Extend libkiwix structures to be contructed/updated from libzim structures

include/book.h

@@ -26,6 +26,10 @@ namespace pugi {
 class xml_node;
 }
 
+namespace zim {
+class Archive;
+}
+
 namespace kiwix
 {
 
@@ -43,6 +47,7 @@ class Book
 
   bool update(const Book& other);
   void update(const Reader& reader);
+  void update(const zim::Archive& archive);
   void updateFromXml(const pugi::xml_node& node, const std::string& baseDir);
   void updateFromOpds(const pugi::xml_node& node, const std::string& urlHost);
   std::string getHumanReadableIdFromPath() const;

include/reader.h

@@ -292,16 +292,6 @@ class Reader
    */
   string getScraper() const;
 
-  /**
-   * Get the origId of the zim file.
-   *
-   * The origId is only used in the case of patch zim file and is the Id
-   * of the original zim file.
-   *
-   * @return The origId of the zim file as specified in the zim metadata.
-   */
-  string getOrigId() const;
-
   /**
    * Get the favicon of the zim file.
    *

src/book.cpp

@@ -27,6 +27,9 @@
 #include "tools/otherTools.h"
 #include "tools/stringTools.h"
 #include "tools/pathTools.h"
+#include "tools/archiveTools.h"
+
+#include <zim/archive.h>
 
 #include <pugixml.hpp>
 
@@ -80,26 +83,28 @@ bool Book::update(const kiwix::Book& other)
 
 void Book::update(const kiwix::Reader& reader)
 {
-  m_path = reader.getZimFilePath();
+  update(*reader.getZimArchive());
+}
+
+void Book::update(const zim::Archive& archive) {
+  m_path = archive.getFilename();
   m_pathValid = true;
-  m_id = reader.getId();
-  m_title = reader.getTitle();
-  m_description = reader.getDescription();
-  m_language = reader.getLanguage();
-  m_creator = reader.getCreator();
-  m_publisher = reader.getPublisher();
-  m_date = reader.getDate();
-  m_name = reader.getName();
-  m_flavour = reader.getFlavour();
-  m_tags = reader.getTags();
+  m_id = getArchiveId(archive);
+  m_title = getArchiveTitle(archive);
+  m_description = getMetaDescription(archive);
+  m_language = getMetaLanguage(archive);
+  m_creator = getMetaCreator(archive);
+  m_publisher = getMetaPublisher(archive);
+  m_date = getMetaDate(archive);
+  m_name = getMetaName(archive);
+  m_flavour = getMetaFlavour(archive);
+  m_tags = getMetaTags(archive);
   m_category = getCategoryFromTags();
-  m_origId = reader.getOrigId();
-  m_articleCount = reader.getArticleCount();
-  m_mediaCount = reader.getMediaCount();
-  m_size = static_cast<uint64_t>(reader.getFileSize()) << 10;
-  m_pathValid = true;
+  m_articleCount = archive.getArticleCount();
+  m_mediaCount = getArchiveMediaCount(archive);
+  m_size = static_cast<uint64_t>(getArchiveFileSize(archive)) << 10;
 
-  reader.getFavicon(m_favicon, m_faviconMimeType);
+  getArchiveFavicon(archive, m_favicon, m_faviconMimeType);
 }
 
 #define ATTR(name) node.attribute(name).value()

src/manager.cpp

@@ -215,8 +215,8 @@ bool Manager::readBookFromPath(const std::string& path, kiwix::Book* book)
     tmp_path = computeAbsolutePath(getCurrentDirectory(), path);
   }
   try {
-    kiwix::Reader reader(tmp_path);
-    book->update(reader);
+    zim::Archive archive(tmp_path);
+    book->update(archive);
     book->setPathValid(true);
   } catch (const std::exception& e) {
     book->setPathValid(false);

src/reader.cpp

@@ -29,44 +29,6 @@
 #include "tools/otherTools.h"
 #include "tools/archiveTools.h"
 
-inline char hi(char v)
-{
-  char hex[] = "0123456789abcdef";
-  return hex[(v >> 4) & 0xf];
-}
-
-inline char lo(char v)
-{
-  char hex[] = "0123456789abcdef";
-  return hex[v & 0xf];
-}
-
-std::string hexUUID(std::string in)
-{
-  std::ostringstream out;
-  for (unsigned n = 0; n < 4; ++n) {
-    out << hi(in[n]) << lo(in[n]);
-  }
-  out << '-';
-  for (unsigned n = 4; n < 6; ++n) {
-    out << hi(in[n]) << lo(in[n]);
-  }
-  out << '-';
-  for (unsigned n = 6; n < 8; ++n) {
-    out << hi(in[n]) << lo(in[n]);
-  }
-  out << '-';
-  for (unsigned n = 8; n < 10; ++n) {
-    out << hi(in[n]) << lo(in[n]);
-  }
-  out << '-';
-  for (unsigned n = 10; n < 16; ++n) {
-    out << hi(in[n]) << lo(in[n]);
-  }
-  std::string op = out.str();
-  return op;
-}
-
 namespace kiwix
 {
 /* Constructor */
@@ -119,12 +81,7 @@ zim::Archive* Reader::getZimArchive() const
 
 MimeCounterType Reader::parseCounterMetadata() const
 {
-  try {
-    auto counterContent = zimArchive->getMetadata("Counter");
-    return parseMimetypeCounter(counterContent);
-  } catch (zim::EntryNotFound& e) {
-    return {};
-  }
+  return kiwix::parseArchiveCounter(*zimArchive);
 }
 
 /* Get the count of articles which can be indexed/displayed */
@@ -146,19 +103,7 @@ unsigned int Reader::getArticleCount() const
 /* Get the count of medias content in the ZIM file */
 unsigned int Reader::getMediaCount() const
 {
-  std::map<const std::string, unsigned int> counterMap
-      = this->parseCounterMetadata();
-  unsigned int counter = 0;
-
-  for (auto &pair:counterMap) {
-    if (startsWith(pair.first, "image/") ||
-        startsWith(pair.first, "video/") ||
-        startsWith(pair.first, "audio/")) {
-      counter += pair.second;
-    }
-  }
-
-  return counter;
+  return kiwix::getArchiveMediaCount(*zimArchive);
 }
 
 /* Get the total of all items of a ZIM file, redirects included */
@@ -170,9 +115,7 @@ unsigned int Reader::getGlobalCount() const
 /* Return the UID of the ZIM file */
 string Reader::getId() const
 {
-  std::ostringstream s;
-  s << zimArchive->getUuid();
-  return s.str();
+  return kiwix::getArchiveId(*zimArchive);
 }
 
 Entry Reader::getRandomPage() const
@@ -281,7 +224,7 @@ string Reader::getRelation() const
 
 string Reader::getFlavour() const
 {
-  METADATA("Flavour")
+  return kiwix::getMetaFlavour(*zimArchive);
 }
 
 string Reader::getSource() const
@@ -295,31 +238,6 @@ string Reader::getScraper() const
 }
 #undef METADATA
 
-string Reader::getOrigId() const
-{
-  string value;
-  this->getMetadata("startfileuid", value);
-  if (value.empty()) {
-    return "";
-  }
-  std::string id = value;
-  std::string origID;
-  std::string temp = "";
-  unsigned int k = 0;
-  char tempArray[16] = "";
-  for (unsigned int i = 0; i < id.size(); i++) {
-    if (id[i] == '\n') {
-      tempArray[k] = atoi(temp.c_str());
-      temp = "";
-      k++;
-    } else {
-      temp += id[i];
-    }
-  }
-  origID = hexUUID(tempArray);
-  return origID;
-}
-
 Entry Reader::getEntryFromPath(const std::string& path) const
 {
   try {
@@ -546,7 +464,7 @@ bool Reader::isCorrupted() const
 /* Return the file size, works also for splitted files */
 unsigned int Reader::getFileSize() const
 {
-  return zimArchive->getFilesize() / 1024;
+  return kiwix::getArchiveFileSize(*zimArchive);
 }
 
 }

src/tools/archiveTools.cpp

@@ -69,18 +69,6 @@ std::string getMetaTags(const zim::Archive& archive, bool original) {
   return join(tags, ";");
 }
 
-bool getArchiveFavicon(const zim::Archive& archive,
-                           std::string& content, std::string& mimeType){
-  try {
-    auto item = archive.getIllustrationItem();
-    content = item.getData();
-    mimeType = item.getMimetype();
-    return true;
-  } catch(zim::EntryNotFound& e) {};
-
-  return false;
-}
-
 std::string getMetaLanguage(const zim::Archive& archive) {
   return getMetadata(archive, "Language");
 }
@@ -101,6 +89,46 @@ std::string getMetaPublisher(const zim::Archive& archive) {
   return getMetadata(archive, "Publisher");
 }
 
+std::string getMetaFlavour(const zim::Archive& archive) {
+  return getMetadata(archive, "Flavour");
+}
+
+std::string getArchiveId(const zim::Archive& archive) {
+  return (std::string) archive.getUuid();
+}
+
+bool getArchiveFavicon(const zim::Archive& archive,
+                           std::string& content, std::string& mimeType){
+  try {
+    auto item = archive.getIllustrationItem();
+    content = item.getData();
+    mimeType = item.getMimetype();
+    return true;
+  } catch(zim::EntryNotFound& e) {};
+
+  return false;
+}
+
+// should this be in libzim
+unsigned int getArchiveMediaCount(const zim::Archive& archive) {
+  std::map<const std::string, unsigned int> counterMap = parseArchiveCounter(archive);
+  unsigned int counter = 0;
+
+  for (auto &pair:counterMap) {
+    if (startsWith(pair.first, "image/") ||
+        startsWith(pair.first, "video/") ||
+        startsWith(pair.first, "audio/")) {
+      counter += pair.second;
+    }
+  }
+
+  return counter;
+}
+
+unsigned int getArchiveFileSize(const zim::Archive& archive) {
+  return archive.getFilesize() / 1024;
+}
+
 zim::Item getFinalItem(const zim::Archive& archive, const zim::Entry& entry)
 {
   return entry.getItem(true);
@@ -118,4 +146,13 @@ zim::Entry getEntryFromPath(const zim::Archive& archive, const std::string& path
   throw zim::EntryNotFound("Cannot find entry for non empty path");
 }
 
+MimeCounterType parseArchiveCounter(const zim::Archive& archive) {
+  try {
+    auto counterContent = archive.getMetadata("Counter");
+    return parseMimetypeCounter(counterContent);
+  } catch (zim::EntryNotFound& e) {
+    return {};
+  }
+}
+
 } // kiwix

src/tools/archiveTools.h

@@ -21,6 +21,7 @@
 #define KIWIX_ARCHIVETOOLS_H
 
 #include <zim/archive.h>
+#include <tools/otherTools.h>
 
 /**
  * This file contains all the functions that would make handling data related to
@@ -33,15 +34,26 @@ namespace kiwix
     std::string getArchiveTitle(const zim::Archive& archive);
     std::string getMetaDescription(const zim::Archive& archive);
     std::string getMetaTags(const zim::Archive& archive, bool original = false);
-    bool getArchiveFavicon(const zim::Archive& archive,
-                           std::string& content, std::string& mimeType);
     std::string getMetaLanguage(const zim::Archive& archive);
     std::string getMetaName(const zim::Archive& archive);
     std::string getMetaDate(const zim::Archive& archive);
     std::string getMetaCreator(const zim::Archive& archive);
     std::string getMetaPublisher(const zim::Archive& archive);
+    std::string getMetaFlavour(const zim::Archive& archive);
+    std::string getArchiveId(const zim::Archive& archive);
+
+    bool getArchiveFavicon(const zim::Archive& archive,
+                           std::string& content, std::string& mimeType);
+
+    unsigned int getArchiveMediaCount(const zim::Archive& archive);
+    unsigned int getArchiveFileSize(const zim::Archive& archive);
+
     zim::Item getFinalItem(const zim::Archive& archive, const zim::Entry& entry);
+
     zim::Entry getEntryFromPath(const zim::Archive& archive, const std::string& path);
+
+    MimeCounterType parseArchiveCounter(const zim::Archive& archive);
+
 }
 
 #endif