Function main

app/parse.cpp:108–287

}

int main(int argc, char* argv[]) {
  int orig_argc = argc;

  // Initialize loguru
  loguru::init(argc, argv);

  try {
    cxxopts::Options options("PDFProcessor", "A program to process PDF files or configuration files");

    // Define the options
    options.add_options()
      ("i,input",        "Input PDF file",                                                        cxxopts::value<std::string>())
      ("c,config",       "Config file",                                                           cxxopts::value<std::string>())
      ("create-config",  "Create config file",                                                    cxxopts::value<std::string>())
      ("p,page",         "Pages to process (default: -1 for all)",                               cxxopts::value<int>()->default_value("-1"))
      ("page-range",     "Inclusive page range to process, e.g. 10-20",                          cxxopts::value<std::string>())
      ("password",       "Password for accessing encrypted, password-protected files",            cxxopts::value<std::string>())
      ("o,output",       "Output file",                                                           cxxopts::value<std::string>())
      ("export-images",  "Export images to directory",                                            cxxopts::value<std::string>())
      ("print-cells",    "Print cells to stdout [char, word, line, all] (default: none)",        cxxopts::value<std::string>())
      ("l,loglevel",     "Log level [error, warning, info]",                                     cxxopts::value<std::string>())
      ("h,help",         "Print usage")

      // ---- decode_config ----
      ("page-boundary",   "Page boundary [crop_box, media_box, ...] (default: crop_box)",        cxxopts::value<std::string>())
      ("do-sanitization", "Run post-parse sanitization (default: true)",                         cxxopts::value<bool>()->implicit_value("true"))
      ("keep-char-cells", "Keep individual character cells (default: true)",                     cxxopts::value<bool>()->implicit_value("true"))
      ("keep-shapes",     "Keep shape items (default: true)",                                    cxxopts::value<bool>()->implicit_value("true"))
      ("keep-bitmaps",    "Keep bitmap items (default: true)",                                   cxxopts::value<bool>()->implicit_value("true"))
      ("max-num-lines",   "Cap on number of lines per page (-1 = no cap)",                       cxxopts::value<int>())
      ("max-num-bitmaps", "Cap on number of bitmaps per page (-1 = no cap)",                    cxxopts::value<int>())
      ("create-word-cells",  "Build word-level cells (default: true)",                           cxxopts::value<bool>()->implicit_value("true"))
      ("create-line-cells",  "Build line-level cells (default: true)",                           cxxopts::value<bool>()->implicit_value("true"))
      ("enforce-same-font",  "Require same font within a word/line cell (default: true)",        cxxopts::value<bool>()->implicit_value("true"))
      ("horizontal-cell-tolerance", "Horizontal merge tolerance (default: 1.0)",                 cxxopts::value<double>())
      ("word-space-factor",  "Space-width factor for word merging (default: 0.33)",              cxxopts::value<double>())
      ("line-space-factor",  "Space-width factor for line merging (default: 1.0)",               cxxopts::value<double>())
      ("line-space-factor-with-space", "Space-width factor for line merging with space (default: 0.33)", cxxopts::value<double>())
      ("keep-glyphs",        "Keep unmapped GLYPH<...> tokens (default: false)",                 cxxopts::value<bool>()->implicit_value("true"))
      ("keep-qpdf-warnings", "Emit QPDF warnings (default: false)",                              cxxopts::value<bool>()->implicit_value("true"))
      ("populate-json",      "Populate JSON objects during decode (default: false)",              cxxopts::value<bool>()->implicit_value("true"));

    // Parse command line arguments
    auto result = options.parse(argc, argv);

    // Check if either input or config file is provided (mandatory)
    if (orig_argc == 1) {
      LOG_F(ERROR, "Either input (-i) or config (-c) must be specified.");
      LOG_F(INFO, "%s", options.help().c_str());
      return 1;
    }

    std::string level = "warning";
    if (result.count("loglevel")) {
      level = result["loglevel"].as<std::string>();

      // Convert the string to lowercase
      std::transform(level.begin(), level.end(), level.begin(), [](unsigned char c) {