def test_exclude_encodings(self):
# This is UTF-8.
utf8_data = u"Räksmörgås".encode("utf-8")
# But if we exclude UTF-8 from consideration, the guess is
# Windows-1252.
dammit = UnicodeDammit(utf8_data, exclude_encodings=["utf-8"])
self.assertEqual(dammit.original_encoding.lower(), 'windows-1252')
# And if we exclude that, there is no valid guess at all.
dammit = UnicodeDammit(
utf8_data, exclude_encodings=["utf-8", "windows-1252"])
self.assertEqual(dammit.original_encoding, None)
def test_last_ditch_entity_replacement(self):
# This is a UTF-8 document that contains bytestrings
# completely incompatible with UTF-8 (ie. encoded with some other
# encoding).
#
# Since there is no consistent encoding for the document,
# Unicode, Dammit will eventually encode the document as UTF-8
# and encode the incompatible characters as REPLACEMENT
# CHARACTER.
#
# If chardet is installed, it will detect that the document
# can be converted into ISO-8859-1 without errors. This happens
# to be the wrong encoding, but it is a consistent encoding, so the
# code we're testing here won't run.
#
# So we temporarily disable chardet if it's present.
doc = b"""\357\273\277<?xml version="1.0" encoding="UTF-8"?>
<html><b>\330\250\330\252\330\261</b>
<i>\310\322\321\220\312\321\355\344</i></html>"""
chardet = thoughtfulsoup.dammit.chardet_dammit
logging.disable(logging.WARNING)
try:
def noop(str):
return None
thoughtfulsoup.dammit.chardet_dammit = noop
dammit = UnicodeDammit(doc)
self.assertEqual(True, dammit.contains_replacement_characters)
self.assertTrue(u"\ufffd" in dammit.unicode_markup)
soup = ThoughtfulSoup(doc, "html.parser")
self.assertTrue(soup.contains_replacement_characters)
finally:
logging.disable(logging.NOTSET)
thoughtfulsoup.dammit.chardet_dammit = chardet
def test_smart_quote_substitution(self):
# MS smart quotes are a common source of frustration, so we
# give them a special test.
quotes = b"\x91\x92foo\x93\x94"
dammit = UnicodeDammit(quotes)
self.assertEqual(self.sub.substitute_html(dammit.markup),
"‘’foo“”")
def test_detect_html5_style_meta_tag(self):
for data in (
b'<html><meta charset="euc-jp" /></html>',
b"<html><meta charset='euc-jp' /></html>",
b"<html><meta charset=euc-jp /></html>",
b"<html><meta charset=euc-jp/></html>"):
dammit = UnicodeDammit(data, is_html=True)
self.assertEqual(
"euc-jp", dammit.original_encoding)
def test_detwingle_ignores_multibyte_characters(self):
# Each of these characters has a UTF-8 representation ending
# in \x93. \x93 is a smart quote if interpreted as
# Windows-1252. But our code knows to skip over multibyte
# UTF-8 characters, so they'll survive the process unscathed.
for tricky_unicode_char in (
u"\N{LATIN SMALL LIGATURE OE}", # 2-byte char '\xc5\x93'
u"\N{LATIN SUBSCRIPT SMALL LETTER X}", # 3-byte char '\xe2\x82\x93'
u"\xf0\x90\x90\x93", # This is a CJK character, not sure which one.
):
input = tricky_unicode_char.encode("utf8")
self.assertTrue(input.endswith(b'\x93'))
output = UnicodeDammit.detwingle(input)
self.assertEqual(output, input)
def prepare_markup(self, markup, user_specified_encoding=None,
document_declared_encoding=None, exclude_encodings=None):
"""
:return: A 4-tuple (markup, original encoding, encoding
declared within markup, whether any characters had to be
replaced with REPLACEMENT CHARACTER).
"""
if isinstance(markup, unicode):
yield (markup, None, None, False)
return
try_encodings = [user_specified_encoding, document_declared_encoding]
dammit = UnicodeDammit(markup, try_encodings, is_html=True,
exclude_encodings=exclude_encodings)
yield (dammit.markup, dammit.original_encoding,
dammit.declared_html_encoding,
dammit.contains_replacement_characters)
def test_detwingle(self):
# Here's a UTF8 document.
utf8 = (u"\N{SNOWMAN}" * 3).encode("utf8")
# Here's a Windows-1252 document.
windows_1252 = (
u"\N{LEFT DOUBLE QUOTATION MARK}Hi, I like Windows!"
u"\N{RIGHT DOUBLE QUOTATION MARK}").encode("windows_1252")
# Through some unholy alchemy, they've been stuck together.
doc = utf8 + windows_1252 + utf8
# The document can't be turned into UTF-8:
self.assertRaises(UnicodeDecodeError, doc.decode, "utf8")
# Unicode, Dammit thinks the whole document is Windows-1252,
# and decodes it into "☃☃☃“Hi, I like Windows!”☃☃☃"
# But if we run it through fix_embedded_windows_1252, it's fixed:
fixed = UnicodeDammit.detwingle(doc)
self.assertEqual(
u"☃☃☃“Hi, I like Windows!”☃☃☃", fixed.decode("utf8"))
def test_ignore_inappropriate_codecs(self):
utf8_data = u"Räksmörgås".encode("utf-8")
dammit = UnicodeDammit(utf8_data, ["iso-8859-8"])
self.assertEqual(dammit.original_encoding.lower(), 'utf-8')
def test_smart_quotes_to_unicode(self):
markup = b"<foo>\x91\x92\x93\x94</foo>"
dammit = UnicodeDammit(markup)
self.assertEqual(
dammit.unicode_markup, u"<foo>\u2018\u2019\u201c\u201d</foo>")
def test_smart_quotes_to_xml_entities(self):
markup = b"<foo>\x91\x92\x93\x94</foo>"
dammit = UnicodeDammit(markup, smart_quotes_to="xml")
self.assertEqual(
dammit.unicode_markup, "<foo>‘’“”</foo>")
def test_byte_order_mark_removed(self):
# A document written in UTF-16LE will have its byte order marker stripped.
data = b'\xff\xfe<\x00a\x00>\x00\xe1\x00\xe9\x00<\x00/\x00a\x00>\x00'
dammit = UnicodeDammit(data)
self.assertEqual(u"<a>áé</a>", dammit.unicode_markup)
self.assertEqual("utf-16le", dammit.original_encoding)
def test_smart_quotes_to_ascii(self):
markup = b"<foo>\x91\x92\x93\x94</foo>"
dammit = UnicodeDammit(markup, smart_quotes_to="ascii")
self.assertEqual(
dammit.unicode_markup, """<foo>''""</foo>""")
def test_smart_quotes_to_html_entities(self):
markup = b"<foo>\x91\x92\x93\x94</foo>"
dammit = UnicodeDammit(markup, smart_quotes_to="html")
self.assertEqual(
dammit.unicode_markup, "<foo>‘’“”</foo>")
def test_ignore_invalid_codecs(self):
utf8_data = u"Räksmörgås".encode("utf-8")
for bad_encoding in ['.utf8', '...', 'utF---16.!']:
dammit = UnicodeDammit(utf8_data, [bad_encoding])
self.assertEqual(dammit.original_encoding.lower(), 'utf-8')
def test_unicode_input(self):
markup = u"I'm already Unicode! \N{SNOWMAN}"
dammit = UnicodeDammit(markup)
self.assertEqual(dammit.unicode_markup, markup)
def test_dont_see_smart_quotes_where_there_are_none(self):
utf_8 = b"\343\202\261\343\203\274\343\202\277\343\202\244 Watch"
dammit = UnicodeDammit(utf_8)
self.assertEqual(dammit.original_encoding.lower(), 'utf-8')
self.assertEqual(dammit.unicode_markup.encode("utf-8"), utf_8)
def test_convert_hebrew(self):
hebrew = b"\xed\xe5\xec\xf9"
dammit = UnicodeDammit(hebrew, ["iso-8859-8"])
self.assertEqual(dammit.original_encoding.lower(), 'iso-8859-8')
self.assertEqual(dammit.unicode_markup, u'\u05dd\u05d5\u05dc\u05e9')
def test_detect_utf8(self):
utf8 = b"Sacr\xc3\xa9 bleu! \xe2\x98\x83"
dammit = UnicodeDammit(utf8)
self.assertEqual(dammit.original_encoding.lower(), 'utf-8')
self.assertEqual(dammit.unicode_markup, u'Sacr\xe9 bleu! \N{SNOWMAN}')