字符串搜索算法

include/CMakeLists.txt

@@ -4,6 +4,8 @@ mkheader_h_recurse(libc-base)
 mkheader_hpp_recurse(libc-base)
 mkheader_h_recurse(data-structure)
 mkheader_hpp_recurse(data-structure)
+mkheader_h_recurse(algorithm)
+mkheader_hpp_recurse(algorithm)
 mkheader_recurse(misc)
 mkheader_recurse(libc)
 mkheader_hpp_recurse(libc++)

include/algorithm.h

@@ -0,0 +1,12 @@
+// Generated by macro 'mkheader'.
+// This file is automatically generated, please do not modify it.
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+#include "algorithm/base.h"
+#include "algorithm/bm.h"
+#include "algorithm/kmp.h"
+#ifdef __cplusplus
+}
+#endif

include/algorithm.hpp

@@ -0,0 +1,3 @@
+// Generated by macro 'mkheader'.
+// This file is automatically generated, please do not modify it.
+#pragma once

include/algorithm/bm.h

@@ -0,0 +1,60 @@
+#pragma once
+#include "base.h"
+
+static ssize_t bm_nsearch(cstr text, i32 textlen, cstr pattern, i32 patternlen) {
+  i32 bc[256];
+  for (i32 i = 0; i < 256; i++) {
+    bc[i] = patternlen;
+  }
+  for (i32 i = 0; i < patternlen - 1; i++) {
+    bc[(byte)pattern[i]] = patternlen - i - 1;
+  }
+
+  for (i32 i = patternlen - 1; i < textlen;) {
+    for (i32 j = patternlen - 1; text[i] == pattern[j]; i--, j--) {
+      if (j == 0) return i;
+    }
+    i += bc[(byte)text[i]];
+  }
+
+  return -1;
+}
+
+static ssize_t bm_search(cstr text, cstr pattern) {
+  return bm_nsearch(text, strlen(text), pattern, strlen(pattern));
+}
+
+static void *bm_nprepare(cstr pattern, i32 patternlen) {
+  i32 *data = (i32 *)calloc(1 + PADDING_UP(patternlen, 4) / 4 + 256, 4);
+  data[0]   = patternlen;
+  memcpy(data + 1, pattern, patternlen);
+  i32 *bc = data + 1 + PADDING_UP(patternlen, 4) / 4;
+
+  for (i32 i = 0; i < 256; i++) {
+    bc[i] = patternlen;
+  }
+  for (i32 i = 0; i < patternlen - 1; i++) {
+    bc[(byte)pattern[i]] = patternlen - i - 1;
+  }
+
+  return data;
+}
+
+static void *bm_prepare(cstr pattern) {
+  return bm_nprepare(pattern, strlen(pattern));
+}
+
+static ssize_t bm_run(cstr text, i32 textlen, const void *data) {
+  i32  patternlen = *(const i32 *)data;
+  cstr pattern    = (cstr)data + 4;
+  var  bc         = (const i32 *)data + 1 + PADDING_UP(patternlen, 4) / 4;
+
+  for (i32 i = patternlen - 1; i < textlen;) {
+    for (i32 j = patternlen - 1; text[i] == pattern[j]; i--, j--) {
+      if (j == 0) return i;
+    }
+    i += bc[(byte)text[i]];
+  }
+
+  return -1;
+}

include/algorithm/kmp.h

@@ -22,7 +22,7 @@ static ssize_t kmp_nsearch(cstr text, i32 textlen, cstr pattern, i32 patternlen)
     }
     if (j == patternlen) {
       free(lps);
-      return i - j;
+      return i - j + 1;
     }
   }
 
@@ -35,7 +35,7 @@ static ssize_t kmp_search(cstr text, cstr pattern) {
 }
 
 static void *kmp_nprepare(cstr pattern, i32 patternlen) {
-  i32 *data = (i32 *)malloc(1 + PADDING_UP(patternlen, 4) + patternlen * 4);
+  i32 *data = (i32 *)calloc(1 + PADDING_UP(patternlen, 4) / 4 + patternlen, 4);
   data[0]   = patternlen;
   memcpy(data + 1, pattern, patternlen);
   i32 *lps = data + 1 + PADDING_UP(patternlen, 4) / 4;
@@ -49,17 +49,17 @@ static void *kmp_nprepare(cstr pattern, i32 patternlen) {
     }
   }
 
-  return lps - 1;
+  return data;
 }
 
 static void *kmp_prepare(cstr pattern) {
   return kmp_nprepare(pattern, strlen(pattern));
 }
 
-static ssize_t kmp_nrun(cstr text, i32 textlen, void *data) {
-  i32  patternlen = *(i32 *)data;
+static ssize_t kmp_nrun(cstr text, i32 textlen, const void *data) {
+  i32  patternlen = *(const i32 *)data;
   cstr pattern    = (cstr)data + 4;
-  i32 *lps        = (i32 *)data + 1 + PADDING_UP(patternlen, 4) / 4;
+  var  lps        = (const i32 *)data + 1 + PADDING_UP(patternlen, 4) / 4;
 
   for (int i = 0, j = 0; i < textlen; i++) {
     if (pattern[j] == text[i]) {
@@ -68,7 +68,7 @@ static ssize_t kmp_nrun(cstr text, i32 textlen, void *data) {
       j = lps[j - 1];
       i--;
     }
-    if (j == patternlen) return i - j;
+    if (j == patternlen) return i - j + 1;
   }
 
   return -1;

include/define/config/undef.h

@@ -47,6 +47,7 @@
 #undef __wur
 #undef __nonnull
 #undef __attr_dealloc
+#undef __attr_access
 #undef __attribute_pure__
 
 #undef INFINITY

include/define/define/attribute.h

@@ -47,7 +47,7 @@
 #  define __attr_readonly(...)
 #  define __attr_writeonly(...)
 #else
-#  define __attr_access(x)      __attr(access x) // 这个 redefine 没有问题，忽略掉
+#  define __attr_access(x)      __attr(access x)
 #  define __attr_readonly(...)  __attr(access(read_only, ##__VA_ARGS__))
 #  define __attr_writeonly(...) __attr(access(write_only, ##__VA_ARGS__))
 #endif

test/c/CMakeLists.txt

@@ -1,3 +1,4 @@
 benchmark(c-rbtree rbtree-bench.cpp)
 gtest(c-rbtree rbtree-test.cpp)
 benchmark(c-avltree avltree-bench.cpp)
+testexec(string-search string-search.c)

test/c/string-search.c

@@ -0,0 +1,17 @@
+#include <stdio.h>
+#include <stdlib.h>
+
+#define NO_STD 0
+#include <algorithm.h>
+
+int main() {
+  assert(kmp_search("hello, world!", "world") == 7);
+  assert(kmp_search("hello, world!", "hello") == 0);
+  assert(kmp_search("hello, world!", "lo") == 3);
+  assert(kmp_search("hello, world!", "wl") == -1);
+  assert(bm_search("hello, world!", "world") == 7);
+  assert(bm_search("hello, world!", "hello") == 0);
+  assert(bm_search("hello, world!", "lo") == 3);
+  assert(bm_search("hello, world!", "wl") == -1);
+  return 0;
+}