def test_strip_tags_flag(self):
content = '<div ascii_attribute="some more ascii">λ, λ, λ</div>'
set_strip_tags(True)
detection = detect(content)
self.assertEqual(get_strip_tags(), True)
self.assertEqual('UTF-8', detection['encoding'])
content2 = '<div ascii_attribute="some more ascii">λ, λ, λ</div>'
set_strip_tags(False)
detection2 = detect(content2)
self.assertEqual(get_strip_tags(), False)
self.assertEqual('UTF-8', detection2['encoding'])
def test_detection_works_as_expected(self):
for test in self.TEST_FILES:
file_path = self.TEST_FILES[test][0]
file_result = self.TEST_FILES[test][1]
content = open(file_path).read()
test_result = detect(content)
self.assertEqual(test_result, file_result)
def test_detection_works_as_expected_2(self):
for fixture in self.FIXTURE_FILES:
filename, encoding, type_ = fixture
file_path = 'fixtures/%s' % filename
content = open(file_path).read()
test_result = detect(content)
self.assertEqual(type_, test_result['type'])
check_encoding = (encoding == test_result.get('encoding', None))
self.assertTrue(check_encoding)
def detect_encoding(self):
"""
Try to guess the encoding
Returns: a hash, with :encoding, :confidence, :type
this will return nil if an error occurred during detection or
no valid encoding could be found
"""
if hasattr(self, '_detect_encoding'):
return self._detect_encoding
if self.data:
self._detect_encoding = pycharlockholmes.detect(self.data)
return self._detect_encoding
def get_detection(binary_file):
path = '%s/%s' % (binary_bin, binary_file)
detected = detect(open(path).read())
return detected
def test_detect_accepts_encoding_hint(self):
content = 'test'
detected = detect(content, 'UTF-8')
self.assertEqual('ISO-8859-1', detected['encoding'])
def test_detect_method(self):
content = 'test'
detected = detect(content)
self.assertEqual('ISO-8859-1', detected['encoding'])
