def __test_txt(self, image_file, expected_box_file, lang='eng'):
image_file = "tests/data/" + image_file
expected_box_file = "tests/cuneiform/" + expected_box_file
with codecs.open(expected_box_file, 'r', encoding='utf-8') \
as file_descriptor:
expected_boxes = self.builder.read_file(file_descriptor)
expected_boxes.sort()
boxes = cuneiform.image_to_string(Image.open(image_file),
lang=lang,
builder=self.builder)
boxes.sort()
self.assertEqual(len(boxes), len(expected_boxes))
for i in range(0, min(len(boxes), len(expected_boxes))):
try:
# Python 2.7
self.assertEqual(type(expected_boxes[i].content), unicode)
self.assertEqual(type(boxes[i].content), unicode)
except NameError:
# Python 3.x
self.assertEqual(type(expected_boxes[i].content), str)
self.assertEqual(type(boxes[i].content), str)
self.assertEqual(boxes[i], expected_boxes[i])
def __test_txt(self, image_file, expected_box_file, lang='eng'):
image_file = "tests/data/" + image_file
expected_box_file = "tests/cuneiform/" + expected_box_file
with codecs.open(expected_box_file, 'r', encoding='utf-8') \
as file_descriptor:
expected_boxes = self.builder.read_file(file_descriptor)
expected_boxes.sort()
boxes = cuneiform.image_to_string(Image.open(image_file), lang=lang,
builder=self.builder)
boxes.sort()
self.assertEqual(len(boxes), len(expected_boxes))
for i in range(0, min(len(boxes), len(expected_boxes))):
try:
# Python 2.7
self.assertEqual(type(expected_boxes[i].content), unicode)
self.assertEqual(type(boxes[i].content), unicode)
except NameError:
# Python 3.x
self.assertEqual(type(expected_boxes[i].content), str)
self.assertEqual(type(boxes[i].content), str)
self.assertEqual(boxes[i], expected_boxes[i])
def __test_txt(self, image_file, expected_output_file, lang='eng'):
image_file = "tests/data/" + image_file
expected_output_file = "tests/cuneiform/" + expected_output_file
expected_output = ""
with codecs.open(expected_output_file, 'r', encoding='utf-8') \
as file_descriptor:
for line in file_descriptor:
expected_output += line
expected_output = expected_output.strip()
output = cuneiform.image_to_string(Image.open(image_file), lang=lang)
self.assertEqual(output, expected_output)
def __test_txt(self, image_file, expected_output_file, lang='eng'):
image_file = "tests/data/" + image_file
expected_output_file = "tests/cuneiform/" + expected_output_file
expected_output = ""
with codecs.open(expected_output_file, 'r', encoding='utf-8') \
as file_descriptor:
for line in file_descriptor:
expected_output += line
expected_output = expected_output.strip()
output = cuneiform.image_to_string(Image.open(image_file), lang=lang)
self.assertEqual(output, expected_output)
def test_write_read(self):
original_boxes = cuneiform.image_to_string(
Image.open("tests/data/test.png"), builder=self.builder)
self.assertTrue(len(original_boxes) > 0)
(file_descriptor, tmp_path) = tempfile.mkstemp()
try:
# we must open the file with codecs.open() for utf-8 support
os.close(file_descriptor)
with codecs.open(tmp_path, 'w', encoding='utf-8') as file_descriptor:
self.builder.write_file(file_descriptor, original_boxes)
with codecs.open(tmp_path, 'r', encoding='utf-8') as file_descriptor:
new_boxes = self.builder.read_file(file_descriptor)
self.assertEqual(len(new_boxes), len(original_boxes))
for i in range(0, len(original_boxes)):
self.assertEqual(new_boxes[i], original_boxes[i])
finally:
os.remove(tmp_path)
def test_write_read(self):
original_boxes = cuneiform.image_to_string(
Image.open("tests/data/test.png"), builder=self.builder)
self.assertTrue(len(original_boxes) > 0)
(file_descriptor, tmp_path) = tempfile.mkstemp()
try:
# we must open the file with codecs.open() for utf-8 support
os.close(file_descriptor)
with codecs.open(tmp_path, 'w', encoding='utf-8') as file_descriptor:
self.builder.write_file(file_descriptor, original_boxes)
with codecs.open(tmp_path, 'r', encoding='utf-8') as file_descriptor:
new_boxes = self.builder.read_file(file_descriptor)
self.assertEqual(len(new_boxes), len(original_boxes))
for i in range(0, len(original_boxes)):
self.assertEqual(new_boxes[i], original_boxes[i])
finally:
os.remove(tmp_path)