PDF to HTML (#349)

Author: andiwand

CMakeLists.txt

@@ -114,6 +114,7 @@ add_library(odr
         "src/odr/internal/html/filesystem.cpp"
         "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/text_file.cpp"
 
         "src/odr/internal/json/json_file.cpp"
@@ -152,7 +153,7 @@ add_library(odr
         "src/odr/internal/pdf/pdf_document.cpp"
         "src/odr/internal/pdf/pdf_document_element.cpp"
         "src/odr/internal/pdf/pdf_document_parser.cpp"
-        "src/odr/internal/pdf/pdf_object.cpp"
+        "src/odr/internal/pdf/pdf_file.cpp"
         "src/odr/internal/pdf/pdf_file_parser.cpp"
         "src/odr/internal/pdf/pdf_graphics_operator.cpp"
         "src/odr/internal/pdf/pdf_graphics_operator_parser.cpp"

src/odr/exceptions.cpp

@@ -44,6 +44,8 @@ NoOpenDocumentFile::NoOpenDocumentFile()
 NoOfficeOpenXmlFile::NoOfficeOpenXmlFile()
     : std::runtime_error("not an office open xml file") {}
 
+NoPdfFile::NoPdfFile() : std::runtime_error("not a pdf file") {}
+
 NoXml::NoXml() : std::runtime_error("not xml") {}
 
 UnsupportedCryptoAlgorithm::UnsupportedCryptoAlgorithm()

src/odr/exceptions.hpp

@@ -80,6 +80,10 @@ struct NoOfficeOpenXmlFile final : public std::runtime_error {
   NoOfficeOpenXmlFile();
 };
 
+struct NoPdfFile final : public std::runtime_error {
+  NoPdfFile();
+};
+
 struct NoXml final : public std::runtime_error {
   NoXml();
 };

src/odr/file.cpp

@@ -7,6 +7,7 @@
 #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>
@@ -89,9 +90,7 @@ DecodedFile::DecodedFile(const std::string &path, FileType as)
 
 DecodedFile::operator bool() const { return m_impl.operator bool(); }
 
-FileType DecodedFile::file_type() const noexcept {
-  return m_impl->file_meta().type;
-}
+FileType DecodedFile::file_type() const noexcept { return m_impl->file_type(); }
 
 FileCategory DecodedFile::file_category() const noexcept {
   return m_impl->file_category();
@@ -101,6 +100,30 @@ FileMeta DecodedFile::file_meta() const noexcept { return m_impl->file_meta(); }
 
 File DecodedFile::file() const { return File(m_impl->file()); }
 
+bool DecodedFile::is_text_file() const {
+  return std::dynamic_pointer_cast<internal::abstract::TextFile>(m_impl) !=
+         nullptr;
+}
+
+bool DecodedFile::is_image_file() const {
+  return std::dynamic_pointer_cast<internal::abstract::ImageFile>(m_impl) !=
+         nullptr;
+}
+
+bool DecodedFile::is_archive_file() const {
+  return std::dynamic_pointer_cast<internal::abstract::ArchiveFile>(m_impl) !=
+         nullptr;
+}
+
+bool DecodedFile::is_document_file() const {
+  return std::dynamic_pointer_cast<internal::abstract::DocumentFile>(m_impl) !=
+         nullptr;
+}
+
+bool DecodedFile::is_pdf_file() const {
+  return std::dynamic_pointer_cast<internal::pdf::PdfFile>(m_impl) != nullptr;
+}
+
 TextFile DecodedFile::text_file() const {
   if (auto text_file =
           std::dynamic_pointer_cast<internal::abstract::TextFile>(m_impl)) {
@@ -133,6 +156,14 @@ DocumentFile DecodedFile::document_file() const {
   throw NoDocumentFile();
 }
 
+PdfFile DecodedFile::pdf_file() const {
+  if (auto pdf_file =
+          std::dynamic_pointer_cast<internal::pdf::PdfFile>(m_impl)) {
+    return PdfFile(pdf_file);
+  }
+  throw NoPdfFile();
+}
+
 TextFile::TextFile(std::shared_ptr<internal::abstract::TextFile> impl)
     : DecodedFile(impl), m_impl{std::move(impl)} {}
 
@@ -198,4 +229,7 @@ DocumentMeta DocumentFile::document_meta() const {
 
 Document DocumentFile::document() const { return Document(m_impl->document()); }
 
+PdfFile::PdfFile(std::shared_ptr<internal::pdf::PdfFile> impl)
+    : DecodedFile(impl), m_impl{std::move(impl)} {}
+
 } // namespace odr

src/odr/file.hpp

@@ -16,11 +16,16 @@ class ArchiveFile;
 class DocumentFile;
 } // namespace odr::internal::abstract
 
+namespace odr::internal::pdf {
+class PdfFile;
+}
+
 namespace odr {
 class TextFile;
 class ImageFile;
 class ArchiveFile;
 class DocumentFile;
+class PdfFile;
 
 class Archive;
 class Document;
@@ -166,10 +171,17 @@ class DecodedFile {
 
   [[nodiscard]] File file() const;
 
+  [[nodiscard]] bool is_text_file() const;
+  [[nodiscard]] bool is_image_file() const;
+  [[nodiscard]] bool is_archive_file() const;
+  [[nodiscard]] bool is_document_file() const;
+  [[nodiscard]] bool is_pdf_file() const;
+
   [[nodiscard]] TextFile text_file() const;
   [[nodiscard]] ImageFile image_file() const;
   [[nodiscard]] ArchiveFile archive_file() const;
   [[nodiscard]] DocumentFile document_file() const;
+  [[nodiscard]] PdfFile pdf_file() const;
 
 protected:
   std::shared_ptr<internal::abstract::DecodedFile> m_impl;
@@ -229,6 +241,14 @@ class DocumentFile final : public DecodedFile {
   std::shared_ptr<internal::abstract::DocumentFile> m_impl;
 };
 
+class PdfFile final : public DecodedFile {
+public:
+  explicit PdfFile(std::shared_ptr<internal::pdf::PdfFile>);
+
+private:
+  std::shared_ptr<internal::pdf::PdfFile> m_impl;
+};
+
 } // namespace odr
 
 #endif // ODR_FILE_HPP

src/odr/html.cpp

@@ -9,12 +9,13 @@
 #include <odr/internal/html/document.hpp>
 #include <odr/internal/html/filesystem.hpp>
 #include <odr/internal/html/image_file.hpp>
+#include <odr/internal/html/pdf_file.hpp>
 #include <odr/internal/html/text_file.hpp>
 
-#include <nlohmann/json.hpp>
-
 #include <filesystem>
 
+#include <nlohmann/json.hpp>
+
 using namespace odr::internal;
 namespace fs = std::filesystem;
 
@@ -53,21 +54,32 @@ Html html::translate(const File &file, const std::string &output_path,
                      const PasswordCallback &password_callback) {
   auto decoded_file = DecodedFile(file);
 
-  if (decoded_file.file_category() == FileCategory::text) {
-    return translate(decoded_file.text_file(), output_path, config);
-  } else if (decoded_file.file_category() == FileCategory::image) {
-    return translate(decoded_file.image_file(), output_path, config);
-  } else if (decoded_file.file_category() == FileCategory::archive) {
-    return translate(decoded_file.archive_file().archive(), output_path,
-                     config);
-  } else if (decoded_file.file_category() == FileCategory::document) {
+  if (decoded_file.is_document_file()) {
     DocumentFile document_file = decoded_file.document_file();
     if (document_file.password_encrypted()) {
       if (!document_file.decrypt(password_callback())) {
         throw WrongPassword();
       }
     }
-    return translate(document_file.document(), output_path, config);
+  }
+
+  return translate(decoded_file, output_path, config);
+}
+
+Html html::translate(const DecodedFile &decoded_file,
+                     const std::string &output_path, const HtmlConfig &config) {
+  if (decoded_file.is_text_file()) {
+    return translate(decoded_file.text_file(), output_path, config);
+  } 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);
+  } else if (decoded_file.is_document_file()) {
+    return translate(decoded_file.document_file().document(), output_path,
+                     config);
+  } else if (decoded_file.is_pdf_file()) {
+    return translate(decoded_file.pdf_file(), output_path, config);
   }
 
   throw UnsupportedFileType(decoded_file.file_type());
@@ -98,6 +110,12 @@ 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()) {

src/odr/html.hpp

@@ -90,9 +90,13 @@ struct HtmlPage final {
 using PasswordCallback = std::function<std::string()>;
 
 namespace html {
+
 Html translate(const File &file, const std::string &output_path,
                const HtmlConfig &config,
                const PasswordCallback &password_callback);
+Html translate(const DecodedFile &file, const std::string &output_path,
+               const HtmlConfig &config);
+
 Html translate(const TextFile &text_file, const std::string &output_path,
                const HtmlConfig &config);
 Html translate(const ImageFile &image_file, const std::string &output_path,
@@ -101,7 +105,11 @@ Html translate(const Archive &archive, const std::string &output_path,
                const HtmlConfig &config);
 Html translate(const Document &document, const std::string &output_path,
                const HtmlConfig &config);
+Html translate(const PdfFile &pdf_file, const std::string &output_path,
+               const HtmlConfig &config);
+
 void edit(const Document &document, const char *diff);
+
 } // namespace html
 
 } // namespace odr