add test on XmlInputSource; refactoring

src/colorer/parsers/ParserFactoryImpl.cpp

@@ -44,7 +44,7 @@ void ParserFactory::Impl::loadHrcPath(const UnicodeString& location)
 {
   try {
     COLORER_LOG_DEBUG("try load '%'", location);
-    if (XmlInputSource::isFileURI(*base_catalog_path, &location)) {
+    if (XmlInputSource::isFsURI(*base_catalog_path, &location)) {
       auto files = colorer::Environment::getFilesFromPath(base_catalog_path.get(), &location, ".hrc");
       for (auto& file : files) {
         loadHrc(file, nullptr);

src/colorer/utils/Environment.cpp

@@ -1,11 +1,11 @@
 #include "colorer/utils/Environment.h"
-#include <regex>
 #ifdef WIN32
 #include <windows.h>
 #include <cwchar>
 #endif
 
 namespace colorer {
+
 fs::path Environment::to_filepath(const UnicodeString* str)
 {
 #ifdef _WINDOWS
@@ -18,7 +18,7 @@ fs::path Environment::to_filepath(const UnicodeString* str)
 
 uUnicodeString Environment::getOSVariable(const UnicodeString& name)
 {
-#ifdef WIN32
+#ifdef _WINDOWS
   COLORER_LOG_DEBUG("get system environment '%'", name);
   auto str_name = UStr::to_stdwstr(&name);
   size_t sz = 0;
@@ -44,36 +44,9 @@ uUnicodeString Environment::getOSVariable(const UnicodeString& name)
     COLORER_LOG_DEBUG("'%' not set", name);
     return nullptr;
   }
-  else {
-    COLORER_LOG_DEBUG("'%' = '%'", name, value);
-    return std::make_unique<UnicodeString>(value);
-  }
-#endif
-}
-
-uUnicodeString Environment::expandEnvironment(const UnicodeString* path)
-{
-  COLORER_LOG_DEBUG("expand system environment for '%'", *path);
-#ifdef WIN32
-  std::wstring path_ws = UStr::to_stdwstr(path);
-  size_t i = ExpandEnvironmentStringsW(path_ws.c_str(), nullptr, 0);
-  auto temp = std::make_unique<wchar_t[]>(i);
-  ExpandEnvironmentStringsW(path_ws.c_str(), temp.get(), static_cast<DWORD>(i));
-  return std::make_unique<UnicodeString>(temp.get());
-#else
-  std::smatch matcher;
-  std::string result;
-  auto text = UStr::to_stdstr(path);
-  static const std::regex env_re {R"--(\$\{([^}]+)\})--"};
-  while (std::regex_search(text, matcher, env_re)) {
-    result += matcher.prefix().str();
-    result += std::getenv(matcher[1].str().c_str());
-    text = matcher.suffix().str();
-  }
-  result += text;
 
-  COLORER_LOG_DEBUG("result of expand '%'", result);
-  return std::make_unique<UnicodeString>(result.c_str());
+  COLORER_LOG_DEBUG("'%' = '%'", name, value);
+  return std::make_unique<UnicodeString>(value);
 #endif
 }
 
@@ -84,8 +57,8 @@ uUnicodeString Environment::normalizePath(const UnicodeString* path)
 
 fs::path Environment::normalizeFsPath(const UnicodeString* path)
 {
-  auto expanded_string = Environment::expandEnvironment(path);
-  auto fpath = fs::path(Environment::to_filepath(expanded_string.get()));
+  auto expanded_string = expandEnvironment(*path);
+  auto fpath = fs::path(to_filepath(&expanded_string));
   fpath = fpath.lexically_normal();
   if (fs::is_symlink(fpath)) {
     fpath = fs::read_symlink(fpath);
@@ -178,4 +151,52 @@ UnicodeString Environment::expandSpecialEnvironment(const UnicodeString& path)
   COLORER_LOG_DEBUG("result of expand '%'", result);
   return UnicodeString(result.c_str());
 }
+
+UnicodeString Environment::expandEnvironment(const UnicodeString& path)
+{
+  COLORER_LOG_DEBUG("expand system environment for '%'", path);
+  if (path.isEmpty()) {
+    COLORER_LOG_DEBUG("result of expand ''");
+    return {};
+  }
+
+#ifdef _WINDOWS
+  std::wstring path_ws = UStr::to_stdwstr(&path);
+  size_t i = ExpandEnvironmentStringsW(path_ws.c_str(), nullptr, 0);
+  auto temp = std::make_unique<wchar_t[]>(i);
+  ExpandEnvironmentStringsW(path_ws.c_str(), temp.get(), static_cast<DWORD>(i));
+  COLORER_LOG_DEBUG("result of expand '%'", temp.get());
+  return {temp.get()};
+#else
+  const auto text = UStr::to_stdstr(&path);
+  auto res = expandEnvByRegexp(text, std::regex(R"--(\$\{([[:alpha:]]\w*)\})--"));
+  res = expandEnvByRegexp(res, std::regex(R"--(\$([[:alpha:]]\w*)\b)--"));
+  COLORER_LOG_DEBUG("result of expand '%'", res);
+  return {res.c_str()};
+#endif
+}
+
+std::string Environment::expandEnvByRegexp(const std::string& path, const std::regex& regex)
+{
+  std::smatch matcher;
+  std::string result;
+  auto text = path;
+  while (std::regex_search(text, matcher, regex)) {
+    result += matcher.prefix().str();
+    auto env_value = getOSVariable(matcher[1].str().c_str());
+    if (env_value) {
+      // add expanded value
+      result += UStr::to_stdstr(env_value);
+    }
+    else {
+      // add variable name
+      result += matcher[0].str();
+    }
+    text = matcher.suffix().str();
+  }
+  result += text;
+
+  return result;
+}
+
 }  // namespace colorer

src/colorer/utils/Environment.h

@@ -1,22 +1,25 @@
 #ifndef COLORER_ENVIRONMENT_H
 #define COLORER_ENVIRONMENT_H
 
+#include <regex>
+#include <vector>
 #include "colorer/Common.h"
+
 #ifdef COLORER_FEATURE_OLD_COMPILERS
 #include "colorer/platform/filesystem.hpp"
 namespace fs = ghc::filesystem;
 #else
 #include <filesystem>
 namespace fs = std::filesystem;
 #endif
-#include <vector>
+
+
 namespace colorer {
 
 class Environment
 {
  public:
   static uUnicodeString getOSVariable(const UnicodeString& name);
-  static uUnicodeString expandEnvironment(const UnicodeString* path);
   static uUnicodeString normalizePath(const UnicodeString* path);
   static fs::path normalizeFsPath(const UnicodeString* path);
   static fs::path getClearFilePath(const UnicodeString* basePath, const UnicodeString* relPath);
@@ -28,6 +31,10 @@ class Environment
   static UnicodeString getAbsolutePath(const UnicodeString& basePath, const UnicodeString& relPath);
 
   static UnicodeString expandSpecialEnvironment(const UnicodeString& path);
+  static UnicodeString expandEnvironment(const UnicodeString& path);
+
+private:
+  static std::string expandEnvByRegexp(const std::string& path, const std::regex& regex);
 };
 
 }  // namespace colorer

src/colorer/xml/XmlInputSource.cpp

@@ -18,7 +18,7 @@ UnicodeString& XmlInputSource::getPath() const
   return xml_input_source->getPath();
 }
 
-bool XmlInputSource::isFileURI(const UnicodeString& path, const UnicodeString* base)
+bool XmlInputSource::isFsURI(const UnicodeString& path, const UnicodeString* base)
 {
   const UnicodeString jar(u"jar:");
   if (path.startsWith(jar) || (base && base->startsWith(jar))) {

src/colorer/xml/XmlInputSource.h

@@ -21,7 +21,7 @@ class XmlInputSource
   [[nodiscard]]
   UnicodeString& getPath() const;
 
-  static bool isFileURI(const UnicodeString& path, const UnicodeString* base);
+  static bool isFsURI(const UnicodeString& path, const UnicodeString* base);
 
  private:
   std::unique_ptr<LibXmlInputSource> xml_input_source;

src/colorer/xml/libxml2/LibXmlInputSource.cpp

@@ -5,7 +5,7 @@
 LibXmlInputSource::LibXmlInputSource(const UnicodeString& path, const UnicodeString* base)
 {
   if (path.isEmpty()) {
-    throw InputSourceException("XmlInputSource::newInstance: path is empty");
+    throw InputSourceException("LibXmlInputSource: path is empty");
   }
   UnicodeString full_path;
   if (path.startsWith(jar) || (base != nullptr && base->startsWith(jar))) {
@@ -43,15 +43,16 @@ UnicodeString& LibXmlInputSource::getPath()
 }
 
 #ifdef COLORER_FEATURE_ZIPINPUTSOURCE
+
 void LibXmlInputSource::initZipSource(const UnicodeString& path, const UnicodeString* base)
 {
-  auto paths = getFullPathFromPathJar(path, base);
+  const auto paths = getFullPathsToZip(path, base);
 
   sourcePath = paths.full_path;
   zip_source = SharedXmlInputSource::getSharedInputSource(paths.path_to_jar);
 }
 
-PathInJar LibXmlInputSource::getFullPathFromPathJar(const UnicodeString& path, const UnicodeString* base)
+PathInJar LibXmlInputSource::getFullPathsToZip(const UnicodeString& path, const UnicodeString* base)
 {
   if (path.startsWith(jar)) {
     const auto path_idx = path.lastIndexOf('!');
@@ -66,7 +67,7 @@ PathInJar LibXmlInputSource::getFullPathFromPathJar(const UnicodeString& path, c
     const UnicodeString full_path = jar + path_to_jar + u"!" + path_in_jar;
     return {full_path, path_to_jar, path_in_jar};
   }
-  else {
+  if (base != nullptr && base->startsWith(jar)) {
     const auto base_idx = base->lastIndexOf('!');
     if (base_idx == -1) {
       throw InputSourceException("Bad jar uri format: " + path);
@@ -77,5 +78,7 @@ PathInJar LibXmlInputSource::getFullPathFromPathJar(const UnicodeString& path, c
     const UnicodeString full_path = jar + path_to_jar + u"!" + path_in_jar;
     return {full_path, path_to_jar, path_in_jar};
   }
+  throw InputSourceException("The path to the jar was not found");
 }
+
 #endif

src/colorer/xml/libxml2/LibXmlInputSource.h

@@ -21,6 +21,7 @@ class LibXmlInputSource
   explicit LibXmlInputSource(const UnicodeString& path, const UnicodeString* base = nullptr);
   ~LibXmlInputSource();
 
+  [[nodiscard]]
   LibXmlInputSource createRelative(const UnicodeString& relPath) const;
 
   [[nodiscard]]
@@ -31,12 +32,11 @@ class LibXmlInputSource
 
 #ifdef COLORER_FEATURE_ZIPINPUTSOURCE
  public:
-  void initZipSource(const UnicodeString& path, const UnicodeString* base = nullptr);
-
-  static PathInJar getFullPathFromPathJar(const UnicodeString& path, const UnicodeString* base);
+  static PathInJar getFullPathsToZip(const UnicodeString& path, const UnicodeString* base = nullptr);
 
  private:
   SharedXmlInputSource* zip_source {nullptr};
+  void initZipSource(const UnicodeString& path, const UnicodeString* base = nullptr);
 #endif
 };
 

src/colorer/xml/libxml2/LibXmlReader.cpp

@@ -152,7 +152,7 @@ xmlParserInputPtr LibXmlReader::xmlMyExternalEntityLoader(const char* URL, const
     if (!is_full_path) {
       string_url = colorer::Environment::expandSpecialEnvironment(string_url);
     }
-    auto paths = LibXmlInputSource::getFullPathFromPathJar(string_url, is_full_path ? nullptr : current_file.get());
+    auto paths = LibXmlInputSource::getFullPathsToZip(string_url, is_full_path ? nullptr : current_file.get());
     is_full_path = false;
     xmlParserInputPtr ret = nullptr;
     try {

src/colorer/xml/libxml2/SharedXmlInputSource.cpp

@@ -52,7 +52,7 @@ SharedXmlInputSource::SharedXmlInputSource(const UnicodeString& path)
 
 SharedXmlInputSource::~SharedXmlInputSource()
 {
-  // не нужно удалять объект, удаляемый из массива. мы и так уже в деструкторе
+  // You don't need to delete an object that has been deleted from the array. We are already in the destructor.
   isHash->erase(source_path);
   if (isHash->empty()) {
     delete isHash;

tests/unit/CMakeLists.txt

@@ -5,7 +5,9 @@ set(unit_tests_SRC
     test_exception.cpp
     test_filetype.cpp
     test_environment.cpp
-    test_hrcparsing.cpp)
+    test_hrcparsing.cpp
+    test_xmlinputsource.cpp
+)
 
 add_executable(unit_tests ${unit_tests_SRC})