put(static_cast(0xFFu)); } template static void Put(OutputByteStream& os, CharType c) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put((c >> 8) & 0xFFu); - os.Put(c & 0xFFu); + os.Put(static_cast((static_cast(c) >> 8) & 0xFFu)); + os.Put(static_cast(static_cast(c) & 0xFFu)); } }; @@ -382,6 +428,13 @@ struct UTF32 { os.Put(codepoint); } + template + static void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputStream::Ch) >= 4); + RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); + PutUnsafe(os, codepoint); + } + template static bool Decode(InputStream& is, unsigned* codepoint) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputStream::Ch) >= 4); @@ -406,32 +459,35 @@ struct UTF32LE : UTF32 { static CharType TakeBOM(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); CharType c = Take(is); - return (unsigned)c == 0x0000FEFFu ? Take(is) : c; + return static_cast(c) == 0x0000FEFFu ? Take(is) : c; } template static CharType Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - CharType c = (unsigned char)is.Take(); - c |= (unsigned char)is.Take() << 8; - c |= (unsigned char)is.Take() << 16; - c |= (unsigned char)is.Take() << 24; - return c; + unsigned c = static_cast(is.Take()); + c |= static_cast(static_cast(is.Take())) << 8; + c |= static_cast(static_cast(is.Take())) << 16; + c |= static_cast(static_cast(is.Take())) << 24; + return static_cast(c); } template static void PutBOM(OutputByteStream& os) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(0xFFu); os.Put(0xFEu); os.Put(0x00u); os.Put(0x00u); + os.Put(static_cast(0xFFu)); + os.Put(static_cast(0xFEu)); + os.Put(static_cast(0x00u)); + os.Put(static_cast(0x00u)); } template static void Put(OutputByteStream& os, CharType c) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(c & 0xFFu); - os.Put((c >> 8) & 0xFFu); - os.Put((c >> 16) & 0xFFu); - os.Put((c >> 24) & 0xFFu); + os.Put(static_cast(c & 0xFFu)); + os.Put(static_cast((c >> 8) & 0xFFu)); + os.Put(static_cast((c >> 16) & 0xFFu)); + os.Put(static_cast((c >> 24) & 0xFFu)); } }; @@ -442,32 +498,35 @@ struct UTF32BE : UTF32 { static CharType TakeBOM(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); CharType c = Take(is); - return (unsigned)c == 0x0000FEFFu ? Take(is) : c; + return static_cast(c) == 0x0000FEFFu ? Take(is) : c; } template static CharType Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - CharType c = (unsigned char)is.Take() << 24; - c |= (unsigned char)is.Take() << 16; - c |= (unsigned char)is.Take() << 8; - c |= (unsigned char)is.Take(); - return c; + unsigned c = static_cast(static_cast(is.Take())) << 24; + c |= static_cast(static_cast(is.Take())) << 16; + c |= static_cast(static_cast(is.Take())) << 8; + c |= static_cast(static_cast(is.Take())); + return static_cast(c); } template static void PutBOM(OutputByteStream& os) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put(0x00u); os.Put(0x00u); os.Put(0xFEu); os.Put(0xFFu); + os.Put(static_cast(0x00u)); + os.Put(static_cast(0x00u)); + os.Put(static_cast(0xFEu)); + os.Put(static_cast(0xFFu)); } template static void Put(OutputByteStream& os, CharType c) { RAPIDJSON_STATIC_ASSERT(sizeof(typename OutputByteStream::Ch) == 1); - os.Put((c >> 24) & 0xFFu); - os.Put((c >> 16) & 0xFFu); - os.Put((c >> 8) & 0xFFu); - os.Put(c & 0xFFu); + os.Put(static_cast((c >> 24) & 0xFFu)); + os.Put(static_cast((c >> 16) & 0xFFu)); + os.Put(static_cast((c >> 8) & 0xFFu)); + os.Put(static_cast(c & 0xFFu)); } }; @@ -491,31 +550,37 @@ struct ASCII { os.Put(static_cast(codepoint & 0xFF)); } + template + static void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + RAPIDJSON_ASSERT(codepoint <= 0x7F); + PutUnsafe(os, static_cast(codepoint & 0xFF)); + } + template static bool Decode(InputStream& is, unsigned* codepoint) { - unsigned char c = static_cast(is.Take()); + uint8_t c = static_cast(is.Take()); *codepoint = c; return c <= 0X7F; } template static bool Validate(InputStream& is, OutputStream& os) { - unsigned char c = is.Take(); - os.Put(c); + uint8_t c = static_cast(is.Take()); + os.Put(static_cast(c)); return c <= 0x7F; } template static CharType TakeBOM(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - Ch c = Take(is); - return c; + uint8_t c = static_cast(Take(is)); + return static_cast(c); } template static Ch Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); - return is.Take(); + return static_cast(is.Take()); } template @@ -555,21 +620,28 @@ struct AutoUTF { #define RAPIDJSON_ENCODINGS_FUNC(x) UTF8::x, UTF16LE::x, UTF16BE::x, UTF32LE::x, UTF32BE::x template - RAPIDJSON_FORCEINLINE static void Encode(OutputStream& os, unsigned codepoint) { + static RAPIDJSON_FORCEINLINE void Encode(OutputStream& os, unsigned codepoint) { typedef void (*EncodeFunc)(OutputStream&, unsigned); static const EncodeFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(Encode) }; (*f[os.GetType()])(os, codepoint); } + template + static RAPIDJSON_FORCEINLINE void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + typedef void (*EncodeFunc)(OutputStream&, unsigned); + static const EncodeFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(EncodeUnsafe) }; + (*f[os.GetType()])(os, codepoint); + } + template - RAPIDJSON_FORCEINLINE static bool Decode(InputStream& is, unsigned* codepoint) { + static RAPIDJSON_FORCEINLINE bool Decode(InputStream& is, unsigned* codepoint) { typedef bool (*DecodeFunc)(InputStream&, unsigned*); static const DecodeFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(Decode) }; return (*f[is.GetType()])(is, codepoint); } template - RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { typedef bool (*ValidateFunc)(InputStream&, OutputStream&); static const ValidateFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(Validate) }; return (*f[is.GetType()])(is, os); @@ -586,7 +658,7 @@ template struct Transcoder { //! Take one Unicode codepoint from source encoding, convert it to target encoding and put it to the output stream. template - RAPIDJSON_FORCEINLINE static bool Transcode(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Transcode(InputStream& is, OutputStream& os) { unsigned codepoint; if (!SourceEncoding::Decode(is, &codepoint)) return false; @@ -594,31 +666,50 @@ struct Transcoder { return true; } + template + static RAPIDJSON_FORCEINLINE bool TranscodeUnsafe(InputStream& is, OutputStream& os) { + unsigned codepoint; + if (!SourceEncoding::Decode(is, &codepoint)) + return false; + TargetEncoding::EncodeUnsafe(os, codepoint); + return true; + } + //! Validate one Unicode codepoint from an encoded stream. template - RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { return Transcode(is, os); // Since source/target encoding is different, must transcode. } }; +// Forward declaration. +template +inline void PutUnsafe(Stream& stream, typename Stream::Ch c); + //! Specialization of Transcoder with same source and target encoding. template struct Transcoder { template - RAPIDJSON_FORCEINLINE static bool Transcode(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Transcode(InputStream& is, OutputStream& os) { os.Put(is.Take()); // Just copy one code unit. This semantic is different from primary template class. return true; } template - RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool TranscodeUnsafe(InputStream& is, OutputStream& os) { + PutUnsafe(os, is.Take()); // Just copy one code unit. This semantic is different from primary template class. + return true; + } + + template + static RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { return Encoding::Validate(is, os); // source/target encoding are the same } }; RAPIDJSON_NAMESPACE_END -#if defined(__GNUC__) || defined(_MSV_VER) +#if defined(__GNUC__) || defined(_MSC_VER) RAPIDJSON_DIAG_POP #endif diff --git a/rapidjson/error/en.h b/rapidjson/error/en.h index d5f9caa..2db838b 100644 --- a/rapidjson/error/en.h +++ b/rapidjson/error/en.h @@ -12,11 +12,17 @@ // CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. -#ifndef RAPIDJSON_ERROR_EN_H__ -#define RAPIDJSON_ERROR_EN_H__ +#ifndef RAPIDJSON_ERROR_EN_H_ +#define RAPIDJSON_ERROR_EN_H_ #include "error.h" +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(switch-enum) +RAPIDJSON_DIAG_OFF(covered-switch-default) +#endif + RAPIDJSON_NAMESPACE_BEGIN //! Maps error code of parsing into error message. @@ -32,7 +38,7 @@ inline const RAPIDJSON_ERROR_CHARTYPE* GetParseError_En(ParseErrorCode parseErro case kParseErrorNone: return RAPIDJSON_ERROR_STRING("No error."); case kParseErrorDocumentEmpty: return RAPIDJSON_ERROR_STRING("The document is empty."); - case kParseErrorDocumentRootNotSingular: return RAPIDJSON_ERROR_STRING("The document root must not follow by other values."); + case kParseErrorDocumentRootNotSingular: return RAPIDJSON_ERROR_STRING("The document root must not be followed by other values."); case kParseErrorValueInvalid: return RAPIDJSON_ERROR_STRING("Invalid value."); @@ -55,11 +61,14 @@ inline const RAPIDJSON_ERROR_CHARTYPE* GetParseError_En(ParseErrorCode parseErro case kParseErrorTermination: return RAPIDJSON_ERROR_STRING("Terminate parsing due to Handler error."); case kParseErrorUnspecificSyntaxError: return RAPIDJSON_ERROR_STRING("Unspecific syntax error."); - default: - return RAPIDJSON_ERROR_STRING("Unknown error."); + default: return RAPIDJSON_ERROR_STRING("Unknown error."); } } RAPIDJSON_NAMESPACE_END -#endif // RAPIDJSON_ERROR_EN_H__ +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + +#endif // RAPIDJSON_ERROR_EN_H_ diff --git a/rapidjson/error/error.h b/rapidjson/error/error.h index f9094fb..9311d2f 100644 --- a/rapidjson/error/error.h +++ b/rapidjson/error/error.h @@ -12,11 +12,16 @@ // CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. -#ifndef RAPIDJSON_ERROR_ERROR_H__ -#define RAPIDJSON_ERROR_ERROR_H__ +#ifndef RAPIDJSON_ERROR_ERROR_H_ +#define RAPIDJSON_ERROR_ERROR_H_ #include "../rapidjson.h" +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +#endif + /*! \file error.h */ /*! \defgroup RAPIDJSON_ERRORS RapidJSON error handling */ @@ -99,7 +104,9 @@ enum ParseErrorCode { \see GenericReader::Parse, GenericDocument::Parse */ struct ParseResult { - + //!! Unspecified boolean type + typedef bool (ParseResult::*BooleanType)() const; +public: //! Default constructor, no error. ParseResult() : code_(kParseErrorNone), offset_(0) {} //! Constructor to set an error. @@ -110,8 +117,8 @@ struct ParseResult { //! Get the error offset, if \ref IsError(), 0 otherwise. size_t Offset() const { return offset_; } - //! Conversion to \c bool, returns \c true, iff !\ref IsError(). - operator bool() const { return !IsError(); } + //! Explicit conversion to \c bool, returns \c true, iff !\ref IsError(). + operator BooleanType() const { return !IsError() ? &ParseResult::IsError : NULL; } //! Whether the result is an error. bool IsError() const { return code_ != kParseErrorNone; } @@ -119,6 +126,10 @@ struct ParseResult { bool operator==(ParseErrorCode code) const { return code_ == code; } friend bool operator==(ParseErrorCode code, const ParseResult & err) { return code == err.code_; } + bool operator!=(const ParseResult& that) const { return !(*this == that); } + bool operator!=(ParseErrorCode code) const { return !(*this == code); } + friend bool operator!=(ParseErrorCode code, const ParseResult & err) { return err != code; } + //! Reset error code. void Clear() { Set(kParseErrorNone); } //! Update error code and offset. @@ -143,4 +154,8 @@ typedef const RAPIDJSON_ERROR_CHARTYPE* (*GetParseErrorFunc)(ParseErrorCode); RAPIDJSON_NAMESPACE_END -#endif // RAPIDJSON_ERROR_ERROR_H__ +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + +#endif // RAPIDJSON_ERROR_ERROR_H_ diff --git a/rapidjson/filereadstream.h b/rapidjson/filereadstream.h index 3913eb7..b56ea13 100644 --- a/rapidjson/filereadstream.h +++ b/rapidjson/filereadstream.h @@ -15,9 +15,16 @@ #ifndef RAPIDJSON_FILEREADSTREAM_H_ #define RAPIDJSON_FILEREADSTREAM_H_ -#include "rapidjson.h" +#include "stream.h" #include +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +RAPIDJSON_DIAG_OFF(unreachable-code) +RAPIDJSON_DIAG_OFF(missing-noreturn) +#endif + RAPIDJSON_NAMESPACE_BEGIN //! File byte stream for input using fread(). @@ -85,4 +92,8 @@ class FileReadStream { RAPIDJSON_NAMESPACE_END +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + #endif // RAPIDJSON_FILESTREAM_H_ diff --git a/rapidjson/filewritestream.h b/rapidjson/filewritestream.h index dfb9cbd..6378dd6 100644 --- a/rapidjson/filewritestream.h +++ b/rapidjson/filewritestream.h @@ -15,9 +15,14 @@ #ifndef RAPIDJSON_FILEWRITESTREAM_H_ #define RAPIDJSON_FILEWRITESTREAM_H_ -#include "rapidjson.h" +#include "stream.h" #include +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(unreachable-code) +#endif + RAPIDJSON_NAMESPACE_BEGIN //! Wrapper of C file stream for input using fread(). @@ -57,7 +62,11 @@ class FileWriteStream { void Flush() { if (current_ != buffer_) { - fwrite(buffer_, 1, static_cast(current_ - buffer_), fp_); + size_t result = fwrite(buffer_, 1, static_cast(current_ - buffer_), fp_); + if (result < static_cast(current_ - buffer_)) { + // failure deliberately ignored at this time + // added to avoid warn_unused_result build errors + } current_ = buffer_; } } @@ -88,4 +97,8 @@ inline void PutN(FileWriteStream& stream, char c, size_t n) { RAPIDJSON_NAMESPACE_END +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + #endif // RAPIDJSON_FILESTREAM_H_ diff --git a/rapidjson/fwd.h b/rapidjson/fwd.h new file mode 100644 index 0000000..e8104e8 --- /dev/null +++ b/rapidjson/fwd.h @@ -0,0 +1,151 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_FWD_H_ +#define RAPIDJSON_FWD_H_ + +#include "rapidjson.h" + +RAPIDJSON_NAMESPACE_BEGIN + +// encodings.h + +template struct UTF8; +template struct UTF16; +template struct UTF16BE; +template struct UTF16LE; +template struct UTF32; +template struct UTF32BE; +template struct UTF32LE; +template struct ASCII; +template struct AutoUTF; + +template +struct Transcoder; + +// allocators.h + +class CrtAllocator; + +template +class MemoryPoolAllocator; + +// stream.h + +template +struct GenericStringStream; + +typedef GenericStringStream > StringStream; + +template +struct GenericInsituStringStream; + +typedef GenericInsituStringStream > InsituStringStream; + +// stringbuffer.h + +template +class GenericStringBuffer; + +typedef GenericStringBuffer, CrtAllocator> StringBuffer; + +// filereadstream.h + +class FileReadStream; + +// filewritestream.h + +class FileWriteStream; + +// memorybuffer.h + +template +struct GenericMemoryBuffer; + +typedef GenericMemoryBuffer MemoryBuffer; + +// memorystream.h + +struct MemoryStream; + +// reader.h + +template +struct BaseReaderHandler; + +template +class GenericReader; + +typedef GenericReader, UTF8, CrtAllocator> Reader; + +// writer.h + +template +class Writer; + +// prettywriter.h + +template +class PrettyWriter; + +// document.h + +template +struct GenericMember; + +template +class GenericMemberIterator; + +template +struct GenericStringRef; + +template +class GenericValue; + +typedef GenericValue, MemoryPoolAllocator > Value; + +template +class GenericDocument; + +typedef GenericDocument, MemoryPoolAllocator, CrtAllocator> Document; + +// pointer.h + +template +class GenericPointer; + +typedef GenericPointer Pointer; + +// schema.h + +template +class IGenericRemoteSchemaDocumentProvider; + +template +class GenericSchemaDocument; + +typedef GenericSchemaDocument SchemaDocument; +typedef IGenericRemoteSchemaDocumentProvider IRemoteSchemaDocumentProvider; + +template < + typename SchemaDocumentType, + typename OutputHandler, + typename StateAllocator> +class GenericSchemaValidator; + +typedef GenericSchemaValidator, void>, CrtAllocator> SchemaValidator; + +RAPIDJSON_NAMESPACE_END + +#endif // RAPIDJSON_RAPIDJSONFWD_H_ diff --git a/rapidjson/internal/biginteger.h b/rapidjson/internal/biginteger.h index 99a30ac..9d3e88c 100644 --- a/rapidjson/internal/biginteger.h +++ b/rapidjson/internal/biginteger.h @@ -19,6 +19,7 @@ #if defined(_MSC_VER) && defined(_M_AMD64) #include // for _umul128 +#pragma intrinsic(_umul128) #endif RAPIDJSON_NAMESPACE_BEGIN @@ -50,7 +51,16 @@ class BigInteger { if (length > 0) AppendDecimal64(decimals + i, decimals + i + length); } - + + BigInteger& operator=(const BigInteger &rhs) + { + if (this != &rhs) { + count_ = rhs.count_; + std::memcpy(digits_, rhs.digits_, count_ * sizeof(Type)); + } + return *this; + } + BigInteger& operator=(uint64_t u) { digits_[0] = u; count_ = 1; @@ -230,7 +240,7 @@ class BigInteger { uint64_t r = 0; for (const char* p = begin; p != end; ++p) { RAPIDJSON_ASSERT(*p >= '0' && *p <= '9'); - r = r * 10 + (*p - '0'); + r = r * 10u + static_cast(*p - '0'); } return r; } diff --git a/rapidjson/internal/diyfp.h b/rapidjson/internal/diyfp.h index be5903a..29abf80 100644 --- a/rapidjson/internal/diyfp.h +++ b/rapidjson/internal/diyfp.h @@ -21,9 +21,10 @@ #include "../rapidjson.h" -#if defined(_MSC_VER) && defined(_M_AMD64) +#if defined(_MSC_VER) && defined(_M_AMD64) && !defined(__INTEL_COMPILER) #include #pragma intrinsic(_BitScanReverse64) +#pragma intrinsic(_umul128) #endif RAPIDJSON_NAMESPACE_BEGIN @@ -34,8 +35,13 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) #endif +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +#endif + struct DiyFp { - DiyFp() {} + DiyFp() : f(), e() {} DiyFp(uint64_t fp, int exp) : f(fp), e(exp) {} @@ -228,19 +234,24 @@ inline DiyFp GetCachedPower(int e, int* K) { unsigned index = static_cast((k >> 3) + 1); *K = -(-348 + static_cast(index << 3)); // decimal exponent no need lookup table - return GetCachedPowerByIndex((size_t)index); + return GetCachedPowerByIndex(index); } inline DiyFp GetCachedPower10(int exp, int *outExp) { - unsigned index = (exp + 348) / 8; - *outExp = -348 + index * 8; - return GetCachedPowerByIndex((size_t)index); + unsigned index = (static_cast(exp) + 348u) / 8u; + *outExp = -348 + static_cast(index) * 8; + return GetCachedPowerByIndex(index); } #ifdef __GNUC__ RAPIDJSON_DIAG_POP #endif +#ifdef __clang__ +RAPIDJSON_DIAG_POP +RAPIDJSON_DIAG_OFF(padded) +#endif + } // namespace internal RAPIDJSON_NAMESPACE_END diff --git a/rapidjson/internal/dtoa.h b/rapidjson/internal/dtoa.h index 2d8d2e4..bf2e9b2 100644 --- a/rapidjson/internal/dtoa.h +++ b/rapidjson/internal/dtoa.h @@ -29,6 +29,7 @@ namespace internal { #ifdef __GNUC__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) +RAPIDJSON_DIAG_OFF(array-bounds) // some gcc versions generate wrong warnings https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59124 #endif inline void GrisuRound(char* buffer, int len, uint64_t delta, uint64_t rest, uint64_t ten_kappa, uint64_t wp_w) { @@ -40,7 +41,7 @@ inline void GrisuRound(char* buffer, int len, uint64_t delta, uint64_t rest, uin } } -inline unsigned CountDecimalDigit32(uint32_t n) { +inline int CountDecimalDigit32(uint32_t n) { // Simple pure C++ implementation was faster than __builtin_clz version in this situation. if (n < 10) return 1; if (n < 100) return 2; @@ -101,7 +102,8 @@ inline void DigitGen(const DiyFp& W, const DiyFp& Mp, uint64_t delta, char* buff kappa--; if (p2 < delta) { *K += kappa; - GrisuRound(buffer, *len, delta, p2, one.f, wp_w.f * kPow10[-kappa]); + int index = -kappa; + GrisuRound(buffer, *len, delta, p2, one.f, wp_w.f * (index < 9 ? kPow10[index] : 0)); return; } } @@ -145,10 +147,10 @@ inline char* WriteExponent(int K, char* buffer) { return buffer; } -inline char* Prettify(char* buffer, int length, int k) { +inline char* Prettify(char* buffer, int length, int k, int maxDecimalPlaces) { const int kk = length + k; // 10^(kk-1) <= v < 10^kk - if (length <= kk && kk <= 21) { + if (0 <= k && kk <= 21) { // 1234e7 -> 12340000000 for (int i = length; i < kk; i++) buffer[i] = '0'; @@ -158,19 +160,44 @@ inline char* Prettify(char* buffer, int length, int k) { } else if (0 < kk && kk <= 21) { // 1234e-2 -> 12.34 - std::memmove(&buffer[kk + 1], &buffer[kk], length - kk); + std::memmove(&buffer[kk + 1], &buffer[kk], static_cast(length - kk)); buffer[kk] = '.'; - return &buffer[length + 1]; + if (0 > k + maxDecimalPlaces) { + // When maxDecimalPlaces = 2, 1.2345 -> 1.23, 1.102 -> 1.1 + // Remove extra trailing zeros (at least one) after truncation. + for (int i = kk + maxDecimalPlaces; i > kk + 1; i--) + if (buffer[i] != '0') + return &buffer[i + 1]; + return &buffer[kk + 2]; // Reserve one zero + } + else + return &buffer[length + 1]; } else if (-6 < kk && kk <= 0) { // 1234e-6 -> 0.001234 const int offset = 2 - kk; - std::memmove(&buffer[offset], &buffer[0], length); + std::memmove(&buffer[offset], &buffer[0], static_cast(length)); buffer[0] = '0'; buffer[1] = '.'; for (int i = 2; i < offset; i++) buffer[i] = '0'; - return &buffer[length + offset]; + if (length - kk > maxDecimalPlaces) { + // When maxDecimalPlaces = 2, 0.123 -> 0.12, 0.102 -> 0.1 + // Remove extra trailing zeros (at least one) after truncation. + for (int i = maxDecimalPlaces + 1; i > 2; i--) + if (buffer[i] != '0') + return &buffer[i + 1]; + return &buffer[3]; // Reserve one zero + } + else + return &buffer[length + offset]; + } + else if (kk < -maxDecimalPlaces) { + // Truncate to zero + buffer[0] = '0'; + buffer[1] = '.'; + buffer[2] = '0'; + return &buffer[3]; } else if (length == 1) { // 1e30 @@ -179,14 +206,15 @@ inline char* Prettify(char* buffer, int length, int k) { } else { // 1234e30 -> 1.234e33 - std::memmove(&buffer[2], &buffer[1], length - 1); + std::memmove(&buffer[2], &buffer[1], static_cast(length - 1)); buffer[1] = '.'; buffer[length + 1] = 'e'; return WriteExponent(kk - 1, &buffer[0 + length + 2]); } } -inline char* dtoa(double value, char* buffer) { +inline char* dtoa(double value, char* buffer, int maxDecimalPlaces = 324) { + RAPIDJSON_ASSERT(maxDecimalPlaces >= 1); Double d(value); if (d.IsZero()) { if (d.Sign()) @@ -203,7 +231,7 @@ inline char* dtoa(double value, char* buffer) { } int length, K; Grisu2(value, buffer, &length, &K); - return Prettify(buffer, length, K); + return Prettify(buffer, length, K, maxDecimalPlaces); } } diff --git a/rapidjson/internal/ieee754.h b/rapidjson/internal/ieee754.h index e3f0336..c2684ba 100644 --- a/rapidjson/internal/ieee754.h +++ b/rapidjson/internal/ieee754.h @@ -40,6 +40,7 @@ class Double { bool IsNan() const { return (u_ & kExponentMask) == kExponentMask && Significand() != 0; } bool IsInf() const { return (u_ & kExponentMask) == kExponentMask && Significand() == 0; } + bool IsNanOrInf() const { return (u_ & kExponentMask) == kExponentMask; } bool IsNormal() const { return (u_ & kExponentMask) != 0 || Significand() == 0; } bool IsZero() const { return (u_ & (kExponentMask | kSignificandMask)) == 0; } @@ -47,7 +48,7 @@ class Double { int IntegerExponent() const { return (IsNormal() ? Exponent() : kDenormalExponent) - kSignificandSize; } uint64_t ToBias() const { return (u_ & kSignMask) ? ~u_ + 1 : u_ | kSignMask; } - static unsigned EffectiveSignificandSize(int order) { + static int EffectiveSignificandSize(int order) { if (order >= -1021) return 53; else if (order <= -1074) diff --git a/rapidjson/internal/regex.h b/rapidjson/internal/regex.h new file mode 100644 index 0000000..e1a2faa --- /dev/null +++ b/rapidjson/internal/regex.h @@ -0,0 +1,734 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_INTERNAL_REGEX_H_ +#define RAPIDJSON_INTERNAL_REGEX_H_ + +#include "../allocators.h" +#include "../stream.h" +#include "stack.h" + +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +RAPIDJSON_DIAG_OFF(switch-enum) +RAPIDJSON_DIAG_OFF(implicit-fallthrough) +#endif + +#ifdef __GNUC__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(effc++) +#if __GNUC__ >= 7 +RAPIDJSON_DIAG_OFF(implicit-fallthrough) +#endif +#endif + +#ifdef _MSC_VER +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +#ifndef RAPIDJSON_REGEX_VERBOSE +#define RAPIDJSON_REGEX_VERBOSE 0 +#endif + +RAPIDJSON_NAMESPACE_BEGIN +namespace internal { + +/////////////////////////////////////////////////////////////////////////////// +// DecodedStream + +template +class DecodedStream { +public: + DecodedStream(SourceStream& ss) : ss_(ss), codepoint_() { Decode(); } + unsigned Peek() { return codepoint_; } + unsigned Take() { + unsigned c = codepoint_; + if (c) // No further decoding when '\0' + Decode(); + return c; + } + +private: + void Decode() { + if (!Encoding::Decode(ss_, &codepoint_)) + codepoint_ = 0; + } + + SourceStream& ss_; + unsigned codepoint_; +}; + +/////////////////////////////////////////////////////////////////////////////// +// GenericRegex + +static const SizeType kRegexInvalidState = ~SizeType(0); //!< Represents an invalid index in GenericRegex::State::out, out1 +static const SizeType kRegexInvalidRange = ~SizeType(0); + +template +class GenericRegexSearch; + +//! Regular expression engine with subset of ECMAscript grammar. +/*! + Supported regular expression syntax: + - \c ab Concatenation + - \c a|b Alternation + - \c a? Zero or one + - \c a* Zero or more + - \c a+ One or more + - \c a{3} Exactly 3 times + - \c a{3,} At least 3 times + - \c a{3,5} 3 to 5 times + - \c (ab) Grouping + - \c ^a At the beginning + - \c a$ At the end + - \c . Any character + - \c [abc] Character classes + - \c [a-c] Character class range + - \c [a-z0-9_] Character class combination + - \c [^abc] Negated character classes + - \c [^a-c] Negated character class range + - \c [\b] Backspace (U+0008) + - \c \\| \\\\ ... Escape characters + - \c \\f Form feed (U+000C) + - \c \\n Line feed (U+000A) + - \c \\r Carriage return (U+000D) + - \c \\t Tab (U+0009) + - \c \\v Vertical tab (U+000B) + + \note This is a Thompson NFA engine, implemented with reference to + Cox, Russ. "Regular Expression Matching Can Be Simple And Fast (but is slow in Java, Perl, PHP, Python, Ruby,...).", + https://swtch.com/~rsc/regexp/regexp1.html +*/ +template +class GenericRegex { +public: + typedef Encoding EncodingType; + typedef typename Encoding::Ch Ch; + template friend class GenericRegexSearch; + + GenericRegex(const Ch* source, Allocator* allocator = 0) : + states_(allocator, 256), ranges_(allocator, 256), root_(kRegexInvalidState), stateCount_(), rangeCount_(), + anchorBegin_(), anchorEnd_() + { + GenericStringStream ss(source); + DecodedStream, Encoding> ds(ss); + Parse(ds); + } + + ~GenericRegex() {} + + bool IsValid() const { + return root_ != kRegexInvalidState; + } + +private: + enum Operator { + kZeroOrOne, + kZeroOrMore, + kOneOrMore, + kConcatenation, + kAlternation, + kLeftParenthesis + }; + + static const unsigned kAnyCharacterClass = 0xFFFFFFFF; //!< For '.' + static const unsigned kRangeCharacterClass = 0xFFFFFFFE; + static const unsigned kRangeNegationFlag = 0x80000000; + + struct Range { + unsigned start; // + unsigned end; + SizeType next; + }; + + struct State { + SizeType out; //!< Equals to kInvalid for matching state + SizeType out1; //!< Equals to non-kInvalid for split + SizeType rangeStart; + unsigned codepoint; + }; + + struct Frag { + Frag(SizeType s, SizeType o, SizeType m) : start(s), out(o), minIndex(m) {} + SizeType start; + SizeType out; //!< link-list of all output states + SizeType minIndex; + }; + + State& GetState(SizeType index) { + RAPIDJSON_ASSERT(index < stateCount_); + return states_.template Bottom()[index]; + } + + const State& GetState(SizeType index) const { + RAPIDJSON_ASSERT(index < stateCount_); + return states_.template Bottom()[index]; + } + + Range& GetRange(SizeType index) { + RAPIDJSON_ASSERT(index < rangeCount_); + return ranges_.template Bottom()[index]; + } + + const Range& GetRange(SizeType index) const { + RAPIDJSON_ASSERT(index < rangeCount_); + return ranges_.template Bottom()[index]; + } + + template + void Parse(DecodedStream& ds) { + Allocator allocator; + Stack operandStack(&allocator, 256); // Frag + Stack operatorStack(&allocator, 256); // Operator + Stack atomCountStack(&allocator, 256); // unsigned (Atom per parenthesis) + + *atomCountStack.template Push() = 0; + + unsigned codepoint; + while (ds.Peek() != 0) { + switch (codepoint = ds.Take()) { + case '^': + anchorBegin_ = true; + break; + + case '$': + anchorEnd_ = true; + break; + + case '|': + while (!operatorStack.Empty() && *operatorStack.template Top() < kAlternation) + if (!Eval(operandStack, *operatorStack.template Pop(1))) + return; + *operatorStack.template Push() = kAlternation; + *atomCountStack.template Top() = 0; + break; + + case '(': + *operatorStack.template Push() = kLeftParenthesis; + *atomCountStack.template Push() = 0; + break; + + case ')': + while (!operatorStack.Empty() && *operatorStack.template Top() != kLeftParenthesis) + if (!Eval(operandStack, *operatorStack.template Pop(1))) + return; + if (operatorStack.Empty()) + return; + operatorStack.template Pop(1); + atomCountStack.template Pop(1); + ImplicitConcatenation(atomCountStack, operatorStack); + break; + + case '?': + if (!Eval(operandStack, kZeroOrOne)) + return; + break; + + case '*': + if (!Eval(operandStack, kZeroOrMore)) + return; + break; + + case '+': + if (!Eval(operandStack, kOneOrMore)) + return; + break; + + case '{': + { + unsigned n, m; + if (!ParseUnsigned(ds, &n)) + return; + + if (ds.Peek() == ',') { + ds.Take(); + if (ds.Peek() == '}') + m = kInfinityQuantifier; + else if (!ParseUnsigned(ds, &m) || m < n) + return; + } + else + m = n; + + if (!EvalQuantifier(operandStack, n, m) || ds.Peek() != '}') + return; + ds.Take(); + } + break; + + case '.': + PushOperand(operandStack, kAnyCharacterClass); + ImplicitConcatenation(atomCountStack, operatorStack); + break; + + case '[': + { + SizeType range; + if (!ParseRange(ds, &range)) + return; + SizeType s = NewState(kRegexInvalidState, kRegexInvalidState, kRangeCharacterClass); + GetState(s).rangeStart = range; + *operandStack.template Push() = Frag(s, s, s); + } + ImplicitConcatenation(atomCountStack, operatorStack); + break; + + case '\\': // Escape character + if (!CharacterEscape(ds, &codepoint)) + return; // Unsupported escape character + // fall through to default + + default: // Pattern character + PushOperand(operandStack, codepoint); + ImplicitConcatenation(atomCountStack, operatorStack); + } + } + + while (!operatorStack.Empty()) + if (!Eval(operandStack, *operatorStack.template Pop(1))) + return; + + // Link the operand to matching state. + if (operandStack.GetSize() == sizeof(Frag)) { + Frag* e = operandStack.template Pop(1); + Patch(e->out, NewState(kRegexInvalidState, kRegexInvalidState, 0)); + root_ = e->start; + +#if RAPIDJSON_REGEX_VERBOSE + printf("root: %d\n", root_); + for (SizeType i = 0; i < stateCount_ ; i++) { + State& s = GetState(i); + printf("[%2d] out: %2d out1: %2d c: '%c'\n", i, s.out, s.out1, (char)s.codepoint); + } + printf("\n"); +#endif + } + } + + SizeType NewState(SizeType out, SizeType out1, unsigned codepoint) { + State* s = states_.template Push(); + s->out = out; + s->out1 = out1; + s->codepoint = codepoint; + s->rangeStart = kRegexInvalidRange; + return stateCount_++; + } + + void PushOperand(Stack& operandStack, unsigned codepoint) { + SizeType s = NewState(kRegexInvalidState, kRegexInvalidState, codepoint); + *operandStack.template Push() = Frag(s, s, s); + } + + void ImplicitConcatenation(Stack& atomCountStack, Stack& operatorStack) { + if (*atomCountStack.template Top()) + *operatorStack.template Push() = kConcatenation; + (*atomCountStack.template Top())++; + } + + SizeType Append(SizeType l1, SizeType l2) { + SizeType old = l1; + while (GetState(l1).out != kRegexInvalidState) + l1 = GetState(l1).out; + GetState(l1).out = l2; + return old; + } + + void Patch(SizeType l, SizeType s) { + for (SizeType next; l != kRegexInvalidState; l = next) { + next = GetState(l).out; + GetState(l).out = s; + } + } + + bool Eval(Stack& operandStack, Operator op) { + switch (op) { + case kConcatenation: + RAPIDJSON_ASSERT(operandStack.GetSize() >= sizeof(Frag) * 2); + { + Frag e2 = *operandStack.template Pop(1); + Frag e1 = *operandStack.template Pop(1); + Patch(e1.out, e2.start); + *operandStack.template Push() = Frag(e1.start, e2.out, Min(e1.minIndex, e2.minIndex)); + } + return true; + + case kAlternation: + if (operandStack.GetSize() >= sizeof(Frag) * 2) { + Frag e2 = *operandStack.template Pop(1); + Frag e1 = *operandStack.template Pop(1); + SizeType s = NewState(e1.start, e2.start, 0); + *operandStack.template Push() = Frag(s, Append(e1.out, e2.out), Min(e1.minIndex, e2.minIndex)); + return true; + } + return false; + + case kZeroOrOne: + if (operandStack.GetSize() >= sizeof(Frag)) { + Frag e = *operandStack.template Pop(1); + SizeType s = NewState(kRegexInvalidState, e.start, 0); + *operandStack.template Push() = Frag(s, Append(e.out, s), e.minIndex); + return true; + } + return false; + + case kZeroOrMore: + if (operandStack.GetSize() >= sizeof(Frag)) { + Frag e = *operandStack.template Pop(1); + SizeType s = NewState(kRegexInvalidState, e.start, 0); + Patch(e.out, s); + *operandStack.template Push() = Frag(s, s, e.minIndex); + return true; + } + return false; + + default: + RAPIDJSON_ASSERT(op == kOneOrMore); + if (operandStack.GetSize() >= sizeof(Frag)) { + Frag e = *operandStack.template Pop(1); + SizeType s = NewState(kRegexInvalidState, e.start, 0); + Patch(e.out, s); + *operandStack.template Push() = Frag(e.start, s, e.minIndex); + return true; + } + return false; + } + } + + bool EvalQuantifier(Stack& operandStack, unsigned n, unsigned m) { + RAPIDJSON_ASSERT(n <= m); + RAPIDJSON_ASSERT(operandStack.GetSize() >= sizeof(Frag)); + + if (n == 0) { + if (m == 0) // a{0} not support + return false; + else if (m == kInfinityQuantifier) + Eval(operandStack, kZeroOrMore); // a{0,} -> a* + else { + Eval(operandStack, kZeroOrOne); // a{0,5} -> a? + for (unsigned i = 0; i < m - 1; i++) + CloneTopOperand(operandStack); // a{0,5} -> a? a? a? a? a? + for (unsigned i = 0; i < m - 1; i++) + Eval(operandStack, kConcatenation); // a{0,5} -> a?a?a?a?a? + } + return true; + } + + for (unsigned i = 0; i < n - 1; i++) // a{3} -> a a a + CloneTopOperand(operandStack); + + if (m == kInfinityQuantifier) + Eval(operandStack, kOneOrMore); // a{3,} -> a a a+ + else if (m > n) { + CloneTopOperand(operandStack); // a{3,5} -> a a a a + Eval(operandStack, kZeroOrOne); // a{3,5} -> a a a a? + for (unsigned i = n; i < m - 1; i++) + CloneTopOperand(operandStack); // a{3,5} -> a a a a? a? + for (unsigned i = n; i < m; i++) + Eval(operandStack, kConcatenation); // a{3,5} -> a a aa?a? + } + + for (unsigned i = 0; i < n - 1; i++) + Eval(operandStack, kConcatenation); // a{3} -> aaa, a{3,} -> aaa+, a{3.5} -> aaaa?a? + + return true; + } + + static SizeType Min(SizeType a, SizeType b) { return a < b ? a : b; } + + void CloneTopOperand(Stack& operandStack) { + const Frag src = *operandStack.template Top(); // Copy constructor to prevent invalidation + SizeType count = stateCount_ - src.minIndex; // Assumes top operand contains states in [src->minIndex, stateCount_) + State* s = states_.template Push(count); + memcpy(s, &GetState(src.minIndex), count * sizeof(State)); + for (SizeType j = 0; j < count; j++) { + if (s[j].out != kRegexInvalidState) + s[j].out += count; + if (s[j].out1 != kRegexInvalidState) + s[j].out1 += count; + } + *operandStack.template Push() = Frag(src.start + count, src.out + count, src.minIndex + count); + stateCount_ += count; + } + + template + bool ParseUnsigned(DecodedStream& ds, unsigned* u) { + unsigned r = 0; + if (ds.Peek() < '0' || ds.Peek() > '9') + return false; + while (ds.Peek() >= '0' && ds.Peek() <= '9') { + if (r >= 429496729 && ds.Peek() > '5') // 2^32 - 1 = 4294967295 + return false; // overflow + r = r * 10 + (ds.Take() - '0'); + } + *u = r; + return true; + } + + template + bool ParseRange(DecodedStream& ds, SizeType* range) { + bool isBegin = true; + bool negate = false; + int step = 0; + SizeType start = kRegexInvalidRange; + SizeType current = kRegexInvalidRange; + unsigned codepoint; + while ((codepoint = ds.Take()) != 0) { + if (isBegin) { + isBegin = false; + if (codepoint == '^') { + negate = true; + continue; + } + } + + switch (codepoint) { + case ']': + if (start == kRegexInvalidRange) + return false; // Error: nothing inside [] + if (step == 2) { // Add trailing '-' + SizeType r = NewRange('-'); + RAPIDJSON_ASSERT(current != kRegexInvalidRange); + GetRange(current).next = r; + } + if (negate) + GetRange(start).start |= kRangeNegationFlag; + *range = start; + return true; + + case '\\': + if (ds.Peek() == 'b') { + ds.Take(); + codepoint = 0x0008; // Escape backspace character + } + else if (!CharacterEscape(ds, &codepoint)) + return false; + // fall through to default + + default: + switch (step) { + case 1: + if (codepoint == '-') { + step++; + break; + } + // fall through to step 0 for other characters + + case 0: + { + SizeType r = NewRange(codepoint); + if (current != kRegexInvalidRange) + GetRange(current).next = r; + if (start == kRegexInvalidRange) + start = r; + current = r; + } + step = 1; + break; + + default: + RAPIDJSON_ASSERT(step == 2); + GetRange(current).end = codepoint; + step = 0; + } + } + } + return false; + } + + SizeType NewRange(unsigned codepoint) { + Range* r = ranges_.template Push(); + r->start = r->end = codepoint; + r->next = kRegexInvalidRange; + return rangeCount_++; + } + + template + bool CharacterEscape(DecodedStream& ds, unsigned* escapedCodepoint) { + unsigned codepoint; + switch (codepoint = ds.Take()) { + case '^': + case '$': + case '|': + case '(': + case ')': + case '?': + case '*': + case '+': + case '.': + case '[': + case ']': + case '{': + case '}': + case '\\': + *escapedCodepoint = codepoint; return true; + case 'f': *escapedCodepoint = 0x000C; return true; + case 'n': *escapedCodepoint = 0x000A; return true; + case 'r': *escapedCodepoint = 0x000D; return true; + case 't': *escapedCodepoint = 0x0009; return true; + case 'v': *escapedCodepoint = 0x000B; return true; + default: + return false; // Unsupported escape character + } + } + + Stack states_; + Stack ranges_; + SizeType root_; + SizeType stateCount_; + SizeType rangeCount_; + + static const unsigned kInfinityQuantifier = ~0u; + + // For SearchWithAnchoring() + bool anchorBegin_; + bool anchorEnd_; +}; + +template +class GenericRegexSearch { +public: + typedef typename RegexType::EncodingType Encoding; + typedef typename Encoding::Ch Ch; + + GenericRegexSearch(const RegexType& regex, Allocator* allocator = 0) : + regex_(regex), allocator_(allocator), ownAllocator_(0), + state0_(allocator, 0), state1_(allocator, 0), stateSet_() + { + RAPIDJSON_ASSERT(regex_.IsValid()); + if (!allocator_) + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + stateSet_ = static_cast(allocator_->Malloc(GetStateSetSize())); + state0_.template Reserve(regex_.stateCount_); + state1_.template Reserve(regex_.stateCount_); + } + + ~GenericRegexSearch() { + Allocator::Free(stateSet_); + RAPIDJSON_DELETE(ownAllocator_); + } + + template + bool Match(InputStream& is) { + return SearchWithAnchoring(is, true, true); + } + + bool Match(const Ch* s) { + GenericStringStream is(s); + return Match(is); + } + + template + bool Search(InputStream& is) { + return SearchWithAnchoring(is, regex_.anchorBegin_, regex_.anchorEnd_); + } + + bool Search(const Ch* s) { + GenericStringStream is(s); + return Search(is); + } + +private: + typedef typename RegexType::State State; + typedef typename RegexType::Range Range; + + template + bool SearchWithAnchoring(InputStream& is, bool anchorBegin, bool anchorEnd) { + DecodedStream ds(is); + + state0_.Clear(); + Stack *current = &state0_, *next = &state1_; + const size_t stateSetSize = GetStateSetSize(); + std::memset(stateSet_, 0, stateSetSize); + + bool matched = AddState(*current, regex_.root_); + unsigned codepoint; + while (!current->Empty() && (codepoint = ds.Take()) != 0) { + std::memset(stateSet_, 0, stateSetSize); + next->Clear(); + matched = false; + for (const SizeType* s = current->template Bottom(); s != current->template End(); ++s) { + const State& sr = regex_.GetState(*s); + if (sr.codepoint == codepoint || + sr.codepoint == RegexType::kAnyCharacterClass || + (sr.codepoint == RegexType::kRangeCharacterClass && MatchRange(sr.rangeStart, codepoint))) + { + matched = AddState(*next, sr.out) || matched; + if (!anchorEnd && matched) + return true; + } + if (!anchorBegin) + AddState(*next, regex_.root_); + } + internal::Swap(current, next); + } + + return matched; + } + + size_t GetStateSetSize() const { + return (regex_.stateCount_ + 31) / 32 * 4; + } + + // Return whether the added states is a match state + bool AddState(Stack& l, SizeType index) { + RAPIDJSON_ASSERT(index != kRegexInvalidState); + + const State& s = regex_.GetState(index); + if (s.out1 != kRegexInvalidState) { // Split + bool matched = AddState(l, s.out); + return AddState(l, s.out1) || matched; + } + else if (!(stateSet_[index >> 5] & (1u << (index & 31)))) { + stateSet_[index >> 5] |= (1u << (index & 31)); + *l.template PushUnsafe() = index; + } + return s.out == kRegexInvalidState; // by using PushUnsafe() above, we can ensure s is not validated due to reallocation. + } + + bool MatchRange(SizeType rangeIndex, unsigned codepoint) const { + bool yes = (regex_.GetRange(rangeIndex).start & RegexType::kRangeNegationFlag) == 0; + while (rangeIndex != kRegexInvalidRange) { + const Range& r = regex_.GetRange(rangeIndex); + if (codepoint >= (r.start & ~RegexType::kRangeNegationFlag) && codepoint <= r.end) + return yes; + rangeIndex = r.next; + } + return !yes; + } + + const RegexType& regex_; + Allocator* allocator_; + Allocator* ownAllocator_; + Stack state0_; + Stack state1_; + uint32_t* stateSet_; +}; + +typedef GenericRegex > Regex; +typedef GenericRegexSearch RegexSearch; + +} // namespace internal +RAPIDJSON_NAMESPACE_END + +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + +#ifdef _MSC_VER +RAPIDJSON_DIAG_POP +#endif + +#endif // RAPIDJSON_INTERNAL_REGEX_H_ diff --git a/rapidjson/internal/stack.h b/rapidjson/internal/stack.h index 722d569..5c5398c 100644 --- a/rapidjson/internal/stack.h +++ b/rapidjson/internal/stack.h @@ -15,7 +15,13 @@ #ifndef RAPIDJSON_INTERNAL_STACK_H_ #define RAPIDJSON_INTERNAL_STACK_H_ -#include "../rapidjson.h" +#include "../allocators.h" +#include "swap.h" + +#if defined(__clang__) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif RAPIDJSON_NAMESPACE_BEGIN namespace internal { @@ -32,7 +38,6 @@ class Stack { // Optimization note: Do not allocate memory for stack_ in constructor. // Do it lazily when first Push() -> Expand() -> Resize(). Stack(Allocator* allocator, size_t stackCapacity) : allocator_(allocator), ownAllocator_(0), stack_(0), stackTop_(0), stackEnd_(0), initialCapacity_(stackCapacity) { - RAPIDJSON_ASSERT(stackCapacity > 0); } #if RAPIDJSON_HAS_CXX11_RVALUE_REFS @@ -81,6 +86,15 @@ class Stack { } #endif + void Swap(Stack& rhs) RAPIDJSON_NOEXCEPT { + internal::Swap(allocator_, rhs.allocator_); + internal::Swap(ownAllocator_, rhs.ownAllocator_); + internal::Swap(stack_, rhs.stack_); + internal::Swap(stackTop_, rhs.stackTop_); + internal::Swap(stackEnd_, rhs.stackEnd_); + internal::Swap(initialCapacity_, rhs.initialCapacity_); + } + void Clear() { stackTop_ = stack_; } void ShrinkToFit() { @@ -98,11 +112,22 @@ class Stack { // Optimization note: try to minimize the size of this function for force inline. // Expansion is run very infrequently, so it is moved to another (probably non-inline) function. template - RAPIDJSON_FORCEINLINE T* Push(size_t count = 1) { + RAPIDJSON_FORCEINLINE void Reserve(size_t count = 1) { // Expand the stack if needed - if (stackTop_ + sizeof(T) * count >= stackEnd_) + if (RAPIDJSON_UNLIKELY(stackTop_ + sizeof(T) * count > stackEnd_)) Expand(count); + } + + template + RAPIDJSON_FORCEINLINE T* Push(size_t count = 1) { + Reserve(count); + return PushUnsafe(count); + } + template + RAPIDJSON_FORCEINLINE T* PushUnsafe(size_t count = 1) { + RAPIDJSON_ASSERT(stackTop_); + RAPIDJSON_ASSERT(stackTop_ + sizeof(T) * count <= stackEnd_); T* ret = reinterpret_cast(stackTop_); stackTop_ += sizeof(T) * count; return ret; @@ -122,9 +147,32 @@ class Stack { } template - T* Bottom() { return (T*)stack_; } + const T* Top() const { + RAPIDJSON_ASSERT(GetSize() >= sizeof(T)); + return reinterpret_cast(stackTop_ - sizeof(T)); + } + + template + T* End() { return reinterpret_cast(stackTop_); } + + template + const T* End() const { return reinterpret_cast(stackTop_); } + + template + T* Bottom() { return reinterpret_cast(stack_); } + + template + const T* Bottom() const { return reinterpret_cast(stack_); } + + bool HasAllocator() const { + return allocator_ != 0; + } + + Allocator& GetAllocator() { + RAPIDJSON_ASSERT(allocator_); + return *allocator_; + } - Allocator& GetAllocator() { return *allocator_; } bool Empty() const { return stackTop_ == stack_; } size_t GetSize() const { return static_cast(stackTop_ - stack_); } size_t GetCapacity() const { return static_cast(stackEnd_ - stack_); } @@ -136,7 +184,7 @@ class Stack { size_t newCapacity; if (stack_ == 0) { if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); newCapacity = initialCapacity_; } else { newCapacity = GetCapacity(); @@ -151,7 +199,7 @@ class Stack { void Resize(size_t newCapacity) { const size_t size = GetSize(); // Backup the current size - stack_ = (char*)allocator_->Realloc(stack_, GetCapacity(), newCapacity); + stack_ = static_cast(allocator_->Realloc(stack_, GetCapacity(), newCapacity)); stackTop_ = stack_ + size; stackEnd_ = stack_ + newCapacity; } @@ -176,4 +224,8 @@ class Stack { } // namespace internal RAPIDJSON_NAMESPACE_END +#if defined(__clang__) +RAPIDJSON_DIAG_POP +#endif + #endif // RAPIDJSON_STACK_H_ diff --git a/rapidjson/internal/strfunc.h b/rapidjson/internal/strfunc.h index 8440506..226439a 100644 --- a/rapidjson/internal/strfunc.h +++ b/rapidjson/internal/strfunc.h @@ -15,7 +15,8 @@ #ifndef RAPIDJSON_INTERNAL_STRFUNC_H_ #define RAPIDJSON_INTERNAL_STRFUNC_H_ -#include "../rapidjson.h" +#include "../stream.h" +#include RAPIDJSON_NAMESPACE_BEGIN namespace internal { @@ -28,11 +29,40 @@ namespace internal { */ template inline SizeType StrLen(const Ch* s) { + RAPIDJSON_ASSERT(s != 0); const Ch* p = s; while (*p) ++p; return SizeType(p - s); } +template <> +inline SizeType StrLen(const char* s) { + return SizeType(std::strlen(s)); +} + +template <> +inline SizeType StrLen(const wchar_t* s) { + return SizeType(std::wcslen(s)); +} + +//! Returns number of code points in a encoded string. +template +bool CountStringCodePoint(const typename Encoding::Ch* s, SizeType length, SizeType* outCount) { + RAPIDJSON_ASSERT(s != 0); + RAPIDJSON_ASSERT(outCount != 0); + GenericStringStream is(s); + const typename Encoding::Ch* end = s + length; + SizeType count = 0; + while (is.src_ < end) { + unsigned codepoint; + if (!Encoding::Decode(is, &codepoint)) + return false; + count++; + } + *outCount = count; + return true; +} + } // namespace internal RAPIDJSON_NAMESPACE_END diff --git a/rapidjson/internal/strtod.h b/rapidjson/internal/strtod.h index bb2e3cd..adf49e3 100644 --- a/rapidjson/internal/strtod.h +++ b/rapidjson/internal/strtod.h @@ -15,7 +15,6 @@ #ifndef RAPIDJSON_STRTOD_ #define RAPIDJSON_STRTOD_ -#include "../rapidjson.h" #include "ieee754.h" #include "biginteger.h" #include "diyfp.h" @@ -95,13 +94,13 @@ inline int CheckWithinHalfULP(double b, const BigInteger& d, int dExp) { hS_Exp2 -= common_Exp2; BigInteger dS = d; - dS.MultiplyPow5(dS_Exp5) <<= dS_Exp2; + dS.MultiplyPow5(static_cast(dS_Exp5)) <<= static_cast(dS_Exp2); BigInteger bS(bInt); - bS.MultiplyPow5(bS_Exp5) <<= bS_Exp2; + bS.MultiplyPow5(static_cast(bS_Exp5)) <<= static_cast(bS_Exp2); BigInteger hS(1); - hS.MultiplyPow5(hS_Exp5) <<= hS_Exp2; + hS.MultiplyPow5(static_cast(hS_Exp5)) <<= static_cast(hS_Exp2); BigInteger delta(0); dS.Difference(bS, &delta); @@ -134,22 +133,22 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit if (significand > RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) || (significand == RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) && decimals[i] > '5')) break; - significand = significand * 10 + (decimals[i] - '0'); + significand = significand * 10u + static_cast(decimals[i] - '0'); } if (i < length && decimals[i] >= '5') // Rounding significand++; size_t remaining = length - i; - const unsigned kUlpShift = 3; - const unsigned kUlp = 1 << kUlpShift; - int error = (remaining == 0) ? 0 : kUlp / 2; + const int kUlpShift = 3; + const int kUlp = 1 << kUlpShift; + int64_t error = (remaining == 0) ? 0 : kUlp / 2; DiyFp v(significand, 0); v = v.Normalize(); error <<= -v.e; - const int dExp = (int)decimalPosition - (int)i + exp; + const int dExp = static_cast(decimalPosition) - static_cast(i) + exp; int actualExp; DiyFp cachedPower = GetCachedPower10(dExp, &actualExp); @@ -163,10 +162,10 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit DiyFp(RAPIDJSON_UINT64_C2(0xf4240000, 00000000), -44), // 10^6 DiyFp(RAPIDJSON_UINT64_C2(0x98968000, 00000000), -40) // 10^7 }; - int adjustment = dExp - actualExp - 1; + int adjustment = dExp - actualExp - 1; RAPIDJSON_ASSERT(adjustment >= 0 && adjustment < 7); v = v * kPow10[adjustment]; - if (length + adjustment > 19) // has more digits than decimal digits in 64-bit + if (length + static_cast(adjustment)> 19u) // has more digits than decimal digits in 64-bit error += kUlp / 2; } @@ -178,10 +177,10 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit v = v.Normalize(); error <<= oldExp - v.e; - const unsigned effectiveSignificandSize = Double::EffectiveSignificandSize(64 + v.e); - unsigned precisionSize = 64 - effectiveSignificandSize; + const int effectiveSignificandSize = Double::EffectiveSignificandSize(64 + v.e); + int precisionSize = 64 - effectiveSignificandSize; if (precisionSize + kUlpShift >= 64) { - unsigned scaleExp = (precisionSize + kUlpShift) - 63; + int scaleExp = (precisionSize + kUlpShift) - 63; v.f >>= scaleExp; v.e += scaleExp; error = (error >> scaleExp) + 1 + kUlp; @@ -191,7 +190,7 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit DiyFp rounded(v.f >> precisionSize, v.e + precisionSize); const uint64_t precisionBits = (v.f & ((uint64_t(1) << precisionSize) - 1)) * kUlp; const uint64_t halfWay = (uint64_t(1) << (precisionSize - 1)) * kUlp; - if (precisionBits >= halfWay + error) { + if (precisionBits >= halfWay + static_cast(error)) { rounded.f++; if (rounded.f & (DiyFp::kDpHiddenBit << 1)) { // rounding overflows mantissa (issue #340) rounded.f >>= 1; @@ -201,12 +200,12 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit *result = rounded.ToDouble(); - return halfWay - error >= precisionBits || precisionBits >= halfWay + error; + return halfWay - static_cast(error) >= precisionBits || precisionBits >= halfWay + static_cast(error); } inline double StrtodBigInteger(double approx, const char* decimals, size_t length, size_t decimalPosition, int exp) { const BigInteger dInt(decimals, length); - const int dExp = (int)decimalPosition - (int)length + exp; + const int dExp = static_cast(decimalPosition) - static_cast(length) + exp; Double a(approx); int cmp = CheckWithinHalfULP(a.Value(), dInt, dExp); if (cmp < 0) @@ -246,10 +245,10 @@ inline double StrtodFullPrecision(double d, int p, const char* decimals, size_t // Trim right-most digits const int kMaxDecimalDigit = 780; - if ((int)length > kMaxDecimalDigit) { - int delta = (int(length) - kMaxDecimalDigit); + if (static_cast(length) > kMaxDecimalDigit) { + int delta = (static_cast(length) - kMaxDecimalDigit); exp += delta; - decimalPosition -= (size_t)delta; + decimalPosition -= static_cast(delta); length = kMaxDecimalDigit; } diff --git a/rapidjson/internal/swap.h b/rapidjson/internal/swap.h new file mode 100644 index 0000000..666e49f --- /dev/null +++ b/rapidjson/internal/swap.h @@ -0,0 +1,46 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_INTERNAL_SWAP_H_ +#define RAPIDJSON_INTERNAL_SWAP_H_ + +#include "../rapidjson.h" + +#if defined(__clang__) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif + +RAPIDJSON_NAMESPACE_BEGIN +namespace internal { + +//! Custom swap() to avoid dependency on C++ header +/*! \tparam T Type of the arguments to swap, should be instantiated with primitive C++ types only. + \note This has the same semantics as std::swap(). +*/ +template +inline void Swap(T& a, T& b) RAPIDJSON_NOEXCEPT { + T tmp = a; + a = b; + b = tmp; +} + +} // namespace internal +RAPIDJSON_NAMESPACE_END + +#if defined(__clang__) +RAPIDJSON_DIAG_POP +#endif + +#endif // RAPIDJSON_INTERNAL_SWAP_H_ diff --git a/rapidjson/istreamwrapper.h b/rapidjson/istreamwrapper.h new file mode 100644 index 0000000..8639c8c --- /dev/null +++ b/rapidjson/istreamwrapper.h @@ -0,0 +1,115 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_ISTREAMWRAPPER_H_ +#define RAPIDJSON_ISTREAMWRAPPER_H_ + +#include "stream.h" +#include + +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +#endif + +#ifdef _MSC_VER +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4351) // new behavior: elements of array 'array' will be default initialized +#endif + +RAPIDJSON_NAMESPACE_BEGIN + +//! Wrapper of \c std::basic_istream into RapidJSON's Stream concept. +/*! + The classes can be wrapped including but not limited to: + + - \c std::istringstream + - \c std::stringstream + - \c std::wistringstream + - \c std::wstringstream + - \c std::ifstream + - \c std::fstream + - \c std::wifstream + - \c std::wfstream + + \tparam StreamType Class derived from \c std::basic_istream. +*/ + +template +class BasicIStreamWrapper { +public: + typedef typename StreamType::char_type Ch; + BasicIStreamWrapper(StreamType& stream) : stream_(stream), count_(), peekBuffer_() {} + + Ch Peek() const { + typename StreamType::int_type c = stream_.peek(); + return RAPIDJSON_LIKELY(c != StreamType::traits_type::eof()) ? static_cast(c) : static_cast('\0'); + } + + Ch Take() { + typename StreamType::int_type c = stream_.get(); + if (RAPIDJSON_LIKELY(c != StreamType::traits_type::eof())) { + count_++; + return static_cast(c); + } + else + return '\0'; + } + + // tellg() may return -1 when failed. So we count by ourself. + size_t Tell() const { return count_; } + + Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } + void Put(Ch) { RAPIDJSON_ASSERT(false); } + void Flush() { RAPIDJSON_ASSERT(false); } + size_t PutEnd(Ch*) { RAPIDJSON_ASSERT(false); return 0; } + + // For encoding detection only. + const Ch* Peek4() const { + RAPIDJSON_ASSERT(sizeof(Ch) == 1); // Only usable for byte stream. + int i; + bool hasError = false; + for (i = 0; i < 4; ++i) { + typename StreamType::int_type c = stream_.get(); + if (c == StreamType::traits_type::eof()) { + hasError = true; + stream_.clear(); + break; + } + peekBuffer_[i] = static_cast(c); + } + for (--i; i >= 0; --i) + stream_.putback(peekBuffer_[i]); + return !hasError ? peekBuffer_ : 0; + } + +private: + BasicIStreamWrapper(const BasicIStreamWrapper&); + BasicIStreamWrapper& operator=(const BasicIStreamWrapper&); + + StreamType& stream_; + size_t count_; //!< Number of characters read. Note: + mutable Ch peekBuffer_[4]; +}; + +typedef BasicIStreamWrapper IStreamWrapper; +typedef BasicIStreamWrapper WIStreamWrapper; + +#if defined(__clang__) || defined(_MSC_VER) +RAPIDJSON_DIAG_POP +#endif + +RAPIDJSON_NAMESPACE_END + +#endif // RAPIDJSON_ISTREAMWRAPPER_H_ diff --git a/rapidjson/memorybuffer.h b/rapidjson/memorybuffer.h index 2484b21..39bee1d 100644 --- a/rapidjson/memorybuffer.h +++ b/rapidjson/memorybuffer.h @@ -15,7 +15,7 @@ #ifndef RAPIDJSON_MEMORYBUFFER_H_ #define RAPIDJSON_MEMORYBUFFER_H_ -#include "rapidjson.h" +#include "stream.h" #include "internal/stack.h" RAPIDJSON_NAMESPACE_BEGIN diff --git a/rapidjson/memorystream.h b/rapidjson/memorystream.h index 99feae5..1d71d8a 100644 --- a/rapidjson/memorystream.h +++ b/rapidjson/memorystream.h @@ -15,7 +15,13 @@ #ifndef RAPIDJSON_MEMORYSTREAM_H_ #define RAPIDJSON_MEMORYSTREAM_H_ -#include "rapidjson.h" +#include "stream.h" + +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(unreachable-code) +RAPIDJSON_DIAG_OFF(missing-noreturn) +#endif RAPIDJSON_NAMESPACE_BEGIN @@ -36,8 +42,8 @@ struct MemoryStream { MemoryStream(const Ch *src, size_t size) : src_(src), begin_(src), end_(src + size), size_(size) {} - Ch Peek() const { return (src_ == end_) ? '\0' : *src_; } - Ch Take() { return (src_ == end_) ? '\0' : *src_++; } + Ch Peek() const { return RAPIDJSON_UNLIKELY(src_ == end_) ? '\0' : *src_; } + Ch Take() { return RAPIDJSON_UNLIKELY(src_ == end_) ? '\0' : *src_++; } size_t Tell() const { return static_cast(src_ - begin_); } Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } @@ -58,4 +64,8 @@ struct MemoryStream { RAPIDJSON_NAMESPACE_END +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + #endif // RAPIDJSON_MEMORYBUFFER_H_ diff --git a/rapidjson/msinttypes/stdint.h b/rapidjson/msinttypes/stdint.h index a26fff4..3d4477b 100644 --- a/rapidjson/msinttypes/stdint.h +++ b/rapidjson/msinttypes/stdint.h @@ -89,14 +89,14 @@ #include // For Visual Studio 6 in C++ mode and for many Visual Studio versions when -// compiling for ARM we should wrap include with 'extern "C++" {}' -// or compiler give many errors like this: +// compiling for ARM we have to wrap include with 'extern "C++" {}' +// or compiler would give many errors like this: // error C2733: second C linkage of overloaded function 'wmemchr' not allowed -#ifdef __cplusplus +#if defined(__cplusplus) && !defined(_M_ARM) extern "C" { #endif # include -#ifdef __cplusplus +#if defined(__cplusplus) && !defined(_M_ARM) } #endif diff --git a/rapidjson/ostreamwrapper.h b/rapidjson/ostreamwrapper.h new file mode 100644 index 0000000..6f4667c --- /dev/null +++ b/rapidjson/ostreamwrapper.h @@ -0,0 +1,81 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_OSTREAMWRAPPER_H_ +#define RAPIDJSON_OSTREAMWRAPPER_H_ + +#include "stream.h" +#include + +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +#endif + +RAPIDJSON_NAMESPACE_BEGIN + +//! Wrapper of \c std::basic_ostream into RapidJSON's Stream concept. +/*! + The classes can be wrapped including but not limited to: + + - \c std::ostringstream + - \c std::stringstream + - \c std::wpstringstream + - \c std::wstringstream + - \c std::ifstream + - \c std::fstream + - \c std::wofstream + - \c std::wfstream + + \tparam StreamType Class derived from \c std::basic_ostream. +*/ + +template +class BasicOStreamWrapper { +public: + typedef typename StreamType::char_type Ch; + BasicOStreamWrapper(StreamType& stream) : stream_(stream) {} + + void Put(Ch c) { + stream_.put(c); + } + + void Flush() { + stream_.flush(); + } + + // Not implemented + char Peek() const { RAPIDJSON_ASSERT(false); return 0; } + char Take() { RAPIDJSON_ASSERT(false); return 0; } + size_t Tell() const { RAPIDJSON_ASSERT(false); return 0; } + char* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } + size_t PutEnd(char*) { RAPIDJSON_ASSERT(false); return 0; } + +private: + BasicOStreamWrapper(const BasicOStreamWrapper&); + BasicOStreamWrapper& operator=(const BasicOStreamWrapper&); + + StreamType& stream_; +}; + +typedef BasicOStreamWrapper OStreamWrapper; +typedef BasicOStreamWrapper WOStreamWrapper; + +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + +RAPIDJSON_NAMESPACE_END + +#endif // RAPIDJSON_OSTREAMWRAPPER_H_ diff --git a/rapidjson/pointer.h b/rapidjson/pointer.h new file mode 100644 index 0000000..0f377ef --- /dev/null +++ b/rapidjson/pointer.h @@ -0,0 +1,1358 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_POINTER_H_ +#define RAPIDJSON_POINTER_H_ + +#include "document.h" +#include "internal/itoa.h" + +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(switch-enum) +#endif + +#ifdef _MSC_VER +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +RAPIDJSON_NAMESPACE_BEGIN + +static const SizeType kPointerInvalidIndex = ~SizeType(0); //!< Represents an invalid index in GenericPointer::Token + +//! Error code of parsing. +/*! \ingroup RAPIDJSON_ERRORS + \see GenericPointer::GenericPointer, GenericPointer::GetParseErrorCode +*/ +enum PointerParseErrorCode { + kPointerParseErrorNone = 0, //!< The parse is successful + + kPointerParseErrorTokenMustBeginWithSolidus, //!< A token must begin with a '/' + kPointerParseErrorInvalidEscape, //!< Invalid escape + kPointerParseErrorInvalidPercentEncoding, //!< Invalid percent encoding in URI fragment + kPointerParseErrorCharacterMustPercentEncode //!< A character must percent encoded in URI fragment +}; + +/////////////////////////////////////////////////////////////////////////////// +// GenericPointer + +//! Represents a JSON Pointer. Use Pointer for UTF8 encoding and default allocator. +/*! + This class implements RFC 6901 "JavaScript Object Notation (JSON) Pointer" + (https://tools.ietf.org/html/rfc6901). + + A JSON pointer is for identifying a specific value in a JSON document + (GenericDocument). It can simplify coding of DOM tree manipulation, because it + can access multiple-level depth of DOM tree with single API call. + + After it parses a string representation (e.g. "/foo/0" or URI fragment + representation (e.g. "#/foo/0") into its internal representation (tokens), + it can be used to resolve a specific value in multiple documents, or sub-tree + of documents. + + Contrary to GenericValue, Pointer can be copy constructed and copy assigned. + Apart from assignment, a Pointer cannot be modified after construction. + + Although Pointer is very convenient, please aware that constructing Pointer + involves parsing and dynamic memory allocation. A special constructor with user- + supplied tokens eliminates these. + + GenericPointer depends on GenericDocument and GenericValue. + + \tparam ValueType The value type of the DOM tree. E.g. GenericValue > + \tparam Allocator The allocator type for allocating memory for internal representation. + + \note GenericPointer uses same encoding of ValueType. + However, Allocator of GenericPointer is independent of Allocator of Value. +*/ +template +class GenericPointer { +public: + typedef typename ValueType::EncodingType EncodingType; //!< Encoding type from Value + typedef typename ValueType::Ch Ch; //!< Character type from Value + + //! A token is the basic units of internal representation. + /*! + A JSON pointer string representation "/foo/123" is parsed to two tokens: + "foo" and 123. 123 will be represented in both numeric form and string form. + They are resolved according to the actual value type (object or array). + + For token that are not numbers, or the numeric value is out of bound + (greater than limits of SizeType), they are only treated as string form + (i.e. the token's index will be equal to kPointerInvalidIndex). + + This struct is public so that user can create a Pointer without parsing and + allocation, using a special constructor. + */ + struct Token { + const Ch* name; //!< Name of the token. It has null character at the end but it can contain null character. + SizeType length; //!< Length of the name. + SizeType index; //!< A valid array index, if it is not equal to kPointerInvalidIndex. + }; + + //!@name Constructors and destructor. + //@{ + + //! Default constructor. + GenericPointer(Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) {} + + //! Constructor that parses a string or URI fragment representation. + /*! + \param source A null-terminated, string or URI fragment representation of JSON pointer. + \param allocator User supplied allocator for this pointer. If no allocator is provided, it creates a self-owned one. + */ + explicit GenericPointer(const Ch* source, Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + Parse(source, internal::StrLen(source)); + } + +#if RAPIDJSON_HAS_STDSTRING + //! Constructor that parses a string or URI fragment representation. + /*! + \param source A string or URI fragment representation of JSON pointer. + \param allocator User supplied allocator for this pointer. If no allocator is provided, it creates a self-owned one. + \note Requires the definition of the preprocessor symbol \ref RAPIDJSON_HAS_STDSTRING. + */ + explicit GenericPointer(const std::basic_string& source, Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + Parse(source.c_str(), source.size()); + } +#endif + + //! Constructor that parses a string or URI fragment representation, with length of the source string. + /*! + \param source A string or URI fragment representation of JSON pointer. + \param length Length of source. + \param allocator User supplied allocator for this pointer. If no allocator is provided, it creates a self-owned one. + \note Slightly faster than the overload without length. + */ + GenericPointer(const Ch* source, size_t length, Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + Parse(source, length); + } + + //! Constructor with user-supplied tokens. + /*! + This constructor let user supplies const array of tokens. + This prevents the parsing process and eliminates allocation. + This is preferred for memory constrained environments. + + \param tokens An constant array of tokens representing the JSON pointer. + \param tokenCount Number of tokens. + + \b Example + \code + #define NAME(s) { s, sizeof(s) / sizeof(s[0]) - 1, kPointerInvalidIndex } + #define INDEX(i) { #i, sizeof(#i) - 1, i } + + static const Pointer::Token kTokens[] = { NAME("foo"), INDEX(123) }; + static const Pointer p(kTokens, sizeof(kTokens) / sizeof(kTokens[0])); + // Equivalent to static const Pointer p("/foo/123"); + + #undef NAME + #undef INDEX + \endcode + */ + GenericPointer(const Token* tokens, size_t tokenCount) : allocator_(), ownAllocator_(), nameBuffer_(), tokens_(const_cast(tokens)), tokenCount_(tokenCount), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) {} + + //! Copy constructor. + GenericPointer(const GenericPointer& rhs, Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + *this = rhs; + } + + //! Destructor. + ~GenericPointer() { + if (nameBuffer_) // If user-supplied tokens constructor is used, nameBuffer_ is nullptr and tokens_ are not deallocated. + Allocator::Free(tokens_); + RAPIDJSON_DELETE(ownAllocator_); + } + + //! Assignment operator. + GenericPointer& operator=(const GenericPointer& rhs) { + if (this != &rhs) { + // Do not delete ownAllcator + if (nameBuffer_) + Allocator::Free(tokens_); + + tokenCount_ = rhs.tokenCount_; + parseErrorOffset_ = rhs.parseErrorOffset_; + parseErrorCode_ = rhs.parseErrorCode_; + + if (rhs.nameBuffer_) + CopyFromRaw(rhs); // Normally parsed tokens. + else { + tokens_ = rhs.tokens_; // User supplied const tokens. + nameBuffer_ = 0; + } + } + return *this; + } + + //@} + + //!@name Append token + //@{ + + //! Append a token and return a new Pointer + /*! + \param token Token to be appended. + \param allocator Allocator for the newly return Pointer. + \return A new Pointer with appended token. + */ + GenericPointer Append(const Token& token, Allocator* allocator = 0) const { + GenericPointer r; + r.allocator_ = allocator; + Ch *p = r.CopyFromRaw(*this, 1, token.length + 1); + std::memcpy(p, token.name, (token.length + 1) * sizeof(Ch)); + r.tokens_[tokenCount_].name = p; + r.tokens_[tokenCount_].length = token.length; + r.tokens_[tokenCount_].index = token.index; + return r; + } + + //! Append a name token with length, and return a new Pointer + /*! + \param name Name to be appended. + \param length Length of name. + \param allocator Allocator for the newly return Pointer. + \return A new Pointer with appended token. + */ + GenericPointer Append(const Ch* name, SizeType length, Allocator* allocator = 0) const { + Token token = { name, length, kPointerInvalidIndex }; + return Append(token, allocator); + } + + //! Append a name token without length, and return a new Pointer + /*! + \param name Name (const Ch*) to be appended. + \param allocator Allocator for the newly return Pointer. + \return A new Pointer with appended token. + */ + template + RAPIDJSON_DISABLEIF_RETURN((internal::NotExpr::Type, Ch> >), (GenericPointer)) + Append(T* name, Allocator* allocator = 0) const { + return Append(name, internal::StrLen(name), allocator); + } + +#if RAPIDJSON_HAS_STDSTRING + //! Append a name token, and return a new Pointer + /*! + \param name Name to be appended. + \param allocator Allocator for the newly return Pointer. + \return A new Pointer with appended token. + */ + GenericPointer Append(const std::basic_string& name, Allocator* allocator = 0) const { + return Append(name.c_str(), static_cast(name.size()), allocator); + } +#endif + + //! Append a index token, and return a new Pointer + /*! + \param index Index to be appended. + \param allocator Allocator for the newly return Pointer. + \return A new Pointer with appended token. + */ + GenericPointer Append(SizeType index, Allocator* allocator = 0) const { + char buffer[21]; + char* end = sizeof(SizeType) == 4 ? internal::u32toa(index, buffer) : internal::u64toa(index, buffer); + SizeType length = static_cast(end - buffer); + buffer[length] = '\0'; + + if (sizeof(Ch) == 1) { + Token token = { reinterpret_cast(buffer), length, index }; + return Append(token, allocator); + } + else { + Ch name[21]; + for (size_t i = 0; i <= length; i++) + name[i] = static_cast(buffer[i]); + Token token = { name, length, index }; + return Append(token, allocator); + } + } + + //! Append a token by value, and return a new Pointer + /*! + \param token token to be appended. + \param allocator Allocator for the newly return Pointer. + \return A new Pointer with appended token. + */ + GenericPointer Append(const ValueType& token, Allocator* allocator = 0) const { + if (token.IsString()) + return Append(token.GetString(), token.GetStringLength(), allocator); + else { + RAPIDJSON_ASSERT(token.IsUint64()); + RAPIDJSON_ASSERT(token.GetUint64() <= SizeType(~0)); + return Append(static_cast(token.GetUint64()), allocator); + } + } + + //!@name Handling Parse Error + //@{ + + //! Check whether this is a valid pointer. + bool IsValid() const { return parseErrorCode_ == kPointerParseErrorNone; } + + //! Get the parsing error offset in code unit. + size_t GetParseErrorOffset() const { return parseErrorOffset_; } + + //! Get the parsing error code. + PointerParseErrorCode GetParseErrorCode() const { return parseErrorCode_; } + + //@} + + //! Get the allocator of this pointer. + Allocator& GetAllocator() { return *allocator_; } + + //!@name Tokens + //@{ + + //! Get the token array (const version only). + const Token* GetTokens() const { return tokens_; } + + //! Get the number of tokens. + size_t GetTokenCount() const { return tokenCount_; } + + //@} + + //!@name Equality/inequality operators + //@{ + + //! Equality operator. + /*! + \note When any pointers are invalid, always returns false. + */ + bool operator==(const GenericPointer& rhs) const { + if (!IsValid() || !rhs.IsValid() || tokenCount_ != rhs.tokenCount_) + return false; + + for (size_t i = 0; i < tokenCount_; i++) { + if (tokens_[i].index != rhs.tokens_[i].index || + tokens_[i].length != rhs.tokens_[i].length || + (tokens_[i].length != 0 && std::memcmp(tokens_[i].name, rhs.tokens_[i].name, sizeof(Ch)* tokens_[i].length) != 0)) + { + return false; + } + } + + return true; + } + + //! Inequality operator. + /*! + \note When any pointers are invalid, always returns true. + */ + bool operator!=(const GenericPointer& rhs) const { return !(*this == rhs); } + + //@} + + //!@name Stringify + //@{ + + //! Stringify the pointer into string representation. + /*! + \tparam OutputStream Type of output stream. + \param os The output stream. + */ + template + bool Stringify(OutputStream& os) const { + return Stringify(os); + } + + //! Stringify the pointer into URI fragment representation. + /*! + \tparam OutputStream Type of output stream. + \param os The output stream. + */ + template + bool StringifyUriFragment(OutputStream& os) const { + return Stringify(os); + } + + //@} + + //!@name Create value + //@{ + + //! Create a value in a subtree. + /*! + If the value is not exist, it creates all parent values and a JSON Null value. + So it always succeed and return the newly created or existing value. + + Remind that it may change types of parents according to tokens, so it + potentially removes previously stored values. For example, if a document + was an array, and "/foo" is used to create a value, then the document + will be changed to an object, and all existing array elements are lost. + + \param root Root value of a DOM subtree to be resolved. It can be any value other than document root. + \param allocator Allocator for creating the values if the specified value or its parents are not exist. + \param alreadyExist If non-null, it stores whether the resolved value is already exist. + \return The resolved newly created (a JSON Null value), or already exists value. + */ + ValueType& Create(ValueType& root, typename ValueType::AllocatorType& allocator, bool* alreadyExist = 0) const { + RAPIDJSON_ASSERT(IsValid()); + ValueType* v = &root; + bool exist = true; + for (const Token *t = tokens_; t != tokens_ + tokenCount_; ++t) { + if (v->IsArray() && t->name[0] == '-' && t->length == 1) { + v->PushBack(ValueType().Move(), allocator); + v = &((*v)[v->Size() - 1]); + exist = false; + } + else { + if (t->index == kPointerInvalidIndex) { // must be object name + if (!v->IsObject()) + v->SetObject(); // Change to Object + } + else { // object name or array index + if (!v->IsArray() && !v->IsObject()) + v->SetArray(); // Change to Array + } + + if (v->IsArray()) { + if (t->index >= v->Size()) { + v->Reserve(t->index + 1, allocator); + while (t->index >= v->Size()) + v->PushBack(ValueType().Move(), allocator); + exist = false; + } + v = &((*v)[t->index]); + } + else { + typename ValueType::MemberIterator m = v->FindMember(GenericStringRef(t->name, t->length)); + if (m == v->MemberEnd()) { + v->AddMember(ValueType(t->name, t->length, allocator).Move(), ValueType().Move(), allocator); + v = &(--v->MemberEnd())->value; // Assumes AddMember() appends at the end + exist = false; + } + else + v = &m->value; + } + } + } + + if (alreadyExist) + *alreadyExist = exist; + + return *v; + } + + //! Creates a value in a document. + /*! + \param document A document to be resolved. + \param alreadyExist If non-null, it stores whether the resolved value is already exist. + \return The resolved newly created, or already exists value. + */ + template + ValueType& Create(GenericDocument& document, bool* alreadyExist = 0) const { + return Create(document, document.GetAllocator(), alreadyExist); + } + + //@} + + //!@name Query value + //@{ + + //! Query a value in a subtree. + /*! + \param root Root value of a DOM sub-tree to be resolved. It can be any value other than document root. + \param unresolvedTokenIndex If the pointer cannot resolve a token in the pointer, this parameter can obtain the index of unresolved token. + \return Pointer to the value if it can be resolved. Otherwise null. + + \note + There are only 3 situations when a value cannot be resolved: + 1. A value in the path is not an array nor object. + 2. An object value does not contain the token. + 3. A token is out of range of an array value. + + Use unresolvedTokenIndex to retrieve the token index. + */ + ValueType* Get(ValueType& root, size_t* unresolvedTokenIndex = 0) const { + RAPIDJSON_ASSERT(IsValid()); + ValueType* v = &root; + for (const Token *t = tokens_; t != tokens_ + tokenCount_; ++t) { + switch (v->GetType()) { + case kObjectType: + { + typename ValueType::MemberIterator m = v->FindMember(GenericStringRef(t->name, t->length)); + if (m == v->MemberEnd()) + break; + v = &m->value; + } + continue; + case kArrayType: + if (t->index == kPointerInvalidIndex || t->index >= v->Size()) + break; + v = &((*v)[t->index]); + continue; + default: + break; + } + + // Error: unresolved token + if (unresolvedTokenIndex) + *unresolvedTokenIndex = static_cast(t - tokens_); + return 0; + } + return v; + } + + //! Query a const value in a const subtree. + /*! + \param root Root value of a DOM sub-tree to be resolved. It can be any value other than document root. + \return Pointer to the value if it can be resolved. Otherwise null. + */ + const ValueType* Get(const ValueType& root, size_t* unresolvedTokenIndex = 0) const { + return Get(const_cast(root), unresolvedTokenIndex); + } + + //@} + + //!@name Query a value with default + //@{ + + //! Query a value in a subtree with default value. + /*! + Similar to Get(), but if the specified value do not exists, it creates all parents and clone the default value. + So that this function always succeed. + + \param root Root value of a DOM sub-tree to be resolved. It can be any value other than document root. + \param defaultValue Default value to be cloned if the value was not exists. + \param allocator Allocator for creating the values if the specified value or its parents are not exist. + \see Create() + */ + ValueType& GetWithDefault(ValueType& root, const ValueType& defaultValue, typename ValueType::AllocatorType& allocator) const { + bool alreadyExist; + Value& v = Create(root, allocator, &alreadyExist); + return alreadyExist ? v : v.CopyFrom(defaultValue, allocator); + } + + //! Query a value in a subtree with default null-terminated string. + ValueType& GetWithDefault(ValueType& root, const Ch* defaultValue, typename ValueType::AllocatorType& allocator) const { + bool alreadyExist; + Value& v = Create(root, allocator, &alreadyExist); + return alreadyExist ? v : v.SetString(defaultValue, allocator); + } + +#if RAPIDJSON_HAS_STDSTRING + //! Query a value in a subtree with default std::basic_string. + ValueType& GetWithDefault(ValueType& root, const std::basic_string& defaultValue, typename ValueType::AllocatorType& allocator) const { + bool alreadyExist; + Value& v = Create(root, allocator, &alreadyExist); + return alreadyExist ? v : v.SetString(defaultValue, allocator); + } +#endif + + //! Query a value in a subtree with default primitive value. + /*! + \tparam T Either \ref Type, \c int, \c unsigned, \c int64_t, \c uint64_t, \c bool + */ + template + RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (ValueType&)) + GetWithDefault(ValueType& root, T defaultValue, typename ValueType::AllocatorType& allocator) const { + return GetWithDefault(root, ValueType(defaultValue).Move(), allocator); + } + + //! Query a value in a document with default value. + template + ValueType& GetWithDefault(GenericDocument& document, const ValueType& defaultValue) const { + return GetWithDefault(document, defaultValue, document.GetAllocator()); + } + + //! Query a value in a document with default null-terminated string. + template + ValueType& GetWithDefault(GenericDocument& document, const Ch* defaultValue) const { + return GetWithDefault(document, defaultValue, document.GetAllocator()); + } + +#if RAPIDJSON_HAS_STDSTRING + //! Query a value in a document with default std::basic_string. + template + ValueType& GetWithDefault(GenericDocument& document, const std::basic_string& defaultValue) const { + return GetWithDefault(document, defaultValue, document.GetAllocator()); + } +#endif + + //! Query a value in a document with default primitive value. + /*! + \tparam T Either \ref Type, \c int, \c unsigned, \c int64_t, \c uint64_t, \c bool + */ + template + RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (ValueType&)) + GetWithDefault(GenericDocument& document, T defaultValue) const { + return GetWithDefault(document, defaultValue, document.GetAllocator()); + } + + //@} + + //!@name Set a value + //@{ + + //! Set a value in a subtree, with move semantics. + /*! + It creates all parents if they are not exist or types are different to the tokens. + So this function always succeeds but potentially remove existing values. + + \param root Root value of a DOM sub-tree to be resolved. It can be any value other than document root. + \param value Value to be set. + \param allocator Allocator for creating the values if the specified value or its parents are not exist. + \see Create() + */ + ValueType& Set(ValueType& root, ValueType& value, typename ValueType::AllocatorType& allocator) const { + return Create(root, allocator) = value; + } + + //! Set a value in a subtree, with copy semantics. + ValueType& Set(ValueType& root, const ValueType& value, typename ValueType::AllocatorType& allocator) const { + return Create(root, allocator).CopyFrom(value, allocator); + } + + //! Set a null-terminated string in a subtree. + ValueType& Set(ValueType& root, const Ch* value, typename ValueType::AllocatorType& allocator) const { + return Create(root, allocator) = ValueType(value, allocator).Move(); + } + +#if RAPIDJSON_HAS_STDSTRING + //! Set a std::basic_string in a subtree. + ValueType& Set(ValueType& root, const std::basic_string& value, typename ValueType::AllocatorType& allocator) const { + return Create(root, allocator) = ValueType(value, allocator).Move(); + } +#endif + + //! Set a primitive value in a subtree. + /*! + \tparam T Either \ref Type, \c int, \c unsigned, \c int64_t, \c uint64_t, \c bool + */ + template + RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (ValueType&)) + Set(ValueType& root, T value, typename ValueType::AllocatorType& allocator) const { + return Create(root, allocator) = ValueType(value).Move(); + } + + //! Set a value in a document, with move semantics. + template + ValueType& Set(GenericDocument& document, ValueType& value) const { + return Create(document) = value; + } + + //! Set a value in a document, with copy semantics. + template + ValueType& Set(GenericDocument& document, const ValueType& value) const { + return Create(document).CopyFrom(value, document.GetAllocator()); + } + + //! Set a null-terminated string in a document. + template + ValueType& Set(GenericDocument& document, const Ch* value) const { + return Create(document) = ValueType(value, document.GetAllocator()).Move(); + } + +#if RAPIDJSON_HAS_STDSTRING + //! Sets a std::basic_string in a document. + template + ValueType& Set(GenericDocument& document, const std::basic_string& value) const { + return Create(document) = ValueType(value, document.GetAllocator()).Move(); + } +#endif + + //! Set a primitive value in a document. + /*! + \tparam T Either \ref Type, \c int, \c unsigned, \c int64_t, \c uint64_t, \c bool + */ + template + RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (ValueType&)) + Set(GenericDocument& document, T value) const { + return Create(document) = value; + } + + //@} + + //!@name Swap a value + //@{ + + //! Swap a value with a value in a subtree. + /*! + It creates all parents if they are not exist or types are different to the tokens. + So this function always succeeds but potentially remove existing values. + + \param root Root value of a DOM sub-tree to be resolved. It can be any value other than document root. + \param value Value to be swapped. + \param allocator Allocator for creating the values if the specified value or its parents are not exist. + \see Create() + */ + ValueType& Swap(ValueType& root, ValueType& value, typename ValueType::AllocatorType& allocator) const { + return Create(root, allocator).Swap(value); + } + + //! Swap a value with a value in a document. + template + ValueType& Swap(GenericDocument& document, ValueType& value) const { + return Create(document).Swap(value); + } + + //@} + + //! Erase a value in a subtree. + /*! + \param root Root value of a DOM sub-tree to be resolved. It can be any value other than document root. + \return Whether the resolved value is found and erased. + + \note Erasing with an empty pointer \c Pointer(""), i.e. the root, always fail and return false. + */ + bool Erase(ValueType& root) const { + RAPIDJSON_ASSERT(IsValid()); + if (tokenCount_ == 0) // Cannot erase the root + return false; + + ValueType* v = &root; + const Token* last = tokens_ + (tokenCount_ - 1); + for (const Token *t = tokens_; t != last; ++t) { + switch (v->GetType()) { + case kObjectType: + { + typename ValueType::MemberIterator m = v->FindMember(GenericStringRef(t->name, t->length)); + if (m == v->MemberEnd()) + return false; + v = &m->value; + } + break; + case kArrayType: + if (t->index == kPointerInvalidIndex || t->index >= v->Size()) + return false; + v = &((*v)[t->index]); + break; + default: + return false; + } + } + + switch (v->GetType()) { + case kObjectType: + return v->EraseMember(GenericStringRef(last->name, last->length)); + case kArrayType: + if (last->index == kPointerInvalidIndex || last->index >= v->Size()) + return false; + v->Erase(v->Begin() + last->index); + return true; + default: + return false; + } + } + +private: + //! Clone the content from rhs to this. + /*! + \param rhs Source pointer. + \param extraToken Extra tokens to be allocated. + \param extraNameBufferSize Extra name buffer size (in number of Ch) to be allocated. + \return Start of non-occupied name buffer, for storing extra names. + */ + Ch* CopyFromRaw(const GenericPointer& rhs, size_t extraToken = 0, size_t extraNameBufferSize = 0) { + if (!allocator_) // allocator is independently owned. + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + + size_t nameBufferSize = rhs.tokenCount_; // null terminators for tokens + for (Token *t = rhs.tokens_; t != rhs.tokens_ + rhs.tokenCount_; ++t) + nameBufferSize += t->length; + + tokenCount_ = rhs.tokenCount_ + extraToken; + tokens_ = static_cast(allocator_->Malloc(tokenCount_ * sizeof(Token) + (nameBufferSize + extraNameBufferSize) * sizeof(Ch))); + nameBuffer_ = reinterpret_cast(tokens_ + tokenCount_); + if (rhs.tokenCount_ > 0) { + std::memcpy(tokens_, rhs.tokens_, rhs.tokenCount_ * sizeof(Token)); + } + if (nameBufferSize > 0) { + std::memcpy(nameBuffer_, rhs.nameBuffer_, nameBufferSize * sizeof(Ch)); + } + + // Adjust pointers to name buffer + std::ptrdiff_t diff = nameBuffer_ - rhs.nameBuffer_; + for (Token *t = tokens_; t != tokens_ + rhs.tokenCount_; ++t) + t->name += diff; + + return nameBuffer_ + nameBufferSize; + } + + //! Check whether a character should be percent-encoded. + /*! + According to RFC 3986 2.3 Unreserved Characters. + \param c The character (code unit) to be tested. + */ + bool NeedPercentEncode(Ch c) const { + return !((c >= '0' && c <= '9') || (c >= 'A' && c <='Z') || (c >= 'a' && c <= 'z') || c == '-' || c == '.' || c == '_' || c =='~'); + } + + //! Parse a JSON String or its URI fragment representation into tokens. +#ifndef __clang__ // -Wdocumentation + /*! + \param source Either a JSON Pointer string, or its URI fragment representation. Not need to be null terminated. + \param length Length of the source string. + \note Source cannot be JSON String Representation of JSON Pointer, e.g. In "/\u0000", \u0000 will not be unescaped. + */ +#endif + void Parse(const Ch* source, size_t length) { + RAPIDJSON_ASSERT(source != NULL); + RAPIDJSON_ASSERT(nameBuffer_ == 0); + RAPIDJSON_ASSERT(tokens_ == 0); + + // Create own allocator if user did not supply. + if (!allocator_) + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + + // Count number of '/' as tokenCount + tokenCount_ = 0; + for (const Ch* s = source; s != source + length; s++) + if (*s == '/') + tokenCount_++; + + Token* token = tokens_ = static_cast(allocator_->Malloc(tokenCount_ * sizeof(Token) + length * sizeof(Ch))); + Ch* name = nameBuffer_ = reinterpret_cast(tokens_ + tokenCount_); + size_t i = 0; + + // Detect if it is a URI fragment + bool uriFragment = false; + if (source[i] == '#') { + uriFragment = true; + i++; + } + + if (i != length && source[i] != '/') { + parseErrorCode_ = kPointerParseErrorTokenMustBeginWithSolidus; + goto error; + } + + while (i < length) { + RAPIDJSON_ASSERT(source[i] == '/'); + i++; // consumes '/' + + token->name = name; + bool isNumber = true; + + while (i < length && source[i] != '/') { + Ch c = source[i]; + if (uriFragment) { + // Decoding percent-encoding for URI fragment + if (c == '%') { + PercentDecodeStream is(&source[i], source + length); + GenericInsituStringStream os(name); + Ch* begin = os.PutBegin(); + if (!Transcoder, EncodingType>().Validate(is, os) || !is.IsValid()) { + parseErrorCode_ = kPointerParseErrorInvalidPercentEncoding; + goto error; + } + size_t len = os.PutEnd(begin); + i += is.Tell() - 1; + if (len == 1) + c = *name; + else { + name += len; + isNumber = false; + i++; + continue; + } + } + else if (NeedPercentEncode(c)) { + parseErrorCode_ = kPointerParseErrorCharacterMustPercentEncode; + goto error; + } + } + + i++; + + // Escaping "~0" -> '~', "~1" -> '/' + if (c == '~') { + if (i < length) { + c = source[i]; + if (c == '0') c = '~'; + else if (c == '1') c = '/'; + else { + parseErrorCode_ = kPointerParseErrorInvalidEscape; + goto error; + } + i++; + } + else { + parseErrorCode_ = kPointerParseErrorInvalidEscape; + goto error; + } + } + + // First check for index: all of characters are digit + if (c < '0' || c > '9') + isNumber = false; + + *name++ = c; + } + token->length = static_cast(name - token->name); + if (token->length == 0) + isNumber = false; + *name++ = '\0'; // Null terminator + + // Second check for index: more than one digit cannot have leading zero + if (isNumber && token->length > 1 && token->name[0] == '0') + isNumber = false; + + // String to SizeType conversion + SizeType n = 0; + if (isNumber) { + for (size_t j = 0; j < token->length; j++) { + SizeType m = n * 10 + static_cast(token->name[j] - '0'); + if (m < n) { // overflow detection + isNumber = false; + break; + } + n = m; + } + } + + token->index = isNumber ? n : kPointerInvalidIndex; + token++; + } + + RAPIDJSON_ASSERT(name <= nameBuffer_ + length); // Should not overflow buffer + parseErrorCode_ = kPointerParseErrorNone; + return; + + error: + Allocator::Free(tokens_); + nameBuffer_ = 0; + tokens_ = 0; + tokenCount_ = 0; + parseErrorOffset_ = i; + return; + } + + //! Stringify to string or URI fragment representation. + /*! + \tparam uriFragment True for stringifying to URI fragment representation. False for string representation. + \tparam OutputStream type of output stream. + \param os The output stream. + */ + template + bool Stringify(OutputStream& os) const { + RAPIDJSON_ASSERT(IsValid()); + + if (uriFragment) + os.Put('#'); + + for (Token *t = tokens_; t != tokens_ + tokenCount_; ++t) { + os.Put('/'); + for (size_t j = 0; j < t->length; j++) { + Ch c = t->name[j]; + if (c == '~') { + os.Put('~'); + os.Put('0'); + } + else if (c == '/') { + os.Put('~'); + os.Put('1'); + } + else if (uriFragment && NeedPercentEncode(c)) { + // Transcode to UTF8 sequence + GenericStringStream source(&t->name[j]); + PercentEncodeStream target(os); + if (!Transcoder >().Validate(source, target)) + return false; + j += source.Tell() - 1; + } + else + os.Put(c); + } + } + return true; + } + + //! A helper stream for decoding a percent-encoded sequence into code unit. + /*! + This stream decodes %XY triplet into code unit (0-255). + If it encounters invalid characters, it sets output code unit as 0 and + mark invalid, and to be checked by IsValid(). + */ + class PercentDecodeStream { + public: + typedef typename ValueType::Ch Ch; + + //! Constructor + /*! + \param source Start of the stream + \param end Past-the-end of the stream. + */ + PercentDecodeStream(const Ch* source, const Ch* end) : src_(source), head_(source), end_(end), valid_(true) {} + + Ch Take() { + if (*src_ != '%' || src_ + 3 > end_) { // %XY triplet + valid_ = false; + return 0; + } + src_++; + Ch c = 0; + for (int j = 0; j < 2; j++) { + c = static_cast(c << 4); + Ch h = *src_; + if (h >= '0' && h <= '9') c = static_cast(c + h - '0'); + else if (h >= 'A' && h <= 'F') c = static_cast(c + h - 'A' + 10); + else if (h >= 'a' && h <= 'f') c = static_cast(c + h - 'a' + 10); + else { + valid_ = false; + return 0; + } + src_++; + } + return c; + } + + size_t Tell() const { return static_cast(src_ - head_); } + bool IsValid() const { return valid_; } + + private: + const Ch* src_; //!< Current read position. + const Ch* head_; //!< Original head of the string. + const Ch* end_; //!< Past-the-end position. + bool valid_; //!< Whether the parsing is valid. + }; + + //! A helper stream to encode character (UTF-8 code unit) into percent-encoded sequence. + template + class PercentEncodeStream { + public: + PercentEncodeStream(OutputStream& os) : os_(os) {} + void Put(char c) { // UTF-8 must be byte + unsigned char u = static_cast(c); + static const char hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; + os_.Put('%'); + os_.Put(static_cast(hexDigits[u >> 4])); + os_.Put(static_cast(hexDigits[u & 15])); + } + private: + OutputStream& os_; + }; + + Allocator* allocator_; //!< The current allocator. It is either user-supplied or equal to ownAllocator_. + Allocator* ownAllocator_; //!< Allocator owned by this Pointer. + Ch* nameBuffer_; //!< A buffer containing all names in tokens. + Token* tokens_; //!< A list of tokens. + size_t tokenCount_; //!< Number of tokens in tokens_. + size_t parseErrorOffset_; //!< Offset in code unit when parsing fail. + PointerParseErrorCode parseErrorCode_; //!< Parsing error code. +}; + +//! GenericPointer for Value (UTF-8, default allocator). +typedef GenericPointer Pointer; + +//!@name Helper functions for GenericPointer +//@{ + +////////////////////////////////////////////////////////////////////////////// + +template +typename T::ValueType& CreateValueByPointer(T& root, const GenericPointer& pointer, typename T::AllocatorType& a) { + return pointer.Create(root, a); +} + +template +typename T::ValueType& CreateValueByPointer(T& root, const CharType(&source)[N], typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Create(root, a); +} + +// No allocator parameter + +template +typename DocumentType::ValueType& CreateValueByPointer(DocumentType& document, const GenericPointer& pointer) { + return pointer.Create(document); +} + +template +typename DocumentType::ValueType& CreateValueByPointer(DocumentType& document, const CharType(&source)[N]) { + return GenericPointer(source, N - 1).Create(document); +} + +////////////////////////////////////////////////////////////////////////////// + +template +typename T::ValueType* GetValueByPointer(T& root, const GenericPointer& pointer, size_t* unresolvedTokenIndex = 0) { + return pointer.Get(root, unresolvedTokenIndex); +} + +template +const typename T::ValueType* GetValueByPointer(const T& root, const GenericPointer& pointer, size_t* unresolvedTokenIndex = 0) { + return pointer.Get(root, unresolvedTokenIndex); +} + +template +typename T::ValueType* GetValueByPointer(T& root, const CharType (&source)[N], size_t* unresolvedTokenIndex = 0) { + return GenericPointer(source, N - 1).Get(root, unresolvedTokenIndex); +} + +template +const typename T::ValueType* GetValueByPointer(const T& root, const CharType(&source)[N], size_t* unresolvedTokenIndex = 0) { + return GenericPointer(source, N - 1).Get(root, unresolvedTokenIndex); +} + +////////////////////////////////////////////////////////////////////////////// + +template +typename T::ValueType& GetValueByPointerWithDefault(T& root, const GenericPointer& pointer, const typename T::ValueType& defaultValue, typename T::AllocatorType& a) { + return pointer.GetWithDefault(root, defaultValue, a); +} + +template +typename T::ValueType& GetValueByPointerWithDefault(T& root, const GenericPointer& pointer, const typename T::Ch* defaultValue, typename T::AllocatorType& a) { + return pointer.GetWithDefault(root, defaultValue, a); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename T::ValueType& GetValueByPointerWithDefault(T& root, const GenericPointer& pointer, const std::basic_string& defaultValue, typename T::AllocatorType& a) { + return pointer.GetWithDefault(root, defaultValue, a); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename T::ValueType&)) +GetValueByPointerWithDefault(T& root, const GenericPointer& pointer, T2 defaultValue, typename T::AllocatorType& a) { + return pointer.GetWithDefault(root, defaultValue, a); +} + +template +typename T::ValueType& GetValueByPointerWithDefault(T& root, const CharType(&source)[N], const typename T::ValueType& defaultValue, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).GetWithDefault(root, defaultValue, a); +} + +template +typename T::ValueType& GetValueByPointerWithDefault(T& root, const CharType(&source)[N], const typename T::Ch* defaultValue, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).GetWithDefault(root, defaultValue, a); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename T::ValueType& GetValueByPointerWithDefault(T& root, const CharType(&source)[N], const std::basic_string& defaultValue, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).GetWithDefault(root, defaultValue, a); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename T::ValueType&)) +GetValueByPointerWithDefault(T& root, const CharType(&source)[N], T2 defaultValue, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).GetWithDefault(root, defaultValue, a); +} + +// No allocator parameter + +template +typename DocumentType::ValueType& GetValueByPointerWithDefault(DocumentType& document, const GenericPointer& pointer, const typename DocumentType::ValueType& defaultValue) { + return pointer.GetWithDefault(document, defaultValue); +} + +template +typename DocumentType::ValueType& GetValueByPointerWithDefault(DocumentType& document, const GenericPointer& pointer, const typename DocumentType::Ch* defaultValue) { + return pointer.GetWithDefault(document, defaultValue); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename DocumentType::ValueType& GetValueByPointerWithDefault(DocumentType& document, const GenericPointer& pointer, const std::basic_string& defaultValue) { + return pointer.GetWithDefault(document, defaultValue); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename DocumentType::ValueType&)) +GetValueByPointerWithDefault(DocumentType& document, const GenericPointer& pointer, T2 defaultValue) { + return pointer.GetWithDefault(document, defaultValue); +} + +template +typename DocumentType::ValueType& GetValueByPointerWithDefault(DocumentType& document, const CharType(&source)[N], const typename DocumentType::ValueType& defaultValue) { + return GenericPointer(source, N - 1).GetWithDefault(document, defaultValue); +} + +template +typename DocumentType::ValueType& GetValueByPointerWithDefault(DocumentType& document, const CharType(&source)[N], const typename DocumentType::Ch* defaultValue) { + return GenericPointer(source, N - 1).GetWithDefault(document, defaultValue); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename DocumentType::ValueType& GetValueByPointerWithDefault(DocumentType& document, const CharType(&source)[N], const std::basic_string& defaultValue) { + return GenericPointer(source, N - 1).GetWithDefault(document, defaultValue); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename DocumentType::ValueType&)) +GetValueByPointerWithDefault(DocumentType& document, const CharType(&source)[N], T2 defaultValue) { + return GenericPointer(source, N - 1).GetWithDefault(document, defaultValue); +} + +////////////////////////////////////////////////////////////////////////////// + +template +typename T::ValueType& SetValueByPointer(T& root, const GenericPointer& pointer, typename T::ValueType& value, typename T::AllocatorType& a) { + return pointer.Set(root, value, a); +} + +template +typename T::ValueType& SetValueByPointer(T& root, const GenericPointer& pointer, const typename T::ValueType& value, typename T::AllocatorType& a) { + return pointer.Set(root, value, a); +} + +template +typename T::ValueType& SetValueByPointer(T& root, const GenericPointer& pointer, const typename T::Ch* value, typename T::AllocatorType& a) { + return pointer.Set(root, value, a); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename T::ValueType& SetValueByPointer(T& root, const GenericPointer& pointer, const std::basic_string& value, typename T::AllocatorType& a) { + return pointer.Set(root, value, a); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename T::ValueType&)) +SetValueByPointer(T& root, const GenericPointer& pointer, T2 value, typename T::AllocatorType& a) { + return pointer.Set(root, value, a); +} + +template +typename T::ValueType& SetValueByPointer(T& root, const CharType(&source)[N], typename T::ValueType& value, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Set(root, value, a); +} + +template +typename T::ValueType& SetValueByPointer(T& root, const CharType(&source)[N], const typename T::ValueType& value, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Set(root, value, a); +} + +template +typename T::ValueType& SetValueByPointer(T& root, const CharType(&source)[N], const typename T::Ch* value, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Set(root, value, a); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename T::ValueType& SetValueByPointer(T& root, const CharType(&source)[N], const std::basic_string& value, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Set(root, value, a); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename T::ValueType&)) +SetValueByPointer(T& root, const CharType(&source)[N], T2 value, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Set(root, value, a); +} + +// No allocator parameter + +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const GenericPointer& pointer, typename DocumentType::ValueType& value) { + return pointer.Set(document, value); +} + +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const GenericPointer& pointer, const typename DocumentType::ValueType& value) { + return pointer.Set(document, value); +} + +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const GenericPointer& pointer, const typename DocumentType::Ch* value) { + return pointer.Set(document, value); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const GenericPointer& pointer, const std::basic_string& value) { + return pointer.Set(document, value); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename DocumentType::ValueType&)) +SetValueByPointer(DocumentType& document, const GenericPointer& pointer, T2 value) { + return pointer.Set(document, value); +} + +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const CharType(&source)[N], typename DocumentType::ValueType& value) { + return GenericPointer(source, N - 1).Set(document, value); +} + +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const CharType(&source)[N], const typename DocumentType::ValueType& value) { + return GenericPointer(source, N - 1).Set(document, value); +} + +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const CharType(&source)[N], const typename DocumentType::Ch* value) { + return GenericPointer(source, N - 1).Set(document, value); +} + +#if RAPIDJSON_HAS_STDSTRING +template +typename DocumentType::ValueType& SetValueByPointer(DocumentType& document, const CharType(&source)[N], const std::basic_string& value) { + return GenericPointer(source, N - 1).Set(document, value); +} +#endif + +template +RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (typename DocumentType::ValueType&)) +SetValueByPointer(DocumentType& document, const CharType(&source)[N], T2 value) { + return GenericPointer(source, N - 1).Set(document, value); +} + +////////////////////////////////////////////////////////////////////////////// + +template +typename T::ValueType& SwapValueByPointer(T& root, const GenericPointer& pointer, typename T::ValueType& value, typename T::AllocatorType& a) { + return pointer.Swap(root, value, a); +} + +template +typename T::ValueType& SwapValueByPointer(T& root, const CharType(&source)[N], typename T::ValueType& value, typename T::AllocatorType& a) { + return GenericPointer(source, N - 1).Swap(root, value, a); +} + +template +typename DocumentType::ValueType& SwapValueByPointer(DocumentType& document, const GenericPointer& pointer, typename DocumentType::ValueType& value) { + return pointer.Swap(document, value); +} + +template +typename DocumentType::ValueType& SwapValueByPointer(DocumentType& document, const CharType(&source)[N], typename DocumentType::ValueType& value) { + return GenericPointer(source, N - 1).Swap(document, value); +} + +////////////////////////////////////////////////////////////////////////////// + +template +bool EraseValueByPointer(T& root, const GenericPointer& pointer) { + return pointer.Erase(root); +} + +template +bool EraseValueByPointer(T& root, const CharType(&source)[N]) { + return GenericPointer(source, N - 1).Erase(root); +} + +//@} + +RAPIDJSON_NAMESPACE_END + +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + +#ifdef _MSC_VER +RAPIDJSON_DIAG_POP +#endif + +#endif // RAPIDJSON_POINTER_H_ diff --git a/rapidjson/prettywriter.h b/rapidjson/prettywriter.h index 416dd49..98dfb30 100644 --- a/rapidjson/prettywriter.h +++ b/rapidjson/prettywriter.h @@ -22,8 +22,21 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) #endif +#if defined(__clang__) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif + RAPIDJSON_NAMESPACE_BEGIN +//! Combination of PrettyWriter format flags. +/*! \see PrettyWriter::SetFormatOptions + */ +enum PrettyFormatOptions { + kFormatDefault = 0, //!< Default pretty formatting. + kFormatSingleLineArray = 1 //!< Format arrays on a single line. +}; + //! Writer with indentation and spacing. /*! \tparam OutputStream Type of ouptut os. @@ -31,10 +44,10 @@ RAPIDJSON_NAMESPACE_BEGIN \tparam TargetEncoding Encoding of output stream. \tparam StackAllocator Type of allocator for allocating memory of stack. */ -template, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator> -class PrettyWriter : public Writer { +template, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator, unsigned writeFlags = kWriteDefaultFlags> +class PrettyWriter : public Writer { public: - typedef Writer Base; + typedef Writer Base; typedef typename Base::Ch Ch; //! Constructor @@ -42,8 +55,17 @@ class PrettyWriter : public Writer(rhs)), indentChar_(rhs.indentChar_), indentCharCount_(rhs.indentCharCount_), formatOptions_(rhs.formatOptions_) {} +#endif //! Set custom indentation. /*! \param indentChar Character for indentation. Must be whitespace character (' ', '\\t', '\\n', '\\r'). @@ -57,6 +79,14 @@ class PrettyWriter : public Writer& str) { + return Key(str.data(), SizeType(str.size())); + } +#endif bool EndObject(SizeType memberCount = 0) { (void)memberCount; - RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); - RAPIDJSON_ASSERT(!Base::level_stack_.template Top()->inArray); + RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); // not inside an Object + RAPIDJSON_ASSERT(!Base::level_stack_.template Top()->inArray); // currently inside an Array, not Object + RAPIDJSON_ASSERT(0 == Base::level_stack_.template Top()->valueCount % 2); // Object has a Key without a Value + bool empty = Base::level_stack_.template Pop(1)->valueCount == 0; if (!empty) { @@ -104,7 +150,7 @@ class PrettyWriter : public WriterFlush(); + Base::Flush(); return true; } @@ -120,7 +166,7 @@ class PrettyWriter : public Writer()->inArray); bool empty = Base::level_stack_.template Pop(1)->valueCount == 0; - if (!empty) { + if (!empty && !(formatOptions_ & kFormatSingleLineArray)) { Base::os_->Put('\n'); WriteIndent(); } @@ -128,7 +174,7 @@ class PrettyWriter : public WriterFlush(); + Base::Flush(); return true; } @@ -142,6 +188,22 @@ class PrettyWriter : public WriterinArray) { if (level->valueCount > 0) { Base::os_->Put(','); // add comma if it is not the first element in array - Base::os_->Put('\n'); + if (formatOptions_ & kFormatSingleLineArray) + Base::os_->Put(' '); } - else + + if (!(formatOptions_ & kFormatSingleLineArray)) { Base::os_->Put('\n'); - WriteIndent(); + WriteIndent(); + } } else { // in object if (level->valueCount > 0) { @@ -186,11 +251,12 @@ class PrettyWriter : public Writer(indentChar_), count); } Ch indentChar_; unsigned indentCharCount_; + PrettyFormatOptions formatOptions_; private: // Prohibit copy constructor & assignment operator. @@ -200,6 +266,10 @@ class PrettyWriter : public Writer.." string format. */ #define RAPIDJSON_MAJOR_VERSION 1 -#define RAPIDJSON_MINOR_VERSION 0 -#define RAPIDJSON_PATCH_VERSION 2 +#define RAPIDJSON_MINOR_VERSION 1 +#define RAPIDJSON_PATCH_VERSION 0 #define RAPIDJSON_VERSION_STRING \ RAPIDJSON_STRINGIFY(RAPIDJSON_MAJOR_VERSION.RAPIDJSON_MINOR_VERSION.RAPIDJSON_PATCH_VERSION) @@ -119,6 +124,31 @@ #define RAPIDJSON_NAMESPACE_END } #endif +/////////////////////////////////////////////////////////////////////////////// +// RAPIDJSON_HAS_STDSTRING + +#ifndef RAPIDJSON_HAS_STDSTRING +#ifdef RAPIDJSON_DOXYGEN_RUNNING +#define RAPIDJSON_HAS_STDSTRING 1 // force generation of documentation +#else +#define RAPIDJSON_HAS_STDSTRING 0 // no std::string support by default +#endif +/*! \def RAPIDJSON_HAS_STDSTRING + \ingroup RAPIDJSON_CONFIG + \brief Enable RapidJSON support for \c std::string + + By defining this preprocessor symbol to \c 1, several convenience functions for using + \ref rapidjson::GenericValue with \c std::string are enabled, especially + for construction and comparison. + + \hideinitializer +*/ +#endif // !defined(RAPIDJSON_HAS_STDSTRING) + +#if RAPIDJSON_HAS_STDSTRING +#include +#endif // RAPIDJSON_HAS_STDSTRING + /////////////////////////////////////////////////////////////////////////////// // RAPIDJSON_NO_INT64DEFINE @@ -134,7 +164,7 @@ */ #ifndef RAPIDJSON_NO_INT64DEFINE //!@cond RAPIDJSON_HIDDEN_FROM_DOXYGEN -#ifdef _MSC_VER +#if defined(_MSC_VER) && (_MSC_VER < 1800) // Visual Studio 2013 #include "msinttypes/stdint.h" #include "msinttypes/inttypes.h" #else @@ -153,9 +183,9 @@ #ifndef RAPIDJSON_FORCEINLINE //!@cond RAPIDJSON_HIDDEN_FROM_DOXYGEN -#if defined(_MSC_VER) && !defined(NDEBUG) +#if defined(_MSC_VER) && defined(NDEBUG) #define RAPIDJSON_FORCEINLINE __forceinline -#elif defined(__GNUC__) && __GNUC__ >= 4 && !defined(NDEBUG) +#elif defined(__GNUC__) && __GNUC__ >= 4 && defined(NDEBUG) #define RAPIDJSON_FORCEINLINE __attribute__((always_inline)) #else #define RAPIDJSON_FORCEINLINE @@ -211,6 +241,8 @@ # define RAPIDJSON_ENDIAN RAPIDJSON_BIGENDIAN # elif defined(__i386__) || defined(__alpha__) || defined(__ia64) || defined(__ia64__) || defined(_M_IX86) || defined(_M_IA64) || defined(_M_ALPHA) || defined(__amd64) || defined(__amd64__) || defined(_M_AMD64) || defined(__x86_64) || defined(__x86_64__) || defined(_M_X64) || defined(__bfin__) # define RAPIDJSON_ENDIAN RAPIDJSON_LITTLEENDIAN +# elif defined(_MSC_VER) && defined(_M_ARM) +# define RAPIDJSON_ENDIAN RAPIDJSON_LITTLEENDIAN # elif defined(RAPIDJSON_DOXYGEN_RUNNING) # define RAPIDJSON_ENDIAN # else @@ -223,7 +255,7 @@ //! Whether using 64-bit architecture #ifndef RAPIDJSON_64BIT -#if defined(__LP64__) || defined(_WIN64) +#if defined(__LP64__) || (defined(__x86_64__) && defined(__ILP32__)) || defined(_WIN64) || defined(__EMSCRIPTEN__) #define RAPIDJSON_64BIT 1 #else #define RAPIDJSON_64BIT 0 @@ -238,13 +270,14 @@ \param x pointer to align Some machines require strict data alignment. Currently the default uses 4 bytes - alignment. User can customize by defining the RAPIDJSON_ALIGN function macro., + alignment on 32-bit platforms and 8 bytes alignment for 64-bit platforms. + User can customize by defining the RAPIDJSON_ALIGN function macro. */ #ifndef RAPIDJSON_ALIGN #if RAPIDJSON_64BIT == 1 -#define RAPIDJSON_ALIGN(x) ((x + 7u) & ~7u) +#define RAPIDJSON_ALIGN(x) (((x) + static_cast(7u)) & ~static_cast(7u)) #else -#define RAPIDJSON_ALIGN(x) ((x + 3u) & ~3u) +#define RAPIDJSON_ALIGN(x) (((x) + 3u) & ~3u) #endif #endif @@ -262,17 +295,47 @@ #endif /////////////////////////////////////////////////////////////////////////////// -// RAPIDJSON_SSE2/RAPIDJSON_SSE42/RAPIDJSON_SIMD +// RAPIDJSON_48BITPOINTER_OPTIMIZATION + +//! Use only lower 48-bit address for some pointers. +/*! + \ingroup RAPIDJSON_CONFIG + + This optimization uses the fact that current X86-64 architecture only implement lower 48-bit virtual address. + The higher 16-bit can be used for storing other data. + \c GenericValue uses this optimization to reduce its size form 24 bytes to 16 bytes in 64-bit architecture. +*/ +#ifndef RAPIDJSON_48BITPOINTER_OPTIMIZATION +#if defined(__amd64__) || defined(__amd64) || defined(__x86_64__) || defined(__x86_64) || defined(_M_X64) || defined(_M_AMD64) +#define RAPIDJSON_48BITPOINTER_OPTIMIZATION 1 +#else +#define RAPIDJSON_48BITPOINTER_OPTIMIZATION 0 +#endif +#endif // RAPIDJSON_48BITPOINTER_OPTIMIZATION + +#if RAPIDJSON_48BITPOINTER_OPTIMIZATION == 1 +#if RAPIDJSON_64BIT != 1 +#error RAPIDJSON_48BITPOINTER_OPTIMIZATION can only be set to 1 when RAPIDJSON_64BIT=1 +#endif +#define RAPIDJSON_SETPOINTER(type, p, x) (p = reinterpret_cast((reinterpret_cast(p) & static_cast(RAPIDJSON_UINT64_C2(0xFFFF0000, 0x00000000))) | reinterpret_cast(reinterpret_cast(x)))) +#define RAPIDJSON_GETPOINTER(type, p) (reinterpret_cast(reinterpret_cast(p) & static_cast(RAPIDJSON_UINT64_C2(0x0000FFFF, 0xFFFFFFFF)))) +#else +#define RAPIDJSON_SETPOINTER(type, p, x) (p = (x)) +#define RAPIDJSON_GETPOINTER(type, p) (p) +#endif + +/////////////////////////////////////////////////////////////////////////////// +// RAPIDJSON_SSE2/RAPIDJSON_SSE42/RAPIDJSON_NEON/RAPIDJSON_SIMD /*! \def RAPIDJSON_SIMD \ingroup RAPIDJSON_CONFIG - \brief Enable SSE2/SSE4.2 optimization. + \brief Enable SSE2/SSE4.2/Neon optimization. RapidJSON supports optimized implementations for some parsing operations - based on the SSE2 or SSE4.2 SIMD extensions on modern Intel-compatible - processors. + based on the SSE2, SSE4.2 or NEon SIMD extensions on modern Intel + or ARM compatible processors. - To enable these optimizations, two different symbols can be defined; + To enable these optimizations, three different symbols can be defined; \code // Enable SSE2 optimization. #define RAPIDJSON_SSE2 @@ -281,13 +344,17 @@ #define RAPIDJSON_SSE42 \endcode - \c RAPIDJSON_SSE42 takes precedence, if both are defined. + // Enable ARM Neon optimization. + #define RAPIDJSON_NEON + \endcode + + \c RAPIDJSON_SSE42 takes precedence over SSE2, if both are defined. If any of these symbols is defined, RapidJSON defines the macro \c RAPIDJSON_SIMD to indicate the availability of the optimized code. */ #if defined(RAPIDJSON_SSE2) || defined(RAPIDJSON_SSE42) \ - || defined(RAPIDJSON_DOXYGEN_RUNNING) + || defined(RAPIDJSON_NEON) || defined(RAPIDJSON_DOXYGEN_RUNNING) #define RAPIDJSON_SIMD #endif @@ -347,25 +414,33 @@ RAPIDJSON_NAMESPACE_END /////////////////////////////////////////////////////////////////////////////// // RAPIDJSON_STATIC_ASSERT -// Adopt from boost +// Prefer C++11 static_assert, if available #ifndef RAPIDJSON_STATIC_ASSERT +#if __cplusplus >= 201103L || ( defined(_MSC_VER) && _MSC_VER >= 1800 ) +#define RAPIDJSON_STATIC_ASSERT(x) \ + static_assert(x, RAPIDJSON_STRINGIFY(x)) +#endif // C++11 +#endif // RAPIDJSON_STATIC_ASSERT + +// Adopt C++03 implementation from boost +#ifndef RAPIDJSON_STATIC_ASSERT +#ifndef __clang__ //!@cond RAPIDJSON_HIDDEN_FROM_DOXYGEN +#endif RAPIDJSON_NAMESPACE_BEGIN template struct STATIC_ASSERTION_FAILURE; template <> struct STATIC_ASSERTION_FAILURE { enum { value = 1 }; }; -template struct StaticAssertTest {}; +template struct StaticAssertTest {}; RAPIDJSON_NAMESPACE_END -#define RAPIDJSON_JOIN(X, Y) RAPIDJSON_DO_JOIN(X, Y) -#define RAPIDJSON_DO_JOIN(X, Y) RAPIDJSON_DO_JOIN2(X, Y) -#define RAPIDJSON_DO_JOIN2(X, Y) X##Y - #if defined(__GNUC__) #define RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE __attribute__((unused)) #else #define RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE #endif +#ifndef __clang__ //!@endcond +#endif /*! \def RAPIDJSON_STATIC_ASSERT \brief (Internal) macro to check for conditions at compile-time @@ -376,6 +451,35 @@ RAPIDJSON_NAMESPACE_END typedef ::RAPIDJSON_NAMESPACE::StaticAssertTest< \ sizeof(::RAPIDJSON_NAMESPACE::STATIC_ASSERTION_FAILURE)> \ RAPIDJSON_JOIN(StaticAssertTypedef, __LINE__) RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE +#endif // RAPIDJSON_STATIC_ASSERT + +/////////////////////////////////////////////////////////////////////////////// +// RAPIDJSON_LIKELY, RAPIDJSON_UNLIKELY + +//! Compiler branching hint for expression with high probability to be true. +/*! + \ingroup RAPIDJSON_CONFIG + \param x Boolean expression likely to be true. +*/ +#ifndef RAPIDJSON_LIKELY +#if defined(__GNUC__) || defined(__clang__) +#define RAPIDJSON_LIKELY(x) __builtin_expect(!!(x), 1) +#else +#define RAPIDJSON_LIKELY(x) (x) +#endif +#endif + +//! Compiler branching hint for expression with low probability to be true. +/*! + \ingroup RAPIDJSON_CONFIG + \param x Boolean expression unlikely to be true. +*/ +#ifndef RAPIDJSON_UNLIKELY +#if defined(__GNUC__) || defined(__clang__) +#define RAPIDJSON_UNLIKELY(x) __builtin_expect(!!(x), 0) +#else +#define RAPIDJSON_UNLIKELY(x) (x) +#endif #endif /////////////////////////////////////////////////////////////////////////////// @@ -438,8 +542,12 @@ RAPIDJSON_NAMESPACE_END #ifndef RAPIDJSON_HAS_CXX11_RVALUE_REFS #if defined(__clang__) -#define RAPIDJSON_HAS_CXX11_RVALUE_REFS __has_feature(cxx_rvalue_references) && \ +#if __has_feature(cxx_rvalue_references) && \ (defined(_LIBCPP_VERSION) || defined(__GLIBCXX__) && __GLIBCXX__ >= 20080306) +#define RAPIDJSON_HAS_CXX11_RVALUE_REFS 1 +#else +#define RAPIDJSON_HAS_CXX11_RVALUE_REFS 0 +#endif #elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,3,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ (defined(_MSC_VER) && _MSC_VER >= 1600) @@ -470,6 +578,17 @@ RAPIDJSON_NAMESPACE_END #define RAPIDJSON_HAS_CXX11_TYPETRAITS 0 #endif +#ifndef RAPIDJSON_HAS_CXX11_RANGE_FOR +#if defined(__clang__) +#define RAPIDJSON_HAS_CXX11_RANGE_FOR __has_feature(cxx_range_for) +#elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ + (defined(_MSC_VER) && _MSC_VER >= 1700) +#define RAPIDJSON_HAS_CXX11_RANGE_FOR 1 +#else +#define RAPIDJSON_HAS_CXX11_RANGE_FOR 0 +#endif +#endif // RAPIDJSON_HAS_CXX11_RANGE_FOR + //!@endcond /////////////////////////////////////////////////////////////////////////////// @@ -477,7 +596,7 @@ RAPIDJSON_NAMESPACE_END #ifndef RAPIDJSON_NEW ///! customization point for global \c new -#define RAPIDJSON_NEW(x) new x +#define RAPIDJSON_NEW(TypeName) new TypeName #endif #ifndef RAPIDJSON_DELETE ///! customization point for global \c delete @@ -485,10 +604,7 @@ RAPIDJSON_NAMESPACE_END #endif /////////////////////////////////////////////////////////////////////////////// -// Allocators and Encodings - -#include "allocators.h" -#include "encodings.h" +// Type /*! \namespace rapidjson \brief main RapidJSON namespace @@ -496,148 +612,6 @@ RAPIDJSON_NAMESPACE_END */ RAPIDJSON_NAMESPACE_BEGIN -/////////////////////////////////////////////////////////////////////////////// -// Stream - -/*! \class rapidjson::Stream - \brief Concept for reading and writing characters. - - For read-only stream, no need to implement PutBegin(), Put(), Flush() and PutEnd(). - - For write-only stream, only need to implement Put() and Flush(). - -\code -concept Stream { - typename Ch; //!< Character type of the stream. - - //! Read the current character from stream without moving the read cursor. - Ch Peek() const; - - //! Read the current character from stream and moving the read cursor to next character. - Ch Take(); - - //! Get the current read cursor. - //! \return Number of characters read from start. - size_t Tell(); - - //! Begin writing operation at the current read pointer. - //! \return The begin writer pointer. - Ch* PutBegin(); - - //! Write a character. - void Put(Ch c); - - //! Flush the buffer. - void Flush(); - - //! End the writing operation. - //! \param begin The begin write pointer returned by PutBegin(). - //! \return Number of characters written. - size_t PutEnd(Ch* begin); -} -\endcode -*/ - -//! Provides additional information for stream. -/*! - By using traits pattern, this type provides a default configuration for stream. - For custom stream, this type can be specialized for other configuration. - See TEST(Reader, CustomStringStream) in readertest.cpp for example. -*/ -template -struct StreamTraits { - //! Whether to make local copy of stream for optimization during parsing. - /*! - By default, for safety, streams do not use local copy optimization. - Stream that can be copied fast should specialize this, like StreamTraits. - */ - enum { copyOptimization = 0 }; -}; - -//! Put N copies of a character to a stream. -template -inline void PutN(Stream& stream, Ch c, size_t n) { - for (size_t i = 0; i < n; i++) - stream.Put(c); -} - -/////////////////////////////////////////////////////////////////////////////// -// StringStream - -//! Read-only string stream. -/*! \note implements Stream concept -*/ -template -struct GenericStringStream { - typedef typename Encoding::Ch Ch; - - GenericStringStream(const Ch *src) : src_(src), head_(src) {} - - Ch Peek() const { return *src_; } - Ch Take() { return *src_++; } - size_t Tell() const { return static_cast(src_ - head_); } - - Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } - void Put(Ch) { RAPIDJSON_ASSERT(false); } - void Flush() { RAPIDJSON_ASSERT(false); } - size_t PutEnd(Ch*) { RAPIDJSON_ASSERT(false); return 0; } - - const Ch* src_; //!< Current read position. - const Ch* head_; //!< Original head of the string. -}; - -template -struct StreamTraits > { - enum { copyOptimization = 1 }; -}; - -//! String stream with UTF8 encoding. -typedef GenericStringStream > StringStream; - -/////////////////////////////////////////////////////////////////////////////// -// InsituStringStream - -//! A read-write string stream. -/*! This string stream is particularly designed for in-situ parsing. - \note implements Stream concept -*/ -template -struct GenericInsituStringStream { - typedef typename Encoding::Ch Ch; - - GenericInsituStringStream(Ch *src) : src_(src), dst_(0), head_(src) {} - - // Read - Ch Peek() { return *src_; } - Ch Take() { return *src_++; } - size_t Tell() { return static_cast(src_ - head_); } - - // Write - void Put(Ch c) { RAPIDJSON_ASSERT(dst_ != 0); *dst_++ = c; } - - Ch* PutBegin() { return dst_ = src_; } - size_t PutEnd(Ch* begin) { return static_cast(dst_ - begin); } - void Flush() {} - - Ch* Push(size_t count) { Ch* begin = dst_; dst_ += count; return begin; } - void Pop(size_t count) { dst_ -= count; } - - Ch* src_; - Ch* dst_; - Ch* head_; -}; - -template -struct StreamTraits > { - enum { copyOptimization = 1 }; -}; - -//! Insitu string stream with UTF8 encoding. -typedef GenericInsituStringStream > InsituStringStream; - -/////////////////////////////////////////////////////////////////////////////// -// Type - //! Type of JSON value enum Type { kNullType = 0, //!< null diff --git a/rapidjson/reader.h b/rapidjson/reader.h index c5ecf4b..120c311 100644 --- a/rapidjson/reader.h +++ b/rapidjson/reader.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #ifndef RAPIDJSON_READER_H_ @@ -17,11 +17,13 @@ /*! \file reader.h */ -#include "rapidjson.h" -#include "encodings.h" +#include "allocators.h" +#include "stream.h" +#include "encodedstream.h" #include "internal/meta.h" #include "internal/stack.h" #include "internal/strtod.h" +#include #if defined(RAPIDJSON_SIMD) && defined(_MSC_VER) #include @@ -31,6 +33,8 @@ #include #elif defined(RAPIDJSON_SSE2) #include +#elif defined(RAPIDJSON_NEON) +#include #endif #ifdef _MSC_VER @@ -39,6 +43,13 @@ RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant RAPIDJSON_DIAG_OFF(4702) // unreachable code #endif +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(old-style-cast) +RAPIDJSON_DIAG_OFF(padded) +RAPIDJSON_DIAG_OFF(switch-enum) +#endif + #ifdef __GNUC__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) @@ -49,7 +60,7 @@ RAPIDJSON_DIAG_OFF(effc++) #ifndef RAPIDJSON_PARSE_ERROR_EARLY_RETURN #define RAPIDJSON_PARSE_ERROR_EARLY_RETURN(value) \ RAPIDJSON_MULTILINEMACRO_BEGIN \ - if (HasParseError()) { return value; } \ + if (RAPIDJSON_UNLIKELY(HasParseError())) { return value; } \ RAPIDJSON_MULTILINEMACRO_END #endif #define RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID \ @@ -120,7 +131,7 @@ RAPIDJSON_NAMESPACE_BEGIN /////////////////////////////////////////////////////////////////////////////// // ParseFlag -/*! \def RAPIDJSON_PARSE_DEFAULT_FLAGS +/*! \def RAPIDJSON_PARSE_DEFAULT_FLAGS \ingroup RAPIDJSON_CONFIG \brief User-defined kParseDefaultFlags definition. @@ -140,6 +151,10 @@ enum ParseFlag { kParseIterativeFlag = 4, //!< Iterative(constant complexity in terms of function call stack size) parsing. kParseStopWhenDoneFlag = 8, //!< After parsing a complete JSON root from stream, stop further processing the rest of stream. When this flag is used, parser will not generate kParseErrorDocumentRootNotSingular error. kParseFullPrecisionFlag = 16, //!< Parse number in full precision (but slower). + kParseCommentsFlag = 32, //!< Allow one-line (//) and multi-line (/**/) comments. + kParseNumbersAsStringsFlag = 64, //!< Parse all numbers (ints/doubles) as strings. + kParseTrailingCommasFlag = 128, //!< Allow trailing commas at the end of objects and arrays. + kParseNanAndInfFlag = 256, //!< Allow parsing NaN, Inf, Infinity, -Inf and -Infinity as doubles. kParseDefaultFlags = RAPIDJSON_PARSE_DEFAULT_FLAGS //!< Default parse flags. Can be customized by defining RAPIDJSON_PARSE_DEFAULT_FLAGS }; @@ -148,7 +163,7 @@ enum ParseFlag { /*! \class rapidjson::Handler \brief Concept for receiving events from GenericReader upon parsing. - The functions return true if no error occurs. If they return false, + The functions return true if no error occurs. If they return false, the event publisher should terminate the process. \code concept Handler { @@ -161,6 +176,8 @@ concept Handler { bool Int64(int64_t i); bool Uint64(uint64_t i); bool Double(double d); + /// enabled via kParseNumbersAsStringsFlag, string is not null-terminated (use length) + bool RawNumber(const Ch* str, SizeType length, bool copy); bool String(const Ch* str, SizeType length, bool copy); bool StartObject(); bool Key(const Ch* str, SizeType length, bool copy); @@ -191,6 +208,8 @@ struct BaseReaderHandler { bool Int64(int64_t) { return static_cast(*this).Default(); } bool Uint64(uint64_t) { return static_cast(*this).Default(); } bool Double(double) { return static_cast(*this).Default(); } + /// enabled via kParseNumbersAsStringsFlag, string is not null-terminated (use length) + bool RawNumber(const Ch* str, SizeType len, bool copy) { return static_cast(*this).String(str, len, copy); } bool String(const Ch*, SizeType, bool) { return static_cast(*this).Default(); } bool StartObject() { return static_cast(*this).Default(); } bool Key(const Ch* str, SizeType len, bool copy) { return static_cast(*this).String(str, len, copy); } @@ -248,10 +267,17 @@ void SkipWhitespace(InputStream& is) { internal::StreamLocalCopy copy(is); InputStream& s(copy.s); - while (s.Peek() == ' ' || s.Peek() == '\n' || s.Peek() == '\r' || s.Peek() == '\t') + typename InputStream::Ch c; + while ((c = s.Peek()) == ' ' || c == '\n' || c == '\r' || c == '\t') s.Take(); } +inline const char* SkipWhitespace(const char* p, const char* end) { + while (p != end && (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t')) + ++p; + return p; +} + #ifdef RAPIDJSON_SSE42 //! Skip whitespace with SSE 4.2 pcmpistrm instruction, testing 16 8-byte characters at once. inline const char *SkipWhitespace_SIMD(const char* p) { @@ -262,7 +288,7 @@ inline const char *SkipWhitespace_SIMD(const char* p) { return p; // 16-byte align to the next boundary - const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & ~15); + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); while (p != nextAligned) if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') ++p; @@ -271,23 +297,37 @@ inline const char *SkipWhitespace_SIMD(const char* p) { // The rest of string using SIMD static const char whitespace[16] = " \n\r\t"; - const __m128i w = _mm_load_si128((const __m128i *)&whitespace[0]); + const __m128i w = _mm_loadu_si128(reinterpret_cast(&whitespace[0])); for (;; p += 16) { - const __m128i s = _mm_load_si128((const __m128i *)p); - const unsigned r = _mm_cvtsi128_si32(_mm_cmpistrm(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK | _SIDD_NEGATIVE_POLARITY)); - if (r != 0) { // some of characters is non-whitespace -#ifdef _MSC_VER // Find the index of first non-whitespace - unsigned long offset; - _BitScanForward(&offset, r); - return p + offset; -#else - return p + __builtin_ffs(r) - 1; -#endif - } + const __m128i s = _mm_load_si128(reinterpret_cast(p)); + const int r = _mm_cmpistri(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT | _SIDD_NEGATIVE_POLARITY); + if (r != 16) // some of characters is non-whitespace + return p + r; } } +inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { + // Fast return for single non-whitespace + if (p != end && (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t')) + ++p; + else + return p; + + // The middle of string using SIMD + static const char whitespace[16] = " \n\r\t"; + const __m128i w = _mm_loadu_si128(reinterpret_cast(&whitespace[0])); + + for (; p <= end - 16; p += 16) { + const __m128i s = _mm_loadu_si128(reinterpret_cast(p)); + const int r = _mm_cmpistri(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT | _SIDD_NEGATIVE_POLARITY); + if (r != 16) // some of characters is non-whitespace + return p + r; + } + + return SkipWhitespace(p, end); +} + #elif defined(RAPIDJSON_SSE2) //! Skip whitespace with SSE2 instructions, testing 16 8-byte characters at once. @@ -299,7 +339,7 @@ inline const char *SkipWhitespace_SIMD(const char* p) { return p; // 16-byte align to the next boundary - const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & ~15); + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); while (p != nextAligned) if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') ++p; @@ -307,24 +347,58 @@ inline const char *SkipWhitespace_SIMD(const char* p) { return p; // The rest of string - static const char whitespaces[4][17] = { - " ", - "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n", - "\r\r\r\r\r\r\r\r\r\r\r\r\r\r\r\r", - "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t"}; + #define C16(c) { c, c, c, c, c, c, c, c, c, c, c, c, c, c, c, c } + static const char whitespaces[4][16] = { C16(' '), C16('\n'), C16('\r'), C16('\t') }; + #undef C16 - const __m128i w0 = _mm_loadu_si128((const __m128i *)&whitespaces[0][0]); - const __m128i w1 = _mm_loadu_si128((const __m128i *)&whitespaces[1][0]); - const __m128i w2 = _mm_loadu_si128((const __m128i *)&whitespaces[2][0]); - const __m128i w3 = _mm_loadu_si128((const __m128i *)&whitespaces[3][0]); + const __m128i w0 = _mm_loadu_si128(reinterpret_cast(&whitespaces[0][0])); + const __m128i w1 = _mm_loadu_si128(reinterpret_cast(&whitespaces[1][0])); + const __m128i w2 = _mm_loadu_si128(reinterpret_cast(&whitespaces[2][0])); + const __m128i w3 = _mm_loadu_si128(reinterpret_cast(&whitespaces[3][0])); for (;; p += 16) { - const __m128i s = _mm_load_si128((const __m128i *)p); + const __m128i s = _mm_load_si128(reinterpret_cast(p)); + __m128i x = _mm_cmpeq_epi8(s, w0); + x = _mm_or_si128(x, _mm_cmpeq_epi8(s, w1)); + x = _mm_or_si128(x, _mm_cmpeq_epi8(s, w2)); + x = _mm_or_si128(x, _mm_cmpeq_epi8(s, w3)); + unsigned short r = static_cast(~_mm_movemask_epi8(x)); + if (r != 0) { // some of characters may be non-whitespace +#ifdef _MSC_VER // Find the index of first non-whitespace + unsigned long offset; + _BitScanForward(&offset, r); + return p + offset; +#else + return p + __builtin_ffs(r) - 1; +#endif + } + } +} + +inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { + // Fast return for single non-whitespace + if (p != end && (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t')) + ++p; + else + return p; + + // The rest of string + #define C16(c) { c, c, c, c, c, c, c, c, c, c, c, c, c, c, c, c } + static const char whitespaces[4][16] = { C16(' '), C16('\n'), C16('\r'), C16('\t') }; + #undef C16 + + const __m128i w0 = _mm_loadu_si128(reinterpret_cast(&whitespaces[0][0])); + const __m128i w1 = _mm_loadu_si128(reinterpret_cast(&whitespaces[1][0])); + const __m128i w2 = _mm_loadu_si128(reinterpret_cast(&whitespaces[2][0])); + const __m128i w3 = _mm_loadu_si128(reinterpret_cast(&whitespaces[3][0])); + + for (; p <= end - 16; p += 16) { + const __m128i s = _mm_loadu_si128(reinterpret_cast(p)); __m128i x = _mm_cmpeq_epi8(s, w0); x = _mm_or_si128(x, _mm_cmpeq_epi8(s, w1)); x = _mm_or_si128(x, _mm_cmpeq_epi8(s, w2)); x = _mm_or_si128(x, _mm_cmpeq_epi8(s, w3)); - unsigned short r = (unsigned short)~_mm_movemask_epi8(x); + unsigned short r = static_cast(~_mm_movemask_epi8(x)); if (r != 0) { // some of characters may be non-whitespace #ifdef _MSC_VER // Find the index of first non-whitespace unsigned long offset; @@ -335,13 +409,100 @@ inline const char *SkipWhitespace_SIMD(const char* p) { #endif } } + + return SkipWhitespace(p, end); +} + +#elif defined(RAPIDJSON_NEON) + +//! Skip whitespace with ARM Neon instructions, testing 16 8-byte characters at once. +inline const char *SkipWhitespace_SIMD(const char* p) { + // Fast return for single non-whitespace + if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') + ++p; + else + return p; + + // 16-byte align to the next boundary + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') + ++p; + else + return p; + + const uint8x16_t w0 = vmovq_n_u8(' '); + const uint8x16_t w1 = vmovq_n_u8('\n'); + const uint8x16_t w2 = vmovq_n_u8('\r'); + const uint8x16_t w3 = vmovq_n_u8('\t'); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, w0); + x = vorrq_u8(x, vceqq_u8(s, w1)); + x = vorrq_u8(x, vceqq_u8(s, w2)); + x = vorrq_u8(x, vceqq_u8(s, w3)); + + x = vmvnq_u8(x); // Negate + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz =__builtin_clzll(high);; + return p + 8 + (lz >> 3); + } + } else { + int lz = __builtin_clzll(low);; + return p + (lz >> 3); + } + } +} + +inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { + // Fast return for single non-whitespace + if (p != end && (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t')) + ++p; + else + return p; + + const uint8x16_t w0 = vmovq_n_u8(' '); + const uint8x16_t w1 = vmovq_n_u8('\n'); + const uint8x16_t w2 = vmovq_n_u8('\r'); + const uint8x16_t w3 = vmovq_n_u8('\t'); + + for (; p <= end - 16; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, w0); + x = vorrq_u8(x, vceqq_u8(s, w1)); + x = vorrq_u8(x, vceqq_u8(s, w2)); + x = vorrq_u8(x, vceqq_u8(s, w3)); + + x = vmvnq_u8(x); // Negate + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz = __builtin_clzll(high); + return p + 8 + (lz >> 3); + } + } else { + int lz = __builtin_clzll(low); + return p + (lz >> 3); + } + } + + return SkipWhitespace(p, end); } -#endif // RAPIDJSON_SSE2 +#endif // RAPIDJSON_NEON #ifdef RAPIDJSON_SIMD //! Template function specialization for InsituStringStream -template<> inline void SkipWhitespace(InsituStringStream& is) { +template<> inline void SkipWhitespace(InsituStringStream& is) { is.src_ = const_cast(SkipWhitespace_SIMD(is.src_)); } @@ -349,23 +510,27 @@ template<> inline void SkipWhitespace(InsituStringStream& is) { template<> inline void SkipWhitespace(StringStream& is) { is.src_ = SkipWhitespace_SIMD(is.src_); } + +template<> inline void SkipWhitespace(EncodedInputStream, MemoryStream>& is) { + is.is_.src_ = SkipWhitespace_SIMD(is.is_.src_, is.is_.end_); +} #endif // RAPIDJSON_SIMD /////////////////////////////////////////////////////////////////////////////// // GenericReader //! SAX-style JSON parser. Use \ref Reader for UTF8 encoding and default allocator. -/*! GenericReader parses JSON text from a stream, and send events synchronously to an +/*! GenericReader parses JSON text from a stream, and send events synchronously to an object implementing Handler concept. - It needs to allocate a stack for storing a single decoded string during + It needs to allocate a stack for storing a single decoded string during non-destructive parsing. - For in-situ parsing, the decoded string is directly written to the source + For in-situ parsing, the decoded string is directly written to the source text string, no temporary buffer is required. A GenericReader object can be reused for parsing multiple JSON text. - + \tparam SourceEncoding Encoding of the input stream. \tparam TargetEncoding Encoding of the parse output. \tparam StackAllocator Allocator type for stack. @@ -398,9 +563,10 @@ class GenericReader { ClearStackOnExit scope(*this); - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); - if (is.Peek() == '\0') { + if (RAPIDJSON_UNLIKELY(is.Peek() == '\0')) { RAPIDJSON_PARSE_ERROR_NORETURN(kParseErrorDocumentEmpty, is.Tell()); RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); } @@ -409,9 +575,10 @@ class GenericReader { RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); if (!(parseFlags & kParseStopWhenDoneFlag)) { - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); - if (is.Peek() != '\0') { + if (RAPIDJSON_UNLIKELY(is.Peek() != '\0')) { RAPIDJSON_PARSE_ERROR_NORETURN(kParseErrorDocumentRootNotSingular, is.Tell()); RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); } @@ -433,9 +600,86 @@ class GenericReader { return Parse(is, handler); } + //! Initialize JSON text token-by-token parsing + /*! + */ + void IterativeParseInit() { + parseResult_.Clear(); + state_ = IterativeParsingStartState; + } + + //! Parse one token from JSON text + /*! \tparam InputStream Type of input stream, implementing Stream concept + \tparam Handler Type of handler, implementing Handler concept. + \param is Input stream to be parsed. + \param handler The handler to receive events. + \return Whether the parsing is successful. + */ + template + bool IterativeParseNext(InputStream& is, Handler& handler) { + while (RAPIDJSON_LIKELY(is.Peek() != '\0')) { + SkipWhitespaceAndComments(is); + + Token t = Tokenize(is.Peek()); + IterativeParsingState n = Predict(state_, t); + IterativeParsingState d = Transit(state_, t, n, is, handler); + + // If we've finished or hit an error... + if (RAPIDJSON_UNLIKELY(IsIterativeParsingCompleteState(d))) { + // Report errors. + if (d == IterativeParsingErrorState) { + HandleError(state_, is); + return false; + } + + // Transition to the finish state. + RAPIDJSON_ASSERT(d == IterativeParsingFinishState); + state_ = d; + + // If StopWhenDone is not set... + if (!(parseFlags & kParseStopWhenDoneFlag)) { + // ... and extra non-whitespace data is found... + SkipWhitespaceAndComments(is); + if (is.Peek() != '\0') { + // ... this is considered an error. + HandleError(state_, is); + return false; + } + } + + // Success! We are done! + return true; + } + + // Transition to the new state. + state_ = d; + + // If we parsed anything other than a delimiter, we invoked the handler, so we can return true now. + if (!IsIterativeParsingDelimiterState(n)) + return true; + } + + // We reached the end of file. + stack_.Clear(); + + if (state_ != IterativeParsingFinishState) { + HandleError(state_, is); + return false; + } + + return true; + } + + //! Check if token-by-token parsing JSON text is complete + /*! \return Whether the JSON has been fully decoded. + */ + RAPIDJSON_FORCEINLINE bool IterativeParseComplete() { + return IsIterativeParsingCompleteState(state_); + } + //! Whether a parse error has occured in the last parsing. bool HasParseError() const { return parseResult_.IsError(); } - + //! Get the \ref ParseErrorCode of last parsing. ParseErrorCode GetParseErrorCode() const { return parseResult_.Code(); } @@ -462,52 +706,98 @@ class GenericReader { ClearStackOnExit& operator=(const ClearStackOnExit&); }; + template + void SkipWhitespaceAndComments(InputStream& is) { + SkipWhitespace(is); + + if (parseFlags & kParseCommentsFlag) { + while (RAPIDJSON_UNLIKELY(Consume(is, '/'))) { + if (Consume(is, '*')) { + while (true) { + if (RAPIDJSON_UNLIKELY(is.Peek() == '\0')) + RAPIDJSON_PARSE_ERROR(kParseErrorUnspecificSyntaxError, is.Tell()); + else if (Consume(is, '*')) { + if (Consume(is, '/')) + break; + } + else + is.Take(); + } + } + else if (RAPIDJSON_LIKELY(Consume(is, '/'))) + while (is.Peek() != '\0' && is.Take() != '\n') {} + else + RAPIDJSON_PARSE_ERROR(kParseErrorUnspecificSyntaxError, is.Tell()); + + SkipWhitespace(is); + } + } + } + // Parse object: { string : value, ... } template void ParseObject(InputStream& is, Handler& handler) { RAPIDJSON_ASSERT(is.Peek() == '{'); is.Take(); // Skip '{' - - if (!handler.StartObject()) + + if (RAPIDJSON_UNLIKELY(!handler.StartObject())) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - if (is.Peek() == '}') { - is.Take(); - if (!handler.EndObject(0)) // empty object + if (Consume(is, '}')) { + if (RAPIDJSON_UNLIKELY(!handler.EndObject(0))) // empty object RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); return; } for (SizeType memberCount = 0;;) { - if (is.Peek() != '"') + if (RAPIDJSON_UNLIKELY(is.Peek() != '"')) RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissName, is.Tell()); ParseString(is, handler, true); RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - if (is.Take() != ':') + if (RAPIDJSON_UNLIKELY(!Consume(is, ':'))) RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissColon, is.Tell()); - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; ParseValue(is, handler); RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; ++memberCount; - switch (is.Take()) { - case ',': SkipWhitespace(is); break; - case '}': - if (!handler.EndObject(memberCount)) + switch (is.Peek()) { + case ',': + is.Take(); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; + break; + case '}': + is.Take(); + if (RAPIDJSON_UNLIKELY(!handler.EndObject(memberCount))) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); return; - default: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissCommaOrCurlyBracket, is.Tell()); + default: + RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissCommaOrCurlyBracket, is.Tell()); break; // This useless break is only for making warning and coverage happy + } + + if (parseFlags & kParseTrailingCommasFlag) { + if (is.Peek() == '}') { + if (RAPIDJSON_UNLIKELY(!handler.EndObject(memberCount))) + RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); + is.Take(); + return; + } } } } @@ -517,15 +807,15 @@ class GenericReader { void ParseArray(InputStream& is, Handler& handler) { RAPIDJSON_ASSERT(is.Peek() == '['); is.Take(); // Skip '[' - - if (!handler.StartArray()) + + if (RAPIDJSON_UNLIKELY(!handler.StartArray())) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); - - SkipWhitespace(is); - if (is.Peek() == ']') { - is.Take(); - if (!handler.EndArray(0)) // empty array + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; + + if (Consume(is, ']')) { + if (RAPIDJSON_UNLIKELY(!handler.EndArray(0))) // empty array RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); return; } @@ -535,15 +825,28 @@ class GenericReader { RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; ++elementCount; - SkipWhitespace(is); + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - switch (is.Take()) { - case ',': SkipWhitespace(is); break; - case ']': - if (!handler.EndArray(elementCount)) + if (Consume(is, ',')) { + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; + } + else if (Consume(is, ']')) { + if (RAPIDJSON_UNLIKELY(!handler.EndArray(elementCount))) + RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); + return; + } + else + RAPIDJSON_PARSE_ERROR(kParseErrorArrayMissCommaOrSquareBracket, is.Tell()); + + if (parseFlags & kParseTrailingCommasFlag) { + if (is.Peek() == ']') { + if (RAPIDJSON_UNLIKELY(!handler.EndArray(elementCount))) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); + is.Take(); return; - default: RAPIDJSON_PARSE_ERROR(kParseErrorArrayMissCommaOrSquareBracket, is.Tell()); + } } } } @@ -553,12 +856,12 @@ class GenericReader { RAPIDJSON_ASSERT(is.Peek() == 'n'); is.Take(); - if (is.Take() == 'u' && is.Take() == 'l' && is.Take() == 'l') { - if (!handler.Null()) + if (RAPIDJSON_LIKELY(Consume(is, 'u') && Consume(is, 'l') && Consume(is, 'l'))) { + if (RAPIDJSON_UNLIKELY(!handler.Null())) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); } else - RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell() - 1); + RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell()); } template @@ -566,12 +869,12 @@ class GenericReader { RAPIDJSON_ASSERT(is.Peek() == 't'); is.Take(); - if (is.Take() == 'r' && is.Take() == 'u' && is.Take() == 'e') { - if (!handler.Bool(true)) + if (RAPIDJSON_LIKELY(Consume(is, 'r') && Consume(is, 'u') && Consume(is, 'e'))) { + if (RAPIDJSON_UNLIKELY(!handler.Bool(true))) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); } else - RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell() - 1); + RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell()); } template @@ -579,20 +882,30 @@ class GenericReader { RAPIDJSON_ASSERT(is.Peek() == 'f'); is.Take(); - if (is.Take() == 'a' && is.Take() == 'l' && is.Take() == 's' && is.Take() == 'e') { - if (!handler.Bool(false)) + if (RAPIDJSON_LIKELY(Consume(is, 'a') && Consume(is, 'l') && Consume(is, 's') && Consume(is, 'e'))) { + if (RAPIDJSON_UNLIKELY(!handler.Bool(false))) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, is.Tell()); } else - RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell() - 1); + RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell()); + } + + template + RAPIDJSON_FORCEINLINE static bool Consume(InputStream& is, typename InputStream::Ch expect) { + if (RAPIDJSON_LIKELY(is.Peek() == expect)) { + is.Take(); + return true; + } + else + return false; } // Helper function to parse four hexidecimal digits in \uXXXX in ParseString(). template - unsigned ParseHex4(InputStream& is) { + unsigned ParseHex4(InputStream& is, size_t escapeOffset) { unsigned codepoint = 0; for (int i = 0; i < 4; i++) { - Ch c = is.Take(); + Ch c = is.Peek(); codepoint <<= 4; codepoint += static_cast(c); if (c >= '0' && c <= '9') @@ -602,9 +915,10 @@ class GenericReader { else if (c >= 'a' && c <= 'f') codepoint -= 'a' - 10; else { - RAPIDJSON_PARSE_ERROR_NORETURN(kParseErrorStringUnicodeEscapeInvalidHex, is.Tell() - 1); + RAPIDJSON_PARSE_ERROR_NORETURN(kParseErrorStringUnicodeEscapeInvalidHex, escapeOffset); RAPIDJSON_PARSE_ERROR_EARLY_RETURN(0); } + is.Take(); } return codepoint; } @@ -619,7 +933,14 @@ class GenericReader { *stack_.template Push() = c; ++length_; } + + RAPIDJSON_FORCEINLINE void* Push(SizeType count) { + length_ += count; + return stack_.template Push(count); + } + size_t Length() const { return length_; } + Ch* Pop() { return stack_.template Pop(length_); } @@ -638,6 +959,9 @@ class GenericReader { internal::StreamLocalCopy copy(is); InputStream& s(copy.s); + RAPIDJSON_ASSERT(s.Peek() == '\"'); + s.Take(); // Skip '\"' + bool success = false; if (parseFlags & kParseInsituFlag) { typename InputStream::Ch *head = s.PutBegin(); @@ -645,7 +969,7 @@ class GenericReader { RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; size_t length = s.PutEnd(head) - 1; RAPIDJSON_ASSERT(length <= 0xFFFFFFFF); - const typename TargetEncoding::Ch* const str = (typename TargetEncoding::Ch*)head; + const typename TargetEncoding::Ch* const str = reinterpret_cast(head); success = (isKey ? handler.Key(str, SizeType(length), false) : handler.String(str, SizeType(length), false)); } else { @@ -656,7 +980,7 @@ class GenericReader { const typename TargetEncoding::Ch* const str = stackStream.Pop(); success = (isKey ? handler.Key(str, length, true) : handler.String(str, length, true)); } - if (!success) + if (RAPIDJSON_UNLIKELY(!success)) RAPIDJSON_PARSE_ERROR(kParseErrorTermination, s.Tell()); } @@ -667,74 +991,421 @@ class GenericReader { //!@cond RAPIDJSON_HIDDEN_FROM_DOXYGEN #define Z16 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 static const char escape[256] = { - Z16, Z16, 0, 0,'\"', 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,'/', - Z16, Z16, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,'\\', 0, 0, 0, - 0, 0,'\b', 0, 0, 0,'\f', 0, 0, 0, 0, 0, 0, 0,'\n', 0, - 0, 0,'\r', 0,'\t', 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + Z16, Z16, 0, 0,'\"', 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,'/', + Z16, Z16, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,'\\', 0, 0, 0, + 0, 0,'\b', 0, 0, 0,'\f', 0, 0, 0, 0, 0, 0, 0,'\n', 0, + 0, 0,'\r', 0,'\t', 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, Z16, Z16, Z16, Z16, Z16, Z16, Z16, Z16 }; #undef Z16 //!@endcond - RAPIDJSON_ASSERT(is.Peek() == '\"'); - is.Take(); // Skip '\"' - for (;;) { + // Scan and copy string before "\\\"" or < 0x20. This is an optional optimzation. + if (!(parseFlags & kParseValidateEncodingFlag)) + ScanCopyUnescapedString(is, os); + Ch c = is.Peek(); - if (c == '\\') { // Escape + if (RAPIDJSON_UNLIKELY(c == '\\')) { // Escape + size_t escapeOffset = is.Tell(); // For invalid escaping, report the inital '\\' as error offset is.Take(); - Ch e = is.Take(); - if ((sizeof(Ch) == 1 || unsigned(e) < 256) && escape[(unsigned char)e]) { - os.Put(escape[(unsigned char)e]); + Ch e = is.Peek(); + if ((sizeof(Ch) == 1 || unsigned(e) < 256) && RAPIDJSON_LIKELY(escape[static_cast(e)])) { + is.Take(); + os.Put(static_cast(escape[static_cast(e)])); } - else if (e == 'u') { // Unicode - unsigned codepoint = ParseHex4(is); + else if (RAPIDJSON_LIKELY(e == 'u')) { // Unicode + is.Take(); + unsigned codepoint = ParseHex4(is, escapeOffset); RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - if (codepoint >= 0xD800 && codepoint <= 0xDBFF) { + if (RAPIDJSON_UNLIKELY(codepoint >= 0xD800 && codepoint <= 0xDBFF)) { // Handle UTF-16 surrogate pair - if (is.Take() != '\\' || is.Take() != 'u') - RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, is.Tell() - 2); - unsigned codepoint2 = ParseHex4(is); + if (RAPIDJSON_UNLIKELY(!Consume(is, '\\') || !Consume(is, 'u'))) + RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset); + unsigned codepoint2 = ParseHex4(is, escapeOffset); RAPIDJSON_PARSE_ERROR_EARLY_RETURN_VOID; - if (codepoint2 < 0xDC00 || codepoint2 > 0xDFFF) - RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, is.Tell() - 2); + if (RAPIDJSON_UNLIKELY(codepoint2 < 0xDC00 || codepoint2 > 0xDFFF)) + RAPIDJSON_PARSE_ERROR(kParseErrorStringUnicodeSurrogateInvalid, escapeOffset); codepoint = (((codepoint - 0xD800) << 10) | (codepoint2 - 0xDC00)) + 0x10000; } TEncoding::Encode(os, codepoint); } else - RAPIDJSON_PARSE_ERROR(kParseErrorStringEscapeInvalid, is.Tell() - 1); + RAPIDJSON_PARSE_ERROR(kParseErrorStringEscapeInvalid, escapeOffset); } - else if (c == '"') { // Closing double quote + else if (RAPIDJSON_UNLIKELY(c == '"')) { // Closing double quote is.Take(); os.Put('\0'); // null-terminate the string return; } - else if (c == '\0') - RAPIDJSON_PARSE_ERROR(kParseErrorStringMissQuotationMark, is.Tell() - 1); - else if ((unsigned)c < 0x20) // RFC 4627: unescaped = %x20-21 / %x23-5B / %x5D-10FFFF - RAPIDJSON_PARSE_ERROR(kParseErrorStringEscapeInvalid, is.Tell() - 1); - else { - if (parseFlags & kParseValidateEncodingFlag ? - !Transcoder::Validate(is, os) : - !Transcoder::Transcode(is, os)) + else if (RAPIDJSON_UNLIKELY(static_cast(c) < 0x20)) { // RFC 4627: unescaped = %x20-21 / %x23-5B / %x5D-10FFFF + if (c == '\0') + RAPIDJSON_PARSE_ERROR(kParseErrorStringMissQuotationMark, is.Tell()); + else RAPIDJSON_PARSE_ERROR(kParseErrorStringInvalidEncoding, is.Tell()); } + else { + size_t offset = is.Tell(); + if (RAPIDJSON_UNLIKELY((parseFlags & kParseValidateEncodingFlag ? + !Transcoder::Validate(is, os) : + !Transcoder::Transcode(is, os)))) + RAPIDJSON_PARSE_ERROR(kParseErrorStringInvalidEncoding, offset); + } } } - template + template + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(InputStream&, OutputStream&) { + // Do nothing for generic version + } + +#if defined(RAPIDJSON_SSE2) || defined(RAPIDJSON_SSE42) + // StringStream -> StackStream + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(StringStream& is, StackStream& os) { + const char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + return; + } + else + os.Put(*p++); + + // The rest of string using SIMD + static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; + static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; + const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); + const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); + const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); + + for (;; p += 16) { + const __m128i s = _mm_load_si128(reinterpret_cast(p)); + const __m128i t1 = _mm_cmpeq_epi8(s, dq); + const __m128i t2 = _mm_cmpeq_epi8(s, bs); + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F + const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); + unsigned short r = static_cast(_mm_movemask_epi8(x)); + if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped + SizeType length; + #ifdef _MSC_VER // Find the index of first escaped + unsigned long offset; + _BitScanForward(&offset, r); + length = offset; + #else + length = static_cast(__builtin_ffs(r) - 1); + #endif + if (length != 0) { + char* q = reinterpret_cast(os.Push(length)); + for (size_t i = 0; i < length; i++) + q[i] = p[i]; + + p += length; + } + break; + } + _mm_storeu_si128(reinterpret_cast<__m128i *>(os.Push(16)), s); + } + + is.src_ = p; + } + + // InsituStringStream -> InsituStringStream + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(InsituStringStream& is, InsituStringStream& os) { + RAPIDJSON_ASSERT(&is == &os); + (void)os; + + if (is.src_ == is.dst_) { + SkipUnescapedString(is); + return; + } + + char* p = is.src_; + char *q = is.dst_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + is.dst_ = q; + return; + } + else + *q++ = *p++; + + // The rest of string using SIMD + static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; + static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; + const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); + const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); + const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); + + for (;; p += 16, q += 16) { + const __m128i s = _mm_load_si128(reinterpret_cast(p)); + const __m128i t1 = _mm_cmpeq_epi8(s, dq); + const __m128i t2 = _mm_cmpeq_epi8(s, bs); + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F + const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); + unsigned short r = static_cast(_mm_movemask_epi8(x)); + if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped + size_t length; +#ifdef _MSC_VER // Find the index of first escaped + unsigned long offset; + _BitScanForward(&offset, r); + length = offset; +#else + length = static_cast(__builtin_ffs(r) - 1); +#endif + for (const char* pend = p + length; p != pend; ) + *q++ = *p++; + break; + } + _mm_storeu_si128(reinterpret_cast<__m128i *>(q), s); + } + + is.src_ = p; + is.dst_ = q; + } + + // When read/write pointers are the same for insitu stream, just skip unescaped characters + static RAPIDJSON_FORCEINLINE void SkipUnescapedString(InsituStringStream& is) { + RAPIDJSON_ASSERT(is.src_ == is.dst_); + char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + for (; p != nextAligned; p++) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = is.dst_ = p; + return; + } + + // The rest of string using SIMD + static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; + static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; + const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); + const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); + const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); + + for (;; p += 16) { + const __m128i s = _mm_load_si128(reinterpret_cast(p)); + const __m128i t1 = _mm_cmpeq_epi8(s, dq); + const __m128i t2 = _mm_cmpeq_epi8(s, bs); + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F + const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); + unsigned short r = static_cast(_mm_movemask_epi8(x)); + if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped + size_t length; +#ifdef _MSC_VER // Find the index of first escaped + unsigned long offset; + _BitScanForward(&offset, r); + length = offset; +#else + length = static_cast(__builtin_ffs(r) - 1); +#endif + p += length; + break; + } + } + + is.src_ = is.dst_ = p; + } +#elif defined(RAPIDJSON_NEON) + // StringStream -> StackStream + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(StringStream& is, StackStream& os) { + const char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + return; + } + else + os.Put(*p++); + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType length = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high);; + length = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low);; + length = lz >> 3; + escaped = true; + } + if (RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + if (length != 0) { + char* q = reinterpret_cast(os.Push(length)); + for (size_t i = 0; i < length; i++) + q[i] = p[i]; + + p += length; + } + break; + } + vst1q_u8(reinterpret_cast(os.Push(16)), s); + } + + is.src_ = p; + } + + // InsituStringStream -> InsituStringStream + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(InsituStringStream& is, InsituStringStream& os) { + RAPIDJSON_ASSERT(&is == &os); + (void)os; + + if (is.src_ == is.dst_) { + SkipUnescapedString(is); + return; + } + + char* p = is.src_; + char *q = is.dst_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + is.dst_ = q; + return; + } + else + *q++ = *p++; + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16, q += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType length = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high); + length = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low); + length = lz >> 3; + escaped = true; + } + if (RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + for (const char* pend = p + length; p != pend; ) { + *q++ = *p++; + } + break; + } + vst1q_u8(reinterpret_cast(q), s); + } + + is.src_ = p; + is.dst_ = q; + } + + // When read/write pointers are the same for insitu stream, just skip unescaped characters + static RAPIDJSON_FORCEINLINE void SkipUnescapedString(InsituStringStream& is) { + RAPIDJSON_ASSERT(is.src_ == is.dst_); + char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + for (; p != nextAligned; p++) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = is.dst_ = p; + return; + } + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz = __builtin_clzll(high); + p += 8 + (lz >> 3); + break; + } + } else { + int lz = __builtin_clzll(low); + p += lz >> 3; + break; + } + } + + is.src_ = is.dst_ = p; + } +#endif // RAPIDJSON_NEON + + template class NumberStream; template - class NumberStream { + class NumberStream { public: + typedef typename InputStream::Ch Ch; + NumberStream(GenericReader& reader, InputStream& s) : is(s) { (void)reader; } - ~NumberStream() {} RAPIDJSON_FORCEINLINE Ch Peek() const { return is.Peek(); } RAPIDJSON_FORCEINLINE Ch TakePush() { return is.Take(); } RAPIDJSON_FORCEINLINE Ch Take() { return is.Take(); } + RAPIDJSON_FORCEINLINE void Push(char) {} + size_t Tell() { return is.Tell(); } size_t Length() { return 0; } const char* Pop() { return 0; } @@ -746,17 +1417,20 @@ class GenericReader { }; template - class NumberStream : public NumberStream { - typedef NumberStream Base; + class NumberStream : public NumberStream { + typedef NumberStream Base; public: - NumberStream(GenericReader& reader, InputStream& is) : NumberStream(reader, is), stackStream(reader.stack_) {} - ~NumberStream() {} + NumberStream(GenericReader& reader, InputStream& is) : Base(reader, is), stackStream(reader.stack_) {} RAPIDJSON_FORCEINLINE Ch TakePush() { - stackStream.Put((char)Base::is.Peek()); + stackStream.Put(static_cast(Base::is.Peek())); return Base::is.Take(); } + RAPIDJSON_FORCEINLINE void Push(char c) { + stackStream.Put(c); + } + size_t Length() { return stackStream.Length(); } const char* Pop() { @@ -768,34 +1442,48 @@ class GenericReader { StackStream stackStream; }; + template + class NumberStream : public NumberStream { + typedef NumberStream Base; + public: + NumberStream(GenericReader& reader, InputStream& is) : Base(reader, is) {} + + RAPIDJSON_FORCEINLINE Ch Take() { return Base::TakePush(); } + }; + template void ParseNumber(InputStream& is, Handler& handler) { internal::StreamLocalCopy copy(is); - NumberStream s(*this, copy.s); + NumberStream s(*this, copy.s); + + size_t startOffset = s.Tell(); + double d = 0.0; + bool useNanOrInf = false; // Parse minus - bool minus = false; - if (s.Peek() == '-') { - minus = true; - s.Take(); - } + bool minus = Consume(s, '-'); // Parse int: zero / ( digit1-9 *DIGIT ) unsigned i = 0; uint64_t i64 = 0; bool use64bit = false; int significandDigit = 0; - if (s.Peek() == '0') { + if (RAPIDJSON_UNLIKELY(s.Peek() == '0')) { i = 0; s.TakePush(); } - else if (s.Peek() >= '1' && s.Peek() <= '9') { + else if (RAPIDJSON_LIKELY(s.Peek() >= '1' && s.Peek() <= '9')) { i = static_cast(s.TakePush() - '0'); if (minus) - while (s.Peek() >= '0' && s.Peek() <= '9') { - if (i >= 214748364) { // 2^31 = 2147483648 - if (i != 214748364 || s.Peek() > '8') { + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + if (RAPIDJSON_UNLIKELY(i >= 214748364)) { // 2^31 = 2147483648 + if (RAPIDJSON_LIKELY(i != 214748364 || s.Peek() > '8')) { i64 = i; use64bit = true; break; @@ -805,9 +1493,9 @@ class GenericReader { significandDigit++; } else - while (s.Peek() >= '0' && s.Peek() <= '9') { - if (i >= 429496729) { // 2^32 - 1 = 4294967295 - if (i != 429496729 || s.Peek() > '5') { + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + if (RAPIDJSON_UNLIKELY(i >= 429496729)) { // 2^32 - 1 = 4294967295 + if (RAPIDJSON_LIKELY(i != 429496729 || s.Peek() > '5')) { i64 = i; use64bit = true; break; @@ -817,18 +1505,41 @@ class GenericReader { significandDigit++; } } + // Parse NaN or Infinity here + else if ((parseFlags & kParseNanAndInfFlag) && RAPIDJSON_LIKELY((s.Peek() == 'I' || s.Peek() == 'N'))) { + if (Consume(s, 'N')) { + if (Consume(s, 'a') && Consume(s, 'N')) { + d = std::numeric_limits::quiet_NaN(); + useNanOrInf = true; + } + } + else if (RAPIDJSON_LIKELY(Consume(s, 'I'))) { + if (Consume(s, 'n') && Consume(s, 'f')) { + d = (minus ? -std::numeric_limits::infinity() : std::numeric_limits::infinity()); + useNanOrInf = true; + + if (RAPIDJSON_UNLIKELY(s.Peek() == 'i' && !(Consume(s, 'i') && Consume(s, 'n') + && Consume(s, 'i') && Consume(s, 't') && Consume(s, 'y')))) { + RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + } + } + } + + if (RAPIDJSON_UNLIKELY(!useNanOrInf)) { + RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + } + } else RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); // Parse 64bit int bool useDouble = false; - double d = 0.0; if (use64bit) { - if (minus) - while (s.Peek() >= '0' && s.Peek() <= '9') { - if (i64 >= RAPIDJSON_UINT64_C2(0x0CCCCCCC, 0xCCCCCCCC)) // 2^63 = 9223372036854775808 - if (i64 != RAPIDJSON_UINT64_C2(0x0CCCCCCC, 0xCCCCCCCC) || s.Peek() > '8') { - d = i64; + if (minus) + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + if (RAPIDJSON_UNLIKELY(i64 >= RAPIDJSON_UINT64_C2(0x0CCCCCCC, 0xCCCCCCCC))) // 2^63 = 9223372036854775808 + if (RAPIDJSON_LIKELY(i64 != RAPIDJSON_UINT64_C2(0x0CCCCCCC, 0xCCCCCCCC) || s.Peek() > '8')) { + d = static_cast(i64); useDouble = true; break; } @@ -836,10 +1547,10 @@ class GenericReader { significandDigit++; } else - while (s.Peek() >= '0' && s.Peek() <= '9') { - if (i64 >= RAPIDJSON_UINT64_C2(0x19999999, 0x99999999)) // 2^64 - 1 = 18446744073709551615 - if (i64 != RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) || s.Peek() > '5') { - d = i64; + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + if (RAPIDJSON_UNLIKELY(i64 >= RAPIDJSON_UINT64_C2(0x19999999, 0x99999999))) // 2^64 - 1 = 18446744073709551615 + if (RAPIDJSON_LIKELY(i64 != RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) || s.Peek() > '5')) { + d = static_cast(i64); useDouble = true; break; } @@ -850,9 +1561,9 @@ class GenericReader { // Force double for big integer if (useDouble) { - while (s.Peek() >= '0' && s.Peek() <= '9') { - if (d >= 1.7976931348623157e307) // DBL_MAX / 10.0 - RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, s.Tell()); + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + if (RAPIDJSON_UNLIKELY(d >= 1.7976931348623157e307)) // DBL_MAX / 10.0 + RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, startOffset); d = d * 10 + (s.TakePush() - '0'); } } @@ -860,11 +1571,10 @@ class GenericReader { // Parse frac = decimal-point 1*DIGIT int expFrac = 0; size_t decimalPosition; - if (s.Peek() == '.') { - s.Take(); + if (Consume(s, '.')) { decimalPosition = s.Length(); - if (!(s.Peek() >= '0' && s.Peek() <= '9')) + if (RAPIDJSON_UNLIKELY(!(s.Peek() >= '0' && s.Peek() <= '9'))) RAPIDJSON_PARSE_ERROR(kParseErrorNumberMissFraction, s.Tell()); if (!useDouble) { @@ -872,8 +1582,8 @@ class GenericReader { // Use i64 to store significand in 64-bit architecture if (!use64bit) i64 = i; - - while (s.Peek() >= '0' && s.Peek() <= '9') { + + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { if (i64 > RAPIDJSON_UINT64_C2(0x1FFFFF, 0xFFFFFFFF)) // 2^53 - 1 for fast path break; else { @@ -884,19 +1594,19 @@ class GenericReader { } } - d = (double)i64; + d = static_cast(i64); #else // Use double to store significand in 32-bit architecture - d = use64bit ? (double)i64 : (double)i; + d = static_cast(use64bit ? i64 : i); #endif useDouble = true; } - while (s.Peek() >= '0' && s.Peek() <= '9') { + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { if (significandDigit < 17) { d = d * 10.0 + (s.TakePush() - '0'); --expFrac; - if (d > 0.0) + if (RAPIDJSON_LIKELY(d > 0.0)) significandDigit++; } else @@ -908,38 +1618,35 @@ class GenericReader { // Parse exp = e [ minus / plus ] 1*DIGIT int exp = 0; - if (s.Peek() == 'e' || s.Peek() == 'E') { + if (Consume(s, 'e') || Consume(s, 'E')) { if (!useDouble) { - d = use64bit ? i64 : i; + d = static_cast(use64bit ? i64 : i); useDouble = true; } - s.Take(); bool expMinus = false; - if (s.Peek() == '+') - s.Take(); - else if (s.Peek() == '-') { - s.Take(); + if (Consume(s, '+')) + ; + else if (Consume(s, '-')) expMinus = true; - } - if (s.Peek() >= '0' && s.Peek() <= '9') { - exp = s.Take() - '0'; + if (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + exp = static_cast(s.Take() - '0'); if (expMinus) { - while (s.Peek() >= '0' && s.Peek() <= '9') { - exp = exp * 10 + (s.Take() - '0'); + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + exp = exp * 10 + static_cast(s.Take() - '0'); if (exp >= 214748364) { // Issue #313: prevent overflow exponent - while (s.Peek() >= '0' && s.Peek() <= '9') // Consume the rest of exponent + while (RAPIDJSON_UNLIKELY(s.Peek() >= '0' && s.Peek() <= '9')) // Consume the rest of exponent s.Take(); } } } else { // positive exp int maxExp = 308 - expFrac; - while (s.Peek() >= '0' && s.Peek() <= '9') { - exp = exp * 10 + (s.Take() - '0'); - if (exp > maxExp) - RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, s.Tell()); + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { + exp = exp * 10 + static_cast(s.Take() - '0'); + if (RAPIDJSON_UNLIKELY(exp > maxExp)) + RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, startOffset); } } } @@ -952,34 +1659,63 @@ class GenericReader { // Finish parsing, call event according to the type of number. bool cont = true; - size_t length = s.Length(); - const char* decimal = s.Pop(); // Pop stack no matter if it will be used or not. - - if (useDouble) { - int p = exp + expFrac; - if (parseFlags & kParseFullPrecisionFlag) - d = internal::StrtodFullPrecision(d, p, decimal, length, decimalPosition, exp); - else - d = internal::StrtodNormalPrecision(d, p); - cont = handler.Double(minus ? -d : d); - } - else { - if (use64bit) { - if (minus) - cont = handler.Int64(-(int64_t)i64); - else - cont = handler.Uint64(i64); + if (parseFlags & kParseNumbersAsStringsFlag) { + if (parseFlags & kParseInsituFlag) { + s.Pop(); // Pop stack no matter if it will be used or not. + typename InputStream::Ch* head = is.PutBegin(); + const size_t length = s.Tell() - startOffset; + RAPIDJSON_ASSERT(length <= 0xFFFFFFFF); + // unable to insert the \0 character here, it will erase the comma after this number + const typename TargetEncoding::Ch* const str = reinterpret_cast(head); + cont = handler.RawNumber(str, SizeType(length), false); } else { - if (minus) - cont = handler.Int(-(int)i); - else - cont = handler.Uint(i); + SizeType numCharsToCopy = static_cast(s.Length()); + StringStream srcStream(s.Pop()); + StackStream dstStream(stack_); + while (numCharsToCopy--) { + Transcoder, TargetEncoding>::Transcode(srcStream, dstStream); + } + dstStream.Put('\0'); + const typename TargetEncoding::Ch* str = dstStream.Pop(); + const SizeType length = static_cast(dstStream.Length()) - 1; + cont = handler.RawNumber(str, SizeType(length), true); } } - if (!cont) - RAPIDJSON_PARSE_ERROR(kParseErrorTermination, s.Tell()); + else { + size_t length = s.Length(); + const char* decimal = s.Pop(); // Pop stack no matter if it will be used or not. + + if (useDouble) { + int p = exp + expFrac; + if (parseFlags & kParseFullPrecisionFlag) + d = internal::StrtodFullPrecision(d, p, decimal, length, decimalPosition, exp); + else + d = internal::StrtodNormalPrecision(d, p); + + cont = handler.Double(minus ? -d : d); + } + else if (useNanOrInf) { + cont = handler.Double(d); + } + else { + if (use64bit) { + if (minus) + cont = handler.Int64(static_cast(~i64 + 1)); + else + cont = handler.Uint64(i64); + } + else { + if (minus) + cont = handler.Int(static_cast(~i + 1)); + else + cont = handler.Uint(i); + } + } + } + if (RAPIDJSON_UNLIKELY(!cont)) + RAPIDJSON_PARSE_ERROR(kParseErrorTermination, startOffset); } // Parse any JSON value @@ -992,7 +1728,10 @@ class GenericReader { case '"': ParseString(is, handler); break; case '{': ParseObject(is, handler); break; case '[': ParseArray (is, handler); break; - default : ParseNumber(is, handler); + default : + ParseNumber(is, handler); + break; + } } @@ -1000,27 +1739,29 @@ class GenericReader { // States enum IterativeParsingState { - IterativeParsingStartState = 0, - IterativeParsingFinishState, - IterativeParsingErrorState, + IterativeParsingFinishState = 0, // sink states at top + IterativeParsingErrorState, // sink states at top + IterativeParsingStartState, // Object states IterativeParsingObjectInitialState, IterativeParsingMemberKeyState, - IterativeParsingKeyValueDelimiterState, IterativeParsingMemberValueState, - IterativeParsingMemberDelimiterState, IterativeParsingObjectFinishState, // Array states IterativeParsingArrayInitialState, IterativeParsingElementState, - IterativeParsingElementDelimiterState, IterativeParsingArrayFinishState, // Single value state IterativeParsingValueState, - + + // Delimiter states (at bottom) + IterativeParsingElementDelimiterState, + IterativeParsingMemberDelimiterState, + IterativeParsingKeyValueDelimiterState, + cIterativeParsingStateCount }; @@ -1064,9 +1805,9 @@ class GenericReader { #undef N #undef N16 //!@endcond - - if (sizeof(Ch) == 1 || unsigned(c) < 256) - return (Token)tokenMap[(unsigned char)c]; + + if (sizeof(Ch) == 1 || static_cast(c) < 256) + return static_cast(tokenMap[static_cast(c)]); else return NumberToken; } @@ -1074,6 +1815,18 @@ class GenericReader { RAPIDJSON_FORCEINLINE IterativeParsingState Predict(IterativeParsingState state, Token token) { // current state x one lookahead token -> new state static const char G[cIterativeParsingStateCount][kTokenCount] = { + // Finish(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, + // Error(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, // Start { IterativeParsingArrayInitialState, // Left bracket @@ -1088,18 +1841,6 @@ class GenericReader { IterativeParsingValueState, // Null IterativeParsingValueState // Number }, - // Finish(sink state) - { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - }, - // Error(sink state) - { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - }, // ObjectInitial { IterativeParsingErrorState, // Left bracket @@ -1128,20 +1869,6 @@ class GenericReader { IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, - // KeyValueDelimiter - { - IterativeParsingArrayInitialState, // Left bracket(push MemberValue state) - IterativeParsingErrorState, // Right bracket - IterativeParsingObjectInitialState, // Left curly bracket(push MemberValue state) - IterativeParsingErrorState, // Right curly bracket - IterativeParsingErrorState, // Comma - IterativeParsingErrorState, // Colon - IterativeParsingMemberValueState, // String - IterativeParsingMemberValueState, // False - IterativeParsingMemberValueState, // True - IterativeParsingMemberValueState, // Null - IterativeParsingMemberValueState // Number - }, // MemberValue { IterativeParsingErrorState, // Left bracket @@ -1156,20 +1883,6 @@ class GenericReader { IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, - // MemberDelimiter - { - IterativeParsingErrorState, // Left bracket - IterativeParsingErrorState, // Right bracket - IterativeParsingErrorState, // Left curly bracket - IterativeParsingErrorState, // Right curly bracket - IterativeParsingErrorState, // Comma - IterativeParsingErrorState, // Colon - IterativeParsingMemberKeyState, // String - IterativeParsingErrorState, // False - IterativeParsingErrorState, // True - IterativeParsingErrorState, // Null - IterativeParsingErrorState // Number - }, // ObjectFinish(sink state) { IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, @@ -1204,10 +1917,22 @@ class GenericReader { IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, + // ArrayFinish(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, + // Single Value (sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, // ElementDelimiter { IterativeParsingArrayInitialState, // Left bracket(push Element state) - IterativeParsingErrorState, // Right bracket + IterativeParsingArrayFinishState, // Right bracket IterativeParsingObjectInitialState, // Left curly bracket(push Element state) IterativeParsingErrorState, // Right curly bracket IterativeParsingErrorState, // Comma @@ -1218,21 +1943,37 @@ class GenericReader { IterativeParsingElementState, // Null IterativeParsingElementState // Number }, - // ArrayFinish(sink state) + // MemberDelimiter { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState + IterativeParsingErrorState, // Left bracket + IterativeParsingErrorState, // Right bracket + IterativeParsingErrorState, // Left curly bracket + IterativeParsingObjectFinishState, // Right curly bracket + IterativeParsingErrorState, // Comma + IterativeParsingErrorState, // Colon + IterativeParsingMemberKeyState, // String + IterativeParsingErrorState, // False + IterativeParsingErrorState, // True + IterativeParsingErrorState, // Null + IterativeParsingErrorState // Number }, - // Single Value (sink state) + // KeyValueDelimiter { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - } + IterativeParsingArrayInitialState, // Left bracket(push MemberValue state) + IterativeParsingErrorState, // Right bracket + IterativeParsingObjectInitialState, // Left curly bracket(push MemberValue state) + IterativeParsingErrorState, // Right curly bracket + IterativeParsingErrorState, // Comma + IterativeParsingErrorState, // Colon + IterativeParsingMemberValueState, // String + IterativeParsingMemberValueState, // False + IterativeParsingMemberValueState, // True + IterativeParsingMemberValueState, // Null + IterativeParsingMemberValueState // Number + }, }; // End of G - return (IterativeParsingState)G[state][token]; + return static_cast(G[state][token]); } // Make an advance in the token stream and state based on the candidate destination state which was returned by Transit(). @@ -1309,6 +2050,11 @@ class GenericReader { case IterativeParsingObjectFinishState: { + // Transit from delimiter is only allowed when trailing commas are enabled + if (!(parseFlags & kParseTrailingCommasFlag) && src == IterativeParsingMemberDelimiterState) { + RAPIDJSON_PARSE_ERROR_NORETURN(kParseErrorObjectMissName, is.Tell()); + return IterativeParsingErrorState; + } // Get member count. SizeType c = *stack_.template Pop(1); // If the object is not empty, count the last member. @@ -1334,6 +2080,11 @@ class GenericReader { case IterativeParsingArrayFinishState: { + // Transit from delimiter is only allowed when trailing commas are enabled + if (!(parseFlags & kParseTrailingCommasFlag) && src == IterativeParsingElementDelimiterState) { + RAPIDJSON_PARSE_ERROR_NORETURN(kParseErrorValueInvalid, is.Tell()); + return IterativeParsingErrorState; + } // Get element count. SizeType c = *stack_.template Pop(1); // If the array is not empty, count the last element. @@ -1385,55 +2136,68 @@ class GenericReader { // Error flag has been set. return; } - + switch (src) { - case IterativeParsingStartState: RAPIDJSON_PARSE_ERROR(kParseErrorDocumentEmpty, is.Tell()); - case IterativeParsingFinishState: RAPIDJSON_PARSE_ERROR(kParseErrorDocumentRootNotSingular, is.Tell()); + case IterativeParsingStartState: RAPIDJSON_PARSE_ERROR(kParseErrorDocumentEmpty, is.Tell()); return; + case IterativeParsingFinishState: RAPIDJSON_PARSE_ERROR(kParseErrorDocumentRootNotSingular, is.Tell()); return; case IterativeParsingObjectInitialState: - case IterativeParsingMemberDelimiterState: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissName, is.Tell()); - case IterativeParsingMemberKeyState: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissColon, is.Tell()); - case IterativeParsingMemberValueState: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissCommaOrCurlyBracket, is.Tell()); - case IterativeParsingElementState: RAPIDJSON_PARSE_ERROR(kParseErrorArrayMissCommaOrSquareBracket, is.Tell()); - default: RAPIDJSON_PARSE_ERROR(kParseErrorUnspecificSyntaxError, is.Tell()); - } + case IterativeParsingMemberDelimiterState: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissName, is.Tell()); return; + case IterativeParsingMemberKeyState: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissColon, is.Tell()); return; + case IterativeParsingMemberValueState: RAPIDJSON_PARSE_ERROR(kParseErrorObjectMissCommaOrCurlyBracket, is.Tell()); return; + case IterativeParsingKeyValueDelimiterState: + case IterativeParsingArrayInitialState: + case IterativeParsingElementDelimiterState: RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, is.Tell()); return; + default: RAPIDJSON_ASSERT(src == IterativeParsingElementState); RAPIDJSON_PARSE_ERROR(kParseErrorArrayMissCommaOrSquareBracket, is.Tell()); return; + } } + RAPIDJSON_FORCEINLINE bool IsIterativeParsingDelimiterState(IterativeParsingState s) { + return s >= IterativeParsingElementDelimiterState; + } + + RAPIDJSON_FORCEINLINE bool IsIterativeParsingCompleteState(IterativeParsingState s) { + return s <= IterativeParsingErrorState; + } + template ParseResult IterativeParse(InputStream& is, Handler& handler) { parseResult_.Clear(); ClearStackOnExit scope(*this); IterativeParsingState state = IterativeParsingStartState; - - SkipWhitespace(is); + + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); while (is.Peek() != '\0') { Token t = Tokenize(is.Peek()); IterativeParsingState n = Predict(state, t); IterativeParsingState d = Transit(state, t, n, is, handler); - + if (d == IterativeParsingErrorState) { HandleError(state, is); break; } - + state = d; - + // Do not further consume streams if a root JSON has been parsed. if ((parseFlags & kParseStopWhenDoneFlag) && state == IterativeParsingFinishState) break; - - SkipWhitespace(is); + + SkipWhitespaceAndComments(is); + RAPIDJSON_PARSE_ERROR_EARLY_RETURN(parseResult_); } - + // Handle the end of file. if (state != IterativeParsingFinishState) HandleError(state, is); - + return parseResult_; } static const size_t kDefaultStackCapacity = 256; //!< Default stack capacity in bytes for storing a single decoded string. internal::Stack stack_; //!< A stack for storing decoded string temporarily during non-destructive parsing. ParseResult parseResult_; + IterativeParsingState state_; }; // class GenericReader //! Reader with UTF8 encoding and default allocator. @@ -1441,6 +2205,11 @@ typedef GenericReader, UTF8<> > Reader; RAPIDJSON_NAMESPACE_END +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + + #ifdef __GNUC__ RAPIDJSON_DIAG_POP #endif diff --git a/rapidjson/schema.h b/rapidjson/schema.h new file mode 100644 index 0000000..abcf1a1 --- /dev/null +++ b/rapidjson/schema.h @@ -0,0 +1,2016 @@ +// Tencent is pleased to support the open source community by making RapidJSON available-> +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip-> All rights reserved-> +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License-> You may obtain a copy of the License at +// +// http://opensource->org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied-> See the License for the +// specific language governing permissions and limitations under the License-> + +#ifndef RAPIDJSON_SCHEMA_H_ +#define RAPIDJSON_SCHEMA_H_ + +#include "document.h" +#include "pointer.h" +#include // abs, floor + +#if !defined(RAPIDJSON_SCHEMA_USE_INTERNALREGEX) +#define RAPIDJSON_SCHEMA_USE_INTERNALREGEX 1 +#else +#define RAPIDJSON_SCHEMA_USE_INTERNALREGEX 0 +#endif + +#if !RAPIDJSON_SCHEMA_USE_INTERNALREGEX && !defined(RAPIDJSON_SCHEMA_USE_STDREGEX) && (__cplusplus >=201103L || (defined(_MSC_VER) && _MSC_VER >= 1800)) +#define RAPIDJSON_SCHEMA_USE_STDREGEX 1 +#else +#define RAPIDJSON_SCHEMA_USE_STDREGEX 0 +#endif + +#if RAPIDJSON_SCHEMA_USE_INTERNALREGEX +#include "internal/regex.h" +#elif RAPIDJSON_SCHEMA_USE_STDREGEX +#include +#endif + +#if RAPIDJSON_SCHEMA_USE_INTERNALREGEX || RAPIDJSON_SCHEMA_USE_STDREGEX +#define RAPIDJSON_SCHEMA_HAS_REGEX 1 +#else +#define RAPIDJSON_SCHEMA_HAS_REGEX 0 +#endif + +#ifndef RAPIDJSON_SCHEMA_VERBOSE +#define RAPIDJSON_SCHEMA_VERBOSE 0 +#endif + +#if RAPIDJSON_SCHEMA_VERBOSE +#include "stringbuffer.h" +#endif + +RAPIDJSON_DIAG_PUSH + +#if defined(__GNUC__) +RAPIDJSON_DIAG_OFF(effc++) +#endif + +#ifdef __clang__ +RAPIDJSON_DIAG_OFF(weak-vtables) +RAPIDJSON_DIAG_OFF(exit-time-destructors) +RAPIDJSON_DIAG_OFF(c++98-compat-pedantic) +RAPIDJSON_DIAG_OFF(variadic-macros) +#endif + +#ifdef _MSC_VER +RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +RAPIDJSON_NAMESPACE_BEGIN + +/////////////////////////////////////////////////////////////////////////////// +// Verbose Utilities + +#if RAPIDJSON_SCHEMA_VERBOSE + +namespace internal { + +inline void PrintInvalidKeyword(const char* keyword) { + printf("Fail keyword: %s\n", keyword); +} + +inline void PrintInvalidKeyword(const wchar_t* keyword) { + wprintf(L"Fail keyword: %ls\n", keyword); +} + +inline void PrintInvalidDocument(const char* document) { + printf("Fail document: %s\n\n", document); +} + +inline void PrintInvalidDocument(const wchar_t* document) { + wprintf(L"Fail document: %ls\n\n", document); +} + +inline void PrintValidatorPointers(unsigned depth, const char* s, const char* d) { + printf("S: %*s%s\nD: %*s%s\n\n", depth * 4, " ", s, depth * 4, " ", d); +} + +inline void PrintValidatorPointers(unsigned depth, const wchar_t* s, const wchar_t* d) { + wprintf(L"S: %*ls%ls\nD: %*ls%ls\n\n", depth * 4, L" ", s, depth * 4, L" ", d); +} + +} // namespace internal + +#endif // RAPIDJSON_SCHEMA_VERBOSE + +/////////////////////////////////////////////////////////////////////////////// +// RAPIDJSON_INVALID_KEYWORD_RETURN + +#if RAPIDJSON_SCHEMA_VERBOSE +#define RAPIDJSON_INVALID_KEYWORD_VERBOSE(keyword) internal::PrintInvalidKeyword(keyword) +#else +#define RAPIDJSON_INVALID_KEYWORD_VERBOSE(keyword) +#endif + +#define RAPIDJSON_INVALID_KEYWORD_RETURN(keyword)\ +RAPIDJSON_MULTILINEMACRO_BEGIN\ + context.invalidKeyword = keyword.GetString();\ + RAPIDJSON_INVALID_KEYWORD_VERBOSE(keyword.GetString());\ + return false;\ +RAPIDJSON_MULTILINEMACRO_END + +/////////////////////////////////////////////////////////////////////////////// +// Forward declarations + +template +class GenericSchemaDocument; + +namespace internal { + +template +class Schema; + +/////////////////////////////////////////////////////////////////////////////// +// ISchemaValidator + +class ISchemaValidator { +public: + virtual ~ISchemaValidator() {} + virtual bool IsValid() const = 0; +}; + +/////////////////////////////////////////////////////////////////////////////// +// ISchemaStateFactory + +template +class ISchemaStateFactory { +public: + virtual ~ISchemaStateFactory() {} + virtual ISchemaValidator* CreateSchemaValidator(const SchemaType&) = 0; + virtual void DestroySchemaValidator(ISchemaValidator* validator) = 0; + virtual void* CreateHasher() = 0; + virtual uint64_t GetHashCode(void* hasher) = 0; + virtual void DestroryHasher(void* hasher) = 0; + virtual void* MallocState(size_t size) = 0; + virtual void FreeState(void* p) = 0; +}; + +/////////////////////////////////////////////////////////////////////////////// +// Hasher + +// For comparison of compound value +template +class Hasher { +public: + typedef typename Encoding::Ch Ch; + + Hasher(Allocator* allocator = 0, size_t stackCapacity = kDefaultSize) : stack_(allocator, stackCapacity) {} + + bool Null() { return WriteType(kNullType); } + bool Bool(bool b) { return WriteType(b ? kTrueType : kFalseType); } + bool Int(int i) { Number n; n.u.i = i; n.d = static_cast(i); return WriteNumber(n); } + bool Uint(unsigned u) { Number n; n.u.u = u; n.d = static_cast(u); return WriteNumber(n); } + bool Int64(int64_t i) { Number n; n.u.i = i; n.d = static_cast(i); return WriteNumber(n); } + bool Uint64(uint64_t u) { Number n; n.u.u = u; n.d = static_cast(u); return WriteNumber(n); } + bool Double(double d) { + Number n; + if (d < 0) n.u.i = static_cast(d); + else n.u.u = static_cast(d); + n.d = d; + return WriteNumber(n); + } + + bool RawNumber(const Ch* str, SizeType len, bool) { + WriteBuffer(kNumberType, str, len * sizeof(Ch)); + return true; + } + + bool String(const Ch* str, SizeType len, bool) { + WriteBuffer(kStringType, str, len * sizeof(Ch)); + return true; + } + + bool StartObject() { return true; } + bool Key(const Ch* str, SizeType len, bool copy) { return String(str, len, copy); } + bool EndObject(SizeType memberCount) { + uint64_t h = Hash(0, kObjectType); + uint64_t* kv = stack_.template Pop(memberCount * 2); + for (SizeType i = 0; i < memberCount; i++) + h ^= Hash(kv[i * 2], kv[i * 2 + 1]); // Use xor to achieve member order insensitive + *stack_.template Push() = h; + return true; + } + + bool StartArray() { return true; } + bool EndArray(SizeType elementCount) { + uint64_t h = Hash(0, kArrayType); + uint64_t* e = stack_.template Pop(elementCount); + for (SizeType i = 0; i < elementCount; i++) + h = Hash(h, e[i]); // Use hash to achieve element order sensitive + *stack_.template Push() = h; + return true; + } + + bool IsValid() const { return stack_.GetSize() == sizeof(uint64_t); } + + uint64_t GetHashCode() const { + RAPIDJSON_ASSERT(IsValid()); + return *stack_.template Top(); + } + +private: + static const size_t kDefaultSize = 256; + struct Number { + union U { + uint64_t u; + int64_t i; + }u; + double d; + }; + + bool WriteType(Type type) { return WriteBuffer(type, 0, 0); } + + bool WriteNumber(const Number& n) { return WriteBuffer(kNumberType, &n, sizeof(n)); } + + bool WriteBuffer(Type type, const void* data, size_t len) { + // FNV-1a from http://isthe.com/chongo/tech/comp/fnv/ + uint64_t h = Hash(RAPIDJSON_UINT64_C2(0x84222325, 0xcbf29ce4), type); + const unsigned char* d = static_cast(data); + for (size_t i = 0; i < len; i++) + h = Hash(h, d[i]); + *stack_.template Push() = h; + return true; + } + + static uint64_t Hash(uint64_t h, uint64_t d) { + static const uint64_t kPrime = RAPIDJSON_UINT64_C2(0x00000100, 0x000001b3); + h ^= d; + h *= kPrime; + return h; + } + + Stack stack_; +}; + +/////////////////////////////////////////////////////////////////////////////// +// SchemaValidationContext + +template +struct SchemaValidationContext { + typedef Schema SchemaType; + typedef ISchemaStateFactory SchemaValidatorFactoryType; + typedef typename SchemaType::ValueType ValueType; + typedef typename ValueType::Ch Ch; + + enum PatternValidatorType { + kPatternValidatorOnly, + kPatternValidatorWithProperty, + kPatternValidatorWithAdditionalProperty + }; + + SchemaValidationContext(SchemaValidatorFactoryType& f, const SchemaType* s) : + factory(f), + schema(s), + valueSchema(), + invalidKeyword(), + hasher(), + arrayElementHashCodes(), + validators(), + validatorCount(), + patternPropertiesValidators(), + patternPropertiesValidatorCount(), + patternPropertiesSchemas(), + patternPropertiesSchemaCount(), + valuePatternValidatorType(kPatternValidatorOnly), + propertyExist(), + inArray(false), + valueUniqueness(false), + arrayUniqueness(false) + { + } + + ~SchemaValidationContext() { + if (hasher) + factory.DestroryHasher(hasher); + if (validators) { + for (SizeType i = 0; i < validatorCount; i++) + factory.DestroySchemaValidator(validators[i]); + factory.FreeState(validators); + } + if (patternPropertiesValidators) { + for (SizeType i = 0; i < patternPropertiesValidatorCount; i++) + factory.DestroySchemaValidator(patternPropertiesValidators[i]); + factory.FreeState(patternPropertiesValidators); + } + if (patternPropertiesSchemas) + factory.FreeState(patternPropertiesSchemas); + if (propertyExist) + factory.FreeState(propertyExist); + } + + SchemaValidatorFactoryType& factory; + const SchemaType* schema; + const SchemaType* valueSchema; + const Ch* invalidKeyword; + void* hasher; // Only validator access + void* arrayElementHashCodes; // Only validator access this + ISchemaValidator** validators; + SizeType validatorCount; + ISchemaValidator** patternPropertiesValidators; + SizeType patternPropertiesValidatorCount; + const SchemaType** patternPropertiesSchemas; + SizeType patternPropertiesSchemaCount; + PatternValidatorType valuePatternValidatorType; + PatternValidatorType objectPatternValidatorType; + SizeType arrayElementIndex; + bool* propertyExist; + bool inArray; + bool valueUniqueness; + bool arrayUniqueness; +}; + +/////////////////////////////////////////////////////////////////////////////// +// Schema + +template +class Schema { +public: + typedef typename SchemaDocumentType::ValueType ValueType; + typedef typename SchemaDocumentType::AllocatorType AllocatorType; + typedef typename SchemaDocumentType::PointerType PointerType; + typedef typename ValueType::EncodingType EncodingType; + typedef typename EncodingType::Ch Ch; + typedef SchemaValidationContext Context; + typedef Schema SchemaType; + typedef GenericValue SValue; + friend class GenericSchemaDocument; + + Schema(SchemaDocumentType* schemaDocument, const PointerType& p, const ValueType& value, const ValueType& document, AllocatorType* allocator) : + allocator_(allocator), + typeless_(schemaDocument->GetTypeless()), + enum_(), + enumCount_(), + not_(), + type_((1 << kTotalSchemaType) - 1), // typeless + validatorCount_(), + properties_(), + additionalPropertiesSchema_(), + patternProperties_(), + patternPropertyCount_(), + propertyCount_(), + minProperties_(), + maxProperties_(SizeType(~0)), + additionalProperties_(true), + hasDependencies_(), + hasRequired_(), + hasSchemaDependencies_(), + additionalItemsSchema_(), + itemsList_(), + itemsTuple_(), + itemsTupleCount_(), + minItems_(), + maxItems_(SizeType(~0)), + additionalItems_(true), + uniqueItems_(false), + pattern_(), + minLength_(0), + maxLength_(~SizeType(0)), + exclusiveMinimum_(false), + exclusiveMaximum_(false) + { + typedef typename SchemaDocumentType::ValueType ValueType; + typedef typename ValueType::ConstValueIterator ConstValueIterator; + typedef typename ValueType::ConstMemberIterator ConstMemberIterator; + + if (!value.IsObject()) + return; + + if (const ValueType* v = GetMember(value, GetTypeString())) { + type_ = 0; + if (v->IsString()) + AddType(*v); + else if (v->IsArray()) + for (ConstValueIterator itr = v->Begin(); itr != v->End(); ++itr) + AddType(*itr); + } + + if (const ValueType* v = GetMember(value, GetEnumString())) + if (v->IsArray() && v->Size() > 0) { + enum_ = static_cast(allocator_->Malloc(sizeof(uint64_t) * v->Size())); + for (ConstValueIterator itr = v->Begin(); itr != v->End(); ++itr) { + typedef Hasher > EnumHasherType; + char buffer[256 + 24]; + MemoryPoolAllocator<> hasherAllocator(buffer, sizeof(buffer)); + EnumHasherType h(&hasherAllocator, 256); + itr->Accept(h); + enum_[enumCount_++] = h.GetHashCode(); + } + } + + if (schemaDocument) { + AssignIfExist(allOf_, *schemaDocument, p, value, GetAllOfString(), document); + AssignIfExist(anyOf_, *schemaDocument, p, value, GetAnyOfString(), document); + AssignIfExist(oneOf_, *schemaDocument, p, value, GetOneOfString(), document); + } + + if (const ValueType* v = GetMember(value, GetNotString())) { + schemaDocument->CreateSchema(¬_, p.Append(GetNotString(), allocator_), *v, document); + notValidatorIndex_ = validatorCount_; + validatorCount_++; + } + + // Object + + const ValueType* properties = GetMember(value, GetPropertiesString()); + const ValueType* required = GetMember(value, GetRequiredString()); + const ValueType* dependencies = GetMember(value, GetDependenciesString()); + { + // Gather properties from properties/required/dependencies + SValue allProperties(kArrayType); + + if (properties && properties->IsObject()) + for (ConstMemberIterator itr = properties->MemberBegin(); itr != properties->MemberEnd(); ++itr) + AddUniqueElement(allProperties, itr->name); + + if (required && required->IsArray()) + for (ConstValueIterator itr = required->Begin(); itr != required->End(); ++itr) + if (itr->IsString()) + AddUniqueElement(allProperties, *itr); + + if (dependencies && dependencies->IsObject()) + for (ConstMemberIterator itr = dependencies->MemberBegin(); itr != dependencies->MemberEnd(); ++itr) { + AddUniqueElement(allProperties, itr->name); + if (itr->value.IsArray()) + for (ConstValueIterator i = itr->value.Begin(); i != itr->value.End(); ++i) + if (i->IsString()) + AddUniqueElement(allProperties, *i); + } + + if (allProperties.Size() > 0) { + propertyCount_ = allProperties.Size(); + properties_ = static_cast(allocator_->Malloc(sizeof(Property) * propertyCount_)); + for (SizeType i = 0; i < propertyCount_; i++) { + new (&properties_[i]) Property(); + properties_[i].name = allProperties[i]; + properties_[i].schema = typeless_; + } + } + } + + if (properties && properties->IsObject()) { + PointerType q = p.Append(GetPropertiesString(), allocator_); + for (ConstMemberIterator itr = properties->MemberBegin(); itr != properties->MemberEnd(); ++itr) { + SizeType index; + if (FindPropertyIndex(itr->name, &index)) + schemaDocument->CreateSchema(&properties_[index].schema, q.Append(itr->name, allocator_), itr->value, document); + } + } + + if (const ValueType* v = GetMember(value, GetPatternPropertiesString())) { + PointerType q = p.Append(GetPatternPropertiesString(), allocator_); + patternProperties_ = static_cast(allocator_->Malloc(sizeof(PatternProperty) * v->MemberCount())); + patternPropertyCount_ = 0; + + for (ConstMemberIterator itr = v->MemberBegin(); itr != v->MemberEnd(); ++itr) { + new (&patternProperties_[patternPropertyCount_]) PatternProperty(); + patternProperties_[patternPropertyCount_].pattern = CreatePattern(itr->name); + schemaDocument->CreateSchema(&patternProperties_[patternPropertyCount_].schema, q.Append(itr->name, allocator_), itr->value, document); + patternPropertyCount_++; + } + } + + if (required && required->IsArray()) + for (ConstValueIterator itr = required->Begin(); itr != required->End(); ++itr) + if (itr->IsString()) { + SizeType index; + if (FindPropertyIndex(*itr, &index)) { + properties_[index].required = true; + hasRequired_ = true; + } + } + + if (dependencies && dependencies->IsObject()) { + PointerType q = p.Append(GetDependenciesString(), allocator_); + hasDependencies_ = true; + for (ConstMemberIterator itr = dependencies->MemberBegin(); itr != dependencies->MemberEnd(); ++itr) { + SizeType sourceIndex; + if (FindPropertyIndex(itr->name, &sourceIndex)) { + if (itr->value.IsArray()) { + properties_[sourceIndex].dependencies = static_cast(allocator_->Malloc(sizeof(bool) * propertyCount_)); + std::memset(properties_[sourceIndex].dependencies, 0, sizeof(bool)* propertyCount_); + for (ConstValueIterator targetItr = itr->value.Begin(); targetItr != itr->value.End(); ++targetItr) { + SizeType targetIndex; + if (FindPropertyIndex(*targetItr, &targetIndex)) + properties_[sourceIndex].dependencies[targetIndex] = true; + } + } + else if (itr->value.IsObject()) { + hasSchemaDependencies_ = true; + schemaDocument->CreateSchema(&properties_[sourceIndex].dependenciesSchema, q.Append(itr->name, allocator_), itr->value, document); + properties_[sourceIndex].dependenciesValidatorIndex = validatorCount_; + validatorCount_++; + } + } + } + } + + if (const ValueType* v = GetMember(value, GetAdditionalPropertiesString())) { + if (v->IsBool()) + additionalProperties_ = v->GetBool(); + else if (v->IsObject()) + schemaDocument->CreateSchema(&additionalPropertiesSchema_, p.Append(GetAdditionalPropertiesString(), allocator_), *v, document); + } + + AssignIfExist(minProperties_, value, GetMinPropertiesString()); + AssignIfExist(maxProperties_, value, GetMaxPropertiesString()); + + // Array + if (const ValueType* v = GetMember(value, GetItemsString())) { + PointerType q = p.Append(GetItemsString(), allocator_); + if (v->IsObject()) // List validation + schemaDocument->CreateSchema(&itemsList_, q, *v, document); + else if (v->IsArray()) { // Tuple validation + itemsTuple_ = static_cast(allocator_->Malloc(sizeof(const Schema*) * v->Size())); + SizeType index = 0; + for (ConstValueIterator itr = v->Begin(); itr != v->End(); ++itr, index++) + schemaDocument->CreateSchema(&itemsTuple_[itemsTupleCount_++], q.Append(index, allocator_), *itr, document); + } + } + + AssignIfExist(minItems_, value, GetMinItemsString()); + AssignIfExist(maxItems_, value, GetMaxItemsString()); + + if (const ValueType* v = GetMember(value, GetAdditionalItemsString())) { + if (v->IsBool()) + additionalItems_ = v->GetBool(); + else if (v->IsObject()) + schemaDocument->CreateSchema(&additionalItemsSchema_, p.Append(GetAdditionalItemsString(), allocator_), *v, document); + } + + AssignIfExist(uniqueItems_, value, GetUniqueItemsString()); + + // String + AssignIfExist(minLength_, value, GetMinLengthString()); + AssignIfExist(maxLength_, value, GetMaxLengthString()); + + if (const ValueType* v = GetMember(value, GetPatternString())) + pattern_ = CreatePattern(*v); + + // Number + if (const ValueType* v = GetMember(value, GetMinimumString())) + if (v->IsNumber()) + minimum_.CopyFrom(*v, *allocator_); + + if (const ValueType* v = GetMember(value, GetMaximumString())) + if (v->IsNumber()) + maximum_.CopyFrom(*v, *allocator_); + + AssignIfExist(exclusiveMinimum_, value, GetExclusiveMinimumString()); + AssignIfExist(exclusiveMaximum_, value, GetExclusiveMaximumString()); + + if (const ValueType* v = GetMember(value, GetMultipleOfString())) + if (v->IsNumber() && v->GetDouble() > 0.0) + multipleOf_.CopyFrom(*v, *allocator_); + } + + ~Schema() { + AllocatorType::Free(enum_); + if (properties_) { + for (SizeType i = 0; i < propertyCount_; i++) + properties_[i].~Property(); + AllocatorType::Free(properties_); + } + if (patternProperties_) { + for (SizeType i = 0; i < patternPropertyCount_; i++) + patternProperties_[i].~PatternProperty(); + AllocatorType::Free(patternProperties_); + } + AllocatorType::Free(itemsTuple_); +#if RAPIDJSON_SCHEMA_HAS_REGEX + if (pattern_) { + pattern_->~RegexType(); + AllocatorType::Free(pattern_); + } +#endif + } + + bool BeginValue(Context& context) const { + if (context.inArray) { + if (uniqueItems_) + context.valueUniqueness = true; + + if (itemsList_) + context.valueSchema = itemsList_; + else if (itemsTuple_) { + if (context.arrayElementIndex < itemsTupleCount_) + context.valueSchema = itemsTuple_[context.arrayElementIndex]; + else if (additionalItemsSchema_) + context.valueSchema = additionalItemsSchema_; + else if (additionalItems_) + context.valueSchema = typeless_; + else + RAPIDJSON_INVALID_KEYWORD_RETURN(GetItemsString()); + } + else + context.valueSchema = typeless_; + + context.arrayElementIndex++; + } + return true; + } + + RAPIDJSON_FORCEINLINE bool EndValue(Context& context) const { + if (context.patternPropertiesValidatorCount > 0) { + bool otherValid = false; + SizeType count = context.patternPropertiesValidatorCount; + if (context.objectPatternValidatorType != Context::kPatternValidatorOnly) + otherValid = context.patternPropertiesValidators[--count]->IsValid(); + + bool patternValid = true; + for (SizeType i = 0; i < count; i++) + if (!context.patternPropertiesValidators[i]->IsValid()) { + patternValid = false; + break; + } + + if (context.objectPatternValidatorType == Context::kPatternValidatorOnly) { + if (!patternValid) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } + else if (context.objectPatternValidatorType == Context::kPatternValidatorWithProperty) { + if (!patternValid || !otherValid) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } + else if (!patternValid && !otherValid) // kPatternValidatorWithAdditionalProperty) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } + + if (enum_) { + const uint64_t h = context.factory.GetHashCode(context.hasher); + for (SizeType i = 0; i < enumCount_; i++) + if (enum_[i] == h) + goto foundEnum; + RAPIDJSON_INVALID_KEYWORD_RETURN(GetEnumString()); + foundEnum:; + } + + if (allOf_.schemas) + for (SizeType i = allOf_.begin; i < allOf_.begin + allOf_.count; i++) + if (!context.validators[i]->IsValid()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetAllOfString()); + + if (anyOf_.schemas) { + for (SizeType i = anyOf_.begin; i < anyOf_.begin + anyOf_.count; i++) + if (context.validators[i]->IsValid()) + goto foundAny; + RAPIDJSON_INVALID_KEYWORD_RETURN(GetAnyOfString()); + foundAny:; + } + + if (oneOf_.schemas) { + bool oneValid = false; + for (SizeType i = oneOf_.begin; i < oneOf_.begin + oneOf_.count; i++) + if (context.validators[i]->IsValid()) { + if (oneValid) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetOneOfString()); + else + oneValid = true; + } + if (!oneValid) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetOneOfString()); + } + + if (not_ && context.validators[notValidatorIndex_]->IsValid()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetNotString()); + + return true; + } + + bool Null(Context& context) const { + if (!(type_ & (1 << kNullSchemaType))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + return CreateParallelValidator(context); + } + + bool Bool(Context& context, bool) const { + if (!(type_ & (1 << kBooleanSchemaType))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + return CreateParallelValidator(context); + } + + bool Int(Context& context, int i) const { + if (!CheckInt(context, i)) + return false; + return CreateParallelValidator(context); + } + + bool Uint(Context& context, unsigned u) const { + if (!CheckUint(context, u)) + return false; + return CreateParallelValidator(context); + } + + bool Int64(Context& context, int64_t i) const { + if (!CheckInt(context, i)) + return false; + return CreateParallelValidator(context); + } + + bool Uint64(Context& context, uint64_t u) const { + if (!CheckUint(context, u)) + return false; + return CreateParallelValidator(context); + } + + bool Double(Context& context, double d) const { + if (!(type_ & (1 << kNumberSchemaType))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + + if (!minimum_.IsNull() && !CheckDoubleMinimum(context, d)) + return false; + + if (!maximum_.IsNull() && !CheckDoubleMaximum(context, d)) + return false; + + if (!multipleOf_.IsNull() && !CheckDoubleMultipleOf(context, d)) + return false; + + return CreateParallelValidator(context); + } + + bool String(Context& context, const Ch* str, SizeType length, bool) const { + if (!(type_ & (1 << kStringSchemaType))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + + if (minLength_ != 0 || maxLength_ != SizeType(~0)) { + SizeType count; + if (internal::CountStringCodePoint(str, length, &count)) { + if (count < minLength_) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinLengthString()); + if (count > maxLength_) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxLengthString()); + } + } + + if (pattern_ && !IsPatternMatch(pattern_, str, length)) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternString()); + + return CreateParallelValidator(context); + } + + bool StartObject(Context& context) const { + if (!(type_ & (1 << kObjectSchemaType))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + + if (hasDependencies_ || hasRequired_) { + context.propertyExist = static_cast(context.factory.MallocState(sizeof(bool) * propertyCount_)); + std::memset(context.propertyExist, 0, sizeof(bool) * propertyCount_); + } + + if (patternProperties_) { // pre-allocate schema array + SizeType count = patternPropertyCount_ + 1; // extra for valuePatternValidatorType + context.patternPropertiesSchemas = static_cast(context.factory.MallocState(sizeof(const SchemaType*) * count)); + context.patternPropertiesSchemaCount = 0; + std::memset(context.patternPropertiesSchemas, 0, sizeof(SchemaType*) * count); + } + + return CreateParallelValidator(context); + } + + bool Key(Context& context, const Ch* str, SizeType len, bool) const { + if (patternProperties_) { + context.patternPropertiesSchemaCount = 0; + for (SizeType i = 0; i < patternPropertyCount_; i++) + if (patternProperties_[i].pattern && IsPatternMatch(patternProperties_[i].pattern, str, len)) { + context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = patternProperties_[i].schema; + context.valueSchema = typeless_; + } + } + + SizeType index; + if (FindPropertyIndex(ValueType(str, len).Move(), &index)) { + if (context.patternPropertiesSchemaCount > 0) { + context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = properties_[index].schema; + context.valueSchema = typeless_; + context.valuePatternValidatorType = Context::kPatternValidatorWithProperty; + } + else + context.valueSchema = properties_[index].schema; + + if (context.propertyExist) + context.propertyExist[index] = true; + + return true; + } + + if (additionalPropertiesSchema_) { + if (additionalPropertiesSchema_ && context.patternPropertiesSchemaCount > 0) { + context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = additionalPropertiesSchema_; + context.valueSchema = typeless_; + context.valuePatternValidatorType = Context::kPatternValidatorWithAdditionalProperty; + } + else + context.valueSchema = additionalPropertiesSchema_; + return true; + } + else if (additionalProperties_) { + context.valueSchema = typeless_; + return true; + } + + if (context.patternPropertiesSchemaCount == 0) // patternProperties are not additional properties + RAPIDJSON_INVALID_KEYWORD_RETURN(GetAdditionalPropertiesString()); + + return true; + } + + bool EndObject(Context& context, SizeType memberCount) const { + if (hasRequired_) + for (SizeType index = 0; index < propertyCount_; index++) + if (properties_[index].required) + if (!context.propertyExist[index]) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetRequiredString()); + + if (memberCount < minProperties_) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinPropertiesString()); + + if (memberCount > maxProperties_) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxPropertiesString()); + + if (hasDependencies_) { + for (SizeType sourceIndex = 0; sourceIndex < propertyCount_; sourceIndex++) + if (context.propertyExist[sourceIndex]) { + if (properties_[sourceIndex].dependencies) { + for (SizeType targetIndex = 0; targetIndex < propertyCount_; targetIndex++) + if (properties_[sourceIndex].dependencies[targetIndex] && !context.propertyExist[targetIndex]) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); + } + else if (properties_[sourceIndex].dependenciesSchema) + if (!context.validators[properties_[sourceIndex].dependenciesValidatorIndex]->IsValid()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); + } + } + + return true; + } + + bool StartArray(Context& context) const { + if (!(type_ & (1 << kArraySchemaType))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + + context.arrayElementIndex = 0; + context.inArray = true; + + return CreateParallelValidator(context); + } + + bool EndArray(Context& context, SizeType elementCount) const { + context.inArray = false; + + if (elementCount < minItems_) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinItemsString()); + + if (elementCount > maxItems_) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxItemsString()); + + return true; + } + + // Generate functions for string literal according to Ch +#define RAPIDJSON_STRING_(name, ...) \ + static const ValueType& Get##name##String() {\ + static const Ch s[] = { __VA_ARGS__, '\0' };\ + static const ValueType v(s, static_cast(sizeof(s) / sizeof(Ch) - 1));\ + return v;\ + } + + RAPIDJSON_STRING_(Null, 'n', 'u', 'l', 'l') + RAPIDJSON_STRING_(Boolean, 'b', 'o', 'o', 'l', 'e', 'a', 'n') + RAPIDJSON_STRING_(Object, 'o', 'b', 'j', 'e', 'c', 't') + RAPIDJSON_STRING_(Array, 'a', 'r', 'r', 'a', 'y') + RAPIDJSON_STRING_(String, 's', 't', 'r', 'i', 'n', 'g') + RAPIDJSON_STRING_(Number, 'n', 'u', 'm', 'b', 'e', 'r') + RAPIDJSON_STRING_(Integer, 'i', 'n', 't', 'e', 'g', 'e', 'r') + RAPIDJSON_STRING_(Type, 't', 'y', 'p', 'e') + RAPIDJSON_STRING_(Enum, 'e', 'n', 'u', 'm') + RAPIDJSON_STRING_(AllOf, 'a', 'l', 'l', 'O', 'f') + RAPIDJSON_STRING_(AnyOf, 'a', 'n', 'y', 'O', 'f') + RAPIDJSON_STRING_(OneOf, 'o', 'n', 'e', 'O', 'f') + RAPIDJSON_STRING_(Not, 'n', 'o', 't') + RAPIDJSON_STRING_(Properties, 'p', 'r', 'o', 'p', 'e', 'r', 't', 'i', 'e', 's') + RAPIDJSON_STRING_(Required, 'r', 'e', 'q', 'u', 'i', 'r', 'e', 'd') + RAPIDJSON_STRING_(Dependencies, 'd', 'e', 'p', 'e', 'n', 'd', 'e', 'n', 'c', 'i', 'e', 's') + RAPIDJSON_STRING_(PatternProperties, 'p', 'a', 't', 't', 'e', 'r', 'n', 'P', 'r', 'o', 'p', 'e', 'r', 't', 'i', 'e', 's') + RAPIDJSON_STRING_(AdditionalProperties, 'a', 'd', 'd', 'i', 't', 'i', 'o', 'n', 'a', 'l', 'P', 'r', 'o', 'p', 'e', 'r', 't', 'i', 'e', 's') + RAPIDJSON_STRING_(MinProperties, 'm', 'i', 'n', 'P', 'r', 'o', 'p', 'e', 'r', 't', 'i', 'e', 's') + RAPIDJSON_STRING_(MaxProperties, 'm', 'a', 'x', 'P', 'r', 'o', 'p', 'e', 'r', 't', 'i', 'e', 's') + RAPIDJSON_STRING_(Items, 'i', 't', 'e', 'm', 's') + RAPIDJSON_STRING_(MinItems, 'm', 'i', 'n', 'I', 't', 'e', 'm', 's') + RAPIDJSON_STRING_(MaxItems, 'm', 'a', 'x', 'I', 't', 'e', 'm', 's') + RAPIDJSON_STRING_(AdditionalItems, 'a', 'd', 'd', 'i', 't', 'i', 'o', 'n', 'a', 'l', 'I', 't', 'e', 'm', 's') + RAPIDJSON_STRING_(UniqueItems, 'u', 'n', 'i', 'q', 'u', 'e', 'I', 't', 'e', 'm', 's') + RAPIDJSON_STRING_(MinLength, 'm', 'i', 'n', 'L', 'e', 'n', 'g', 't', 'h') + RAPIDJSON_STRING_(MaxLength, 'm', 'a', 'x', 'L', 'e', 'n', 'g', 't', 'h') + RAPIDJSON_STRING_(Pattern, 'p', 'a', 't', 't', 'e', 'r', 'n') + RAPIDJSON_STRING_(Minimum, 'm', 'i', 'n', 'i', 'm', 'u', 'm') + RAPIDJSON_STRING_(Maximum, 'm', 'a', 'x', 'i', 'm', 'u', 'm') + RAPIDJSON_STRING_(ExclusiveMinimum, 'e', 'x', 'c', 'l', 'u', 's', 'i', 'v', 'e', 'M', 'i', 'n', 'i', 'm', 'u', 'm') + RAPIDJSON_STRING_(ExclusiveMaximum, 'e', 'x', 'c', 'l', 'u', 's', 'i', 'v', 'e', 'M', 'a', 'x', 'i', 'm', 'u', 'm') + RAPIDJSON_STRING_(MultipleOf, 'm', 'u', 'l', 't', 'i', 'p', 'l', 'e', 'O', 'f') + +#undef RAPIDJSON_STRING_ + +private: + enum SchemaValueType { + kNullSchemaType, + kBooleanSchemaType, + kObjectSchemaType, + kArraySchemaType, + kStringSchemaType, + kNumberSchemaType, + kIntegerSchemaType, + kTotalSchemaType + }; + +#if RAPIDJSON_SCHEMA_USE_INTERNALREGEX + typedef internal::GenericRegex RegexType; +#elif RAPIDJSON_SCHEMA_USE_STDREGEX + typedef std::basic_regex RegexType; +#else + typedef char RegexType; +#endif + + struct SchemaArray { + SchemaArray() : schemas(), count() {} + ~SchemaArray() { AllocatorType::Free(schemas); } + const SchemaType** schemas; + SizeType begin; // begin index of context.validators + SizeType count; + }; + + template + void AddUniqueElement(V1& a, const V2& v) { + for (typename V1::ConstValueIterator itr = a.Begin(); itr != a.End(); ++itr) + if (*itr == v) + return; + V1 c(v, *allocator_); + a.PushBack(c, *allocator_); + } + + static const ValueType* GetMember(const ValueType& value, const ValueType& name) { + typename ValueType::ConstMemberIterator itr = value.FindMember(name); + return itr != value.MemberEnd() ? &(itr->value) : 0; + } + + static void AssignIfExist(bool& out, const ValueType& value, const ValueType& name) { + if (const ValueType* v = GetMember(value, name)) + if (v->IsBool()) + out = v->GetBool(); + } + + static void AssignIfExist(SizeType& out, const ValueType& value, const ValueType& name) { + if (const ValueType* v = GetMember(value, name)) + if (v->IsUint64() && v->GetUint64() <= SizeType(~0)) + out = static_cast(v->GetUint64()); + } + + void AssignIfExist(SchemaArray& out, SchemaDocumentType& schemaDocument, const PointerType& p, const ValueType& value, const ValueType& name, const ValueType& document) { + if (const ValueType* v = GetMember(value, name)) { + if (v->IsArray() && v->Size() > 0) { + PointerType q = p.Append(name, allocator_); + out.count = v->Size(); + out.schemas = static_cast(allocator_->Malloc(out.count * sizeof(const Schema*))); + memset(out.schemas, 0, sizeof(Schema*)* out.count); + for (SizeType i = 0; i < out.count; i++) + schemaDocument.CreateSchema(&out.schemas[i], q.Append(i, allocator_), (*v)[i], document); + out.begin = validatorCount_; + validatorCount_ += out.count; + } + } + } + +#if RAPIDJSON_SCHEMA_USE_INTERNALREGEX + template + RegexType* CreatePattern(const ValueType& value) { + if (value.IsString()) { + RegexType* r = new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString()); + if (!r->IsValid()) { + r->~RegexType(); + AllocatorType::Free(r); + r = 0; + } + return r; + } + return 0; + } + + static bool IsPatternMatch(const RegexType* pattern, const Ch *str, SizeType) { + GenericRegexSearch rs(*pattern); + return rs.Search(str); + } +#elif RAPIDJSON_SCHEMA_USE_STDREGEX + template + RegexType* CreatePattern(const ValueType& value) { + if (value.IsString()) + try { + return new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString(), std::size_t(value.GetStringLength()), std::regex_constants::ECMAScript); + } + catch (const std::regex_error&) { + } + return 0; + } + + static bool IsPatternMatch(const RegexType* pattern, const Ch *str, SizeType length) { + std::match_results r; + return std::regex_search(str, str + length, r, *pattern); + } +#else + template + RegexType* CreatePattern(const ValueType&) { return 0; } + + static bool IsPatternMatch(const RegexType*, const Ch *, SizeType) { return true; } +#endif // RAPIDJSON_SCHEMA_USE_STDREGEX + + void AddType(const ValueType& type) { + if (type == GetNullString() ) type_ |= 1 << kNullSchemaType; + else if (type == GetBooleanString()) type_ |= 1 << kBooleanSchemaType; + else if (type == GetObjectString() ) type_ |= 1 << kObjectSchemaType; + else if (type == GetArrayString() ) type_ |= 1 << kArraySchemaType; + else if (type == GetStringString() ) type_ |= 1 << kStringSchemaType; + else if (type == GetIntegerString()) type_ |= 1 << kIntegerSchemaType; + else if (type == GetNumberString() ) type_ |= (1 << kNumberSchemaType) | (1 << kIntegerSchemaType); + } + + bool CreateParallelValidator(Context& context) const { + if (enum_ || context.arrayUniqueness) + context.hasher = context.factory.CreateHasher(); + + if (validatorCount_) { + RAPIDJSON_ASSERT(context.validators == 0); + context.validators = static_cast(context.factory.MallocState(sizeof(ISchemaValidator*) * validatorCount_)); + context.validatorCount = validatorCount_; + + if (allOf_.schemas) + CreateSchemaValidators(context, allOf_); + + if (anyOf_.schemas) + CreateSchemaValidators(context, anyOf_); + + if (oneOf_.schemas) + CreateSchemaValidators(context, oneOf_); + + if (not_) + context.validators[notValidatorIndex_] = context.factory.CreateSchemaValidator(*not_); + + if (hasSchemaDependencies_) { + for (SizeType i = 0; i < propertyCount_; i++) + if (properties_[i].dependenciesSchema) + context.validators[properties_[i].dependenciesValidatorIndex] = context.factory.CreateSchemaValidator(*properties_[i].dependenciesSchema); + } + } + + return true; + } + + void CreateSchemaValidators(Context& context, const SchemaArray& schemas) const { + for (SizeType i = 0; i < schemas.count; i++) + context.validators[schemas.begin + i] = context.factory.CreateSchemaValidator(*schemas.schemas[i]); + } + + // O(n) + bool FindPropertyIndex(const ValueType& name, SizeType* outIndex) const { + SizeType len = name.GetStringLength(); + const Ch* str = name.GetString(); + for (SizeType index = 0; index < propertyCount_; index++) + if (properties_[index].name.GetStringLength() == len && + (std::memcmp(properties_[index].name.GetString(), str, sizeof(Ch) * len) == 0)) + { + *outIndex = index; + return true; + } + return false; + } + + bool CheckInt(Context& context, int64_t i) const { + if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + + if (!minimum_.IsNull()) { + if (minimum_.IsInt64()) { + if (exclusiveMinimum_ ? i <= minimum_.GetInt64() : i < minimum_.GetInt64()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } + else if (minimum_.IsUint64()) { + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); // i <= max(int64_t) < minimum.GetUint64() + } + else if (!CheckDoubleMinimum(context, static_cast(i))) + return false; + } + + if (!maximum_.IsNull()) { + if (maximum_.IsInt64()) { + if (exclusiveMaximum_ ? i >= maximum_.GetInt64() : i > maximum_.GetInt64()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } + else if (maximum_.IsUint64()) { } + /* do nothing */ // i <= max(int64_t) < maximum_.GetUint64() + else if (!CheckDoubleMaximum(context, static_cast(i))) + return false; + } + + if (!multipleOf_.IsNull()) { + if (multipleOf_.IsUint64()) { + if (static_cast(i >= 0 ? i : -i) % multipleOf_.GetUint64() != 0) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } + else if (!CheckDoubleMultipleOf(context, static_cast(i))) + return false; + } + + return true; + } + + bool CheckUint(Context& context, uint64_t i) const { + if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + + if (!minimum_.IsNull()) { + if (minimum_.IsUint64()) { + if (exclusiveMinimum_ ? i <= minimum_.GetUint64() : i < minimum_.GetUint64()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } + else if (minimum_.IsInt64()) + /* do nothing */; // i >= 0 > minimum.Getint64() + else if (!CheckDoubleMinimum(context, static_cast(i))) + return false; + } + + if (!maximum_.IsNull()) { + if (maximum_.IsUint64()) { + if (exclusiveMaximum_ ? i >= maximum_.GetUint64() : i > maximum_.GetUint64()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } + else if (maximum_.IsInt64()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); // i >= 0 > maximum_ + else if (!CheckDoubleMaximum(context, static_cast(i))) + return false; + } + + if (!multipleOf_.IsNull()) { + if (multipleOf_.IsUint64()) { + if (i % multipleOf_.GetUint64() != 0) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } + else if (!CheckDoubleMultipleOf(context, static_cast(i))) + return false; + } + + return true; + } + + bool CheckDoubleMinimum(Context& context, double d) const { + if (exclusiveMinimum_ ? d <= minimum_.GetDouble() : d < minimum_.GetDouble()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + return true; + } + + bool CheckDoubleMaximum(Context& context, double d) const { + if (exclusiveMaximum_ ? d >= maximum_.GetDouble() : d > maximum_.GetDouble()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + return true; + } + + bool CheckDoubleMultipleOf(Context& context, double d) const { + double a = std::abs(d), b = std::abs(multipleOf_.GetDouble()); + double q = std::floor(a / b); + double r = a - q * b; + if (r > 0.0) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + return true; + } + + struct Property { + Property() : schema(), dependenciesSchema(), dependenciesValidatorIndex(), dependencies(), required(false) {} + ~Property() { AllocatorType::Free(dependencies); } + SValue name; + const SchemaType* schema; + const SchemaType* dependenciesSchema; + SizeType dependenciesValidatorIndex; + bool* dependencies; + bool required; + }; + + struct PatternProperty { + PatternProperty() : schema(), pattern() {} + ~PatternProperty() { + if (pattern) { + pattern->~RegexType(); + AllocatorType::Free(pattern); + } + } + const SchemaType* schema; + RegexType* pattern; + }; + + AllocatorType* allocator_; + const SchemaType* typeless_; + uint64_t* enum_; + SizeType enumCount_; + SchemaArray allOf_; + SchemaArray anyOf_; + SchemaArray oneOf_; + const SchemaType* not_; + unsigned type_; // bitmask of kSchemaType + SizeType validatorCount_; + SizeType notValidatorIndex_; + + Property* properties_; + const SchemaType* additionalPropertiesSchema_; + PatternProperty* patternProperties_; + SizeType patternPropertyCount_; + SizeType propertyCount_; + SizeType minProperties_; + SizeType maxProperties_; + bool additionalProperties_; + bool hasDependencies_; + bool hasRequired_; + bool hasSchemaDependencies_; + + const SchemaType* additionalItemsSchema_; + const SchemaType* itemsList_; + const SchemaType** itemsTuple_; + SizeType itemsTupleCount_; + SizeType minItems_; + SizeType maxItems_; + bool additionalItems_; + bool uniqueItems_; + + RegexType* pattern_; + SizeType minLength_; + SizeType maxLength_; + + SValue minimum_; + SValue maximum_; + SValue multipleOf_; + bool exclusiveMinimum_; + bool exclusiveMaximum_; +}; + +template +struct TokenHelper { + RAPIDJSON_FORCEINLINE static void AppendIndexToken(Stack& documentStack, SizeType index) { + *documentStack.template Push() = '/'; + char buffer[21]; + size_t length = static_cast((sizeof(SizeType) == 4 ? u32toa(index, buffer) : u64toa(index, buffer)) - buffer); + for (size_t i = 0; i < length; i++) + *documentStack.template Push() = static_cast(buffer[i]); + } +}; + +// Partial specialized version for char to prevent buffer copying. +template +struct TokenHelper { + RAPIDJSON_FORCEINLINE static void AppendIndexToken(Stack& documentStack, SizeType index) { + if (sizeof(SizeType) == 4) { + char *buffer = documentStack.template Push(1 + 10); // '/' + uint + *buffer++ = '/'; + const char* end = internal::u32toa(index, buffer); + documentStack.template Pop(static_cast(10 - (end - buffer))); + } + else { + char *buffer = documentStack.template Push(1 + 20); // '/' + uint64 + *buffer++ = '/'; + const char* end = internal::u64toa(index, buffer); + documentStack.template Pop(static_cast(20 - (end - buffer))); + } + } +}; + +} // namespace internal + +/////////////////////////////////////////////////////////////////////////////// +// IGenericRemoteSchemaDocumentProvider + +template +class IGenericRemoteSchemaDocumentProvider { +public: + typedef typename SchemaDocumentType::Ch Ch; + + virtual ~IGenericRemoteSchemaDocumentProvider() {} + virtual const SchemaDocumentType* GetRemoteDocument(const Ch* uri, SizeType length) = 0; +}; + +/////////////////////////////////////////////////////////////////////////////// +// GenericSchemaDocument + +//! JSON schema document. +/*! + A JSON schema document is a compiled version of a JSON schema. + It is basically a tree of internal::Schema. + + \note This is an immutable class (i.e. its instance cannot be modified after construction). + \tparam ValueT Type of JSON value (e.g. \c Value ), which also determine the encoding. + \tparam Allocator Allocator type for allocating memory of this document. +*/ +template +class GenericSchemaDocument { +public: + typedef ValueT ValueType; + typedef IGenericRemoteSchemaDocumentProvider IRemoteSchemaDocumentProviderType; + typedef Allocator AllocatorType; + typedef typename ValueType::EncodingType EncodingType; + typedef typename EncodingType::Ch Ch; + typedef internal::Schema SchemaType; + typedef GenericPointer PointerType; + friend class internal::Schema; + template + friend class GenericSchemaValidator; + + //! Constructor. + /*! + Compile a JSON document into schema document. + + \param document A JSON document as source. + \param remoteProvider An optional remote schema document provider for resolving remote reference. Can be null. + \param allocator An optional allocator instance for allocating memory. Can be null. + */ + explicit GenericSchemaDocument(const ValueType& document, IRemoteSchemaDocumentProviderType* remoteProvider = 0, Allocator* allocator = 0) : + remoteProvider_(remoteProvider), + allocator_(allocator), + ownAllocator_(), + root_(), + typeless_(), + schemaMap_(allocator, kInitialSchemaMapSize), + schemaRef_(allocator, kInitialSchemaRefSize) + { + if (!allocator_) + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + + typeless_ = static_cast(allocator_->Malloc(sizeof(SchemaType))); + new (typeless_) SchemaType(this, PointerType(), ValueType(kObjectType).Move(), ValueType(kObjectType).Move(), 0); + + // Generate root schema, it will call CreateSchema() to create sub-schemas, + // And call AddRefSchema() if there are $ref. + CreateSchemaRecursive(&root_, PointerType(), document, document); + + // Resolve $ref + while (!schemaRef_.Empty()) { + SchemaRefEntry* refEntry = schemaRef_.template Pop(1); + if (const SchemaType* s = GetSchema(refEntry->target)) { + if (refEntry->schema) + *refEntry->schema = s; + + // Create entry in map if not exist + if (!GetSchema(refEntry->source)) { + new (schemaMap_.template Push()) SchemaEntry(refEntry->source, const_cast(s), false, allocator_); + } + } + else if (refEntry->schema) + *refEntry->schema = typeless_; + + refEntry->~SchemaRefEntry(); + } + + RAPIDJSON_ASSERT(root_ != 0); + + schemaRef_.ShrinkToFit(); // Deallocate all memory for ref + } + +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + //! Move constructor in C++11 + GenericSchemaDocument(GenericSchemaDocument&& rhs) RAPIDJSON_NOEXCEPT : + remoteProvider_(rhs.remoteProvider_), + allocator_(rhs.allocator_), + ownAllocator_(rhs.ownAllocator_), + root_(rhs.root_), + typeless_(rhs.typeless_), + schemaMap_(std::move(rhs.schemaMap_)), + schemaRef_(std::move(rhs.schemaRef_)) + { + rhs.remoteProvider_ = 0; + rhs.allocator_ = 0; + rhs.ownAllocator_ = 0; + rhs.typeless_ = 0; + } +#endif + + //! Destructor + ~GenericSchemaDocument() { + while (!schemaMap_.Empty()) + schemaMap_.template Pop(1)->~SchemaEntry(); + + if (typeless_) { + typeless_->~SchemaType(); + Allocator::Free(typeless_); + } + + RAPIDJSON_DELETE(ownAllocator_); + } + + //! Get the root schema. + const SchemaType& GetRoot() const { return *root_; } + +private: + //! Prohibit copying + GenericSchemaDocument(const GenericSchemaDocument&); + //! Prohibit assignment + GenericSchemaDocument& operator=(const GenericSchemaDocument&); + + struct SchemaRefEntry { + SchemaRefEntry(const PointerType& s, const PointerType& t, const SchemaType** outSchema, Allocator *allocator) : source(s, allocator), target(t, allocator), schema(outSchema) {} + PointerType source; + PointerType target; + const SchemaType** schema; + }; + + struct SchemaEntry { + SchemaEntry(const PointerType& p, SchemaType* s, bool o, Allocator* allocator) : pointer(p, allocator), schema(s), owned(o) {} + ~SchemaEntry() { + if (owned) { + schema->~SchemaType(); + Allocator::Free(schema); + } + } + PointerType pointer; + SchemaType* schema; + bool owned; + }; + + void CreateSchemaRecursive(const SchemaType** schema, const PointerType& pointer, const ValueType& v, const ValueType& document) { + if (schema) + *schema = typeless_; + + if (v.GetType() == kObjectType) { + const SchemaType* s = GetSchema(pointer); + if (!s) + CreateSchema(schema, pointer, v, document); + + for (typename ValueType::ConstMemberIterator itr = v.MemberBegin(); itr != v.MemberEnd(); ++itr) + CreateSchemaRecursive(0, pointer.Append(itr->name, allocator_), itr->value, document); + } + else if (v.GetType() == kArrayType) + for (SizeType i = 0; i < v.Size(); i++) + CreateSchemaRecursive(0, pointer.Append(i, allocator_), v[i], document); + } + + void CreateSchema(const SchemaType** schema, const PointerType& pointer, const ValueType& v, const ValueType& document) { + RAPIDJSON_ASSERT(pointer.IsValid()); + if (v.IsObject()) { + if (!HandleRefSchema(pointer, schema, v, document)) { + SchemaType* s = new (allocator_->Malloc(sizeof(SchemaType))) SchemaType(this, pointer, v, document, allocator_); + new (schemaMap_.template Push()) SchemaEntry(pointer, s, true, allocator_); + if (schema) + *schema = s; + } + } + } + + bool HandleRefSchema(const PointerType& source, const SchemaType** schema, const ValueType& v, const ValueType& document) { + static const Ch kRefString[] = { '$', 'r', 'e', 'f', '\0' }; + static const ValueType kRefValue(kRefString, 4); + + typename ValueType::ConstMemberIterator itr = v.FindMember(kRefValue); + if (itr == v.MemberEnd()) + return false; + + if (itr->value.IsString()) { + SizeType len = itr->value.GetStringLength(); + if (len > 0) { + const Ch* s = itr->value.GetString(); + SizeType i = 0; + while (i < len && s[i] != '#') // Find the first # + i++; + + if (i > 0) { // Remote reference, resolve immediately + if (remoteProvider_) { + if (const GenericSchemaDocument* remoteDocument = remoteProvider_->GetRemoteDocument(s, i)) { + PointerType pointer(&s[i], len - i, allocator_); + if (pointer.IsValid()) { + if (const SchemaType* sc = remoteDocument->GetSchema(pointer)) { + if (schema) + *schema = sc; + return true; + } + } + } + } + } + else if (s[i] == '#') { // Local reference, defer resolution + PointerType pointer(&s[i], len - i, allocator_); + if (pointer.IsValid()) { + if (const ValueType* nv = pointer.Get(document)) + if (HandleRefSchema(source, schema, *nv, document)) + return true; + + new (schemaRef_.template Push()) SchemaRefEntry(source, pointer, schema, allocator_); + return true; + } + } + } + } + return false; + } + + const SchemaType* GetSchema(const PointerType& pointer) const { + for (const SchemaEntry* target = schemaMap_.template Bottom(); target != schemaMap_.template End(); ++target) + if (pointer == target->pointer) + return target->schema; + return 0; + } + + PointerType GetPointer(const SchemaType* schema) const { + for (const SchemaEntry* target = schemaMap_.template Bottom(); target != schemaMap_.template End(); ++target) + if (schema == target->schema) + return target->pointer; + return PointerType(); + } + + const SchemaType* GetTypeless() const { return typeless_; } + + static const size_t kInitialSchemaMapSize = 64; + static const size_t kInitialSchemaRefSize = 64; + + IRemoteSchemaDocumentProviderType* remoteProvider_; + Allocator *allocator_; + Allocator *ownAllocator_; + const SchemaType* root_; //!< Root schema. + SchemaType* typeless_; + internal::Stack schemaMap_; // Stores created Pointer -> Schemas + internal::Stack schemaRef_; // Stores Pointer from $ref and schema which holds the $ref +}; + +//! GenericSchemaDocument using Value type. +typedef GenericSchemaDocument SchemaDocument; +//! IGenericRemoteSchemaDocumentProvider using SchemaDocument. +typedef IGenericRemoteSchemaDocumentProvider IRemoteSchemaDocumentProvider; + +/////////////////////////////////////////////////////////////////////////////// +// GenericSchemaValidator + +//! JSON Schema Validator. +/*! + A SAX style JSON schema validator. + It uses a \c GenericSchemaDocument to validate SAX events. + It delegates the incoming SAX events to an output handler. + The default output handler does nothing. + It can be reused multiple times by calling \c Reset(). + + \tparam SchemaDocumentType Type of schema document. + \tparam OutputHandler Type of output handler. Default handler does nothing. + \tparam StateAllocator Allocator for storing the internal validation states. +*/ +template < + typename SchemaDocumentType, + typename OutputHandler = BaseReaderHandler, + typename StateAllocator = CrtAllocator> +class GenericSchemaValidator : + public internal::ISchemaStateFactory, + public internal::ISchemaValidator +{ +public: + typedef typename SchemaDocumentType::SchemaType SchemaType; + typedef typename SchemaDocumentType::PointerType PointerType; + typedef typename SchemaType::EncodingType EncodingType; + typedef typename EncodingType::Ch Ch; + + //! Constructor without output handler. + /*! + \param schemaDocument The schema document to conform to. + \param allocator Optional allocator for storing internal validation states. + \param schemaStackCapacity Optional initial capacity of schema path stack. + \param documentStackCapacity Optional initial capacity of document path stack. + */ + GenericSchemaValidator( + const SchemaDocumentType& schemaDocument, + StateAllocator* allocator = 0, + size_t schemaStackCapacity = kDefaultSchemaStackCapacity, + size_t documentStackCapacity = kDefaultDocumentStackCapacity) + : + schemaDocument_(&schemaDocument), + root_(schemaDocument.GetRoot()), + stateAllocator_(allocator), + ownStateAllocator_(0), + schemaStack_(allocator, schemaStackCapacity), + documentStack_(allocator, documentStackCapacity), + outputHandler_(0), + valid_(true) +#if RAPIDJSON_SCHEMA_VERBOSE + , depth_(0) +#endif + { + } + + //! Constructor with output handler. + /*! + \param schemaDocument The schema document to conform to. + \param allocator Optional allocator for storing internal validation states. + \param schemaStackCapacity Optional initial capacity of schema path stack. + \param documentStackCapacity Optional initial capacity of document path stack. + */ + GenericSchemaValidator( + const SchemaDocumentType& schemaDocument, + OutputHandler& outputHandler, + StateAllocator* allocator = 0, + size_t schemaStackCapacity = kDefaultSchemaStackCapacity, + size_t documentStackCapacity = kDefaultDocumentStackCapacity) + : + schemaDocument_(&schemaDocument), + root_(schemaDocument.GetRoot()), + stateAllocator_(allocator), + ownStateAllocator_(0), + schemaStack_(allocator, schemaStackCapacity), + documentStack_(allocator, documentStackCapacity), + outputHandler_(&outputHandler), + valid_(true) +#if RAPIDJSON_SCHEMA_VERBOSE + , depth_(0) +#endif + { + } + + //! Destructor. + ~GenericSchemaValidator() { + Reset(); + RAPIDJSON_DELETE(ownStateAllocator_); + } + + //! Reset the internal states. + void Reset() { + while (!schemaStack_.Empty()) + PopSchema(); + documentStack_.Clear(); + valid_ = true; + } + + //! Checks whether the current state is valid. + // Implementation of ISchemaValidator + virtual bool IsValid() const { return valid_; } + + //! Gets the JSON pointer pointed to the invalid schema. + PointerType GetInvalidSchemaPointer() const { + return schemaStack_.Empty() ? PointerType() : schemaDocument_->GetPointer(&CurrentSchema()); + } + + //! Gets the keyword of invalid schema. + const Ch* GetInvalidSchemaKeyword() const { + return schemaStack_.Empty() ? 0 : CurrentContext().invalidKeyword; + } + + //! Gets the JSON pointer pointed to the invalid value. + PointerType GetInvalidDocumentPointer() const { + return documentStack_.Empty() ? PointerType() : PointerType(documentStack_.template Bottom(), documentStack_.GetSize() / sizeof(Ch)); + } + +#if RAPIDJSON_SCHEMA_VERBOSE +#define RAPIDJSON_SCHEMA_HANDLE_BEGIN_VERBOSE_() \ +RAPIDJSON_MULTILINEMACRO_BEGIN\ + *documentStack_.template Push() = '\0';\ + documentStack_.template Pop(1);\ + internal::PrintInvalidDocument(documentStack_.template Bottom());\ +RAPIDJSON_MULTILINEMACRO_END +#else +#define RAPIDJSON_SCHEMA_HANDLE_BEGIN_VERBOSE_() +#endif + +#define RAPIDJSON_SCHEMA_HANDLE_BEGIN_(method, arg1)\ + if (!valid_) return false; \ + if (!BeginValue() || !CurrentSchema().method arg1) {\ + RAPIDJSON_SCHEMA_HANDLE_BEGIN_VERBOSE_();\ + return valid_ = false;\ + } + +#define RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(method, arg2)\ + for (Context* context = schemaStack_.template Bottom(); context != schemaStack_.template End(); context++) {\ + if (context->hasher)\ + static_cast(context->hasher)->method arg2;\ + if (context->validators)\ + for (SizeType i_ = 0; i_ < context->validatorCount; i_++)\ + static_cast(context->validators[i_])->method arg2;\ + if (context->patternPropertiesValidators)\ + for (SizeType i_ = 0; i_ < context->patternPropertiesValidatorCount; i_++)\ + static_cast(context->patternPropertiesValidators[i_])->method arg2;\ + } + +#define RAPIDJSON_SCHEMA_HANDLE_END_(method, arg2)\ + return valid_ = EndValue() && (!outputHandler_ || outputHandler_->method arg2) + +#define RAPIDJSON_SCHEMA_HANDLE_VALUE_(method, arg1, arg2) \ + RAPIDJSON_SCHEMA_HANDLE_BEGIN_ (method, arg1);\ + RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(method, arg2);\ + RAPIDJSON_SCHEMA_HANDLE_END_ (method, arg2) + + bool Null() { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Null, (CurrentContext() ), ( )); } + bool Bool(bool b) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Bool, (CurrentContext(), b), (b)); } + bool Int(int i) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Int, (CurrentContext(), i), (i)); } + bool Uint(unsigned u) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Uint, (CurrentContext(), u), (u)); } + bool Int64(int64_t i) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Int64, (CurrentContext(), i), (i)); } + bool Uint64(uint64_t u) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Uint64, (CurrentContext(), u), (u)); } + bool Double(double d) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Double, (CurrentContext(), d), (d)); } + bool RawNumber(const Ch* str, SizeType length, bool copy) + { RAPIDJSON_SCHEMA_HANDLE_VALUE_(String, (CurrentContext(), str, length, copy), (str, length, copy)); } + bool String(const Ch* str, SizeType length, bool copy) + { RAPIDJSON_SCHEMA_HANDLE_VALUE_(String, (CurrentContext(), str, length, copy), (str, length, copy)); } + + bool StartObject() { + RAPIDJSON_SCHEMA_HANDLE_BEGIN_(StartObject, (CurrentContext())); + RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(StartObject, ()); + return valid_ = !outputHandler_ || outputHandler_->StartObject(); + } + + bool Key(const Ch* str, SizeType len, bool copy) { + if (!valid_) return false; + AppendToken(str, len); + if (!CurrentSchema().Key(CurrentContext(), str, len, copy)) return valid_ = false; + RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(Key, (str, len, copy)); + return valid_ = !outputHandler_ || outputHandler_->Key(str, len, copy); + } + + bool EndObject(SizeType memberCount) { + if (!valid_) return false; + RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(EndObject, (memberCount)); + if (!CurrentSchema().EndObject(CurrentContext(), memberCount)) return valid_ = false; + RAPIDJSON_SCHEMA_HANDLE_END_(EndObject, (memberCount)); + } + + bool StartArray() { + RAPIDJSON_SCHEMA_HANDLE_BEGIN_(StartArray, (CurrentContext())); + RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(StartArray, ()); + return valid_ = !outputHandler_ || outputHandler_->StartArray(); + } + + bool EndArray(SizeType elementCount) { + if (!valid_) return false; + RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(EndArray, (elementCount)); + if (!CurrentSchema().EndArray(CurrentContext(), elementCount)) return valid_ = false; + RAPIDJSON_SCHEMA_HANDLE_END_(EndArray, (elementCount)); + } + +#undef RAPIDJSON_SCHEMA_HANDLE_BEGIN_VERBOSE_ +#undef RAPIDJSON_SCHEMA_HANDLE_BEGIN_ +#undef RAPIDJSON_SCHEMA_HANDLE_PARALLEL_ +#undef RAPIDJSON_SCHEMA_HANDLE_VALUE_ + + // Implementation of ISchemaStateFactory + virtual ISchemaValidator* CreateSchemaValidator(const SchemaType& root) { + return new (GetStateAllocator().Malloc(sizeof(GenericSchemaValidator))) GenericSchemaValidator(*schemaDocument_, root, +#if RAPIDJSON_SCHEMA_VERBOSE + depth_ + 1, +#endif + &GetStateAllocator()); + } + + virtual void DestroySchemaValidator(ISchemaValidator* validator) { + GenericSchemaValidator* v = static_cast(validator); + v->~GenericSchemaValidator(); + StateAllocator::Free(v); + } + + virtual void* CreateHasher() { + return new (GetStateAllocator().Malloc(sizeof(HasherType))) HasherType(&GetStateAllocator()); + } + + virtual uint64_t GetHashCode(void* hasher) { + return static_cast(hasher)->GetHashCode(); + } + + virtual void DestroryHasher(void* hasher) { + HasherType* h = static_cast(hasher); + h->~HasherType(); + StateAllocator::Free(h); + } + + virtual void* MallocState(size_t size) { + return GetStateAllocator().Malloc(size); + } + + virtual void FreeState(void* p) { + StateAllocator::Free(p); + } + +private: + typedef typename SchemaType::Context Context; + typedef GenericValue, StateAllocator> HashCodeArray; + typedef internal::Hasher HasherType; + + GenericSchemaValidator( + const SchemaDocumentType& schemaDocument, + const SchemaType& root, +#if RAPIDJSON_SCHEMA_VERBOSE + unsigned depth, +#endif + StateAllocator* allocator = 0, + size_t schemaStackCapacity = kDefaultSchemaStackCapacity, + size_t documentStackCapacity = kDefaultDocumentStackCapacity) + : + schemaDocument_(&schemaDocument), + root_(root), + stateAllocator_(allocator), + ownStateAllocator_(0), + schemaStack_(allocator, schemaStackCapacity), + documentStack_(allocator, documentStackCapacity), + outputHandler_(0), + valid_(true) +#if RAPIDJSON_SCHEMA_VERBOSE + , depth_(depth) +#endif + { + } + + StateAllocator& GetStateAllocator() { + if (!stateAllocator_) + stateAllocator_ = ownStateAllocator_ = RAPIDJSON_NEW(StateAllocator)(); + return *stateAllocator_; + } + + bool BeginValue() { + if (schemaStack_.Empty()) + PushSchema(root_); + else { + if (CurrentContext().inArray) + internal::TokenHelper, Ch>::AppendIndexToken(documentStack_, CurrentContext().arrayElementIndex); + + if (!CurrentSchema().BeginValue(CurrentContext())) + return false; + + SizeType count = CurrentContext().patternPropertiesSchemaCount; + const SchemaType** sa = CurrentContext().patternPropertiesSchemas; + typename Context::PatternValidatorType patternValidatorType = CurrentContext().valuePatternValidatorType; + bool valueUniqueness = CurrentContext().valueUniqueness; + RAPIDJSON_ASSERT(CurrentContext().valueSchema); + PushSchema(*CurrentContext().valueSchema); + + if (count > 0) { + CurrentContext().objectPatternValidatorType = patternValidatorType; + ISchemaValidator**& va = CurrentContext().patternPropertiesValidators; + SizeType& validatorCount = CurrentContext().patternPropertiesValidatorCount; + va = static_cast(MallocState(sizeof(ISchemaValidator*) * count)); + for (SizeType i = 0; i < count; i++) + va[validatorCount++] = CreateSchemaValidator(*sa[i]); + } + + CurrentContext().arrayUniqueness = valueUniqueness; + } + return true; + } + + bool EndValue() { + if (!CurrentSchema().EndValue(CurrentContext())) + return false; + +#if RAPIDJSON_SCHEMA_VERBOSE + GenericStringBuffer sb; + schemaDocument_->GetPointer(&CurrentSchema()).Stringify(sb); + + *documentStack_.template Push() = '\0'; + documentStack_.template Pop(1); + internal::PrintValidatorPointers(depth_, sb.GetString(), documentStack_.template Bottom()); +#endif + + uint64_t h = CurrentContext().arrayUniqueness ? static_cast(CurrentContext().hasher)->GetHashCode() : 0; + + PopSchema(); + + if (!schemaStack_.Empty()) { + Context& context = CurrentContext(); + if (context.valueUniqueness) { + HashCodeArray* a = static_cast(context.arrayElementHashCodes); + if (!a) + CurrentContext().arrayElementHashCodes = a = new (GetStateAllocator().Malloc(sizeof(HashCodeArray))) HashCodeArray(kArrayType); + for (typename HashCodeArray::ConstValueIterator itr = a->Begin(); itr != a->End(); ++itr) + if (itr->GetUint64() == h) + RAPIDJSON_INVALID_KEYWORD_RETURN(SchemaType::GetUniqueItemsString()); + a->PushBack(h, GetStateAllocator()); + } + } + + // Remove the last token of document pointer + while (!documentStack_.Empty() && *documentStack_.template Pop(1) != '/') + ; + + return true; + } + + void AppendToken(const Ch* str, SizeType len) { + documentStack_.template Reserve(1 + len * 2); // worst case all characters are escaped as two characters + *documentStack_.template PushUnsafe() = '/'; + for (SizeType i = 0; i < len; i++) { + if (str[i] == '~') { + *documentStack_.template PushUnsafe() = '~'; + *documentStack_.template PushUnsafe() = '0'; + } + else if (str[i] == '/') { + *documentStack_.template PushUnsafe() = '~'; + *documentStack_.template PushUnsafe() = '1'; + } + else + *documentStack_.template PushUnsafe() = str[i]; + } + } + + RAPIDJSON_FORCEINLINE void PushSchema(const SchemaType& schema) { new (schemaStack_.template Push()) Context(*this, &schema); } + + RAPIDJSON_FORCEINLINE void PopSchema() { + Context* c = schemaStack_.template Pop(1); + if (HashCodeArray* a = static_cast(c->arrayElementHashCodes)) { + a->~HashCodeArray(); + StateAllocator::Free(a); + } + c->~Context(); + } + + const SchemaType& CurrentSchema() const { return *schemaStack_.template Top()->schema; } + Context& CurrentContext() { return *schemaStack_.template Top(); } + const Context& CurrentContext() const { return *schemaStack_.template Top(); } + + static const size_t kDefaultSchemaStackCapacity = 1024; + static const size_t kDefaultDocumentStackCapacity = 256; + const SchemaDocumentType* schemaDocument_; + const SchemaType& root_; + StateAllocator* stateAllocator_; + StateAllocator* ownStateAllocator_; + internal::Stack schemaStack_; //!< stack to store the current path of schema (BaseSchemaType *) + internal::Stack documentStack_; //!< stack to store the current path of validating document (Ch) + OutputHandler* outputHandler_; + bool valid_; +#if RAPIDJSON_SCHEMA_VERBOSE + unsigned depth_; +#endif +}; + +typedef GenericSchemaValidator SchemaValidator; + +/////////////////////////////////////////////////////////////////////////////// +// SchemaValidatingReader + +//! A helper class for parsing with validation. +/*! + This helper class is a functor, designed as a parameter of \ref GenericDocument::Populate(). + + \tparam parseFlags Combination of \ref ParseFlag. + \tparam InputStream Type of input stream, implementing Stream concept. + \tparam SourceEncoding Encoding of the input stream. + \tparam SchemaDocumentType Type of schema document. + \tparam StackAllocator Allocator type for stack. +*/ +template < + unsigned parseFlags, + typename InputStream, + typename SourceEncoding, + typename SchemaDocumentType = SchemaDocument, + typename StackAllocator = CrtAllocator> +class SchemaValidatingReader { +public: + typedef typename SchemaDocumentType::PointerType PointerType; + typedef typename InputStream::Ch Ch; + + //! Constructor + /*! + \param is Input stream. + \param sd Schema document. + */ + SchemaValidatingReader(InputStream& is, const SchemaDocumentType& sd) : is_(is), sd_(sd), invalidSchemaKeyword_(), isValid_(true) {} + + template + bool operator()(Handler& handler) { + GenericReader reader; + GenericSchemaValidator validator(sd_, handler); + parseResult_ = reader.template Parse(is_, validator); + + isValid_ = validator.IsValid(); + if (isValid_) { + invalidSchemaPointer_ = PointerType(); + invalidSchemaKeyword_ = 0; + invalidDocumentPointer_ = PointerType(); + } + else { + invalidSchemaPointer_ = validator.GetInvalidSchemaPointer(); + invalidSchemaKeyword_ = validator.GetInvalidSchemaKeyword(); + invalidDocumentPointer_ = validator.GetInvalidDocumentPointer(); + } + + return parseResult_; + } + + const ParseResult& GetParseResult() const { return parseResult_; } + bool IsValid() const { return isValid_; } + const PointerType& GetInvalidSchemaPointer() const { return invalidSchemaPointer_; } + const Ch* GetInvalidSchemaKeyword() const { return invalidSchemaKeyword_; } + const PointerType& GetInvalidDocumentPointer() const { return invalidDocumentPointer_; } + +private: + InputStream& is_; + const SchemaDocumentType& sd_; + + ParseResult parseResult_; + PointerType invalidSchemaPointer_; + const Ch* invalidSchemaKeyword_; + PointerType invalidDocumentPointer_; + bool isValid_; +}; + +RAPIDJSON_NAMESPACE_END +RAPIDJSON_DIAG_POP + +#endif // RAPIDJSON_SCHEMA_H_ diff --git a/rapidjson/stream.h b/rapidjson/stream.h new file mode 100644 index 0000000..fef82c2 --- /dev/null +++ b/rapidjson/stream.h @@ -0,0 +1,179 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#include "rapidjson.h" + +#ifndef RAPIDJSON_STREAM_H_ +#define RAPIDJSON_STREAM_H_ + +#include "encodings.h" + +RAPIDJSON_NAMESPACE_BEGIN + +/////////////////////////////////////////////////////////////////////////////// +// Stream + +/*! \class rapidjson::Stream + \brief Concept for reading and writing characters. + + For read-only stream, no need to implement PutBegin(), Put(), Flush() and PutEnd(). + + For write-only stream, only need to implement Put() and Flush(). + +\code +concept Stream { + typename Ch; //!< Character type of the stream. + + //! Read the current character from stream without moving the read cursor. + Ch Peek() const; + + //! Read the current character from stream and moving the read cursor to next character. + Ch Take(); + + //! Get the current read cursor. + //! \return Number of characters read from start. + size_t Tell(); + + //! Begin writing operation at the current read pointer. + //! \return The begin writer pointer. + Ch* PutBegin(); + + //! Write a character. + void Put(Ch c); + + //! Flush the buffer. + void Flush(); + + //! End the writing operation. + //! \param begin The begin write pointer returned by PutBegin(). + //! \return Number of characters written. + size_t PutEnd(Ch* begin); +} +\endcode +*/ + +//! Provides additional information for stream. +/*! + By using traits pattern, this type provides a default configuration for stream. + For custom stream, this type can be specialized for other configuration. + See TEST(Reader, CustomStringStream) in readertest.cpp for example. +*/ +template +struct StreamTraits { + //! Whether to make local copy of stream for optimization during parsing. + /*! + By default, for safety, streams do not use local copy optimization. + Stream that can be copied fast should specialize this, like StreamTraits. + */ + enum { copyOptimization = 0 }; +}; + +//! Reserve n characters for writing to a stream. +template +inline void PutReserve(Stream& stream, size_t count) { + (void)stream; + (void)count; +} + +//! Write character to a stream, presuming buffer is reserved. +template +inline void PutUnsafe(Stream& stream, typename Stream::Ch c) { + stream.Put(c); +} + +//! Put N copies of a character to a stream. +template +inline void PutN(Stream& stream, Ch c, size_t n) { + PutReserve(stream, n); + for (size_t i = 0; i < n; i++) + PutUnsafe(stream, c); +} + +/////////////////////////////////////////////////////////////////////////////// +// StringStream + +//! Read-only string stream. +/*! \note implements Stream concept +*/ +template +struct GenericStringStream { + typedef typename Encoding::Ch Ch; + + GenericStringStream(const Ch *src) : src_(src), head_(src) {} + + Ch Peek() const { return *src_; } + Ch Take() { return *src_++; } + size_t Tell() const { return static_cast(src_ - head_); } + + Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } + void Put(Ch) { RAPIDJSON_ASSERT(false); } + void Flush() { RAPIDJSON_ASSERT(false); } + size_t PutEnd(Ch*) { RAPIDJSON_ASSERT(false); return 0; } + + const Ch* src_; //!< Current read position. + const Ch* head_; //!< Original head of the string. +}; + +template +struct StreamTraits > { + enum { copyOptimization = 1 }; +}; + +//! String stream with UTF8 encoding. +typedef GenericStringStream > StringStream; + +/////////////////////////////////////////////////////////////////////////////// +// InsituStringStream + +//! A read-write string stream. +/*! This string stream is particularly designed for in-situ parsing. + \note implements Stream concept +*/ +template +struct GenericInsituStringStream { + typedef typename Encoding::Ch Ch; + + GenericInsituStringStream(Ch *src) : src_(src), dst_(0), head_(src) {} + + // Read + Ch Peek() { return *src_; } + Ch Take() { return *src_++; } + size_t Tell() { return static_cast(src_ - head_); } + + // Write + void Put(Ch c) { RAPIDJSON_ASSERT(dst_ != 0); *dst_++ = c; } + + Ch* PutBegin() { return dst_ = src_; } + size_t PutEnd(Ch* begin) { return static_cast(dst_ - begin); } + void Flush() {} + + Ch* Push(size_t count) { Ch* begin = dst_; dst_ += count; return begin; } + void Pop(size_t count) { dst_ -= count; } + + Ch* src_; + Ch* dst_; + Ch* head_; +}; + +template +struct StreamTraits > { + enum { copyOptimization = 1 }; +}; + +//! Insitu string stream with UTF8 encoding. +typedef GenericInsituStringStream > InsituStringStream; + +RAPIDJSON_NAMESPACE_END + +#endif // RAPIDJSON_STREAM_H_ diff --git a/rapidjson/stringbuffer.h b/rapidjson/stringbuffer.h index 1c9c80b..4e38b82 100644 --- a/rapidjson/stringbuffer.h +++ b/rapidjson/stringbuffer.h @@ -15,7 +15,8 @@ #ifndef RAPIDJSON_STRINGBUFFER_H_ #define RAPIDJSON_STRINGBUFFER_H_ -#include "rapidjson.h" +#include "stream.h" +#include "internal/stack.h" #if RAPIDJSON_HAS_CXX11_RVALUE_REFS #include // std::move @@ -23,6 +24,11 @@ #include "internal/stack.h" +#if defined(__clang__) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif + RAPIDJSON_NAMESPACE_BEGIN //! Represents an in-memory output stream. @@ -48,6 +54,7 @@ class GenericStringBuffer { #endif void Put(Ch c) { *stack_.template Push() = c; } + void PutUnsafe(Ch c) { *stack_.template PushUnsafe() = c; } void Flush() {} void Clear() { stack_.Clear(); } @@ -57,7 +64,10 @@ class GenericStringBuffer { stack_.ShrinkToFit(); stack_.template Pop(1); } + + void Reserve(size_t count) { stack_.template Reserve(count); } Ch* Push(size_t count) { return stack_.template Push(count); } + Ch* PushUnsafe(size_t count) { return stack_.template PushUnsafe(count); } void Pop(size_t count) { stack_.template Pop(count); } const Ch* GetString() const { @@ -68,8 +78,12 @@ class GenericStringBuffer { return stack_.template Bottom(); } + //! Get the size of string in bytes in the string buffer. size_t GetSize() const { return stack_.GetSize(); } + //! Get the length of string in Ch in the string buffer. + size_t GetLength() const { return stack_.GetSize() / sizeof(Ch); } + static const size_t kDefaultCapacity = 256; mutable internal::Stack stack_; @@ -82,6 +96,16 @@ class GenericStringBuffer { //! String buffer with UTF8 encoding typedef GenericStringBuffer > StringBuffer; +template +inline void PutReserve(GenericStringBuffer& stream, size_t count) { + stream.Reserve(count); +} + +template +inline void PutUnsafe(GenericStringBuffer& stream, typename Encoding::Ch c) { + stream.PutUnsafe(c); +} + //! Implement specialized version of PutN() with memset() for better performance. template<> inline void PutN(GenericStringBuffer >& stream, char c, size_t n) { @@ -90,4 +114,8 @@ inline void PutN(GenericStringBuffer >& stream, char c, size_t n) { RAPIDJSON_NAMESPACE_END +#if defined(__clang__) +RAPIDJSON_DIAG_POP +#endif + #endif // RAPIDJSON_STRINGBUFFER_H_ diff --git a/rapidjson/writer.h b/rapidjson/writer.h index e1eea38..e610ebb 100644 --- a/rapidjson/writer.h +++ b/rapidjson/writer.h @@ -15,7 +15,8 @@ #ifndef RAPIDJSON_WRITER_H_ #define RAPIDJSON_WRITER_H_ -#include "rapidjson.h" +#include "stream.h" +#include "internal/meta.h" #include "internal/stack.h" #include "internal/strfunc.h" #include "internal/dtoa.h" @@ -23,8 +24,16 @@ #include "stringbuffer.h" #include // placement new -#if RAPIDJSON_HAS_STDSTRING -#include +#if defined(RAPIDJSON_SIMD) && defined(_MSC_VER) +#include +#pragma intrinsic(_BitScanForward) +#endif +#ifdef RAPIDJSON_SSE42 +#include +#elif defined(RAPIDJSON_SSE2) +#include +#elif defined(RAPIDJSON_NEON) +#include #endif #ifdef _MSC_VER @@ -32,8 +41,36 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant #endif +#ifdef __clang__ +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(padded) +RAPIDJSON_DIAG_OFF(unreachable-code) +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif + RAPIDJSON_NAMESPACE_BEGIN +/////////////////////////////////////////////////////////////////////////////// +// WriteFlag + +/*! \def RAPIDJSON_WRITE_DEFAULT_FLAGS + \ingroup RAPIDJSON_CONFIG + \brief User-defined kWriteDefaultFlags definition. + + User can define this as any \c WriteFlag combinations. +*/ +#ifndef RAPIDJSON_WRITE_DEFAULT_FLAGS +#define RAPIDJSON_WRITE_DEFAULT_FLAGS kWriteNoFlags +#endif + +//! Combination of writeFlags +enum WriteFlag { + kWriteNoFlags = 0, //!< No flags are set. + kWriteValidateEncodingFlag = 1, //!< Validate encoding of JSON strings. + kWriteNanAndInfFlag = 2, //!< Allow writing of Infinity, -Infinity and NaN. + kWriteDefaultFlags = RAPIDJSON_WRITE_DEFAULT_FLAGS //!< Default write flags. Can be customized by defining RAPIDJSON_WRITE_DEFAULT_FLAGS +}; + //! JSON writer /*! Writer implements the concept Handler. It generates JSON text by events to an output os. @@ -50,11 +87,13 @@ RAPIDJSON_NAMESPACE_BEGIN \tparam StackAllocator Type of allocator for allocating memory of stack. \note implements Handler concept */ -template, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator> +template, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator, unsigned writeFlags = kWriteDefaultFlags> class Writer { public: typedef typename SourceEncoding::Ch Ch; + static const int kDefaultMaxDecimalPlaces = 324; + //! Constructor /*! \param os Output stream. \param stackAllocator User supplied allocator. If it is null, it will create a private one. @@ -62,11 +101,18 @@ class Writer { */ explicit Writer(OutputStream& os, StackAllocator* stackAllocator = 0, size_t levelDepth = kDefaultLevelDepth) : - os_(&os), level_stack_(stackAllocator, levelDepth * sizeof(Level)), hasRoot_(false) {} + os_(&os), level_stack_(stackAllocator, levelDepth * sizeof(Level)), maxDecimalPlaces_(kDefaultMaxDecimalPlaces), hasRoot_(false) {} explicit Writer(StackAllocator* allocator = 0, size_t levelDepth = kDefaultLevelDepth) : - os_(0), level_stack_(allocator, levelDepth * sizeof(Level)), hasRoot_(false) {} + os_(0), level_stack_(allocator, levelDepth * sizeof(Level)), maxDecimalPlaces_(kDefaultMaxDecimalPlaces), hasRoot_(false) {} + +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + Writer(Writer&& rhs) : + os_(rhs.os_), level_stack_(std::move(rhs.level_stack_)), maxDecimalPlaces_(rhs.maxDecimalPlaces_), hasRoot_(rhs.hasRoot_) { + rhs.os_ = 0; + } +#endif //! Reset the writer with a new stream. /*! @@ -100,29 +146,66 @@ class Writer { return hasRoot_ && level_stack_.Empty(); } + int GetMaxDecimalPlaces() const { + return maxDecimalPlaces_; + } + + //! Sets the maximum number of decimal places for double output. + /*! + This setting truncates the output with specified number of decimal places. + + For example, + + \code + writer.SetMaxDecimalPlaces(3); + writer.StartArray(); + writer.Double(0.12345); // "0.123" + writer.Double(0.0001); // "0.0" + writer.Double(1.234567890123456e30); // "1.234567890123456e30" (do not truncate significand for positive exponent) + writer.Double(1.23e-4); // "0.0" (do truncate significand for negative exponent) + writer.EndArray(); + \endcode + + The default setting does not truncate any decimal places. You can restore to this setting by calling + \code + writer.SetMaxDecimalPlaces(Writer::kDefaultMaxDecimalPlaces); + \endcode + */ + void SetMaxDecimalPlaces(int maxDecimalPlaces) { + maxDecimalPlaces_ = maxDecimalPlaces; + } + /*!@name Implementation of Handler \see Handler */ //@{ - bool Null() { Prefix(kNullType); return WriteNull(); } - bool Bool(bool b) { Prefix(b ? kTrueType : kFalseType); return WriteBool(b); } - bool Int(int i) { Prefix(kNumberType); return WriteInt(i); } - bool Uint(unsigned u) { Prefix(kNumberType); return WriteUint(u); } - bool Int64(int64_t i64) { Prefix(kNumberType); return WriteInt64(i64); } - bool Uint64(uint64_t u64) { Prefix(kNumberType); return WriteUint64(u64); } + bool Null() { Prefix(kNullType); return EndValue(WriteNull()); } + bool Bool(bool b) { Prefix(b ? kTrueType : kFalseType); return EndValue(WriteBool(b)); } + bool Int(int i) { Prefix(kNumberType); return EndValue(WriteInt(i)); } + bool Uint(unsigned u) { Prefix(kNumberType); return EndValue(WriteUint(u)); } + bool Int64(int64_t i64) { Prefix(kNumberType); return EndValue(WriteInt64(i64)); } + bool Uint64(uint64_t u64) { Prefix(kNumberType); return EndValue(WriteUint64(u64)); } //! Writes the given \c double value to the stream /*! \param d The value to be written. \return Whether it is succeed. */ - bool Double(double d) { Prefix(kNumberType); return WriteDouble(d); } + bool Double(double d) { Prefix(kNumberType); return EndValue(WriteDouble(d)); } + + bool RawNumber(const Ch* str, SizeType length, bool copy = false) { + RAPIDJSON_ASSERT(str != 0); + (void)copy; + Prefix(kNumberType); + return EndValue(WriteString(str, length)); + } bool String(const Ch* str, SizeType length, bool copy = false) { + RAPIDJSON_ASSERT(str != 0); (void)copy; Prefix(kStringType); - return WriteString(str, length); + return EndValue(WriteString(str, length)); } #if RAPIDJSON_HAS_STDSTRING @@ -138,16 +221,21 @@ class Writer { } bool Key(const Ch* str, SizeType length, bool copy = false) { return String(str, length, copy); } + +#if RAPIDJSON_HAS_STDSTRING + bool Key(const std::basic_string& str) + { + return Key(str.data(), SizeType(str.size())); + } +#endif bool EndObject(SizeType memberCount = 0) { (void)memberCount; - RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); - RAPIDJSON_ASSERT(!level_stack_.template Top()->inArray); + RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); // not inside an Object + RAPIDJSON_ASSERT(!level_stack_.template Top()->inArray); // currently inside an Array, not Object + RAPIDJSON_ASSERT(0 == level_stack_.template Top()->valueCount % 2); // Object has a Key without a Value level_stack_.template Pop(1); - bool ret = WriteEndObject(); - if (level_stack_.Empty()) // end of json text - os_->Flush(); - return ret; + return EndValue(WriteEndObject()); } bool StartArray() { @@ -161,10 +249,7 @@ class Writer { RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); RAPIDJSON_ASSERT(level_stack_.template Top()->inArray); level_stack_.template Pop(1); - bool ret = WriteEndArray(); - if (level_stack_.Empty()) // end of json text - os_->Flush(); - return ret; + return EndValue(WriteEndArray()); } //@} @@ -172,11 +257,33 @@ class Writer { //@{ //! Simpler but slower overload. - bool String(const Ch* str) { return String(str, internal::StrLen(str)); } - bool Key(const Ch* str) { return Key(str, internal::StrLen(str)); } - + bool String(const Ch* const& str) { return String(str, internal::StrLen(str)); } + bool Key(const Ch* const& str) { return Key(str, internal::StrLen(str)); } + //@} + //! Write a raw JSON value. + /*! + For user to write a stringified JSON as a value. + + \param json A well-formed JSON value. It should not contain null character within [0, length - 1] range. + \param length Length of the json. + \param type Type of the root of json. + */ + bool RawValue(const Ch* json, size_t length, Type type) { + RAPIDJSON_ASSERT(json != 0); + Prefix(type); + return EndValue(WriteRawValue(json, length)); + } + + //! Flush the output stream. + /*! + Allows the user to flush the output stream immediately. + */ + void Flush() { + os_->Flush(); + } + protected: //! Information for each nested level struct Level { @@ -188,15 +295,18 @@ class Writer { static const size_t kDefaultLevelDepth = 32; bool WriteNull() { - os_->Put('n'); os_->Put('u'); os_->Put('l'); os_->Put('l'); return true; + PutReserve(*os_, 4); + PutUnsafe(*os_, 'n'); PutUnsafe(*os_, 'u'); PutUnsafe(*os_, 'l'); PutUnsafe(*os_, 'l'); return true; } bool WriteBool(bool b) { if (b) { - os_->Put('t'); os_->Put('r'); os_->Put('u'); os_->Put('e'); + PutReserve(*os_, 4); + PutUnsafe(*os_, 't'); PutUnsafe(*os_, 'r'); PutUnsafe(*os_, 'u'); PutUnsafe(*os_, 'e'); } else { - os_->Put('f'); os_->Put('a'); os_->Put('l'); os_->Put('s'); os_->Put('e'); + PutReserve(*os_, 5); + PutUnsafe(*os_, 'f'); PutUnsafe(*os_, 'a'); PutUnsafe(*os_, 'l'); PutUnsafe(*os_, 's'); PutUnsafe(*os_, 'e'); } return true; } @@ -204,45 +314,69 @@ class Writer { bool WriteInt(int i) { char buffer[11]; const char* end = internal::i32toa(i, buffer); + PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - os_->Put(*p); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteUint(unsigned u) { char buffer[10]; const char* end = internal::u32toa(u, buffer); + PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - os_->Put(*p); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteInt64(int64_t i64) { char buffer[21]; const char* end = internal::i64toa(i64, buffer); + PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - os_->Put(*p); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteUint64(uint64_t u64) { char buffer[20]; char* end = internal::u64toa(u64, buffer); + PutReserve(*os_, static_cast(end - buffer)); for (char* p = buffer; p != end; ++p) - os_->Put(*p); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteDouble(double d) { + if (internal::Double(d).IsNanOrInf()) { + if (!(writeFlags & kWriteNanAndInfFlag)) + return false; + if (internal::Double(d).IsNan()) { + PutReserve(*os_, 3); + PutUnsafe(*os_, 'N'); PutUnsafe(*os_, 'a'); PutUnsafe(*os_, 'N'); + return true; + } + if (internal::Double(d).Sign()) { + PutReserve(*os_, 9); + PutUnsafe(*os_, '-'); + } + else + PutReserve(*os_, 8); + PutUnsafe(*os_, 'I'); PutUnsafe(*os_, 'n'); PutUnsafe(*os_, 'f'); + PutUnsafe(*os_, 'i'); PutUnsafe(*os_, 'n'); PutUnsafe(*os_, 'i'); PutUnsafe(*os_, 't'); PutUnsafe(*os_, 'y'); + return true; + } + char buffer[25]; - char* end = internal::dtoa(d, buffer); + char* end = internal::dtoa(d, buffer, maxDecimalPlaces_); + PutReserve(*os_, static_cast(end - buffer)); for (char* p = buffer; p != end; ++p) - os_->Put(*p); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteString(const Ch* str, SizeType length) { - static const char hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; + static const typename OutputStream::Ch hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; static const char escape[256] = { #define Z16 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 //0 1 2 3 4 5 6 7 8 9 A B C D E F @@ -255,22 +389,27 @@ class Writer { #undef Z16 }; - os_->Put('\"'); + if (TargetEncoding::supportUnicode) + PutReserve(*os_, 2 + length * 6); // "\uxxxx..." + else + PutReserve(*os_, 2 + length * 12); // "\uxxxx\uyyyy..." + + PutUnsafe(*os_, '\"'); GenericStringStream is(str); - while (is.Tell() < length) { + while (ScanWriteUnescapedString(is, length)) { const Ch c = is.Peek(); - if (!TargetEncoding::supportUnicode && (unsigned)c >= 0x80) { + if (!TargetEncoding::supportUnicode && static_cast(c) >= 0x80) { // Unicode escaping unsigned codepoint; - if (!SourceEncoding::Decode(is, &codepoint)) + if (RAPIDJSON_UNLIKELY(!SourceEncoding::Decode(is, &codepoint))) return false; - os_->Put('\\'); - os_->Put('u'); + PutUnsafe(*os_, '\\'); + PutUnsafe(*os_, 'u'); if (codepoint <= 0xD7FF || (codepoint >= 0xE000 && codepoint <= 0xFFFF)) { - os_->Put(hexDigits[(codepoint >> 12) & 15]); - os_->Put(hexDigits[(codepoint >> 8) & 15]); - os_->Put(hexDigits[(codepoint >> 4) & 15]); - os_->Put(hexDigits[(codepoint ) & 15]); + PutUnsafe(*os_, hexDigits[(codepoint >> 12) & 15]); + PutUnsafe(*os_, hexDigits[(codepoint >> 8) & 15]); + PutUnsafe(*os_, hexDigits[(codepoint >> 4) & 15]); + PutUnsafe(*os_, hexDigits[(codepoint ) & 15]); } else { RAPIDJSON_ASSERT(codepoint >= 0x010000 && codepoint <= 0x10FFFF); @@ -278,45 +417,59 @@ class Writer { unsigned s = codepoint - 0x010000; unsigned lead = (s >> 10) + 0xD800; unsigned trail = (s & 0x3FF) + 0xDC00; - os_->Put(hexDigits[(lead >> 12) & 15]); - os_->Put(hexDigits[(lead >> 8) & 15]); - os_->Put(hexDigits[(lead >> 4) & 15]); - os_->Put(hexDigits[(lead ) & 15]); - os_->Put('\\'); - os_->Put('u'); - os_->Put(hexDigits[(trail >> 12) & 15]); - os_->Put(hexDigits[(trail >> 8) & 15]); - os_->Put(hexDigits[(trail >> 4) & 15]); - os_->Put(hexDigits[(trail ) & 15]); + PutUnsafe(*os_, hexDigits[(lead >> 12) & 15]); + PutUnsafe(*os_, hexDigits[(lead >> 8) & 15]); + PutUnsafe(*os_, hexDigits[(lead >> 4) & 15]); + PutUnsafe(*os_, hexDigits[(lead ) & 15]); + PutUnsafe(*os_, '\\'); + PutUnsafe(*os_, 'u'); + PutUnsafe(*os_, hexDigits[(trail >> 12) & 15]); + PutUnsafe(*os_, hexDigits[(trail >> 8) & 15]); + PutUnsafe(*os_, hexDigits[(trail >> 4) & 15]); + PutUnsafe(*os_, hexDigits[(trail ) & 15]); } } - else if ((sizeof(Ch) == 1 || (unsigned)c < 256) && escape[(unsigned char)c]) { + else if ((sizeof(Ch) == 1 || static_cast(c) < 256) && RAPIDJSON_UNLIKELY(escape[static_cast(c)])) { is.Take(); - os_->Put('\\'); - os_->Put(escape[(unsigned char)c]); - if (escape[(unsigned char)c] == 'u') { - os_->Put('0'); - os_->Put('0'); - os_->Put(hexDigits[(unsigned char)c >> 4]); - os_->Put(hexDigits[(unsigned char)c & 0xF]); + PutUnsafe(*os_, '\\'); + PutUnsafe(*os_, static_cast(escape[static_cast(c)])); + if (escape[static_cast(c)] == 'u') { + PutUnsafe(*os_, '0'); + PutUnsafe(*os_, '0'); + PutUnsafe(*os_, hexDigits[static_cast(c) >> 4]); + PutUnsafe(*os_, hexDigits[static_cast(c) & 0xF]); } } - else - if (!Transcoder::Transcode(is, *os_)) - return false; + else if (RAPIDJSON_UNLIKELY(!(writeFlags & kWriteValidateEncodingFlag ? + Transcoder::Validate(is, *os_) : + Transcoder::TranscodeUnsafe(is, *os_)))) + return false; } - os_->Put('\"'); + PutUnsafe(*os_, '\"'); return true; } + bool ScanWriteUnescapedString(GenericStringStream& is, size_t length) { + return RAPIDJSON_LIKELY(is.Tell() < length); + } + bool WriteStartObject() { os_->Put('{'); return true; } bool WriteEndObject() { os_->Put('}'); return true; } bool WriteStartArray() { os_->Put('['); return true; } bool WriteEndArray() { os_->Put(']'); return true; } + bool WriteRawValue(const Ch* json, size_t length) { + PutReserve(*os_, length); + for (size_t i = 0; i < length; i++) { + RAPIDJSON_ASSERT(json[i] != '\0'); + PutUnsafe(*os_, json[i]); + } + return true; + } + void Prefix(Type type) { (void)type; - if (level_stack_.GetSize() != 0) { // this value is not at root + if (RAPIDJSON_LIKELY(level_stack_.GetSize() != 0)) { // this value is not at root Level* level = level_stack_.template Top(); if (level->valueCount > 0) { if (level->inArray) @@ -334,8 +487,16 @@ class Writer { } } + // Flush the value if it is the top level one. + bool EndValue(bool ret) { + if (RAPIDJSON_UNLIKELY(level_stack_.Empty())) // end of json text + Flush(); + return ret; + } + OutputStream* os_; internal::Stack level_stack_; + int maxDecimalPlaces_; bool hasRoot_; private: @@ -350,7 +511,7 @@ template<> inline bool Writer::WriteInt(int i) { char *buffer = os_->Push(11); const char* end = internal::i32toa(i, buffer); - os_->Pop(11 - (end - buffer)); + os_->Pop(static_cast(11 - (end - buffer))); return true; } @@ -358,7 +519,7 @@ template<> inline bool Writer::WriteUint(unsigned u) { char *buffer = os_->Push(10); const char* end = internal::u32toa(u, buffer); - os_->Pop(10 - (end - buffer)); + os_->Pop(static_cast(10 - (end - buffer))); return true; } @@ -366,7 +527,7 @@ template<> inline bool Writer::WriteInt64(int64_t i64) { char *buffer = os_->Push(21); const char* end = internal::i64toa(i64, buffer); - os_->Pop(21 - (end - buffer)); + os_->Pop(static_cast(21 - (end - buffer))); return true; } @@ -374,22 +535,177 @@ template<> inline bool Writer::WriteUint64(uint64_t u) { char *buffer = os_->Push(20); const char* end = internal::u64toa(u, buffer); - os_->Pop(20 - (end - buffer)); + os_->Pop(static_cast(20 - (end - buffer))); return true; } template<> inline bool Writer::WriteDouble(double d) { + if (internal::Double(d).IsNanOrInf()) { + // Note: This code path can only be reached if (RAPIDJSON_WRITE_DEFAULT_FLAGS & kWriteNanAndInfFlag). + if (!(kWriteDefaultFlags & kWriteNanAndInfFlag)) + return false; + if (internal::Double(d).IsNan()) { + PutReserve(*os_, 3); + PutUnsafe(*os_, 'N'); PutUnsafe(*os_, 'a'); PutUnsafe(*os_, 'N'); + return true; + } + if (internal::Double(d).Sign()) { + PutReserve(*os_, 9); + PutUnsafe(*os_, '-'); + } + else + PutReserve(*os_, 8); + PutUnsafe(*os_, 'I'); PutUnsafe(*os_, 'n'); PutUnsafe(*os_, 'f'); + PutUnsafe(*os_, 'i'); PutUnsafe(*os_, 'n'); PutUnsafe(*os_, 'i'); PutUnsafe(*os_, 't'); PutUnsafe(*os_, 'y'); + return true; + } + char *buffer = os_->Push(25); - char* end = internal::dtoa(d, buffer); - os_->Pop(25 - (end - buffer)); + char* end = internal::dtoa(d, buffer, maxDecimalPlaces_); + os_->Pop(static_cast(25 - (end - buffer))); return true; } +#if defined(RAPIDJSON_SSE2) || defined(RAPIDJSON_SSE42) +template<> +inline bool Writer::ScanWriteUnescapedString(StringStream& is, size_t length) { + if (length < 16) + return RAPIDJSON_LIKELY(is.Tell() < length); + + if (!RAPIDJSON_LIKELY(is.Tell() < length)) + return false; + + const char* p = is.src_; + const char* end = is.head_ + length; + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + const char* endAligned = reinterpret_cast(reinterpret_cast(end) & static_cast(~15)); + if (nextAligned > end) + return true; + + while (p != nextAligned) + if (*p < 0x20 || *p == '\"' || *p == '\\') { + is.src_ = p; + return RAPIDJSON_LIKELY(is.Tell() < length); + } + else + os_->PutUnsafe(*p++); + + // The rest of string using SIMD + static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; + static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; + const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); + const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); + const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); + + for (; p != endAligned; p += 16) { + const __m128i s = _mm_load_si128(reinterpret_cast(p)); + const __m128i t1 = _mm_cmpeq_epi8(s, dq); + const __m128i t2 = _mm_cmpeq_epi8(s, bs); + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F + const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); + unsigned short r = static_cast(_mm_movemask_epi8(x)); + if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped + SizeType len; +#ifdef _MSC_VER // Find the index of first escaped + unsigned long offset; + _BitScanForward(&offset, r); + len = offset; +#else + len = static_cast(__builtin_ffs(r) - 1); +#endif + char* q = reinterpret_cast(os_->PushUnsafe(len)); + for (size_t i = 0; i < len; i++) + q[i] = p[i]; + + p += len; + break; + } + _mm_storeu_si128(reinterpret_cast<__m128i *>(os_->PushUnsafe(16)), s); + } + + is.src_ = p; + return RAPIDJSON_LIKELY(is.Tell() < length); +} +#elif defined(RAPIDJSON_NEON) +template<> +inline bool Writer::ScanWriteUnescapedString(StringStream& is, size_t length) { + if (length < 16) + return RAPIDJSON_LIKELY(is.Tell() < length); + + if (!RAPIDJSON_LIKELY(is.Tell() < length)) + return false; + + const char* p = is.src_; + const char* end = is.head_ + length; + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + const char* endAligned = reinterpret_cast(reinterpret_cast(end) & static_cast(~15)); + if (nextAligned > end) + return true; + + while (p != nextAligned) + if (*p < 0x20 || *p == '\"' || *p == '\\') { + is.src_ = p; + return RAPIDJSON_LIKELY(is.Tell() < length); + } + else + os_->PutUnsafe(*p++); + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (; p != endAligned; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType len = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high); + len = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low); + len = lz >> 3; + escaped = true; + } + if (RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + char* q = reinterpret_cast(os_->PushUnsafe(len)); + for (size_t i = 0; i < len; i++) + q[i] = p[i]; + + p += len; + break; + } + vst1q_u8(reinterpret_cast(os_->PushUnsafe(16)), s); + } + + is.src_ = p; + return RAPIDJSON_LIKELY(is.Tell() < length); +} +#endif // RAPIDJSON_NEON + RAPIDJSON_NAMESPACE_END #ifdef _MSC_VER RAPIDJSON_DIAG_POP #endif +#ifdef __clang__ +RAPIDJSON_DIAG_POP +#endif + #endif // RAPIDJSON_RAPIDJSON_H_