Method parse_file

src/parse/parser.h:209–280

  }

  bool parser::parse_file(std::string inp_filename,
                          std::string out_filename,
                          nlohmann::json& task,
			  pdflib::decode_config page_config,
			  bool pretty_print)
  {
    pdflib::pdf_timings pdf_timings;
    document_decoder = std::make_shared<pdflib::pdf_decoder<pdflib::DOCUMENT>>(pdf_timings);

    if(pdf_timings.has_key("fonts-initialisation"))
      {
        LOG_S(ERROR) << "fonts are not initialised";
        return false;
      }

    std::optional<std::string> password;
    if (input_file["password"].is_null())
      {
        password = std::nullopt;
      }
    else
      {
        password = input_file["password"];
      }
    if(not document_decoder->process_document_from_file(inp_filename, password))
      {
        LOG_S(ERROR) << "aborting the parse of file "<< inp_filename;
        return false;
      }

    if(task.count("page-numbers")==0)
      {
        document_decoder->decode_document(page_config);
      }
    else
      {
        std::vector<int> page_numbers = task["page-numbers"];
        document_decoder->decode_document(page_numbers, page_config);
      }

    // Build the output JSON from the typed API
    nlohmann::json json_document;

    json_document["info"]["filename"] = inp_filename;
    json_document["info"]["#-pages"] = document_decoder->get_number_of_pages();
    json_document["annotations"] = document_decoder->get_annotations();

    nlohmann::json json_pages = nlohmann::json::array({});
    for(int p = 0; p < document_decoder->get_number_of_pages(); ++p)
      {
	if(document_decoder->has_page_decoder(p))
	  {
	    auto page_dec = document_decoder->get_page_decoder(p);
	    json_pages.push_back(page_dec->get(page_config));
	  }
      }
    json_document["pages"] = json_pages;