def test_first_second_separation_invalid(self):
# Tests the case where only 2 out of 3 digits are read by OCR. Should return 0.
captcha_solver = CaptchaSolver(None)
test_img = cv2.imread(os.path.join(os.getcwd(), 'test_ocr_invalid.png'))
assert captcha_solver.solve_captcha(test_img) == 0
assert captcha_solver.first_number is None
assert captcha_solver.second_number is None
def test_first_second_separation(self):
# Tests correct separation of first_number (first 2 digits) and second_number (last digit)
captcha_solver = CaptchaSolver(None)
test_img = cv2.imread(os.path.join(os.getcwd(), 'test_ocr_valid.png'))
assert captcha_solver.solve_captcha(test_img) == 15
assert captcha_solver.first_number == 12
assert captcha_solver.second_number == 3
def test_ocr(self):
# Do OCR on easy image to ensure it's working
captcha_solver = CaptchaSolver(None)
test_img1 = cv2.imread(os.path.join(os.getcwd(), 'test_ocr_valid.png'))
test_img2 = cv2.imread(os.path.join(os.getcwd(), 'test_ocr_invalid.png'))
captcha_text1 = captcha_solver.read_captcha(test_img1)
captcha_text2 = captcha_solver.read_captcha(test_img2)
assert captcha_text1 == '123'
assert captcha_text2 == '12'
def test_first_second_separation_invalid(self, tmpdir, testdatadir):
# Tests the case where only 2 out of 3 digits are read by OCR. Should return 0.
td = tmpdir.mkdir('captcha')
captcha_solver = CaptchaSolver(None, outdir=td)
test_img = cv2.imread(testdatadir.join('test_ocr_invalid.png').strpath)
assert captcha_solver.solve_captcha(test_img) == 0
assert captcha_solver.first_number is None
assert captcha_solver.second_number is None
def test_first_second_separation(self, tmpdir, testdatadir):
# Tests correct separation of first_number (first 2 digits) and second_number (last digit)
td = tmpdir.mkdir('captcha')
captcha_solver = CaptchaSolver(None, outdir=td)
test_img = cv2.imread(testdatadir.join('test_ocr_valid.png').strpath)
assert captcha_solver.solve_captcha(test_img) == 15
assert captcha_solver.first_number == 12
assert captcha_solver.second_number == 3
def test_ocr(self, tmpdir, testdatadir):
# Do OCR on easy image to ensure it's working
td = tmpdir.mkdir('captcha')
captcha_solver = CaptchaSolver(None, outdir=td)
test_img1 = cv2.imread(testdatadir.join('test_ocr_valid.png').strpath)
test_img2 = cv2.imread(
testdatadir.join('test_ocr_invalid.png').strpath)
captcha_text1 = captcha_solver.read_captcha(test_img1)
captcha_text2 = captcha_solver.read_captcha(test_img2)
assert captcha_text1 == '123'
assert captcha_text2 == '12'
def test_captcha_fail_save(self):
captcha_solver = CaptchaSolver(None)
test_img = cv2.imread(os.path.join(os.getcwd(), 'test_ocr_valid.png'))
captcha_solver.solve_captcha(test_img)
captcha_solver.notify_last_captcha_fail()
assert os.path.exists(os.path.join(os.getcwd(), 'captcha', 'incorrect', 'captcha1.png'))
captcha_solver.notify_last_captcha_fail()
assert os.path.exists(os.path.join(os.getcwd(), 'captcha', 'incorrect', 'captcha2.png'))
os.remove(os.path.join(os.getcwd(), 'captcha', 'incorrect', 'captcha1.png'))
os.remove(os.path.join(os.getcwd(), 'captcha', 'incorrect', 'captcha2.png'))
def test_captcha_fail_save(self, tmpdir, testdatadir):
td = tmpdir.mkdir('captcha')
captcha_solver = CaptchaSolver(None, outdir=td)
test_img = cv2.imread(testdatadir.join('test_ocr_valid.png').strpath)
captcha_solver.solve_captcha(test_img)
captcha_solver.notify_last_captcha_fail()
assert os.path.exists(td.join('incorrect', 'captcha1.png'))
captcha_solver.notify_last_captcha_fail()
assert os.path.exists(td.join('incorrect', 'captcha2.png'))
def test_folder_creation(self):
# Note, this only removes the captcha folder in the tests directory.
os.rmdir(os.path.join(os.getcwd(), 'captcha', 'incorrect'))
os.rmdir(os.path.join(os.getcwd(), 'captcha', 'correct'))
captcha_solver = CaptchaSolver(None)
assert os.path.exists(os.path.join(os.getcwd(), 'captcha', 'incorrect'))
assert os.path.exists(os.path.join(os.getcwd(), 'captcha', 'correct'))
# Cleanup folders
os.rmdir(os.path.join(os.getcwd(), 'captcha', 'incorrect'))
os.rmdir(os.path.join(os.getcwd(), 'captcha', 'correct'))
'collect-pii': False,
'connect-thresh': 10,
'output': 'bay-county-scraped.csv',
'save-attachments': 'none',
'solve-captchas': False,
'verbose': False
}
output_attachments = os.path.join(os.getcwd(), 'attachments')
output_file = os.path.join(os.getcwd(), settings['output'])
ffx_profile = webdriver.FirefoxOptions()
# Automatically dismiss unexpected alerts.
ffx_profile.set_capability('unexpectedAlertBehaviour', 'dismiss')
driver = webdriver.Firefox(options=ffx_profile)
captcha_solver = CaptchaSolver(driver)
def main():
# Parse Arguments
args = sys.argv[1:]
short_args = 'p:s:c:y:e:t:pc:o:a:uv'
long_args = ['portal-base=', 'state=', 'county', 'start-year=', 'end-year=', 'missing-thresh=', 'collect-pii',
'connect-thresh=', 'output=', 'save-attachments=','solve-captchas', 'verbose']
try:
args, vals = getopt.getopt(args, short_args, long_args)
for arg, val in args:
if arg in ('-p', '--portal-base'):
settings['portal-base'] = val
elif arg in ('-s', '--state'):
def test_folder_creation(self, tmpdir):
td = tmpdir.mkdir('captcha')
captcha_solver = CaptchaSolver(None, outdir=td)
assert os.path.exists(td.join('incorrect'))
assert os.path.exists(td.join('correct'))