def test_difference_in_ascii():
text1 = os.path.join(os.path.dirname(__file__), '../data/text_ascii1')
text2 = os.path.join(os.path.dirname(__file__), '../data/text_ascii2')
difference = compare_text_files(text1, text2)
assert difference is not None
expected_diff = open(os.path.join(os.path.dirname(__file__), '../data/text_ascii_expected_diff')).read()
assert difference.unified_diff == expected_diff
assert difference.comment is None
assert len(difference.details) == 0
def compare_unknown(path1, path2, source=None):
logger.debug("compare unknown path: %s and %s", path1, path2)
mime_type1 = guess_mime_type(path1)
mime_type2 = guess_mime_type(path2)
logger.debug("mime_type1: %s | mime_type2: %s", mime_type1, mime_type2)
if mime_type1.startswith('text/') and mime_type2.startswith('text/'):
encodings1 = re.findall(r'; charset=([^ ]+)', mime_type1)
encodings2 = re.findall(r'; charset=([^ ]+)', mime_type2)
if len(encodings1) > 0 and encodings1 == encodings2:
encoding = encodings1[0]
else:
encoding = None
return compare_text_files(path1, path2, encoding, source)
return compare_binary_files(path1, path2, source)
def compare_unknown(path1, path2, source=None):
logger.debug("compare unknown path: %s and %s", path1, path2)
mime_type1 = guess_mime_type(path1)
mime_type2 = guess_mime_type(path2)
logger.debug("mime_type1: %s | mime_type2: %s", mime_type1, mime_type2)
if mime_type1.startswith('text/') and mime_type2.startswith('text/'):
encodings1 = re.findall(r'; charset=([^ ]+)', mime_type1)
encodings2 = re.findall(r'; charset=([^ ]+)', mime_type2)
if len(encodings1) > 0 and encodings1 == encodings2:
encoding = encodings1[0]
else:
encoding = None
return compare_text_files(path1, path2, encoding, source)
return compare_binary_files(path1, path2, source)
def test_difference_between_iso88591_and_unicode():
text1 = os.path.join(os.path.dirname(__file__), '../data/text_unicode1')
text2 = os.path.join(os.path.dirname(__file__), '../data/text_iso8859')
difference = compare_text_files(text1, text2)
expected_diff = codecs.open(os.path.join(os.path.dirname(__file__), '../data/text_iso8859_expected_diff'), encoding='utf-8').read()
assert differences.unified_diff == expected_diff
def test_fallback_to_binary():
text1 = os.path.join(os.path.dirname(__file__), '../data/text_unicode1')
text2 = os.path.join(os.path.dirname(__file__), '../data/text_unicode2')
difference = compare_text_files(text1, text2, encoding='ascii')
expected_diff = open(os.path.join(os.path.dirname(__file__), '../data/text_unicode_binary_fallback')).read()
assert difference.unified_diff == expected_diff
def test_no_differences():
text1 = os.path.join(os.path.dirname(__file__), '../data/text_ascii1')
text2 = os.path.join(os.path.dirname(__file__), '../data/text_ascii1')
assert compare_text_files(text1, text2) is None