Rename *Url* symbols to *Path*.

Dirents' "url" are not really url. They are u8 arrays which serve as the main "key". Specification says that this u8 array store a utf-8 encoded string and we use this key as a path. Public API already use path semantic so rename all internal symbols from url to path. Fix #868
openzim · Mar 29, 2024 · 54a3e6c · 54a3e6c
1 parent 96afb38
commit 54a3e6c
Show file tree

Hide file tree

Showing 22 changed files with 112 additions and 108 deletions.
diff --git a/include/zim/zim.h b/include/zim/zim.h
@@ -95,7 +95,7 @@ namespace zim
     CHECKSUM,
 
     /**
-     * Checks that offsets in UrlPtrList are valid.
+     * Checks that offsets in PathPtrList are valid.
      */
     DIRENT_PTRS,
 

diff --git a/src/_dirent.h b/src/_dirent.h
@@ -46,7 +46,7 @@ namespace zim
 
       char ns;
       std::string title;
-      std::string url;
+      std::string path;
       std::string parameter;
 
     public:
@@ -79,15 +79,15 @@ namespace zim
       entry_index_t getRedirectIndex() const      { return isRedirect() ? redirectIndex : entry_index_t(0); }
 
       char getNamespace() const               { return ns; }
-      const std::string& getTitle() const     { return title.empty() ? url : title; }
-      const std::string& getUrl() const       { return url; }
-      std::string getLongUrl() const;
+      const std::string &getTitle() const     { return title.empty() ? path : title; }
+      const std::string &getPath() const      { return path; }
+      std::string getLongPath() const;
       const std::string& getParameter() const { return parameter; }
 
       size_t getDirentSize() const
       {
-        size_t ret = (isRedirect() ? 12 : 16) + url.size() + parameter.size() + 2;
-        if (title != url)
+        size_t ret = (isRedirect() ? 12 : 16) + path.size() + parameter.size() + 2;
+        if (title != path)
           ret += title.size();
         return ret;
       }
@@ -97,10 +97,10 @@ namespace zim
         title = title_;
       }
 
-      void setUrl(char ns_, const std::string& url_)
+      void setPath(char ns_, const std::string &path_)
       {
         ns = ns_;
-        url = url_;
+        path = path_;
       }
 
       void setParameter(const std::string& parameter_)

diff --git a/src/archive.cpp b/src/archive.cpp
@@ -136,7 +136,7 @@ namespace zim
     auto end = m_impl->getNamespaceEndOffset('M');
     for (auto idx=start; idx!=end; idx++) {
       auto dirent = m_impl->getDirent(idx);
-      ret.push_back(dirent->getUrl());
+      ret.push_back(dirent->getPath());
     }
     return ret;
   }

diff --git a/src/dirent.cpp b/src/dirent.cpp
@@ -76,21 +76,21 @@ namespace zim
       dirent.setItem(mimeType, cluster_index_t(clusterNumber), blob_index_t(blobNumber));
     }
 
-    std::string url;
+    std::string path;
     std::string title;
     std::string parameter;
 
-    log_debug("read url, title and parameters");
+    log_debug("read path, title and parameters");
 
-    size_type url_size = strnlen(
+    size_type path_size = strnlen(
       reader.current(),
       reader.left().v - extraLen
     );
-    if (url_size >= reader.left().v) {
+    if (path_size >= reader.left().v) {
       return false;
     }
-    url = std::string(reader.current(), url_size);
-    reader.skip(zsize_t(url_size+1));
+    path = std::string(reader.current(), path_size);
+    reader.skip(zsize_t(path_size + 1));
 
     size_type title_size = strnlen(
       reader.current(),
@@ -106,7 +106,7 @@ namespace zim
       return false;
     }
     parameter = std::string(reader.current(), extraLen);
-    dirent.setUrl(ns, url);
+    dirent.setPath(ns, path);
     dirent.setTitle(title);
     dirent.setParameter(parameter);
     return true;
@@ -120,12 +120,12 @@ namespace zim
     }
 
     // We don't know the size of the dirent because it depends of the size of
-    // the title, url and extra parameters.
+    // the title, path and extra parameters.
     // This is a pity but we have no choice.
     // We cannot take a buffer of the size of the file, it would be really
     // inefficient. Let's do try, catch and retry while chosing a smart value
     // for the buffer size. Most dirent will be "Article" entry (header's size
-    // == 16) without extra parameters. Let's hope that url + title size will
+    // == 16) without extra parameters. Let's hope that path + title size will
     // be < 256 and if not try again with a bigger size.
 
     size_t bufferSize(std::min(size_type(256), mp_zimReader->size().v-offset.v));
@@ -139,12 +139,12 @@ namespace zim
     }
   }
 
-  std::string Dirent::getLongUrl() const
+  std::string Dirent::getLongPath() const
   {
-    log_trace("Dirent::getLongUrl()");
+    log_trace("Dirent::getLongPath()");
     log_debug("namespace=" << getNamespace() << " title=" << getTitle());
 
-    return std::string(1, getNamespace()) + '/' + getUrl();
+    return std::string(1, getNamespace()) + '/' + getPath();
   }
 
 }
diff --git a/src/dirent_accessor.cpp b/src/dirent_accessor.cpp
@@ -29,9 +29,12 @@
 
 using namespace zim;
 
-DirectDirentAccessor::DirectDirentAccessor(std::shared_ptr<DirentReader> direntReader, std::unique_ptr<const Reader> urlPtrReader, entry_index_t direntCount)
+DirectDirentAccessor::DirectDirentAccessor(
+  std::shared_ptr<DirentReader> direntReader,
+  std::unique_ptr<const Reader> pathPtrReader,
+  entry_index_t direntCount)
   : mp_direntReader(direntReader),
-    mp_urlPtrReader(std::move(urlPtrReader)),
+    mp_pathPtrReader(std::move(pathPtrReader)),
     m_direntCount(direntCount),
     m_direntCache(envValue("ZIM_DIRENTCACHE", DIRENT_CACHE_SIZE)),
     m_bufferDirentZone(256)
@@ -60,7 +63,7 @@ offset_t DirectDirentAccessor::getOffset(entry_index_t idx) const
   if (idx >= m_direntCount) {
     throw std::out_of_range("entry index out of range");
   }
-  offset_t offset(mp_urlPtrReader->read_uint<offset_type>(offset_t(sizeof(offset_type)*idx.v)));
+  offset_t offset(mp_pathPtrReader->read_uint<offset_type>(offset_t(sizeof(offset_type)*idx.v)));
   return offset;
 }
 

diff --git a/src/dirent_accessor.h b/src/dirent_accessor.h
@@ -45,7 +45,9 @@ class DirentReader;
 class DirectDirentAccessor
 {
 public: // functions
-  DirectDirentAccessor(std::shared_ptr<DirentReader> direntReader, std::unique_ptr<const Reader> urlPtrReader, entry_index_t direntCount);
+  DirectDirentAccessor(std::shared_ptr<DirentReader> direntReader,
+                       std::unique_ptr<const Reader> pathPtrReader,
+                       entry_index_t direntCount);
 
   offset_t    getOffset(entry_index_t idx) const;
   std::shared_ptr<const Dirent> getDirent(entry_index_t idx) const;
@@ -56,7 +58,7 @@ class DirectDirentAccessor
 
 private: // data
   std::shared_ptr<DirentReader>  mp_direntReader;
-  std::unique_ptr<const Reader>  mp_urlPtrReader;
+  std::unique_ptr<const Reader>  mp_pathPtrReader;
   entry_index_t                  m_direntCount;
 
   mutable lru_cache<entry_index_type, std::shared_ptr<const Dirent>> m_direntCache;

diff --git a/src/entry.cpp b/src/entry.cpp
@@ -43,9 +43,9 @@ std::string Entry::getTitle() const
 std::string Entry::getPath() const
 {
   if (m_file->hasNewNamespaceScheme()) {
-    return m_dirent->getUrl();
+    return m_dirent->getPath();
   } else {
-    return m_dirent->getLongUrl();
+    return m_dirent->getLongPath();
   }
 }
 

diff --git a/src/fileheader.cpp b/src/fileheader.cpp
@@ -53,7 +53,7 @@ namespace zim
     std::copy(getUuid().data, getUuid().data + sizeof(Uuid), header + 8);
     toLittleEndian(getArticleCount(), header + 24);
     toLittleEndian(getClusterCount(), header + 28);
-    toLittleEndian(getUrlPtrPos(), header + 32);
+    toLittleEndian(getPathPtrPos(), header + 32);
     toLittleEndian(getTitleIdxPos(), header + 40);
     toLittleEndian(getClusterPtrPos(), header + 48);
     toLittleEndian(getMimeListPos(), header + 56);
@@ -100,7 +100,7 @@ namespace zim
 
     setArticleCount(seqReader.read<uint32_t>());
     setClusterCount(seqReader.read<uint32_t>());
-    setUrlPtrPos(seqReader.read<uint64_t>());
+    setPathPtrPos(seqReader.read<uint64_t>());
     setTitleIdxPos(seqReader.read<uint64_t>());
     setClusterPtrPos(seqReader.read<uint64_t>());
     setMimeListPos(seqReader.read<uint64_t>());
@@ -120,8 +120,8 @@ namespace zim
       throw ZimFileFormatError("mimelistPos must be 80.");
     }
 
-    if (urlPtrPos < mimeListPos) {
-      throw ZimFileFormatError("urlPtrPos must be > mimelistPos.");
+    if (pathPtrPos < mimeListPos) {
+      throw ZimFileFormatError("pathPtrPos must be > mimelistPos.");
     }
     if (titleIdxPos < mimeListPos) {
       throw ZimFileFormatError("titleIdxPos must be > mimelistPos.");

diff --git a/src/fileheader.h b/src/fileheader.h
@@ -48,7 +48,7 @@ namespace zim
       Uuid uuid;
       entry_index_type articleCount;
       offset_type titleIdxPos;
-      offset_type urlPtrPos;
+      offset_type pathPtrPos;
       offset_type mimeListPos;
       cluster_index_type clusterCount;
       offset_type clusterPtrPos;
@@ -62,7 +62,7 @@ namespace zim
           minorVersion(zimMinorVersion),
           articleCount(0),
           titleIdxPos(0),
-          urlPtrPos(0),
+          pathPtrPos(0),
           clusterCount(0),
           clusterPtrPos(0),
           mainPage(std::numeric_limits<entry_index_type>::max()),
@@ -92,8 +92,8 @@ namespace zim
       offset_type getTitleIdxPos() const           { return titleIdxPos; }
       void        setTitleIdxPos(offset_type p)    { titleIdxPos = p; }
 
-      offset_type getUrlPtrPos() const             { return urlPtrPos; }
-      void        setUrlPtrPos(offset_type p)      { urlPtrPos = p; }
+      offset_type getPathPtrPos() const            { return pathPtrPos; }
+      void        setPathPtrPos(offset_type p)     { pathPtrPos = p; }
 
       offset_type getMimeListPos() const           { return mimeListPos; }
       void        setMimeListPos(offset_type p)    { mimeListPos = p; }