def run_img_rec(data):
'''
Runs OCR. Takes in an ImgPack, and returns an image
descriptor object with the text object filled out.
'''
import praxyk
print('Transferring image data to %s...' % (STORE_BASENAME + data.img_name))
img_out = open(STORE_BASENAME + data.img_name, 'w+')
img_out.write(data.data)
img_out.close()
print('{Running OCR...}')
data.img_text = praxyk.get_string_from_image(STORE_BASENAME + data.img_name)
data.dump_data()
print('This is where data can be returned to the user')
return data.img_text
def do_ocr(trans, fileh, file_num) :
trans_id = trans.id
# this if statement checks if the trans has been canceled, if so we mark this result as canceled
# and return right away, avoiding further computation
if trans.status == Transaction.TRANSACTION_CANCELED :
this_result = Result_POD_OCR.query.filter(transaction_id=trans_id).filter(item_number=file_num).first()
if this_result :
this_result.finished_at = datetime.datetime.now()
this_result.status = Result_POD_OCR.RESULT_CANCELED
this_result.result_string = ""
this_result.save(full=True)
app_context.pop()
return this_result
# get the individual result struct from redis that this queue task is processing
this_result = Result_POD_OCR.query.filter(transaction_id=trans_id).filter(item_number=file_num).first()
if not this_result :
print "POD Worker Error, Can't Find This Result"
return False
# if we're the first image in the queue, create the working directory for this transaction
imgs_dir = STORE_BASENAME + str(trans_id)
if not os.path.exists(imgs_dir):
os.makedirs(imgs_dir)
# put the image data into a temporary file for the OCR program to use
file_img = imgs_dir+fileh['name']+'_'+str(file_num)
with open(file_img, 'wr+') as fh :
fh.write(fileh['data'])
# get the actual ocr string from the praxyk pytrhon library
this_result.result_string = praxyk.get_string_from_image(file_img)
this_result.finished_at = datetime.datetime.now()
this_result.status = Result_POD_OCR.RESULT_FINISHED
# update this result object in the redis db
this_result.save(full=True)
# clean up the image space used
os.remove(file_img)
print "POD_OCR Result String : (%s) " % this_result.result_string
return this_result
def test_empty_image():
try:
# If system not set up, don't necessarily fail
if "PRAXYK_IMAGES_DIR" not in os.environ:
return True
images_dir = os.environ["PRAXYK_IMAGES_DIR"]
empty_image = os.path.join(images_dir, "empty.png")
image_str = praxyk.get_string_from_image(empty_image)
if image_str != "":
raise RuntimeError("Praxyk detected a string in an empty image.")
return True
except:
print
exc_type, exc_value, exc_traceback = sys.exc_info()
traceback.print_exception(exc_type, exc_value, exc_traceback,
limit=5, file=sys.stdout)
return False
def test_typed_text():
try:
# If system not set up, don't necessarily fail
if "PRAXYK_IMAGES_DIR" not in os.environ:
return True
images_dir = os.environ["PRAXYK_IMAGES_DIR"]
typed_image = os.path.join(images_dir, "typed_text.png")
image_str = praxyk.get_string_from_image(typed_image)
if image_str == "":
raise RuntimeError("Praxyk detected no string in image with typed text.")
should_be = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
if editdistance.eval(image_str, should_be) > len(should_be)/4:
raise RuntimeError("Praxyk failed to detect correct string.")
return True
except:
print
exc_type, exc_value, exc_traceback = sys.exc_info()
traceback.print_exception(exc_type, exc_value, exc_traceback,
limit=5, file=sys.stdout)
return False
