improve poppler encryption handling (#457)

Author: andiwand

scripts/conan

@@ -98,6 +98,12 @@ struct HtmlConfig {
   bool format_html{false};
   std::uint8_t html_indent{2};
 
+  // background image
+  std::string background_image_format{"png"};
+
+  // drm
+  bool no_drm{false};
+
   std::optional<std::string> output_path;
   HtmlResourceLocator resource_locator;
 

src/odr/html.hpp

@@ -82,14 +82,14 @@ pdf2htmlEX::Param create_params(PDFDoc &pdf_doc, const HtmlConfig &config,
   param.text_dpi = 300;
 
   // background
-  param.bg_format = "png";
+  param.bg_format = config.background_image_format;
   param.svg_node_count_limit = -1;
   param.svg_embed_bitmap = 1;
 
   // encryption
   param.owner_password = "";
   param.user_password = "";
-  param.no_drm = 1;
+  param.no_drm = config.no_drm ? 1 : 0;
 
   // misc
   param.clean_tmp = 1;
@@ -312,10 +312,6 @@ odr::HtmlService html::create_poppler_pdf_service(
     html_renderer_param->delay_background = 1;
   }
 
-  if (!pdf_doc.okToCopy() && html_renderer_param->no_drm == 0) {
-    throw DocumentCopyProtectedException();
-  }
-
   // TODO not sure what the `progPath` is used for. it cannot be `nullptr`
   // TODO potentially just a cache dir?
   auto html_renderer = std::make_shared<pdf2htmlEX::HTMLRenderer>(

src/odr/internal/html/pdf2htmlex_wrapper.cpp

@@ -54,7 +54,11 @@ void PopplerPdfFile::open(const std::optional<std::string> &password) {
 
   m_file_meta.type = FileType::portable_document_format;
   m_file_meta.mimetype = "application/pdf";
-  m_file_meta.password_encrypted = m_pdf_doc->isEncrypted();
+  // for some reason `isEncrypted` can return `true` even if the file is opened
+  // without password
+  m_file_meta.password_encrypted =
+      m_encryption_state == EncryptionState::encrypted ||
+      (password_goo.has_value() && m_pdf_doc->isEncrypted());
   m_file_meta.document_meta.emplace();
   m_file_meta.document_meta->document_type = DocumentType::text;
   if (m_encryption_state != EncryptionState::encrypted) {
@@ -73,8 +77,7 @@ DecoderEngine PopplerPdfFile::decoder_engine() const noexcept {
 FileMeta PopplerPdfFile::file_meta() const noexcept { return m_file_meta; }
 
 bool PopplerPdfFile::password_encrypted() const noexcept {
-  return m_encryption_state == EncryptionState::encrypted ||
-         m_encryption_state == EncryptionState::decrypted;
+  return m_file_meta.password_encrypted;
 }
 
 EncryptionState PopplerPdfFile::encryption_state() const noexcept {

src/odr/internal/pdf_poppler/poppler_pdf_file.cpp

@@ -9,7 +9,6 @@
 
 #include <filesystem>
 #include <fstream>
-#include <iostream>
 #include <optional>
 #include <string>
 
@@ -54,7 +53,6 @@ TEST_P(HtmlOutputTests, html_meta) {
   const TestParams &params = GetParam();
   const TestFile &test_file = params.test_file;
   const DecoderEngine engine = params.engine;
-  const std::string &test_repo = params.test_repo;
   const std::string &output_path = params.output_path;
   const std::string &output_path_prefix = params.output_path_prefix;
 
@@ -76,7 +74,8 @@ TEST_P(HtmlOutputTests, html_meta) {
   // TODO fix pdf implementation
   if (engine == DecoderEngine::odr &&
       test_file.type == FileType::portable_document_format &&
-      test_repo != "odr-public") {
+      (test_file.short_path.starts_with("odr-private") ||
+       test_file.short_path == "odr-public/pdf/Casio_WVA-M650-7AJF.pdf")) {
     GTEST_SKIP();
   }