Method Build

deps/tesseract/ccstruct/imagedata.cpp:134–162

Builds and returns an ImageData from the basic data. Note that imagedata, truth_text, and box_text are all the actual file data, NOT filenames.

// Builds and returns an ImageData from the basic data. Note that imagedata,
// truth_text, and box_text are all the actual file data, NOT filenames.
ImageData* ImageData::Build(const char* name, int page_number, const char* lang,
                            const char* imagedata, int imagedatasize,
                            const char* truth_text, const char* box_text) {
  ImageData* image_data = new ImageData();
  image_data->imagefilename_ = name;
  image_data->page_number_ = page_number;
  image_data->language_ = lang;
  // Save the imagedata.
  image_data->image_data_.resize_no_init(imagedatasize);
  memcpy(&image_data->image_data_[0], imagedata, imagedatasize);
  if (!image_data->AddBoxes(box_text)) {
    if (truth_text == NULL || truth_text[0] == '\0') {
      tprintf("Error: No text corresponding to page %d from image %s!\n",
              page_number, name);
      delete image_data;
      return NULL;
    }
    image_data->transcription_ = truth_text;
    // If we have no boxes, the transcription is in the 0th box_texts_.
    image_data->box_texts_.push_back(truth_text);
    // We will create a box for the whole image on PreScale, to save unpacking
    // the image now.
  } else if (truth_text != NULL && truth_text[0] != '\0' &&
             image_data->transcription_ != truth_text) {
    // Save the truth text as it is present and disagrees with the box text.
    image_data->transcription_ = truth_text;
  }
  return image_data;
}

// Writes to the given file. Returns false in case of error.
bool ImageData::Serialize(TFile* fp) const {