def test_line_error(self, popen):
message = ("Cuneiform for Linux 1.1.0\n"
"Magick: Improper image header (example.png) reported by "
"coders/png.c:2932 (ReadPNGImage)\n")
self.stdout.stdout.read.return_value = message.encode()
self.stdout.wait.return_value = 1
popen.return_value = self.stdout
with self.assertRaises(cuneiform.CuneiformError) as ce:
cuneiform.image_to_string(self.image, builder=self.builder)
self.assertEqual(ce.exception.status, 1)
self.assertEqual(ce.exception.message, message)
def test_write_read(self):
original_boxes = cuneiform.image_to_string(
Image.open(
os.path.join("tests", "input", "specific", "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_txt(self, image_file, expected_box_file, lang='eng'):
image_file = os.path.join("tests", "input", "specific", image_file)
expected_box_file = os.path.join(
"tests", "output", "specific", "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_text(self, popen, copen, temp_file):
popen.return_value = self.stdout
copen.return_value = self.text_file
temp_file.return_value = self.enter
output = cuneiform.image_to_string(self.image, builder=self.builder)
self.assertEqual(output, self._get_file_content("text").strip())
popen.assert_called_once_with(
["cuneiform", "-f", "text", "-o", self.tmp_filename, "-"],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
def test_line(self, popen, copen, temp_file):
popen.return_value = self.stdout
copen.return_value = self.text_file
temp_file.return_value = self.enter
output = cuneiform.image_to_string(self.image, builder=self.builder)
popen.assert_called_once_with(
["cuneiform", "-f", "hocr", "-o", self.tmp_filename, "-"],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
for box in output:
self.assertIsInstance(box, builders.LineBox)
def test_image_to_string_defaults_to_text_buidler(self, popen, copen,
temp_file, get_version):
get_version.return_value = (4, 0, 0)
popen.return_value = self.stdout
copen.return_value = self.text_file
temp_file.return_value = self.enter
output = cuneiform.image_to_string(self.image)
self.assertEqual(output, self._get_file_content("text").strip())
popen.assert_called_once_with(
["cuneiform", "-f", "text", "-o", self.tmp_filename, "-"],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
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_text_non_rgb_image(self, popen, copen, temp_file):
"""This tests that image_to_string works with non RGB mode images and
that image is converted in function."""
image = self.image.convert("L")
popen.return_value = self.stdout
copen.return_value = self.text_file
temp_file.return_value = self.enter
output = cuneiform.image_to_string(image, builder=self.builder)
self.assertEqual(output, self._get_file_content("text").strip())
popen.assert_called_once_with(
["cuneiform", "-f", "text", "-o", self.tmp_filename, "-"],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
def __test_txt(self, image_file, expected_output_file, lang='eng'):
image_file = os.path.join("tests", "input", "specific", image_file)
expected_output_file = os.path.join("tests", "output", "specific",
"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 = os.path.join("tests", "input", "specific", image_file)
expected_output_file = os.path.join(
"tests", "output", "specific", "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 extract_name_from_image(self, cropped_chunk, counter):
threshold_names = 120
"""
Для имен еще надо:
- отрезать слева иконку и класс
- попытаться примерно угадать, где вторая строка с уровнем и званием, и удалить её тоже
"""
image_with_name = self.get_part_with_name(cropped_chunk)
# cut icon and class
imgwidth, imgheight = image_with_name.size
percentage_value_x = 0.35
percentage_value_y = 0.5
newwidth = floor(imgwidth * percentage_value_x)
newheight = floor(imgheight * percentage_value_y)
box = (newwidth, 0, floor(imgwidth * 0.8), newheight)
image_with_name = image_with_name.crop(box)
converted_names = image_with_name
# converted_names = grayscale(image_with_name, threshold_names)
# converted_names = add_border(converted_names)
cropped_img = Image.new('RGB', converted_names.size, 255)
cropped_img.paste(converted_names)
# cropped_img.save('chunk-{}-{}'.format(counter, threshold_names) + '.png')
try:
# cuneiform
line_and_word_boxes = image_to_string(
converted_names,
lang='eng'
# builder=pyocr.builders.LineBoxBuilder()
)
text_data_cunei = line_and_word_boxes
# for lb in line_and_word_boxes:
# print('LB', lb.content)
except pyocr.error.CuneiformError:
text_data_cunei = None
pass
# text_data = merge_nearest_chars(line_and_word_boxes)
text_data_tesseract = pytesseract.image_to_string(converted_names)
return {'tesseract': text_data_tesseract, 'cunei': text_data_cunei}
def extract_numbers_from_image(self, cropped_chunk, image_name=None):
"""
Работает через cuneiForm.
:param cropped_chunk:
:param image_name:
:return:
"""
threshold_numbers = 70
converted_numbers = self.get_part_with_numbers(cropped_chunk)
converted_numbers = self.grayscale(converted_numbers, threshold_numbers)
converted_numbers = self.add_border(converted_numbers)
if image_name is not None and self.debug:
converted_numbers.save('debug_images/' + 'extracted_numbers-' + image_name)
line_and_word_boxes = image_to_string(
converted_numbers,
lang='ruseng',
builder=pyocr.builders.LineBoxBuilder()
)
text_numbers_data = self.merge_nearest_chars(line_and_word_boxes)
return text_numbers_data
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_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_digits_box_not_implemented(self):
with self.assertRaises(NotImplementedError):
cuneiform.image_to_string(self.image, builder=self.builder)