def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
try:
img = Image.open(StringIO(data))
except DecompressionBombWarning as dce:
log.debug("Image too large: %", dce)
return None
except IOError as ioe:
log.info("Unknown image format: %r", ioe)
return None
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_image(img)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.get_text() or ''
text = text.decode(encoding="UTF-8")
# extractor.clear()
log.debug('OCR done: %s, %s characters extracted', languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def extract_text(self, data, languages=None):
key = sha1(data).hexdigest()
text = Cache.get_cache(key)
if text is not None:
log.info('OCR: %s chars cached', len(text))
return text
# log.info("Size: %s", len(data))
data = self.ensure_size(data)
if data is None:
return
for attempt in range(10):
try:
service = RecognizeTextStub(self.channel)
languages = ensure_list(languages)
image = RPCImage(data=data, languages=languages)
response = service.Recognize(image)
log.info('OCR: %s chars recognized', len(response.text))
if response.text is not None:
Cache.set_cache(key, response.text)
return response.text
except self.Error as exc:
log.exception("gRPC Error: %s", self.SERVICE)
self.reset_channel()
backoff(failures=attempt)
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
try:
img = Image.open(StringIO(data))
except DecompressionBombWarning as dce:
log.debug("Image too large: %", dce)
return None
except IOError as ioe:
log.info("Unknown image format: %r", ioe)
return None
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
extractor.clear()
log.debug('OCR done: %s, %s characters extracted',
languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def extract_text(self, data, languages=None):
key = sha1(data).hexdigest()
text = Cache.get_cache(key)
if text is not None:
# log.info('%s chars cached', len(text))
return text
data = self.ensure_size(data)
if data is None:
return
for attempt in range(1000):
try:
service = RecognizeTextStub(self.channel)
languages = ensure_list(languages)
image = Image(data=data, languages=languages)
response = service.Recognize(image)
log.info('OCR: %s chars', len(response.text))
if response.text is not None:
Cache.set_cache(key, response.text)
return response.text
except self.Error as e:
if e.code() == self.Status.RESOURCE_EXHAUSTED:
continue
log.warning("gRPC [%s]: %s", e.code(), e.details())
backoff(failures=attempt)
self.reset_channel()
def test_cache_basic(self):
assert None is Cache.get_cache('foo'), Cache.get_cache('foo')
assert db.session.query(Cache).count() == 0
Cache.set_cache('foo', 'bar')
assert 'bar' == Cache.get_cache('foo'), Cache.get_cache('foo')
assert db.session.query(Cache).count() == 1
Cache.set_cache('foo', 'quuux')
assert 'quuux' == Cache.get_cache('foo'), Cache.get_cache('foo')
assert db.session.query(Cache).count() == 1
def extract_text(self, data, languages=None):
key = sha1(data).hexdigest()
text = Cache.get_cache(key)
if text is not None:
log.info('Vision API: %s chars cached', len(text))
return text
data = self.ensure_size(data)
if data is not None:
image = types.Image(content=data)
res = self.client.document_text_detection(image)
ann = res.full_text_annotation
log.info('Vision API: %s chars recognized', len(ann.text))
Cache.set_cache(key, ann.text)
return ann.text
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
img = Image.open(StringIO(data))
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
log.debug('OCR done: %s, %s characters extracted', languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
img = Image.open(StringIO(data))
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
log.debug('OCR done: %s, %s characters extracted',
languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def set_cache(self, key, value):
Cache.set_cache(key, value)
def get_cache(self, key):
return Cache.get_cache(key)
