def plaintextlines_service():
image_uri = request.args.get('imageURI')
width = request.args.get('width')
height = request.args.get('height')
s3 = aws.get_s3_resource()
text_data = get_text_index(s3, image_uri)
word_index = text_data.get("word_index")
o_width = text_data.get("width")
o_height = text_data.get("height")
scale_w = float(width) / float(o_width)
scale_h = float(height) / float(o_height)
if 'line_number' not in word_index[0]:
return jsonify({"error": "plaintext lines not available"}), 415
lines = []
line = ""
line_boxes = []
current_line = -1
for word in word_index:
if word['line_number'] != current_line:
if current_line != -1:
lines.append({"text": line, "xywh": box_join(line_boxes)})
line = ""
line_boxes = []
current_line = word['line_number']
line += word['text'] + " "
line_boxes.append(get_box(word, scale_w, scale_h))
return jsonify({"lines": lines})
def run(self):
self.set_logging()
self.sqs = aws.get_sqs_resource()
self.transcoder = aws.get_transcoder_client()
self.s3 = aws.get_s3_resource()
self.notification_queue = self.get_notification_queue()
self.response_queue = self.get_response_queue()
self.preset_id_map = aws.get_preset_map(self.transcoder, inverse=True)
self.inverse_policy_map = self.get_inverse_policy_map()
while True:
try:
while True:
if os.path.exists('/tmp/stop.txt'):
sys.exit()
for message in self.get_messages_from_queue():
if message is not None:
try:
self.process_message(message)
except:
logging.exception("Error processing message")
finally:
message.delete()
except Exception as e:
logging.exception("Error getting messages")
raise e
def source_metadata():
image_uri = request.args.get('imageURI')
s3 = aws.get_s3_resource()
metadata = get_source_metadata(s3, image_uri)
if metadata is None:
return "", 404
return metadata
def get_confidence(image):
s3 = aws.get_s3_resource()
text_data = get_text_index(s3, image)
result = {"imageURI": image}
confidence = text_data.get('confidence')
if confidence is not None:
result['confidence'] = confidence
return result
def init(self):
self.set_logging()
self.sqs = aws.get_sqs_resource()
self.s3 = aws.get_s3_resource()
self.ingest_queue = aws.get_queue_by_name(self.sqs, settings.INGEST_QUEUE)
self.error_queue = aws.get_queue_by_name(self.sqs, settings.ERROR_QUEUE)
self.text_queue = aws.get_queue_by_name(self.sqs, settings.TEXT_QUEUE)
self.iris = iris_client.IrisClient()
def plaintext_service():
image_uri = request.args.get('imageURI')
s3 = aws.get_s3_resource()
text_data = get_text_index(s3, image_uri)
word_index = text_data.get("word_index")
text = " ".join(map(lambda w: w['text'], word_index))
return jsonify({image_uri: text})
def get_error_queue():
return aws.get_queue_by_name(sqs, settings.ERROR_QUEUE)
def get_pipeline():
return aws.get_pipeline_by_name(transcoder, settings.PIPELINE)
def get_random_prefix():
return "x/" + str(random.randint(0, 1000)).zfill(4) + "/"
if __name__ == "__main__":
if settings.DEBUG:
logzero.loglevel(logging.DEBUG)
else:
logzero.loglevel(logging.INFO)
requested_to_quit = False
sqs = aws.get_sqs_resource()
transcoder = aws.get_transcoder_client()
s3 = aws.get_s3_resource()
input_queue = get_input_queue()
error_queue = get_error_queue()
pipeline = get_pipeline()
preset_id_map = aws.get_preset_map(transcoder)
main()
def get_coords(image):
"""
{"positions": [[25, 31], 100, 110], "height": 768, "width": 1024, "imageURI": "http://aplaceforstuff.co.uk/x/"}
"""
boxes = []
s3 = aws.get_s3_resource()
image_uri = image.get("imageURI")
output = {"image_uri": image_uri}
positions = image.get("positions")
single_box = image.get("single-box") is True
text_data = get_text_index(s3, image_uri)
if text_data is None:
return output
o_width = text_data.get("width")
o_height = text_data.get("height")
canvas_width = text_data.get("canvas_width")
canvas_height = text_data.get("canvas_height")
if o_width is None or o_height is None:
output = {"imageURI": image_uri, "message": "no coordinates available"}
return output
word_index = text_data.get("word_index")
start_index = text_data.get("start_index")
# width to project coordinates in
if 'width' in image:
width = image.get("width")
else:
width = canvas_width
# height to project coordinates in
if 'height' in image:
height = image.get("height")
else:
height = canvas_height
scale_w = float(width) / float(o_width)
scale_h = float(height) / float(o_height)
if word_index is None:
raise Exception
# TODO : handle better!
for phrase_or_position in positions:
if isinstance(phrase_or_position, int):
# single word
idx = start_index.get(str(phrase_or_position))
if idx is None:
# we can't generate a box for this single word
p_boxes = [get_null_box_object(1)]
else:
word_data = word_index[idx]
p_boxes = box_join(get_box(word_data, scale_w, scale_h),
single_box=single_box)
else:
# phrase
phrase_boxes = []
p_boxes = None
for position in phrase_or_position:
idx = start_index.get(str(position))
if idx is None:
# we don't have data for this word, generate a null box for the whole phrase
p_boxes = [get_null_box_object(len(phrase_or_position))]
break
else:
word_data = word_index[idx]
position_box = get_box(word_data, scale_w, scale_h)
phrase_boxes.append(position_box)
if not p_boxes:
p_boxes = box_join(phrase_boxes, single_box=single_box)
boxes.append(p_boxes)
output['phrases'] = boxes
return output
def get_coords(image):
"""
{"positions": [[25, 31], 100, 110], "height": 768, "width": 1024, "imageURI": "http://aplaceforstuff.co.uk/x/"}
"""
boxes = []
s3 = aws.get_s3_resource()
image_uri = image.get("imageURI")
positions = image.get("positions")
text_data = get_text_index(s3, image_uri)
o_width = text_data.get("width")
o_height = text_data.get("height")
if o_width is None or o_height is None:
output = {
"imageURI": image_uri,
"message": "no coordinates available"
}
return output
word_index = text_data.get("word_index")
start_index = text_data.get("start_index")
width = image.get("width")
height = image.get("height")
scale_w = float(width) / float(o_width)
scale_h = float(height) / float(o_height)
if word_index is None:
raise Exception
# TODO : handle better!
for phrase_or_position in positions:
if isinstance(phrase_or_position, int):
# single word
idx = start_index.get(str(phrase_or_position))
if idx is None:
# todo handle
pass
word_data = word_index[idx]
p_boxes = box_join(get_box(word_data, scale_w, scale_h))
else:
# phrase
phrase_boxes = []
for position in phrase_or_position:
idx = start_index.get(str(position))
if idx is None:
# todo handle
pass
word_data = word_index[idx]
position_box = get_box(word_data, scale_w, scale_h)
phrase_boxes.append(position_box)
p_boxes = box_join(phrase_boxes)
boxes.append(p_boxes)
output = {
"image_uri": image_uri,
"phrases": boxes
}
return output
