def test_version_error(self, popen):
self.stdout.wait.return_value = 2
popen.return_value = self.stdout
with self.assertRaises(tesseract.TesseractError) as te:
tesseract.get_version()
self.assertEqual(te.exception.status, 2)
self.assertEqual(te.exception.message, self.message)
def test_version_error_nan(self, popen):
message = self.message.replace("tesseract 4.0.0", "tesseract A.B.C")
self.stdout.stdout.read.return_value = message.encode()
popen.return_value = self.stdout
with self.assertRaises(tesseract.TesseractError) as te:
tesseract.get_version()
self.assertEqual(te.exception.status, 0)
self.assertIn("Unable to parse Tesseract version (not a number): ",
te.exception.message)
def test_version_error_splitting(self, popen):
tesseract.g_version = None # drop cached version
message = self.message.replace(b"tesseract 4.0.0", b"tesseract 3")
self.stdout.stdout.read.return_value = message
popen.return_value = self.stdout
with self.assertRaises(tesseract.TesseractError) as te:
tesseract.get_version()
self.assertEqual(te.exception.status, 0)
self.assertIn("Unable to parse Tesseract version (spliting failed): ",
te.exception.message)
def test_version_cache(self, popen):
"""
Make sure Tesseract is not called everytime we need the version.
We need the version *often* in the code, and calling Tesseract
everytime wouldn't be wise.
"""
tesseract.g_version = None # drop cached version
self.stdout.stdout.read.return_value = self.message
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (4, 0, 0))
self.stdout.stdout.read.return_value = "garbage"
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (4, 0, 0))
class TestContext(unittest.TestCase):
"""
These tests make sure the requirements for the tests are met.
"""
def setUp(self):
pass
def test_available(self):
self.assertTrue(tesseract.is_available(),
"Tesseract not found. Is it installed ?")
@unittest.skipIf(tesseract.get_version() != (3, 2, 1),
"This test only works with Tesseract 3.02.1")
def test_version(self):
self.assertEqual(tesseract.get_version(), (3, 2, 1),
("Tesseract does not have the expected version"
" (3.02.1) ! Tests will fail !"))
def test_langs(self):
langs = tesseract.get_available_languages()
self.assertTrue("eng" in langs,
("English training does not appear to be installed."
" (required for the tests)"))
self.assertTrue("fra" in langs,
("French training does not appear to be installed."
" (required for the tests)"))
self.assertTrue("jpn" in langs,
("Japanese training does not appear to be installed."
" (required for the tests)"))
def tearDown(self):
pass
def test_version_windows(self, popen):
tesseract.g_version = None # drop cached version
message = self.message.replace(b"tesseract 4.0.0",
b"tesseract v4.0.0.20181030")
self.stdout.stdout.read.return_value = message
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (4, 0, 0))
def test_version(self):
self.assertTrue(tesseract.get_version() in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
), ("Tesseract does not have the expected version"
" (3.3.0) ! Some tests will be skipped !"))
class TestTxt(unittest.TestCase):
"""
These tests make sure the "usual" OCR works fine. (the one generating
a .txt file)
"""
def setUp(self):
pass
def __test_txt(self, image_file, expected_output_file, lang='eng'):
image_file = "tests/data/" + image_file
expected_output_file = "tests/tesseract/" + 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 = tesseract.image_to_string(Image.open(image_file), lang=lang)
self.assertEqual(output, expected_output)
def test_basic(self):
self.__test_txt('test.png', 'test.txt')
@unittest.skipIf(tesseract.get_version() not in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
), "This test only works with Tesseract 3.02.1")
def test_european(self):
self.__test_txt('test-european.jpg', 'test-european.txt')
@unittest.skipIf(tesseract.get_version() not in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
), "This test only works with Tesseract 3.02.1")
def test_french(self):
self.__test_txt('test-french.jpg', 'test-french.txt', 'fra')
def test_japanese(self):
self.__test_txt('test-japanese.jpg', 'test-japanese.txt', 'jpn')
def tearDown(self):
pass
def test_version(self):
self.assertTrue(tesseract.get_version() in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
(3, 4, 0),
), ("Tesseract does not have the expected version"
" (3.4.0) ! Some tests will be skipped !"))
def test_version(self):
self.assertTrue(
tesseract.get_version() in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
(3, 4, 0),
(3, 4, 1),
(3, 5, 0),
), ("Tesseract does not have the expected version"))
def test_version_tesseract4(self, popen):
tesseract.g_version = None # drop cached version
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (4, 0, 0))
# stderr must be explicitely ignored when calling 'tesseract -v'.
# See https://gitlab.gnome.org/World/OpenPaperwork/pyocr/-/issues/118
popen.assert_called_once()
(args, kwargs) = popen.call_args
self.assertNotIn('stderr', kwargs)
def test_version(self):
self.assertTrue(
tesseract.get_version() in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
(3, 4, 0),
(3, 4, 1),
(3, 5, 0),
),
("Tesseract does not have the expected version")
)
def test_version_windows(self, popen):
message = self.message.replace("tesseract 4.0.0",
"tesseract v4.0.0.20181030")
self.stdout.stdout.read.return_value = message.encode()
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (4, 0, 0))
def test_version_tesseract3_no_minor(self, popen):
message = self.message.replace("tesseract 4.0.0", "tesseract 3.0")
self.stdout.stdout.read.return_value = message.encode()
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (3, 0, 0))
def test_version_tesseract4(self, popen):
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (4, 0, 0))
class TestCharBox(unittest.TestCase):
"""
These tests make sure that Tesseract box handling works fine.
"""
def setUp(self):
self.builder = tesseract.CharBoxBuilder()
def __test_txt(self, image_file, expected_box_file, lang='eng'):
image_file = "tests/data/" + image_file
expected_box_file = "tests/tesseract/" + 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 = tesseract.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))):
self.assertEqual(boxes[i], expected_boxes[i])
def test_basic(self):
self.__test_txt('test.png', 'test.box')
def test_european(self):
self.__test_txt('test-european.jpg', 'test-european.box')
def test_french(self):
self.__test_txt('test-french.jpg', 'test-french.box', 'fra')
@unittest.skipIf(tesseract.get_version() not in (
(3, 2, 1),
(3, 2, 2),
(3, 3, 0),
), "This test requires Tesseract 3.02.1")
def test_japanese(self):
self.__test_txt('test-japanese.jpg', 'test-japanese.box', 'jpn')
def test_write_read(self):
original_boxes = tesseract.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 fdescriptor:
self.builder.write_file(fdescriptor, original_boxes)
with codecs.open(tmp_path, 'r', encoding='utf-8') as fdescriptor:
new_boxes = self.builder.read_file(fdescriptor)
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 tearDown(self):
pass
def test_version_tesseract3_no_minor(self, popen):
tesseract.g_version = None # drop cached version
message = self.message.replace(b"tesseract 4.0.0", b"tesseract 3.0")
self.stdout.stdout.read.return_value = message
popen.return_value = self.stdout
self.assertSequenceEqual(tesseract.get_version(), (3, 0, 0))
def test_version(self):
self.assertEqual(tesseract.get_version(), (3, 2, 1),
("Tesseract does not have the expected version"
" (3.02.1) ! Tests will fail !"))
def test_version(self):
self.assertEqual(tesseract.get_version(), (3, 2, 1),
("Tesseract does not have the expected version"
" (3.02.1) ! Tests will fail !"))