feat: Decoder engines (#387)

.github/workflows/build_test.yml

@@ -72,6 +72,8 @@ jobs:
           -DCMAKE_CXX_FLAGS="-Werror"
           -DCMAKE_INSTALL_PREFIX=install
           -DODR_TEST=ON
+          -DWITH_PDF2HTMLEX=ON
+          -DWITH_WVWARE=ON
 
       - name: cmake
         if: runner.os == 'Windows'
@@ -82,6 +84,8 @@ jobs:
           -DCMAKE_BUILD_TYPE=Release
           -DCMAKE_INSTALL_PREFIX=install
           -DODR_TEST=ON
+          -DWITH_PDF2HTMLEX=OFF
+          -DWITH_WVWARE=OFF
 
       - name: build
         run: cmake --build build --config Release

CMakeLists.txt

@@ -108,6 +108,7 @@ set(ODR_SOURCE_FILES
         "src/odr/internal/html/html_writer.cpp"
         "src/odr/internal/html/image_file.cpp"
         "src/odr/internal/html/pdf_file.cpp"
+        "src/odr/internal/html/pdf2htmlex_wrapper.hpp"
         "src/odr/internal/html/text_file.cpp"
 
         "src/odr/internal/json/json_file.cpp"
@@ -194,16 +195,40 @@ target_link_libraries(odr
         utf8::cpp
 )
 
-if(WITH_PDF2HTMLEX)
-    target_sources(odr PRIVATE "src/odr/internal/html/pdf2htmlEX_wrapper.cpp")
+if (WITH_PDF2HTMLEX)
     find_package(pdf2htmlEX REQUIRED)
-    target_link_libraries(odr PRIVATE pdf2htmlex::pdf2htmlex)
-endif(WITH_PDF2HTMLEX)
-if(WITH_WVWARE)
-    target_sources(odr PRIVATE "src/odr/internal/html/wvWare_wrapper.cpp")
+    find_package(poppler REQUIRED)
+    target_sources(odr
+            PRIVATE
+            "src/odr/internal/html/pdf2htmlex_wrapper.cpp"
+            "src/odr/internal/pdf_poppler/poppler_pdf_file.cpp"
+    )
+    target_link_libraries(odr
+            PRIVATE
+            pdf2htmlex::pdf2htmlex
+            poppler::poppler
+    )
+    target_compile_definitions(odr
+            PRIVATE
+            ODR_WITH_PDF2HTMLEX
+    )
+endif ()
+if (WITH_WVWARE)
     find_package(wvware REQUIRED)
-    target_link_libraries(odr PRIVATE wvware::wvware)
-endif(WITH_WVWARE)
+    target_sources(odr
+            PRIVATE
+            "src/odr/internal/html/wvware_wrapper.cpp"
+            "src/odr/internal/oldms_wvware/wvware_oldms_file.cpp"
+    )
+    target_link_libraries(odr
+            PRIVATE
+            wvware::wvware
+    )
+    target_compile_definitions(odr
+            PRIVATE
+            ODR_WITH_WVWARE
+    )
+endif ()
 
 if (EXISTS "${PROJECT_SOURCE_DIR}/.git")
     add_dependencies(odr check_git)

scripts/conan

@@ -1,5 +1,5 @@
 #!/usr/bin/env bash
 
-conan install . --output-folder=cmake-build-relwithdebinfo --build=missing -s build_type=Release -s "&:build_type=RelWithDebInfo"
-conan install . --output-folder=cmake-build-debug --build=missing -s build_type=Release -s "&:build_type=Debug"
-conan install . --output-folder=cmake-build-release --build=never -s build_type=Release -s "&:build_type=Release"
+conan install . --output-folder=cmake-build-relwithdebinfo --build=missing -s build_type=RelWithDebInfo -s "&:build_type=RelWithDebInfo"
+conan install . --output-folder=cmake-build-debug --build=missing -s build_type=RelWithDebInfo -s "&:build_type=Debug"
+conan install . --output-folder=cmake-build-release --build=missing -s build_type=RelWithDebInfo -s "&:build_type=Release"

src/odr/exceptions.cpp

@@ -14,6 +14,9 @@ UnknownFileType::UnknownFileType() : std::runtime_error("unknown file type") {}
 UnsupportedFileType::UnsupportedFileType(const FileType file_type)
     : std::runtime_error("unknown file type"), file_type{file_type} {}
 
+UnknownDecoderEngine::UnknownDecoderEngine()
+    : std::runtime_error("unknown decoder engine") {}
+
 FileReadError::FileReadError() : std::runtime_error("file read error") {}
 
 FileWriteError::FileWriteError() : std::runtime_error("file write error") {}

src/odr/exceptions.hpp

@@ -28,6 +28,11 @@ struct UnsupportedFileType final : public std::runtime_error {
   explicit UnsupportedFileType(FileType file_type);
 };
 
+/// @brief Unknown decoder engine exception
+struct UnknownDecoderEngine final : public std::runtime_error {
+  UnknownDecoderEngine();
+};
+
 /// @brief File read error
 struct FileReadError final : public std::runtime_error {
   FileReadError();

src/odr/file.cpp

@@ -7,7 +7,6 @@
 #include <odr/internal/abstract/file.hpp>
 #include <odr/internal/common/file.hpp>
 #include <odr/internal/open_strategy.hpp>
-#include <odr/internal/pdf/pdf_file.hpp>
 
 #include <optional>
 #include <utility>
@@ -59,6 +58,12 @@ std::vector<FileType> DecodedFile::types(const std::string &path) {
       std::make_shared<internal::common::DiskFile>(path));
 }
 
+std::vector<DecoderEngine> DecodedFile::engines(const std::string &path,
+                                                FileType as) {
+  return internal::open_strategy::engines(
+      std::make_shared<internal::common::DiskFile>(path), as);
+}
+
 FileType DecodedFile::type(const std::string &path) {
   return DecodedFile(path).file_type();
 }
@@ -69,7 +74,7 @@ FileMeta DecodedFile::meta(const std::string &path) {
 
 DecodedFile::DecodedFile(std::shared_ptr<internal::abstract::DecodedFile> impl)
     : m_impl{std::move(impl)} {
-  if (!m_impl) {
+  if (m_impl == nullptr) {
     throw UnknownFileType();
   }
 }
@@ -88,6 +93,11 @@ DecodedFile::DecodedFile(const std::string &path, FileType as)
     : DecodedFile(internal::open_strategy::open_file(
           std::make_shared<internal::common::DiskFile>(path), as)) {}
 
+DecodedFile::DecodedFile(const std::string &path,
+                         const DecodePreference &preference)
+    : DecodedFile(internal::open_strategy::open_file(
+          std::make_shared<internal::common::DiskFile>(path), preference)) {}
+
 DecodedFile::operator bool() const { return m_impl.operator bool(); }
 
 FileType DecodedFile::file_type() const noexcept { return m_impl->file_type(); }
@@ -98,6 +108,10 @@ FileCategory DecodedFile::file_category() const noexcept {
 
 FileMeta DecodedFile::file_meta() const noexcept { return m_impl->file_meta(); }
 
+DecoderEngine DecodedFile::decoder_engine() const noexcept {
+  return m_impl->decoder_engine();
+}
+
 File DecodedFile::file() const { return File(m_impl->file()); }
 
 bool DecodedFile::is_text_file() const {
@@ -121,7 +135,8 @@ bool DecodedFile::is_document_file() const {
 }
 
 bool DecodedFile::is_pdf_file() const {
-  return std::dynamic_pointer_cast<internal::pdf::PdfFile>(m_impl) != nullptr;
+  return std::dynamic_pointer_cast<internal::abstract::PdfFile>(m_impl) !=
+         nullptr;
 }
 
 TextFile DecodedFile::text_file() const {
@@ -158,7 +173,7 @@ DocumentFile DecodedFile::document_file() const {
 
 PdfFile DecodedFile::pdf_file() const {
   if (auto pdf_file =
-          std::dynamic_pointer_cast<internal::pdf::PdfFile>(m_impl)) {
+          std::dynamic_pointer_cast<internal::abstract::PdfFile>(m_impl)) {
     return PdfFile(pdf_file);
   }
   throw NoPdfFile();
@@ -229,7 +244,27 @@ DocumentMeta DocumentFile::document_meta() const {
 
 Document DocumentFile::document() const { return Document(m_impl->document()); }
 
-PdfFile::PdfFile(std::shared_ptr<internal::pdf::PdfFile> impl)
+std::shared_ptr<internal::abstract::DocumentFile> DocumentFile::impl() const {
+  return m_impl;
+}
+
+PdfFile::PdfFile(std::shared_ptr<internal::abstract::PdfFile> impl)
     : DecodedFile(impl), m_impl{std::move(impl)} {}
 
+bool PdfFile::password_encrypted() const {
+  return m_impl->password_encrypted();
+}
+
+EncryptionState PdfFile::encryption_state() const {
+  return m_impl->encryption_state();
+}
+
+bool PdfFile::decrypt(const std::string &password) {
+  return m_impl->decrypt(password);
+}
+
+std::shared_ptr<internal::abstract::PdfFile> PdfFile::impl() const {
+  return m_impl;
+}
+
 } // namespace odr

src/odr/file.hpp

@@ -14,11 +14,8 @@ class TextFile;
 class ImageFile;
 class ArchiveFile;
 class DocumentFile;
-} // namespace odr::internal::abstract
-
-namespace odr::internal::pdf {
 class PdfFile;
-}
+} // namespace odr::internal::abstract
 
 namespace odr {
 class TextFile;
@@ -97,6 +94,22 @@ enum class FileLocation {
   disk,
 };
 
+/// @brief Collection of decoder engines.
+enum class DecoderEngine {
+  odr,
+  poppler,
+  wvware,
+};
+
+/// @brief Preference for decoding files.
+struct DecodePreference final {
+  std::optional<FileType> as_file_type;
+  std::optional<DecoderEngine> with_engine;
+
+  std::vector<FileType> file_type_priority;
+  std::vector<DecoderEngine> engine_priority;
+};
+
 /// @brief Collection of encryption states.
 enum class EncryptionState {
   unknown,
@@ -163,20 +176,24 @@ class File final {
 class DecodedFile {
 public:
   static std::vector<FileType> types(const std::string &path);
+  static std::vector<DecoderEngine> engines(const std::string &path,
+                                            FileType as);
   static FileType type(const std::string &path);
   static FileMeta meta(const std::string &path);
 
-  explicit DecodedFile(std::shared_ptr<internal::abstract::DecodedFile>);
+  explicit DecodedFile(std::shared_ptr<internal::abstract::DecodedFile> impl);
   explicit DecodedFile(const File &file);
   DecodedFile(const File &file, FileType as);
   explicit DecodedFile(const std::string &path);
   DecodedFile(const std::string &path, FileType as);
+  DecodedFile(const std::string &path, const DecodePreference &preference);
 
   [[nodiscard]] explicit operator bool() const;
 
   [[nodiscard]] FileType file_type() const noexcept;
   [[nodiscard]] FileCategory file_category() const noexcept;
   [[nodiscard]] FileMeta file_meta() const noexcept;
+  [[nodiscard]] DecoderEngine decoder_engine() const noexcept;
 
   [[nodiscard]] File file() const;
 
@@ -250,17 +267,25 @@ class DocumentFile final : public DecodedFile {
 
   [[nodiscard]] Document document() const;
 
+  [[nodiscard]] std::shared_ptr<internal::abstract::DocumentFile> impl() const;
+
 private:
   std::shared_ptr<internal::abstract::DocumentFile> m_impl;
 };
 
 /// @brief Represents a PDF file.
 class PdfFile final : public DecodedFile {
 public:
-  explicit PdfFile(std::shared_ptr<internal::pdf::PdfFile>);
+  explicit PdfFile(std::shared_ptr<internal::abstract::PdfFile>);
+
+  [[nodiscard]] bool password_encrypted() const;
+  [[nodiscard]] EncryptionState encryption_state() const;
+  bool decrypt(const std::string &password);
+
+  [[nodiscard]] std::shared_ptr<internal::abstract::PdfFile> impl() const;
 
 private:
-  std::shared_ptr<internal::pdf::PdfFile> m_impl;
+  std::shared_ptr<internal::abstract::PdfFile> m_impl;
 };
 
 } // namespace odr

src/odr/html.cpp

@@ -9,8 +9,12 @@
 #include <odr/internal/html/document.hpp>
 #include <odr/internal/html/filesystem.hpp>
 #include <odr/internal/html/image_file.hpp>
+#include <odr/internal/html/pdf2htmlex_wrapper.hpp>
 #include <odr/internal/html/pdf_file.hpp>
 #include <odr/internal/html/text_file.hpp>
+#include <odr/internal/html/wvware_wrapper.hpp>
+#include <odr/internal/oldms_wvware/wvware_oldms_file.hpp>
+#include <odr/internal/pdf_poppler/poppler_pdf_file.hpp>
 
 #include <filesystem>
 
@@ -73,11 +77,9 @@ Html html::translate(const DecodedFile &decoded_file,
   } else if (decoded_file.is_image_file()) {
     return translate(decoded_file.image_file(), output_path, config);
   } else if (decoded_file.is_archive_file()) {
-    return translate(decoded_file.archive_file().archive(), output_path,
-                     config);
+    return translate(decoded_file.archive_file(), output_path, config);
   } else if (decoded_file.is_document_file()) {
-    return translate(decoded_file.document_file().document(), output_path,
-                     config);
+    return translate(decoded_file.document_file(), output_path, config);
   } else if (decoded_file.is_pdf_file()) {
     return translate(decoded_file.pdf_file(), output_path, config);
   }
@@ -97,6 +99,44 @@ Html html::translate(const ImageFile &image_file,
   return internal::html::translate_image_file(image_file, output_path, config);
 }
 
+Html html::translate(const ArchiveFile &archive_file,
+                     const std::string &output_path, const HtmlConfig &config) {
+  return translate(archive_file.archive(), output_path, config);
+}
+
+Html html::translate(const DocumentFile &document_file,
+                     const std::string &output_path, const HtmlConfig &config) {
+  auto document_file_impl = document_file.impl();
+
+#ifdef ODR_WITH_WVWARE
+  if (auto wv_document_file =
+          std::dynamic_pointer_cast<internal::WvWareLegacyMicrosoftFile>(
+              document_file_impl)) {
+    fs::create_directories(output_path);
+    return internal::html::translate_wvware_oldms_file(*wv_document_file,
+                                                       output_path, config);
+  }
+#endif
+
+  return translate(document_file.document(), output_path, config);
+}
+
+Html html::translate(const PdfFile &pdf_file, const std::string &output_path,
+                     const HtmlConfig &config) {
+  auto pdf_file_impl = pdf_file.impl();
+
+#ifdef ODR_WITH_PDF2HTMLEX
+  if (auto poppler_pdf_file =
+          std::dynamic_pointer_cast<internal::PopplerPdfFile>(pdf_file_impl)) {
+    fs::create_directories(output_path);
+    return internal::html::translate_poppler_pdf_file(*poppler_pdf_file,
+                                                      output_path, config);
+  }
+#endif
+
+  return internal::html::translate_pdf_file(pdf_file, output_path, config);
+}
+
 Html html::translate(const Archive &archive, const std::string &output_path,
                      const HtmlConfig &config) {
   fs::create_directories(output_path);
@@ -110,12 +150,6 @@ Html html::translate(const Document &document, const std::string &output_path,
   return internal::html::translate_document(document, output_path, config);
 }
 
-Html html::translate(const PdfFile &pdf_file, const std::string &output_path,
-                     const HtmlConfig &config) {
-  fs::create_directories(output_path);
-  return internal::html::translate_pdf_file(pdf_file, output_path, config);
-}
-
 void html::edit(const Document &document, const char *diff) {
   auto json = nlohmann::json::parse(diff);
   for (const auto &[key, value] : json["modifiedText"].items()) {