Function parse_cli

app/run_scaling.cpp:825–925

  }

  cli_options parse_cli(int argc, char* argv[], pdflib::decode_config& decode_config)
  {
    cxxopts::Options options("run_scaling", "Thread-scaling benchmark for docling-parse C++");
    options.add_options()
      ("input", "Local PDF file or directory", cxxopts::value<std::string>())
      ("mode", "Benchmark stage: parse, render, or both", cxxopts::value<std::string>()->default_value("render"))
      ("recursive,r", "Recurse into subdirectories", cxxopts::value<bool>()->default_value("false")->implicit_value("true"))
      ("max-pages,l", "Maximum number of pages to process across all input PDFs", cxxopts::value<int>())
      ("max-concurrent-results", "Max buffered results for threaded processing", cxxopts::value<int>()->default_value("64"))
      ("threads", "Comma-separated thread counts", cxxopts::value<std::string>()->default_value("1,2,4,8,12,16"))
      ("scale", "Render scale for render mode", cxxopts::value<float>()->default_value("1.0"))
      ("enable-timing", "Write one CSV timing row per page result", cxxopts::value<bool>()->default_value("false")->implicit_value("true"))
      ("timing-csv", "CSV path used when --enable-timing is set", cxxopts::value<std::string>()->default_value("timing-cpp.csv"))
      ("loglevel", "Log level [fatal, error, warning, info]", cxxopts::value<std::string>()->default_value("fatal"))
      ("page-boundary", "Page boundary [crop_box, media_box]", cxxopts::value<std::string>())
      ("do-sanitization", "Run post-parse sanitization", cxxopts::value<std::string>())
      ("keep-char-cells", "Keep individual character cells", cxxopts::value<std::string>())
      ("keep-shapes", "Keep shape items", cxxopts::value<std::string>())
      ("keep-bitmaps", "Keep bitmap items", cxxopts::value<std::string>())
      ("max-num-lines", "Cap on number of lines per page", cxxopts::value<int>())
      ("max-num-bitmaps", "Cap on number of bitmaps per page", cxxopts::value<int>())
      ("create-word-cells", "Build word-level cells", cxxopts::value<std::string>())
      ("create-line-cells", "Build line-level cells", cxxopts::value<std::string>())
      ("enforce-same-font", "Require same font within a word/line cell", cxxopts::value<std::string>())
      ("horizontal-cell-tolerance", "Horizontal merge tolerance", cxxopts::value<double>())
      ("word-space-factor", "Space-width factor for word merging", cxxopts::value<double>())
      ("line-space-factor", "Space-width factor for line merging", cxxopts::value<double>())
      ("line-space-factor-with-space", "Space-width factor for line merging with space", cxxopts::value<double>())
      ("keep-glyphs", "Keep unmapped GLYPH<...> tokens", cxxopts::value<std::string>())
      ("keep-qpdf-warnings", "Emit QPDF warnings", cxxopts::value<std::string>())
      ("h,help", "Print usage");

    options.parse_positional({"input"});
    options.positional_help("input");

    auto result = options.parse(argc, argv);
    if(result.count("help") or not result.count("input"))
      {
        std::cout << options.help() << "\n";
        std::exit(result.count("help") ? 0 : 1);
      }

    cli_options cli;
    cli.input = result["input"].as<std::string>();
    cli.recursive = result["recursive"].as<bool>();
    cli.max_concurrent_results = result["max-concurrent-results"].as<int>();
    if(cli.max_concurrent_results <= 0)
      {
        throw std::runtime_error("--max-concurrent-results must be positive");
      }
    cli.threads = parse_thread_counts(result["threads"].as<std::string>());
    cli.scale = result["scale"].as<float>();
    cli.enable_timing = result["enable-timing"].as<bool>();
    cli.timing_csv = result["timing-csv"].as<std::string>();
    cli.loglevel = result["loglevel"].as<std::string>();

    if(result.count("max-pages"))
      {