Method ProcessPagesFileList

deps/tesseract/api/baseapi.cpp:960–1016

If flist exists, get data from there. Otherwise get data from buf. Seems convoluted, but is the easiest way I know of to meet multiple goals. Support streaming from stdin, and also work on platforms lacking fmemopen.

// goals. Support streaming from stdin, and also work on platforms
// lacking fmemopen.
bool TessBaseAPI::ProcessPagesFileList(FILE *flist,
                                       STRING *buf,
                                       const char* retry_config,
                                       int timeout_millisec,
                                       TessResultRenderer* renderer,
                                       int tessedit_page_number) {
  if (!flist && !buf) return false;
  int page = (tessedit_page_number >= 0) ? tessedit_page_number : 0;
  char pagename[MAX_PATH];

  GenericVector<STRING> lines;
  if (!flist) {
    buf->split('\n', &lines);
    if (lines.empty()) return false;
  }

  // Skip to the requested page number.
  for (int i = 0; i < page; i++) {
    if (flist) {
      if (fgets(pagename, sizeof(pagename), flist) == NULL) break;
    }
  }

  // Begin producing output
  if (renderer && !renderer->BeginDocument(unknown_title_)) {
    return false;
  }

  // Loop over all pages - or just the requested one
  while (true) {
    if (flist) {
      if (fgets(pagename, sizeof(pagename), flist) == NULL) break;
    } else {
      if (page >= lines.size()) break;
      snprintf(pagename, sizeof(pagename), "%s", lines[page].c_str());
    }
    chomp_string(pagename);
    Pix *pix = pixRead(pagename);
    if (pix == NULL) {
      tprintf("Image file %s cannot be read!\n", pagename);
      return false;
    }
    tprintf("Page %d : %s\n", page, pagename);
    bool r = ProcessPage(pix, page, pagename, retry_config,
                         timeout_millisec, renderer);
    pixDestroy(&pix);
    if (!r) return false;
    if (tessedit_page_number >= 0) break;
    ++page;
  }

  // Finish producing output
  if (renderer && !renderer->EndDocument()) {
    return false;
  }
  return true;
}