test file and document meta

src/odr/internal/odf/odf_file.cpp

@@ -47,11 +47,11 @@ DecoderEngine OpenDocumentFile::decoder_engine() const noexcept {
 }
 
 DocumentType OpenDocumentFile::document_type() const {
-  return m_file_meta.document_meta->document_type;
+  return m_file_meta.document_meta.value().document_type;
 }
 
 DocumentMeta OpenDocumentFile::document_meta() const {
-  return *m_file_meta.document_meta;
+  return m_file_meta.document_meta.value();
 }
 
 bool OpenDocumentFile::password_encrypted() const noexcept {

src/odr/internal/oldms/oldms_file.cpp

@@ -62,11 +62,11 @@ DecoderEngine LegacyMicrosoftFile::decoder_engine() const noexcept {
 }
 
 DocumentType LegacyMicrosoftFile::document_type() const {
-  return m_file_meta.document_meta->document_type;
+  return m_file_meta.document_meta.value().document_type;
 }
 
 DocumentMeta LegacyMicrosoftFile::document_meta() const {
-  return *m_file_meta.document_meta;
+  return m_file_meta.document_meta.value();
 }
 
 bool LegacyMicrosoftFile::password_encrypted() const noexcept {

src/odr/internal/oldms_wvware/wvware_oldms_file.cpp

@@ -53,8 +53,13 @@ void WvWareLegacyMicrosoftFile::open() {
 
   int ret = wvInitParser_gsf(&m_parser_state->ps, m_parser_state->gsf_input);
 
+  m_file_meta.type = FileType::legacy_word_document;
+  m_file_meta.document_meta = DocumentMeta();
+  m_file_meta.document_meta->document_type = DocumentType::text;
+
   // check if encrypted
   if ((ret & 0x8000) != 0) {
+    m_file_meta.password_encrypted = true;
     m_encryption_state = EncryptionState::encrypted;
     m_parser_state->encryption_flag = ret & 0x7fff;
 
@@ -82,7 +87,7 @@ FileType WvWareLegacyMicrosoftFile::file_type() const noexcept {
 }
 
 FileMeta WvWareLegacyMicrosoftFile::file_meta() const noexcept {
-  return {file_type(), password_encrypted(), document_meta()};
+  return m_file_meta;
 }
 
 DecoderEngine WvWareLegacyMicrosoftFile::decoder_engine() const noexcept {
@@ -93,11 +98,12 @@ DocumentType WvWareLegacyMicrosoftFile::document_type() const {
   return DocumentType::text;
 }
 
-DocumentMeta WvWareLegacyMicrosoftFile::document_meta() const { return {}; }
+DocumentMeta WvWareLegacyMicrosoftFile::document_meta() const {
+  return m_file_meta.document_meta.value();
+}
 
 bool WvWareLegacyMicrosoftFile::password_encrypted() const noexcept {
-  return m_encryption_state == EncryptionState::encrypted ||
-         m_encryption_state == EncryptionState::decrypted;
+  return m_file_meta.password_encrypted;
 }
 
 EncryptionState WvWareLegacyMicrosoftFile::encryption_state() const noexcept {

src/odr/internal/oldms_wvware/wvware_oldms_file.hpp

@@ -46,6 +46,7 @@ class WvWareLegacyMicrosoftFile final : public abstract::DocumentFile {
   std::shared_ptr<abstract::File> m_file;
   std::shared_ptr<ParserState> m_parser_state;
 
+  FileMeta m_file_meta;
   EncryptionState m_encryption_state{EncryptionState::unknown};
 
   void open();

src/odr/internal/ooxml/ooxml_file.cpp

@@ -44,11 +44,11 @@ DecoderEngine OfficeOpenXmlFile::decoder_engine() const noexcept {
 }
 
 DocumentType OfficeOpenXmlFile::document_type() const {
-  return m_file_meta.document_meta->document_type;
+  return m_file_meta.document_meta.value().document_type;
 }
 
 DocumentMeta OfficeOpenXmlFile::document_meta() const {
-  return *m_file_meta.document_meta;
+  return m_file_meta.document_meta.value();
 }
 
 bool OfficeOpenXmlFile::password_encrypted() const noexcept {

src/odr/internal/ooxml/ooxml_meta.cpp

@@ -26,6 +26,7 @@ FileMeta parse_file_meta(abstract::ReadableFilesystem &filesystem) {
   };
 
   FileMeta result;
+  result.document_meta = DocumentMeta();
 
   if (filesystem.is_file(common::Path("/EncryptionInfo")) &&
       filesystem.is_file(common::Path("/EncryptedPackage"))) {
@@ -34,8 +35,6 @@ FileMeta parse_file_meta(abstract::ReadableFilesystem &filesystem) {
     return result;
   }
 
-  result.document_meta.emplace();
-
   for (auto &&t : types) {
     if (filesystem.is_file(t.first)) {
       result.type = t.second.file_type;

src/odr/internal/pdf/pdf_file.cpp

@@ -3,13 +3,15 @@
 namespace odr::internal {
 
 PdfFile::PdfFile(std::shared_ptr<abstract::File> file)
-    : m_file{std::move(file)} {}
+    : m_file{std::move(file)} {
+  m_file_meta.type = FileType::portable_document_format;
+}
 
 std::shared_ptr<abstract::File> PdfFile::file() const noexcept {
   return m_file;
 }
 
-FileMeta PdfFile::file_meta() const noexcept { return {}; }
+FileMeta PdfFile::file_meta() const noexcept { return m_file_meta; }
 
 DecoderEngine PdfFile::decoder_engine() const noexcept {
   return DecoderEngine::odr;

src/odr/internal/pdf/pdf_file.hpp

@@ -20,6 +20,7 @@ class PdfFile final : public abstract::PdfFile {
 
 private:
   std::shared_ptr<abstract::File> m_file;
+  FileMeta m_file_meta;
 };
 
 } // namespace odr::internal

src/odr/odr.cpp

@@ -114,6 +114,34 @@ odr::file_category_by_file_type(const FileType type) noexcept {
   }
 }
 
+odr::DocumentType
+odr::document_type_by_file_type(const FileType type) noexcept {
+  switch (type) {
+  case FileType::opendocument_text:
+    return DocumentType::text;
+  case FileType::opendocument_presentation:
+    return DocumentType::presentation;
+  case FileType::opendocument_spreadsheet:
+    return DocumentType::spreadsheet;
+  case FileType::opendocument_graphics:
+    return DocumentType::drawing;
+  case FileType::office_open_xml_document:
+    return DocumentType::text;
+  case FileType::office_open_xml_presentation:
+    return DocumentType::presentation;
+  case FileType::office_open_xml_workbook:
+    return DocumentType::spreadsheet;
+  case FileType::legacy_word_document:
+    return DocumentType::text;
+  case FileType::legacy_powerpoint_presentation:
+    return DocumentType::presentation;
+  case FileType::legacy_excel_worksheets:
+    return DocumentType::spreadsheet;
+  default:
+    return DocumentType::unknown;
+  }
+}
+
 std::string odr::file_type_to_string(const FileType type) noexcept {
   switch (type) {
   case FileType::unknown:

src/odr/odr.hpp

@@ -39,6 +39,10 @@ file_type_by_file_extension(const std::string &extension) noexcept;
 /// @param type The file type.
 /// @return The file category.
 [[nodiscard]] FileCategory file_category_by_file_type(FileType type) noexcept;
+/// @brief Get the document type by the file type.
+/// @param type The file type.
+/// @return The document type.
+[[nodiscard]] DocumentType document_type_by_file_type(FileType type) noexcept;
 /// @brief Get the file type as a string.
 /// @param type The file type.
 /// @return The file type as a string.

test/src/html_output_test.cpp

@@ -22,6 +22,22 @@ using namespace odr::internal::common;
 using namespace odr::test;
 namespace fs = std::filesystem;
 
+FileType expected_file_type_pre_decryption(const TestFile &test_file) {
+  if (test_file.password.has_value()) {
+    if (test_file.type == FileType::office_open_xml_document ||
+        test_file.type == FileType::office_open_xml_presentation ||
+        test_file.type == FileType::office_open_xml_workbook) {
+      return FileType::office_open_xml_encrypted;
+    }
+  }
+
+  return test_file.type;
+}
+
+FileType expected_file_type_post_decryption(const TestFile &test_file) {
+  return test_file.type;
+}
+
 struct TestParams {
   TestFile test_file;
   std::string path;
@@ -43,9 +59,11 @@ TEST_P(HtmlOutputTests, html_meta) {
   const std::string &output_path = params.output_path;
   const std::string &output_path_prefix = params.output_path_prefix;
 
-  std::cout << test_file.short_path << " to " << output_path << std::endl;
+  const FileCategory file_category = file_category_by_file_type(test_file.type);
 
-  // TODO compare guessed file type VS actual file type
+  ODR_INFO(*logger, "Testing file: " << test_file.short_path << " with engine: "
+                                     << odr::decoder_engine_to_string(engine)
+                                     << " output to: " << output_path);
 
   // these files cannot be opened
   if (util::string::ends_with(test_file.short_path, ".sxw") ||
@@ -71,6 +89,14 @@ TEST_P(HtmlOutputTests, html_meta) {
 
   FileMeta file_meta = file.file_meta();
 
+  EXPECT_EQ(file_meta.type, expected_file_type_pre_decryption(test_file));
+  if (file_category == FileCategory::document) {
+    EXPECT_TRUE(file_meta.document_meta.has_value());
+    EXPECT_EQ(file_meta.document_meta->document_type,
+              document_type_by_file_type(
+                  expected_file_type_pre_decryption(test_file)));
+  }
+
   fs::create_directories(output_path);
 
   {
@@ -110,6 +136,14 @@ TEST_P(HtmlOutputTests, html_meta) {
     // After decryption, the file meta may change
     file_meta = file.file_meta();
 
+    EXPECT_EQ(file_meta.type, expected_file_type_post_decryption(test_file));
+    if (file_category == FileCategory::document) {
+      EXPECT_TRUE(file_meta.document_meta.has_value());
+      EXPECT_EQ(file_meta.document_meta->document_type,
+                document_type_by_file_type(
+                    expected_file_type_post_decryption(test_file)));
+    }
+
     {
       const std::string meta_output = output_path + "/meta-decrypted.json";
       const nlohmann::json json =