# -*- coding: utf-8 -*- """Tests of Beautiful Soup as a whole.""" from pdb import set_trace import logging import unittest import sys import tempfile from bs4 import ( BeautifulSoup, BeautifulStoneSoup, ) from bs4.element import ( CharsetMetaAttributeValue, ContentMetaAttributeValue, SoupStrainer, NamespacedAttribute, ) import bs4.dammit from bs4.dammit import ( EntitySubstitution, UnicodeDammit, EncodingDetector, ) from bs4.testing import ( SoupTest, skipIf, ) import warnings try: from bs4.builder import LXMLTreeBuilder, LXMLTreeBuilderForXML LXML_PRESENT = True except ImportError, e: LXML_PRESENT = False PYTHON_3_PRE_3_2 = (sys.version_info[0] == 3 and sys.version_info < (3,2)) class TestConstructor(SoupTest): def test_short_unicode_input(self): data = u"

éé

" soup = self.soup(data) self.assertEqual(u"éé", soup.h1.string) def test_embedded_null(self): data = u"

foo\0bar

" soup = self.soup(data) self.assertEqual(u"foo\0bar", soup.h1.string) def test_exclude_encodings(self): utf8_data = u"Räksmörgås".encode("utf-8") soup = self.soup(utf8_data, exclude_encodings=["utf-8"]) self.assertEqual("windows-1252", soup.original_encoding) class TestWarnings(SoupTest): def _no_parser_specified(self, s, is_there=True): v = s.startswith(BeautifulSoup.NO_PARSER_SPECIFIED_WARNING[:80]) self.assertTrue(v) def test_warning_if_no_parser_specified(self): with warnings.catch_warnings(record=True) as w: soup = self.soup("") msg = str(w[0].message) self._assert_no_parser_specified(msg) def test_warning_if_parser_specified_too_vague(self): with warnings.catch_warnings(record=True) as w: soup = self.soup("", "html") msg = str(w[0].message) self._assert_no_parser_specified(msg) def test_no_warning_if_explicit_parser_specified(self): with warnings.catch_warnings(record=True) as w: soup = self.soup("", "html.parser") self.assertEqual([], w) def test_parseOnlyThese_renamed_to_parse_only(self): with warnings.catch_warnings(record=True) as w: soup = self.soup("", parseOnlyThese=SoupStrainer("b")) msg = str(w[0].message) self.assertTrue("parseOnlyThese" in msg) self.assertTrue("parse_only" in msg) self.assertEqual(b"", soup.encode()) def test_fromEncoding_renamed_to_from_encoding(self): with warnings.catch_warnings(record=True) as w: utf8 = b"\xc3\xa9" soup = self.soup(utf8, fromEncoding="utf8") msg = str(w[0].message) self.assertTrue("fromEncoding" in msg) self.assertTrue("from_encoding" in msg) self.assertEqual("utf8", soup.original_encoding) def test_unrecognized_keyword_argument(self): self.assertRaises( TypeError, self.soup, "", no_such_argument=True) class TestWarnings(SoupTest): def test_disk_file_warning(self): filehandle = tempfile.NamedTemporaryFile() filename = filehandle.name try: with warnings.catch_warnings(record=True) as w: soup = self.soup(filename) msg = str(w[0].message) self.assertTrue("looks like a filename" in msg) finally: filehandle.close() # The file no longer exists, so Beautiful Soup will no longer issue the warning. with warnings.catch_warnings(record=True) as w: soup = self.soup(filename) self.assertEqual(0, len(w)) def test_url_warning_with_bytes_url(self): with warnings.catch_warnings(record=True) as warning_list: soup = self.soup(b"http://www.crummybytes.com/") # Be aware this isn't the only warning that can be raised during # execution.. self.assertTrue(any("looks like a URL" in str(w.message) for w in warning_list)) def test_url_warning_with_unicode_url(self): with warnings.catch_warnings(record=True) as warning_list: # note - this url must differ from the bytes one otherwise # python's warnings system swallows the second warning soup = self.soup(u"http://www.crummyunicode.com/") self.assertTrue(any("looks like a URL" in str(w.message) for w in warning_list)) def test_url_warning_with_bytes_and_space(self): with warnings.catch_warnings(record=True) as warning_list: soup = self.soup(b"http://www.crummybytes.com/ is great") self.assertFalse(any("looks like a URL" in str(w.message) for w in warning_list)) def test_url_warning_with_unicode_and_space(self): with warnings.catch_warnings(record=True) as warning_list: soup = self.soup(u"http://www.crummyuncode.com/ is great") self.assertFalse(any("looks like a URL" in str(w.message) for w in warning_list)) class TestSelectiveParsing(SoupTest): def test_parse_with_soupstrainer(self): markup = "NoYesNoYes Yes" strainer = SoupStrainer("b") soup = self.soup(markup, parse_only=strainer) self.assertEqual(soup.encode(), b"YesYes Yes") class TestEntitySubstitution(unittest.TestCase): """Standalone tests of the EntitySubstitution class.""" def setUp(self): self.sub = EntitySubstitution def test_simple_html_substitution(self): # Unicode characters corresponding to named HTML entites # are substituted, and no others. s = u"foo\u2200\N{SNOWMAN}\u00f5bar" self.assertEqual(self.sub.substitute_html(s), u"foo∀\N{SNOWMAN}õbar") 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_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_false(self): s = 'Welcome to "my bar"' self.assertEqual(self.sub.substitute_xml(s, False), s) def test_xml_attribute_quoting_normally_uses_double_quotes(self): self.assertEqual(self.sub.substitute_xml("Welcome", True), '"Welcome"') self.assertEqual(self.sub.substitute_xml("Bob's Bar", True), '"Bob\'s Bar"') def test_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotes(self): s = 'Welcome to "my bar"' self.assertEqual(self.sub.substitute_xml(s, True), "'Welcome to \"my bar\"'") def test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotes(self): s = 'Welcome to "Bob\'s Bar"' self.assertEqual( self.sub.substitute_xml(s, True), '"Welcome to "Bob\'s Bar""') def test_xml_quotes_arent_escaped_when_value_is_not_being_quoted(self): quoted = 'Welcome to "Bob\'s Bar"' self.assertEqual(self.sub.substitute_xml(quoted), quoted) def test_xml_quoting_handles_angle_brackets(self): self.assertEqual( self.sub.substitute_xml("foo"), "foo<bar>") def test_xml_quoting_handles_ampersands(self): self.assertEqual(self.sub.substitute_xml("AT&T"), "AT&T") def test_xml_quoting_including_ampersands_when_they_are_part_of_an_entity(self): self.assertEqual( self.sub.substitute_xml("ÁT&T"), "&Aacute;T&T") def test_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entity(self): self.assertEqual( self.sub.substitute_xml_containing_entities("ÁT&T"), "ÁT&T") def test_quotes_not_html_substituted(self): """There's no need to do this except inside attribute values.""" text = 'Bob\'s "bar"' self.assertEqual(self.sub.substitute_html(text), text) class TestEncodingConversion(SoupTest): # Test Beautiful Soup's ability to decode and encode from various # encodings. def setUp(self): super(TestEncodingConversion, self).setUp() self.unicode_data = u'Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!' self.utf8_data = self.unicode_data.encode("utf-8") # Just so you know what it looks like. self.assertEqual( self.utf8_data, b'Sacr\xc3\xa9 bleu!') def test_ascii_in_unicode_out(self): # ASCII input is converted to Unicode. The original_encoding # attribute is set to 'utf-8', a superset of ASCII. chardet = bs4.dammit.chardet_dammit logging.disable(logging.WARNING) try: def noop(str): return None # Disable chardet, which will realize that the ASCII is ASCII. bs4.dammit.chardet_dammit = noop ascii = b"a" soup_from_ascii = self.soup(ascii) unicode_output = soup_from_ascii.decode() self.assertTrue(isinstance(unicode_output, unicode)) self.assertEqual(unicode_output, self.document_for(ascii.decode())) self.assertEqual(soup_from_ascii.original_encoding.lower(), "utf-8") finally: logging.disable(logging.NOTSET) bs4.dammit.chardet_dammit = chardet def test_unicode_in_unicode_out(self): # Unicode input is left alone. The original_encoding attribute # is not set. soup_from_unicode = self.soup(self.unicode_data) self.assertEqual(soup_from_unicode.decode(), self.unicode_data) self.assertEqual(soup_from_unicode.foo.string, u'Sacr\xe9 bleu!') self.assertEqual(soup_from_unicode.original_encoding, None) def test_utf8_in_unicode_out(self): # UTF-8 input is converted to Unicode. The original_encoding # attribute is set. soup_from_utf8 = self.soup(self.utf8_data) self.assertEqual(soup_from_utf8.decode(), self.unicode_data) self.assertEqual(soup_from_utf8.foo.string, u'Sacr\xe9 bleu!') def test_utf8_out(self): # The internal data structures can be encoded as UTF-8. soup_from_unicode = self.soup(self.unicode_data) self.assertEqual(soup_from_unicode.encode('utf-8'), self.utf8_data) @skipIf( PYTHON_3_PRE_3_2, "Bad HTMLParser detected; skipping test of non-ASCII characters in attribute name.") def test_attribute_name_containing_unicode_characters(self): markup = u'
' self.assertEqual(self.soup(markup).div.encode("utf8"), markup.encode("utf8")) class TestUnicodeDammit(unittest.TestCase): """Standalone tests of UnicodeDammit.""" def test_unicode_input(self): markup = u"I'm already Unicode! \N{SNOWMAN}" dammit = UnicodeDammit(markup) self.assertEqual(dammit.unicode_markup, markup) def test_smart_quotes_to_unicode(self): markup = b"\x91\x92\x93\x94" dammit = UnicodeDammit(markup) self.assertEqual( dammit.unicode_markup, u"\u2018\u2019\u201c\u201d") def test_smart_quotes_to_xml_entities(self): markup = b"\x91\x92\x93\x94" dammit = UnicodeDammit(markup, smart_quotes_to="xml") self.assertEqual( dammit.unicode_markup, "‘’“”") def test_smart_quotes_to_html_entities(self): markup = b"\x91\x92\x93\x94" dammit = UnicodeDammit(markup, smart_quotes_to="html") self.assertEqual( dammit.unicode_markup, "‘’“”") def test_smart_quotes_to_ascii(self): markup = b"\x91\x92\x93\x94" dammit = UnicodeDammit(markup, smart_quotes_to="ascii") self.assertEqual( dammit.unicode_markup, """''""""") 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}') 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_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_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_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_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_encoding_detector_replaces_junk_in_encoding_name_with_replacement_character(self): detected = EncodingDetector( b'') encodings = list(detected.encodings) assert u'utf-\N{REPLACEMENT CHARACTER}' in encodings def test_detect_html5_style_meta_tag(self): for data in ( b'', b"", b"", b""): dammit = UnicodeDammit(data, is_html=True) self.assertEqual( "euc-jp", dammit.original_encoding) 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 \330\250\330\252\330\261 \310\322\321\220\312\321\355\344""" chardet = bs4.dammit.chardet_dammit logging.disable(logging.WARNING) try: def noop(str): return None bs4.dammit.chardet_dammit = noop dammit = UnicodeDammit(doc) self.assertEqual(True, dammit.contains_replacement_characters) self.assertTrue(u"\ufffd" in dammit.unicode_markup) soup = BeautifulSoup(doc, "html.parser") self.assertTrue(soup.contains_replacement_characters) finally: logging.disable(logging.NOTSET) bs4.dammit.chardet_dammit = chardet 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"áé", dammit.unicode_markup) self.assertEqual("utf-16le", dammit.original_encoding) 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_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) class TestNamedspacedAttribute(SoupTest): def test_name_may_be_none(self): a = NamespacedAttribute("xmlns", None) self.assertEqual(a, "xmlns") def test_attribute_is_equivalent_to_colon_separated_string(self): a = NamespacedAttribute("a", "b") self.assertEqual("a:b", a) def test_attributes_are_equivalent_if_prefix_and_name_identical(self): a = NamespacedAttribute("a", "b", "c") b = NamespacedAttribute("a", "b", "c") self.assertEqual(a, b) # The actual namespace is not considered. c = NamespacedAttribute("a", "b", None) self.assertEqual(a, c) # But name and prefix are important. d = NamespacedAttribute("a", "z", "c") self.assertNotEqual(a, d) e = NamespacedAttribute("z", "b", "c") self.assertNotEqual(a, e) class TestAttributeValueWithCharsetSubstitution(unittest.TestCase): def test_content_meta_attribute_value(self): value = CharsetMetaAttributeValue("euc-jp") self.assertEqual("euc-jp", value) self.assertEqual("euc-jp", value.original_value) self.assertEqual("utf8", value.encode("utf8")) def test_content_meta_attribute_value(self): value = ContentMetaAttributeValue("text/html; charset=euc-jp") self.assertEqual("text/html; charset=euc-jp", value) self.assertEqual("text/html; charset=euc-jp", value.original_value) self.assertEqual("text/html; charset=utf8", value.encode("utf8"))