Merge branch 'zip-libxml'

.github/workflows/colorer_ci.yml

@@ -40,9 +40,6 @@ jobs:
         icu: [ON, OFF]
         zip: [ON, OFF]
         libxml: [ON, OFF]
-        exclude:
-          - libxml: ON
-            zip: ON
 
     name: linux-${{ matrix.arch }}-icu_${{ matrix.icu }}-zip_${{ matrix.zip }}-libxml_${{ matrix.libxml }}
 
@@ -86,6 +83,7 @@ jobs:
         arch: [ x64, x86 ]
         icu: [ON, OFF]
         zip: [ON, OFF]
+        libxml: [ON, OFF]
         include:
           - arch: x64
             triplet: x64-windows-static-rel
@@ -95,7 +93,7 @@ jobs:
           - arch: x86
             zip: OFF
 
-    name: windows-${{ matrix.arch }}-icu_${{ matrix.icu }}-zip_${{ matrix.zip }}
+    name: windows-${{ matrix.arch }}-icu_${{ matrix.icu }}-zip_${{ matrix.zip }}-libxml_${{ matrix.libxml }}
 
     steps:
       - name: Checkout source
@@ -148,6 +146,7 @@ jobs:
           -DVCPKG_FEATURE_FLAGS=manifests,versions
           -DCOLORER_USE_ICU_STRINGS=${{ matrix.icu }}
           -DCOLORER_USE_ZIPINPUTSOURCE=${{ matrix.zip }}
+          -DCOLORER_USE_LIBXML=${{ matrix.libxml }}
 
       - name: Build
         shell: bash
@@ -166,9 +165,6 @@ jobs:
         icu: [ON, OFF]
         zip: [ON, OFF]
         libxml: [ON, OFF]
-        exclude:
-          - libxml: ON
-            zip: ON
 
     name: macos-${{ matrix.arch }}-icu_${{ matrix.icu }}-zip_${{ matrix.zip }}-libxml_${{ matrix.libxml }}
 

src/CMakeLists.txt

@@ -100,6 +100,7 @@ if(COLORER_USE_LIBXML)
     colorer/xml/libxml2/LibXmlReader.cpp
     colorer/xml/libxml2/LibXmlReader.h
     colorer/xml/libxml2/LibXmlInputSource.h
+    colorer/xml/libxml2/LibXmlInputSource.cpp
   )
 else()
   set(SRC_COLORER_XML
@@ -118,14 +119,23 @@ else()
   )
 endif()
 
-if(COLORER_USE_ZIPINPUTSOURCE AND NOT(COLORER_USE_LIBXML))
+if(COLORER_USE_ZIPINPUTSOURCE)
   set(COLORER_FEATURE_ZIPINPUTSOURCE 1)
-  set(SRC_COLORER_ZIP
-          colorer/xml/xercesc/MemoryFile.cpp
-          colorer/xml/xercesc/MemoryFile.h
-          colorer/xml/xercesc/ZipXmlInputSource.cpp
-          colorer/xml/xercesc/ZipXmlInputSource.h
-  )
+  if(COLORER_USE_LIBXML)
+    set(SRC_COLORER_ZIP
+      colorer/xml/libxml2/SharedXmlInputSource.cpp
+      colorer/xml/libxml2/SharedXmlInputSource.h
+      colorer/zip/MemoryFile.cpp
+      colorer/zip/MemoryFile.h
+    )
+  else ()
+    set(SRC_COLORER_ZIP
+      colorer/xml/xercesc/ZipXmlInputSource.cpp
+      colorer/xml/xercesc/ZipXmlInputSource.h
+      colorer/zip/MemoryFile.cpp
+      colorer/zip/MemoryFile.h
+    )
+  endif ()
 endif()
 
 if(COLORER_USE_ICU_STRINGS)

src/colorer/utils/Environment.cpp

@@ -142,4 +142,21 @@ bool Environment::isRegularFile(const UnicodeString* basePath, const UnicodeStri
   return fs::is_regular_file(clear_path);
 }
 
+UnicodeString Environment::getAbsolutePath(const UnicodeString& basePath, const UnicodeString& relPath)
+{
+  auto root_pos = basePath.lastIndexOf('/');
+  const auto root_pos2 = basePath.lastIndexOf('\\');
+  if (root_pos2 > root_pos) {
+    root_pos = root_pos2;
+  }
+  if (root_pos == -1) {
+    root_pos = 0;
+  }
+  else {
+    root_pos++;
+  }
+  UnicodeString newPath(basePath, 0, root_pos);
+  newPath.append(relPath);
+  return newPath;
+}
 }  // namespace colorer

src/colorer/utils/Environment.h

@@ -25,6 +25,7 @@ class Environment
   static std::vector<UnicodeString> getFilesFromPath(const UnicodeString* basePath, const UnicodeString* relPath,
                                                      const UnicodeString& extension);
   static bool isRegularFile(const UnicodeString* basePath, const UnicodeString* relPath, UnicodeString& fullPath);
+  static UnicodeString getAbsolutePath(const UnicodeString& basePath, const UnicodeString& relPath);
 };
 
 }  // namespace colorer

src/colorer/xml/XMLNode.h

@@ -5,8 +5,6 @@
 #include <unordered_map>
 #include "colorer/Common.h"
 
-inline const UnicodeString empty_string("");
-
 class XMLNode
 {
 public:
@@ -16,6 +14,8 @@ class XMLNode
 
   const UnicodeString& getAttrValue(const UnicodeString& key) const
   {
+    static const UnicodeString empty_string("");
+
     const auto found = attributes.find(key);
     if (found == attributes.end()) {
       return empty_string;

src/colorer/xml/XmlInputSource.cpp

@@ -14,7 +14,7 @@ XmlInputSource::~XmlInputSource()
 XmlInputSource::XmlInputSource(const UnicodeString& source_path, const UnicodeString* source_base)
 {
 #ifdef COLORER_FEATURE_LIBXML
-  xml_input_source = std::make_unique<LibXmlInputSource>(&source_path, source_base);
+  xml_input_source = std::make_unique<LibXmlInputSource>(source_path, source_base);
 #else
   xercesc::XMLPlatformUtils::Initialize();
   xml_input_source = XercesXmlInputSource::newInstance(&source_path, source_base);
@@ -41,7 +41,8 @@ XercesXmlInputSource* XmlInputSource::getInputSource() const
 
 bool XmlInputSource::isFileURI(const UnicodeString& path, const UnicodeString* base)
 {
-  if (path.startsWith(u"jar") || (base && base->startsWith(u"jar"))) {
+  const UnicodeString jar(u"jar:");
+  if (path.startsWith(jar) || (base && base->startsWith(jar))) {
     return false;
   }
   return true;

src/colorer/xml/libxml2/LibXmlInputSource.cpp

@@ -0,0 +1,73 @@
+#include "colorer/xml/libxml2/LibXmlInputSource.h"
+#include "colorer/Exception.h"
+#include "colorer/utils/Environment.h"
+
+LibXmlInputSource::LibXmlInputSource(const UnicodeString& path, const UnicodeString* base)
+{
+  if (path.isEmpty()) {
+    throw InputSourceException("XmlInputSource::newInstance: path is empty");
+  }
+  UnicodeString full_path;
+  if (path.startsWith(jar) || (base != nullptr && base->startsWith(jar))) {
+#ifdef COLORER_FEATURE_ZIPINPUTSOURCE
+    initZipSource(path, base);
+#else
+    throw InputSourceException("zip input source not supported");
+#endif
+  }
+  else if (colorer::Environment::isRegularFile(base, &path, full_path)) {
+    sourcePath = full_path;
+  }
+  else {
+    throw InputSourceException(full_path + " isn't regular file.");
+  }
+}
+
+LibXmlInputSource::~LibXmlInputSource()
+{
+#ifdef COLORER_FEATURE_ZIPINPUTSOURCE
+  if (zip_source) {
+    zip_source->delref();
+  }
+#endif
+}
+
+LibXmlInputSource LibXmlInputSource::createRelative(const UnicodeString& relPath) const
+{
+  return LibXmlInputSource(relPath, &sourcePath);
+}
+
+UnicodeString& LibXmlInputSource::getPath()
+{
+  return sourcePath;
+}
+
+#ifdef COLORER_FEATURE_ZIPINPUTSOURCE
+void LibXmlInputSource::initZipSource(const UnicodeString& path, const UnicodeString* base)
+{
+  if (path.startsWith(jar)) {
+    const auto path_idx = path.lastIndexOf('!');
+    if (path_idx == -1) {
+      throw InputSourceException("Bad jar uri format: " + path);
+    }
+
+    const UnicodeString path_to_jar = colorer::Environment::getAbsolutePath(
+        base ? *base : u"", UnicodeString(path, jar.length(), path_idx - jar.length()));
+    const UnicodeString path_in_jar(path, path_idx + 1);
+
+    sourcePath = jar + path_to_jar + u"!" + path_in_jar;
+    zip_source = SharedXmlInputSource::getSharedInputSource(path_to_jar);
+  }
+  else {
+    const auto base_idx = base->lastIndexOf('!');
+    if (base_idx == -1) {
+      throw InputSourceException("Bad jar uri format: " + path);
+    }
+    const UnicodeString path_to_jar(*base, jar.length(), base_idx - jar.length());
+    const UnicodeString path_in_jar = colorer::Environment::getAbsolutePath(UnicodeString(*base, base_idx + 1), path);
+
+    sourcePath = jar + path_to_jar + u"!" + path_in_jar;
+    zip_source = SharedXmlInputSource::getSharedInputSource(path_to_jar);
+  }
+}
+#endif

src/colorer/xml/libxml2/LibXmlInputSource.h

@@ -2,37 +2,33 @@
 #define COLORER_LIBXMLINPUTSOURCE_H
 
 #include "colorer/Common.h"
-#include "colorer/Exception.h"
-#include "colorer/utils/Environment.h"
+#ifdef COLORER_FEATURE_ZIPINPUTSOURCE
+#include "colorer/xml/libxml2/SharedXmlInputSource.h"
+#endif
+
+static const UnicodeString jar(u"jar:");
 
 class LibXmlInputSource
 {
  public:
-  explicit LibXmlInputSource(const UnicodeString* path, const UnicodeString* base = nullptr)
-  {
-    UnicodeString full_path;
-    if (colorer::Environment::isRegularFile(base, path, full_path)) {
-      sourcePath = full_path;
-      // file is not open yet, only after makeStream
-    }
-    else {
-      throw InputSourceException(full_path + " isn't regular file.");
-    }
-  }
-
-  LibXmlInputSource createRelative(const UnicodeString& relPath) const
-  {
-    return LibXmlInputSource(&relPath, &sourcePath);
-  }
+  explicit LibXmlInputSource(const UnicodeString& path, const UnicodeString* base = nullptr);
+  ~LibXmlInputSource();
+
+  LibXmlInputSource createRelative(const UnicodeString& relPath) const;
 
   [[nodiscard]]
-  UnicodeString& getPath()
-  {
-    return sourcePath;
-  }
+  UnicodeString& getPath();
 
  private:
   UnicodeString sourcePath;
+
+#ifdef COLORER_FEATURE_ZIPINPUTSOURCE
+ public:
+  void initZipSource(const UnicodeString& path, const UnicodeString* base = nullptr);
+
+ private:
+  SharedXmlInputSource* zip_source {nullptr};
+#endif
 };
 
 #endif  // COLORER_LIBXMLINPUTSOURCE_H