def test_azure_ocr_good_file(self):
azure_ocr_good_file = os.path.join(os.path.dirname(os.path.abspath(__file__)),
'test_samples', 'azure_ocr', 'good_text.json')
with open(azure_ocr_good_file, 'rb') as sample_reponse:
reponse_json = json.load(sample_reponse)
box_list = BoundBox.azure_ocr_boxes(reponse_json)
merged_box = BoundBox.void_box()
# google ocr returns a list of list
for box in box_list[0]:
merged_box += box
self.assertEqual(merged_box.text_value, 'Noisy image to test Tesseract OCR')
box_list_for_lines = BoundBox.azure_ocr_boxes(reponse_json, merge_line=True)
merged_box = BoundBox.void_box()
# google ocr returns a list of list
for box in box_list_for_lines[0]:
merged_box += box
self.assertEqual(merged_box.text_value, 'Noisy image to test Tesseract OCR')
def test_azure_ocr_blank_file(self):
azure_ocr_blank_file = os.path.join(os.path.dirname(os.path.abspath(__file__)),
'test_samples', 'azure_ocr', 'blank_image.json')
with open(azure_ocr_blank_file, 'rb') as sample_reponse:
reponse_json = json.load(sample_reponse)
box_list = BoundBox.azure_ocr_boxes(reponse_json)
self.assertEqual(len(box_list[0]), 0)