def requires_tesseract(func):
"""Decorator for tests that require Tesseract to be installed."""
try:
_tesseract_version()
except:
raise SkipTest("tesseract isn't installed")
return func
def test_that_with_old_tesseract_ocr_raises_an_exception_with_patterns():
# pylint: disable=W0212
if _tesseract_version() >= distutils.version.LooseVersion('3.03'):
raise SkipTest('tesseract is too new')
stbt.ocr(frame=load_image('ocr/192.168.10.1.png'),
mode=stbt.OcrMode.SINGLE_WORD,
tesseract_user_patterns=[r'\d\*.\d\*.\d\*.\d\*'])
def test_tesseract_user_patterns(patterns):
# pylint:disable=protected-access
if _tesseract_version() < LooseVersion('3.03'):
raise SkipTest('tesseract is too old')
# Now the real test:
assert u'192.168.10.1' == stbt.ocr(
frame=load_image('ocr/192.168.10.1.png'),
mode=stbt.OcrMode.SINGLE_WORD,
tesseract_user_patterns=patterns)
def test_tesseract_user_patterns(patterns):
# pylint:disable=protected-access
if _tesseract_version() < LooseVersion('3.03'):
raise SkipTest('tesseract is too old')
# Now the real test:
assert u'192.168.10.1' == stbt.ocr(
frame=load_image('ocr/192.168.10.1.png'),
mode=stbt.OcrMode.SINGLE_WORD,
tesseract_user_patterns=patterns)
def test_that_ocr_engine_has_an_effect():
if _tesseract_version() < LooseVersion("4.0"):
raise SkipTest('tesseract is too old')
f = load_image("ocr/ambig.png")
# This is a regression in tesseract 4.0's legacy engine, compared to 3.04:
assert "sillyness" not in stbt.ocr(f, engine=stbt.OcrEngine.TESSERACT)
assert "sillyness" not in stbt.ocr(f)
# ...but the new LSTM engine does read it correctly:
assert "sillyness" in stbt.ocr(f, engine=stbt.OcrEngine.LSTM)
with temporary_config({'ocr.engine': 'LSTM'}):
assert "sillyness" in stbt.ocr(f)
def test_that_ocr_engine_has_an_effect():
if _tesseract_version() < LooseVersion("4.0"):
raise SkipTest('tesseract is too old')
f = load_image("ocr/ambig.png")
# This is a regression in tesseract 4.0's legacy engine, compared to 3.04:
assert "sillyness" not in stbt.ocr(f, engine=stbt.OcrEngine.TESSERACT)
assert "sillyness" not in stbt.ocr(f)
# ...but the new LSTM engine does read it correctly:
assert "sillyness" in stbt.ocr(f, engine=stbt.OcrEngine.LSTM)
with temporary_config({'ocr.engine': 'LSTM'}):
assert "sillyness" in stbt.ocr(f)
def test_that_setting_config_options_has_an_effect():
# Unfortunately there are many tesseract config options and they are very
# complicated so it's difficult to write a test that tests that a config
# option is having the correct effect. Due to the difficulty in determining
# "correctness" instead here we test that setting a config option has an
# effect at all. This at least excercises our code which sets config
# options. I'm not happy about this and I hope to be able to replace this
# once we have more experience with these settings in the real world.
if _tesseract_version() >= LooseVersion('3.04'):
hocr_mode_config = {
"tessedit_create_txt": 0,
"tessedit_create_hocr": 1
}
else:
hocr_mode_config = {"tessedit_create_hocr": 1}
assert (stbt.ocr(frame=load_image('ocr/ambig.png'),
tesseract_config=hocr_mode_config) !=
stbt.ocr(frame=load_image('ocr/ambig.png')))
def test_that_setting_config_options_has_an_effect():
# Unfortunately there are many tesseract config options and they are very
# complicated so it's difficult to write a test that tests that a config
# option is having the correct effect. Due to the difficulty in determining
# "correctness" instead here we test that setting a config option has an
# effect at all. This at least excercises our code which sets config
# options. I'm not happy about this and I hope to be able to replace this
# once we have more experience with these settings in the real world.
if _tesseract_version() >= LooseVersion('3.04'):
hocr_mode_config = {
"tessedit_create_txt": 0,
"tessedit_create_hocr": 1}
else:
hocr_mode_config = {
"tessedit_create_hocr": 1}
assert (stbt.ocr(frame=load_image('ocr/ambig.png'),
tesseract_config=hocr_mode_config) !=
stbt.ocr(frame=load_image('ocr/ambig.png')))
