def extract_text(self, data, languages=None):
"""Extract text from a binary string of data."""
try:
image = Image.open(BytesIO(data))
image.load()
except Exception:
log.exception("Cannot open image data using Pillow")
return ''
with temp_locale(TESSERACT_LOCALE):
languages = self.language_list(languages)
api = self.configure_engine(languages)
try:
# TODO: play with contrast and sharpening the images.
start_time = time.time()
api.SetImage(image)
text = api.GetUTF8Text()
confidence = api.MeanTextConf()
end_time = time.time()
duration = end_time - start_time
log.info("w: %s, h: %s, l: %s, c: %s, took: %.5f",
image.width, image.height, languages,
confidence, duration)
return text
except Exception as exc:
log.warning("OCR error: %s", exc)
return ''
finally:
api.Clear()
def language_list(self, languages):
if not hasattr(settings, 'ocr_supported'):
with temp_locale(TESSERACT_LOCALE):
# Tesseract language types:
from tesserocr import get_languages
_, settings.ocr_supported = get_languages()
# log.info("OCR languages: %r", settings.ocr_supported)
models = [c for c in alpha3(languages) if c in settings.ocr_supported]
if len(models) > self.MAX_MODELS:
log.warning("Too many models, limit: %s", self.MAX_MODELS)
models = models[:self.MAX_MODELS]
models.append('eng')
return '+'.join(sorted(set(models)))