Method ReCachePages

deps/tesseract/ccstruct/imagedata.cpp:505–549

Locks the pages_mutex_ and Loads as many pages can fit in max_memory_ starting at index pages_offset_.

// Locks the pages_mutex_ and Loads as many pages can fit in max_memory_
// starting at index pages_offset_.
bool DocumentData::ReCachePages() {
  SVAutoLock lock(&pages_mutex_);
  // Read the file.
  set_total_pages(0);
  set_memory_used(0);
  int loaded_pages = 0;
  pages_.truncate(0);
  TFile fp;
  if (!fp.Open(document_name_, reader_) ||
      !PointerVector<ImageData>::DeSerializeSize(false, &fp, &loaded_pages) ||
      loaded_pages <= 0) {
    tprintf("Deserialize header failed: %s\n", document_name_.string());
    return false;
  }
  pages_offset_ %= loaded_pages;
  // Skip pages before the first one we want, and load the rest until max
  // memory and skip the rest after that.
  int page;
  for (page = 0; page < loaded_pages; ++page) {
    if (page < pages_offset_ ||
        (max_memory_ > 0 && memory_used() > max_memory_)) {
      if (!PointerVector<ImageData>::DeSerializeSkip(false, &fp)) break;
    } else {
      if (!pages_.DeSerializeElement(false, &fp)) break;
      ImageData* image_data = pages_.back();
      if (image_data->imagefilename().length() == 0) {
        image_data->set_imagefilename(document_name_);
        image_data->set_page_number(page);
      }
      image_data->set_language(lang_);
      set_memory_used(memory_used() + image_data->MemoryUsed());
    }
  }
  if (page < loaded_pages) {
    tprintf("Deserialize failed: %s read %d/%d pages\n",
            document_name_.string(), page, loaded_pages);
    pages_.truncate(0);
  } else {
    tprintf("Loaded %d/%d pages (%d-%d) of document %s\n", pages_.size(),
            loaded_pages, pages_offset_, pages_offset_ + pages_.size(),
            document_name_.string());
  }
  set_total_pages(loaded_pages);
  return !pages_.empty();
}

// A collection of DocumentData that knows roughly how much memory it is using.
DocumentCache::DocumentCache(inT64 max_memory)