def setUp(self):
self.image_path = get_file_path("test_jpg.jpg")
self.img_bytes = azure_services.get_file_bytes(self.image_path)
self.ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
def test_none_returned_when_input_is_none(self):
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
self.assertIsNone(ta.get_results(None))
def test_invalid_api_returns_none(self):
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url="invalid/api/url")
response = ta.submit_image_for_processing()
self.assertIsNone(response)
def test_bad_subscription_key_returns_none(self):
"""
Check that analyse_image returns None when prediction_key == None
"""
ta = azure_services.TextAnalyser(image_data=self.img_bytes,
subscription_key="invalid_subs_key",
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
self.assertIsNone(response)
def test_non_image_input_returns_none(self):
txt_bytes = azure_services.get_file_bytes(
get_file_path("test_file.txt"))
ta = azure_services.TextAnalyser(
image_data=txt_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
self.assertIsNone(response)
def test_successful_reponse_returned_with_correct_parameters(self):
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
# From the docs, a successful response will have status code 202:
# https://westcentralus.dev.cognitive.microsoft.com/docs/services/computer-vision-v3-ga/operations/5d986960601faab4bf452005
self.assertEqual(response.status_code, 202)
def test_correct_results_returned_from_correct_input(self):
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
self.assertEqual(
ta.get_results(response)["analyzeResult"],
RESULTS.OCR_RESULTS["analyzeResult"])
def test_none_returned_with_wrong_response(self):
"""
Input is a valid response object, but not from the Read API
"""
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = requests.get("https://api.github.com")
self.assertIsNone(ta.get_results(response))
def test_too_large_image_returns_none(self):
"""
Check that analyse_image returns None when image_path == None
"""
bmp_bytes = azure_services.get_file_bytes(
get_file_path("test_bmp.bmp"))
ta = azure_services.TextAnalyser(
image_data=bmp_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
self.assertIsNone(response)
def test_correct_results_returned_from_valid_input(self):
"""
Check that analyse_image returns the correct results when provided the correct arguments
"""
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
results = ta.analyse_image()
expected_results = RESULTS.OCR_RESULTS
self.assertDictEqual(results["analyzeResult"],
expected_results["analyzeResult"])
def setUp(self):
self.image_path = get_file_path("test_jpg.jpg")
self.img_bytes = azure_services.get_file_bytes(self.image_path)
self.ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
# This took a long time to puzzle out...
# If I assign without copy.deepcopy(), the changes that the tests
# make on the 'status' key are carried through to RESULTS.OCR_RESULTS;
# and are *NOT* reset for the next test!!
# Particularly important when 'status' is popped...
self.results = copy.deepcopy(RESULTS.OCR_RESULTS)
def test_none_returned_with_missing_key(self):
"""
Input is a valid response object from the Read API, but the value for
the "Operation-Location" key is missing from the header
"""
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
response.headers["Operation-Location"] = None
self.assertIsNone(ta.get_results(response))
def test_use_words_false_extracts_correct_lines(self):
# create a new ta object for the file we want to process
image_path = get_file_path("lines_of_words.jpg")
img_bytes = azure_services.get_file_bytes(image_path)
ta = azure_services.TextAnalyser(
image_data=img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL,
use_words=False)
# do the test
test_input = RESULTS.OCR_TEXT_IN_LINES
expected_results = RESULTS.PROCESSED_OCR_TEXT_IN_LINES
actual_results = ta.process_output(test_input)
self.assertListEqual(expected_results, actual_results)
def test_none_returned_with_malformed_url(self):
"""
Input is a valid response object from the Read API, but the value for
the "Operation-Location" key is missing from the header
"""
ta = azure_services.TextAnalyser(
image_data=self.img_bytes,
subscription_key=settings.OCR_SUBSCRIPTION_KEY,
api_url=settings.OCR_API_URL)
response = ta.submit_image_for_processing()
# get the correct URL and slice off the last digit
new_url = response.headers["Operation-Location"][:-1]
# set the 'Operation-Location' header to be this new URL
response.headers["Operation-Location"] = new_url
self.assertIsNone(ta.get_results(response))