draft

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/pdf_poppler_file.hpp"
         "src/odr/internal/html/text_file.cpp"
 
         "src/odr/internal/json/json_file.cpp"
@@ -194,16 +195,32 @@ 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/html/pdf_poppler_file.cpp"
+            "src/odr/internal/pdf_poppler/poppler_pdf_file.cpp"
+    )
+    target_link_libraries(odr
+            PRIVATE
+            pdf2htmlex::pdf2htmlex
+            poppler::poppler
+    )
+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"
+    )
+    target_link_libraries(odr
+            PRIVATE
+            wvware::wvware
+    )
+endif ()
 
 if (EXISTS "${PROJECT_SOURCE_DIR}/.git")
     add_dependencies(odr check_git)

conanfile.py

@@ -44,7 +44,7 @@ def requirements(self):
         self.requires("uchardet/0.0.8")
         self.requires("utfcpp/4.0.4")
         if self.options.get_safe("with_pdf2htmlEX", False):
-            self.requires("pdf2htmlex/0.18.8.rc1-20240905-git")
+            self.requires("pdf2htmlex/0.18.8.rc1-20240905-git", transitive_headers=True, transitive_libs=True)
         if self.options.get_safe("with_wvWare", False):
             self.requires("wvware/1.2.9")
 

src/odr/file.hpp

@@ -100,7 +100,7 @@ enum class FileLocation {
 /// @brief Collection of decoder engines.
 enum class DecoderEngine {
   odr,
-  pdf2html_ex,
+  poppler,
   wv_ware,
 };
 

src/odr/internal/html/pdf_file.hpp

@@ -1,5 +1,5 @@
-#ifndef ODR_INTERNAL_PDF_FILE_HPP
-#define ODR_INTERNAL_PDF_FILE_HPP
+#ifndef ODR_INTERNAL_HTML_PDF_FILE_HPP
+#define ODR_INTERNAL_HTML_PDF_FILE_HPP
 
 #include <string>
 
@@ -17,4 +17,4 @@ Html translate_pdf_file(const PdfFile &pdf_file, const std::string &output_path,
 
 }
 
-#endif // ODR_INTERNAL_PDF_FILE_HPP
+#endif // ODR_INTERNAL_HTML_PDF_FILE_HPP

src/odr/internal/html/pdf_poppler_file.cpp

@@ -0,0 +1,58 @@
+#include <odr/internal/html/pdf_poppler_file.hpp>
+
+#include <odr/exceptions.hpp>
+#include <odr/file.hpp>
+#include <odr/html.hpp>
+
+#include <odr/internal/common/file.hpp>
+#include <odr/internal/pdf_poppler/poppler_pdf_file.hpp>
+#include <odr/internal/project_info.hpp>
+
+#include <pdf2htmlEX.h>
+
+#include <cstring>
+
+namespace odr::internal {
+
+Html html::translate_pdf_poppler_file(const PopplerPdfFile &pdf_file,
+                                      const std::string &output_path,
+                                      const HtmlConfig &config) {
+  static const char *fontconfig_path = getenv("FONTCONFIG_PATH");
+  if (nullptr == fontconfig_path) {
+    // Storage is allocated and after successful putenv, it will never be freed.
+    // This is the way of putenv.
+    char *storage = strdup("FONTCONFIG_PATH=" FONTCONFIG_PATH);
+    if (0 != putenv(storage)) {
+      free(storage);
+    }
+    fontconfig_path = getenv("FONTCONFIG_PATH");
+  }
+
+  pdf2htmlEX::pdf2htmlEX pdf2htmlEX;
+  pdf2htmlEX.setDataDir(PDF2HTMLEX_DATA_DIR);
+  pdf2htmlEX.setPopplerDataDir(POPPLER_DATA_DIR);
+
+  pdf2htmlEX.setDestinationDir(output_path);
+  auto output_file_name = "document.html";
+  pdf2htmlEX.setOutputFilename(output_file_name);
+
+  pdf2htmlEX.setDRM(false);
+  pdf2htmlEX.setProcessOutline(false);
+  pdf2htmlEX.setProcessAnnotation(true);
+
+  try {
+    pdf2htmlEX.convert();
+  } catch (const pdf2htmlEX::EncryptionPasswordException &e) {
+    throw WrongPassword();
+  } catch (const pdf2htmlEX::DocumentCopyProtectedException &e) {
+    throw std::runtime_error("document is copy protected");
+  } catch (const pdf2htmlEX::ConversionFailedException &e) {
+    throw std::runtime_error(std::string("conversion error ") + e.what());
+  }
+
+  return {FileType::portable_document_format,
+          config,
+          {{"document", output_path + "/" + output_file_name}}};
+}
+
+} // namespace odr::internal

src/odr/internal/html/pdf_poppler_file.hpp

@@ -0,0 +1,21 @@
+#ifndef ODR_INTERNAL_HTML_PDF_POPPLER_FILE_HPP
+#define ODR_INTERNAL_HTML_PDF_POPPLER_FILE_HPP
+
+#include <string>
+
+namespace odr {
+class PopplerPdfFile;
+
+struct HtmlConfig;
+class Html;
+} // namespace odr
+
+namespace odr::internal::html {
+
+Html translate_pdf_poppler_file(const PopplerPdfFile &pdf_file,
+                                const std::string &output_path,
+                                const HtmlConfig &config);
+
+}
+
+#endif // ODR_INTERNAL_HTML_PDF_POPPLER_FILE_HPP

src/odr/internal/pdf_poppler/poppler_pdf_file.cpp

@@ -0,0 +1,36 @@
+#include <odr/internal/pdf_poppler/poppler_pdf_file.hpp>
+
+#include <odr/internal/common/path.hpp>
+
+#include <poppler/PDFDocFactory.h>
+#include <poppler/goo/GooString.h>
+
+namespace odr::internal::poppler_pdf {
+
+PopplerPdfFile::PopplerPdfFile(std::shared_ptr<common::DiskFile> file)
+    : m_file{std::move(file)} {
+  GooString file_path(file->disk_path()->string().c_str());
+  m_pdf_doc = std::unique_ptr<PDFDoc>(PDFDocFactory().createPDFDoc(file_path));
+}
+
+FileCategory PopplerPdfFile::file_category() const noexcept {
+  return FileCategory::document;
+}
+
+std::shared_ptr<abstract::File> PopplerPdfFile::file() const noexcept {
+  return m_file;
+}
+
+FileType PopplerPdfFile::file_type() const noexcept {
+  return FileType::portable_document_format;
+}
+
+FileMeta PopplerPdfFile::file_meta() const noexcept { return {}; }
+
+DecoderEngine PopplerPdfFile::decoder_engine() const noexcept {
+  return DecoderEngine::poppler;
+}
+
+const PDFDoc &PopplerPdfFile::pdf_doc() const { return *m_pdf_doc; }
+
+} // namespace odr::internal::poppler_pdf

src/odr/internal/pdf_poppler/poppler_pdf_file.hpp

@@ -0,0 +1,30 @@
+#ifndef ODR_INTERNAL_POPPLER_PDF_FILE_HPP
+#define ODR_INTERNAL_POPPLER_PDF_FILE_HPP
+
+#include <odr/internal/common/file.hpp>
+
+class PDFDoc;
+
+namespace odr::internal::poppler_pdf {
+
+class PopplerPdfFile : public abstract::DecodedFile {
+public:
+  explicit PopplerPdfFile(std::shared_ptr<common::DiskFile> file);
+
+  [[nodiscard]] std::shared_ptr<abstract::File> file() const noexcept final;
+
+  [[nodiscard]] FileType file_type() const noexcept final;
+  [[nodiscard]] FileCategory file_category() const noexcept final;
+  [[nodiscard]] FileMeta file_meta() const noexcept final;
+  [[nodiscard]] DecoderEngine decoder_engine() const noexcept final;
+
+  [[nodiscard]] const PDFDoc &pdf_doc() const;
+
+private:
+  std::shared_ptr<abstract::File> m_file;
+  std::unique_ptr<PDFDoc> m_pdf_doc;
+};
+
+} // namespace odr::internal::poppler_pdf
+
+#endif // ODR_INTERNAL_POPPLER_PDF_FILE_HPP