diff options
| -rw-r--r-- | bs4/dammit.py | 66 | ||||
| -rw-r--r-- | bs4/element.py | 16 | ||||
| -rw-r--r-- | bs4/tests/test_lxml.py | 586 | ||||
| -rw-r--r-- | bs4/tests/test_soup.py | 130 | ||||
| -rw-r--r-- | bs4/tests/test_tree.py | 982 |
5 files changed, 1742 insertions, 38 deletions
diff --git a/bs4/dammit.py b/bs4/dammit.py index ed5dc29..8897063 100644 --- a/bs4/dammit.py +++ b/bs4/dammit.py @@ -224,7 +224,7 @@ class UnicodeDammit: # that might have them. if (self.smart_quotes_to is not None and proposed.lower() in self.ENCODINGS_WITH_SMART_QUOTES): - smart_quotes_re = "([\x80-\x9f])" + smart_quotes_re = b"([\x80-\x9f])" smart_quotes_compiled = re.compile(smart_quotes_re) markup = smart_quotes_compiled.sub(self._sub_ms_char, markup) @@ -378,35 +378,35 @@ class UnicodeDammit: ''.join(map(chr, list(range(256)))), ''.join(map(chr, emap))) return s.translate(c.EBCDIC_TO_ASCII_MAP) - MS_CHARS = {'\x80': ('euro', '20AC'), - '\x81': ' ', - '\x82': ('sbquo', '201A'), - '\x83': ('fnof', '192'), - '\x84': ('bdquo', '201E'), - '\x85': ('hellip', '2026'), - '\x86': ('dagger', '2020'), - '\x87': ('Dagger', '2021'), - '\x88': ('circ', '2C6'), - '\x89': ('permil', '2030'), - '\x8A': ('Scaron', '160'), - '\x8B': ('lsaquo', '2039'), - '\x8C': ('OElig', '152'), - '\x8D': '?', - '\x8E': ('#x17D', '17D'), - '\x8F': '?', - '\x90': '?', - '\x91': ('lsquo', '2018'), - '\x92': ('rsquo', '2019'), - '\x93': ('ldquo', '201C'), - '\x94': ('rdquo', '201D'), - '\x95': ('bull', '2022'), - '\x96': ('ndash', '2013'), - '\x97': ('mdash', '2014'), - '\x98': ('tilde', '2DC'), - '\x99': ('trade', '2122'), - '\x9a': ('scaron', '161'), - '\x9b': ('rsaquo', '203A'), - '\x9c': ('oelig', '153'), - '\x9d': '?', - '\x9e': ('#x17E', '17E'), - '\x9f': ('Yuml', ''),} + MS_CHARS = {b'\x80': ('euro', '20AC'), + b'\x81': ' ', + b'\x82': ('sbquo', '201A'), + b'\x83': ('fnof', '192'), + b'\x84': ('bdquo', '201E'), + b'\x85': ('hellip', '2026'), + b'\x86': ('dagger', '2020'), + b'\x87': ('Dagger', '2021'), + b'\x88': ('circ', '2C6'), + b'\x89': ('permil', '2030'), + b'\x8A': ('Scaron', '160'), + b'\x8B': ('lsaquo', '2039'), + b'\x8C': ('OElig', '152'), + b'\x8D': '?', + b'\x8E': ('#x17D', '17D'), + b'\x8F': '?', + b'\x90': '?', + b'\x91': ('lsquo', '2018'), + b'\x92': ('rsquo', '2019'), + b'\x93': ('ldquo', '201C'), + b'\x94': ('rdquo', '201D'), + b'\x95': ('bull', '2022'), + b'\x96': ('ndash', '2013'), + b'\x97': ('mdash', '2014'), + b'\x98': ('tilde', '2DC'), + b'\x99': ('trade', '2122'), + b'\x9a': ('scaron', '161'), + b'\x9b': ('rsaquo', '203A'), + b'\x9c': ('oelig', '153'), + b'\x9d': '?', + b'\x9e': ('#x17E', '17E'), + b'\x9f': ('Yuml', ''),} diff --git a/bs4/element.py b/bs4/element.py index 95661ae..5db5b36 100644 --- a/bs4/element.py +++ b/bs4/element.py @@ -1,8 +1,10 @@ import collections import re +import sys from bs4.dammit import EntitySubstitution DEFAULT_OUTPUT_ENCODING = "utf-8" +PY3K = (sys.version_info[0] > 2) def _match_css_class(str): @@ -523,7 +525,7 @@ class Tag(PageElement): self.extract() i = self while i is not None: - next = i.next + next = i.next_element i.__dict__.clear() i = next @@ -599,7 +601,8 @@ class Tag(PageElement): #print "Getattr %s.%s" % (self.__class__, tag) if len(tag) > 3 and tag.endswith('Tag'): return self.find(tag[:-3]) - elif not tag.startswith("__"): + # We special case contents to avoid recursion. + elif not tag.startswith("__") and not tag=="contents": return self.find(tag) raise AttributeError( "'%s' object has no attribute '%s'" % (self.__class__, tag)) @@ -635,6 +638,9 @@ class Tag(PageElement): def __str__(self): return self.encode() + + if PY3K: + __str__ = __repr__ = __unicode__ def encode(self, encoding=DEFAULT_OUTPUT_ENCODING, indent_level=None, substitute_html_entities=False): @@ -872,7 +878,7 @@ class SoupStrainer(object): found = None # If given a list of items, scan it for a text element that # matches. - if hasattr(markup, '__iter__') and not isinstance(markup, Tag): + if hasattr(markup, '__iter__') and not isinstance(markup, (Tag, basestring)): for element in markup: if isinstance(element, NavigableString) \ and self.search(element): @@ -912,8 +918,8 @@ class SoupStrainer(object): # It's a regexp object. result = markup and match_against.search(markup) elif (hasattr(match_against, '__iter__') - and (markup is not None - or not isinstance(match_against, basestring))): + and markup is not None + and not isinstance(match_against, basestring)): result = markup in match_against elif hasattr(match_against, 'items'): result = match_against in markup diff --git a/bs4/tests/test_lxml.py b/bs4/tests/test_lxml.py new file mode 100644 index 0000000..2f64d31 --- /dev/null +++ b/bs4/tests/test_lxml.py @@ -0,0 +1,586 @@ +"""Tests to ensure that the lxml tree builder generates good trees.""" + +import re + +from bs4 import BeautifulSoup +from bs4.builder import LXMLTreeBuilder, LXMLTreeBuilderForXML +from bs4.element import Comment, Doctype, SoupStrainer +from bs4.testing import SoupTest + + +class TestLXMLBuilder(SoupTest): + """A smoke test for the LXML tree builder. + + Subclass this to test some other HTML tree builder. Subclasses of + this test ensure that all of Beautiful Soup's tree builders + generate more or less the same trees. + + It's okay for trees to differ--just override the appropriate test + method to demonstrate how one tree builder differs from the LXML + builder. But in general, all HTML tree builders should generate + trees that make most of these tests pass. + """ + + def test_bare_string(self): + # A bare string is turned into some kind of HTML document or + # fragment recognizable as the original string. + # + # In this case, lxml puts a <p> tag around the bare string. + self.assertSoupEquals( + "A bare string", "<p>A bare string</p>") + + def test_mixed_case_tags(self): + # Mixed-case tags are folded to lowercase. + self.assertSoupEquals( + "<a><B><Cd><EFG></efg></CD></b></A>", + "<a><b><cd><efg></efg></cd></b></a>") + + def test_empty_element(self): + # HTML's empty-element tags are recognized as such. + self.assertSoupEquals( + "<p>A <meta> tag</p>", "<p>A <meta /> tag</p>") + + self.assertSoupEquals( + "<p>Foo<br/>bar</p>", "<p>Foo<br />bar</p>") + + def test_empty_tag_thats_not_an_empty_element_tag(self): + # A tag that is empty but not an HTML empty-element tag + # is not presented as an empty-element tag. + self.assertSoupEquals("<p>", "<p></p>") + + def test_comment(self): + # Comments are represented as Comment objects. + markup = "<p>foo<!--foobar-->baz</p>" + self.assertSoupEquals(markup) + + soup = self.soup(markup) + comment = soup.find(text="foobar") + self.assertEquals(comment.__class__, Comment) + + def test_nested_inline_elements(self): + # Inline tags can be nested indefinitely. + b_tag = "<b>Inside a B tag</b>" + self.assertSoupEquals(b_tag) + + nested_b_tag = "<p>A <i>nested <b>tag</b></i></p>" + self.assertSoupEquals(nested_b_tag) + + double_nested_b_tag = "<p>A <a>doubly <i>nested <b>tag</b></i></a></p>" + self.assertSoupEquals(nested_b_tag) + + def test_nested_block_level_elements(self): + soup = self.soup('<blockquote><p><b>Foo</b></p></blockquote>') + blockquote = soup.blockquote + self.assertEqual(blockquote.p.b.string, 'Foo') + self.assertEqual(blockquote.b.string, 'Foo') + + # This is a <table> tag containing another <table> tag in one of its + # cells. + TABLE_MARKUP_1 = ('<table id="1">' + '<tr>' + "<td>Here's another table:" + '<table id="2">' + '<tr><td>foo</td></tr>' + '</table></td>') + + def test_correctly_nested_tables(self): + markup = ('<table id="1">' + '<tr>' + "<td>Here's another table:" + '<table id="2">' + '<tr><td>foo</td></tr>' + '</table></td>') + + self.assertSoupEquals( + markup, + '<table id="1"><tr><td>Here\'s another table:' + '<table id="2"><tr><td>foo</td></tr></table>' + '</td></tr></table>') + + self.assertSoupEquals( + "<table><thead><tr><td>Foo</td></tr></thead>" + "<tbody><tr><td>Bar</td></tr></tbody>" + "<tfoot><tr><td>Baz</td></tr></tfoot></table>") + + def test_collapsed_whitespace(self): + """In most tags, whitespace is collapsed.""" + self.assertSoupEquals("<p> </p>", "<p> </p>") + + def test_preserved_whitespace_in_pre_and_textarea(self): + """In <pre> and <textarea> tags, whitespace is preserved.""" + self.assertSoupEquals("<pre> </pre>") + self.assertSoupEquals("<textarea> woo </textarea>") + + def test_single_quote_attribute_values_become_double_quotes(self): + self.assertSoupEquals("<foo attr='bar'></foo>", + '<foo attr="bar"></foo>') + + def test_attribute_values_with_nested_quotes_are_left_alone(self): + text = """<foo attr='bar "brawls" happen'>a</foo>""" + self.assertSoupEquals(text) + + def test_attribute_values_with_double_nested_quotes_get_quoted(self): + text = """<foo attr='bar "brawls" happen'>a</foo>""" + soup = self.soup(text) + soup.foo['attr'] = 'Brawls happen at "Bob\'s Bar"' + self.assertSoupEquals( + soup.foo.decode(), + """<foo attr="Brawls happen at "Bob\'s Bar"">a</foo>""") + + def test_ampersand_in_attribute_value_gets_quoted(self): + self.assertSoupEquals('<this is="really messed up & stuff"></this>', + '<this is="really messed up & stuff"></this>') + + def test_literal_in_textarea(self): + # Anything inside a <textarea> is supposed to be treated as + # the literal value of the field, (XXX citation + # needed). html5lib does this correctly. But, lxml does its + # best to parse the contents of a <textarea> as HTML. + text = '<textarea>Junk like <b> tags and <&<&</textarea>' + soup = self.soup(text) + self.assertEquals(len(soup.textarea.contents), 2) + self.assertEquals(soup.textarea.contents[0], u"Junk like ") + self.assertEquals(soup.textarea.contents[1].name, 'b') + self.assertEquals(soup.textarea.b.string, u" tags and ") + + def test_literal_in_script(self): + # The contents of a <script> tag are treated as a literal string, + # even if that string contains HTML. + javascript = 'if (i < 2) { alert("<b>foo</b>"); }' + soup = self.soup('<script>%s</script>' % javascript) + self.assertEquals(soup.script.string, javascript) + + def test_naked_ampersands(self): + # Ampersands are left alone. + text = "<p>AT&T</p>" + soup = self.soup(text) + self.assertEquals(soup.p.string, "AT&T") + + # Even if they're in attribute values. + invalid_url = '<a href="http://example.org?a=1&b=2;3">foo</a>' + soup = self.soup(invalid_url) + self.assertEquals(soup.a['href'], "http://example.org?a=1&b=2;3") + + def test_entities_in_strings_converted_during_parsing(self): + # Both XML and HTML entities are converted to Unicode characters + # during parsing. + text = "<p><<sacré bleu!>></p>" + expected = u"<p><<sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!>></p>" + self.assertSoupEquals(text, expected) + + def test_smart_quotes_converted_on_the_way_in(self): + # Microsoft smart quotes are converted to Unicode characters during + # parsing. + quote = b"<p>\x91Foo\x92</p>" + soup = self.soup(quote) + self.assertEquals( + soup.p.string, + u"\N{LEFT SINGLE QUOTATION MARK}Foo\N{RIGHT SINGLE QUOTATION MARK}") + + def test_non_breaking_spaces_converted_on_the_way_in(self): + soup = self.soup("<a> </a>") + self.assertEquals(soup.a.string, u"\N{NO-BREAK SPACE}" * 2) + + def test_cdata_where_its_ok(self): + # lxml strips CDATA sections, no matter where they occur. + markup = "<svg><![CDATA[foobar]]>" + self.assertSoupEquals(markup, "<svg></svg>") + + def _test_doctype(self, doctype_fragment): + """Run a battery of assertions on a given doctype string.""" + doctype_str = '<!DOCTYPE %s>' % doctype_fragment + markup = doctype_str + '<p>foo</p>' + soup = self.soup(markup) + doctype = soup.contents[0] + self.assertEquals(doctype.__class__, Doctype) + self.assertEquals(doctype, doctype_fragment) + self.assertEquals(str(soup)[:len(doctype_str)], doctype_str) + + # Make sure that the doctype was correctly associated with the + # parse tree and that the rest of the document parsed. + self.assertEquals(soup.p.contents[0], 'foo') + + def test_doctype(self): + # Test a normal HTML doctype you'll commonly see in a real document. + self._test_doctype( + 'html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"') + + def test_namespaced_system_doctype(self): + # Test a namespaced doctype with a system id. + self._test_doctype('xsl:stylesheet SYSTEM "htmlent.dtd"') + + def test_namespaced_public_doctype(self): + # Test a namespaced doctype with a public id. + self._test_doctype('xsl:stylesheet PUBLIC "htmlent.dtd"') + + def test_real_iso_latin_document(self): + # Smoke test of interrelated functionality, using an + # easy-to-understand document. + + # Here it is in Unicode. Note that it claims to be in ISO-Latin-1. + unicode_html = u'<html><head><meta content="text/html; charset=ISO-Latin-1" http-equiv="Content-type" /></head><body><p>Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!</p></body></html>' + + # That's because we're going to encode it into ISO-Latin-1, and use + # that to test. + iso_latin_html = unicode_html.encode("iso-8859-1") + + # Parse the ISO-Latin-1 HTML. + soup = self.soup(iso_latin_html) + # Encode it to UTF-8. + result = soup.encode("utf-8") + + # What do we expect the result to look like? Well, it would + # look like unicode_html, except that the META tag would say + # UTF-8 instead of ISO-Latin-1. + expected = unicode_html.replace("ISO-Latin-1", "utf-8") + + # And, of course, it would be in UTF-8, not Unicode. + expected = expected.encode("utf-8") + + # Ta-da! + self.assertEquals(result, expected) + + def test_real_shift_jis_document(self): + # Smoke test to make sure the parser can handle a document in + # Shift-JIS encoding, without choking. + shift_jis_html = ( + b'<html><head></head><body><pre>' + b'\x82\xb1\x82\xea\x82\xcdShift-JIS\x82\xc5\x83R\x81[\x83f' + b'\x83B\x83\x93\x83O\x82\xb3\x82\xea\x82\xbd\x93\xfa\x96{\x8c' + b'\xea\x82\xcc\x83t\x83@\x83C\x83\x8b\x82\xc5\x82\xb7\x81B' + b'</pre></body></html>') + unicode_html = shift_jis_html.decode("shift-jis") + soup = self.soup(shift_jis_html) + + # Make sure the parse tree is correctly encoded to various + # encodings. + self.assertEquals(soup.encode("utf-8"), unicode_html.encode("utf-8")) + self.assertEquals(soup.encode("euc_jp"), unicode_html.encode("euc_jp")) + + # Tests below this line need work. + + def test_meta_tag_reflects_current_encoding(self): + # Here's the <meta> tag saying that a document is + # encoded in Shift-JIS. + meta_tag = ('<meta content="text/html; charset=x-sjis" ' + 'http-equiv="Content-type" />') + + # Here's a document incorporating that meta tag. + shift_jis_html = ( + '<html><head>\n%s\n' + '<meta http-equiv="Content-language" content="ja" />' + '</head><body>Shift-JIS markup goes here.') % meta_tag + soup = self.soup(shift_jis_html) + + # Parse the document, and the charset is replaced with a + # generic value. + parsed_meta = soup.find('meta', {'http-equiv': 'Content-type'}) + self.assertEquals(parsed_meta['content'], + 'text/html; charset=%SOUP-ENCODING%') + self.assertEquals(parsed_meta.contains_substitutions, True) + + # For the rest of the story, see TestSubstitutions in + # test_tree.py. + + def test_entities_converted_on_the_way_out(self): + text = "<p><<sacré bleu!>></p>" + expected = u"<<sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!>>".encode("utf-8") + soup = self.soup(text) + str = soup.p.string + #self.assertEquals(str.encode("utf-8"), expected) + + def test_br_tag_is_empty_element(self): + """A <br> tag is designated as an empty-element tag.""" + soup = self.soup("<br></br>") + self.assertTrue(soup.br.is_empty_element) + self.assertEquals(str(soup.br), "<br />") + + def test_p_tag_is_not_empty_element(self): + """A <p> tag is not designated as an empty-element tag.""" + soup = self.soup("<p />") + self.assertFalse(soup.p.is_empty_element) + self.assertEquals(str(soup.p), "<p></p>") + + def test_soupstrainer(self): + strainer = SoupStrainer("b") + soup = self.soup("A <b>bold</b> <meta /> <i>statement</i>", + parse_only=strainer) + self.assertEquals(soup.decode(), "<b>bold</b>") + + +class TestLXMLBuilderInvalidMarkup(SoupTest): + """Tests of invalid markup for the LXML tree builder. + + Subclass this to test other builders. + + These are very likely to give different results for different tree + builders. It's not required that a tree builder handle invalid + markup at all. + """ + + def test_table_containing_bare_markup(self): + # Markup should be in table cells, not directly in the table. + self.assertSoupEquals("<table><div>Foo</div></table>") + + def test_incorrectly_nested_table(self): + # The second <table> tag is floating in the <tr> tag + # rather than being inside a <td>. + bad_markup = ('<table id="1">' + '<tr>' + "<td>Here's another table:</td>" + '<table id="2">' + '<tr><td>foo</td></tr>' + '</table></td>') + + def test_unclosed_block_level_elements(self): + # Unclosed block-level elements should be closed. + self.assertSoupEquals( + '<blockquote><p><b>Foo</blockquote><p>Bar', + '<blockquote><p><b>Foo</b></p></blockquote><p>Bar</p>') + + def test_fake_self_closing_tag(self): + # If a self-closing tag presents as a normal tag, the 'open' + # tag is treated as an instance of the self-closing tag and + # the 'close' tag is ignored. + self.assertSoupEquals( + "<item><link>http://foo.com/</link></item>", + "<item><link />http://foo.com/</item>") + + def test_boolean_attribute_with_no_value_gets_empty_value(self): + soup = self.soup("<table><td nowrap>foo</td></table>") + self.assertEquals(soup.table.td['nowrap'], '') + + def test_incorrectly_nested_tables(self): + self.assertSoupEquals( + '<table><tr><table><tr id="nested">', + '<table><tr><table><tr id="nested"></tr></table></tr></table>') + + def test_paragraphs_containing_block_display_elements(self): + markup = self.soup("<p>this is the definition:" + "<dl><dt>first case</dt>") + # The <p> tag is closed before the <dl> tag begins. + self.assertEquals(markup.p.contents, ["this is the definition:"]) + + def test_empty_element_tag_with_contents(self): + self.assertSoupEquals("<br>foo</br>", "<br />foo") + + def test_doctype_in_body(self): + markup = "<p>one<!DOCTYPE foobar>two</p>" + self.assertSoupEquals(markup) + + def test_nonsensical_declaration(self): + # Declarations that don't make any sense are ignored. + self.assertSoupEquals('<! Foo = -8><p>a</p>', "<p>a</p>") + + def test_whitespace_in_doctype(self): + # A declaration that has extra whitespace is ignored. + self.assertSoupEquals( + ('<! DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN">' + '<p>foo</p>'), + '<p>foo</p>') + + def test_incomplete_declaration(self): + # An incomplete declaration will screw up the rest of the document. + self.assertSoupEquals('a<!b <p>c', '<p>a</p>') + + def test_cdata_where_it_doesnt_belong(self): + #CDATA sections are ignored. + markup = "<div><![CDATA[foo]]>" + self.assertSoupEquals(markup, "<div></div>") + + def test_attribute_value_never_got_closed(self): + markup = '<a href="http://foo.com/</a> and blah and blah' + soup = self.soup(markup) + self.assertEquals( + soup.a['href'], "http://foo.com/</a> and blah and blah") + + def test_attribute_value_was_closed_by_subsequent_tag(self): + markup = """<a href="foo</a>, </a><a href="bar">baz</a>""" + soup = self.soup(markup) + # The string between the first and second quotes was interpreted + # as the value of the 'href' attribute. + self.assertEquals(soup.a['href'], 'foo</a>, </a><a href=') + + #The string after the second quote (bar"), was treated as an + #empty attribute called bar. + self.assertEquals(soup.a['bar'], '') + self.assertEquals(soup.a.string, "baz") + + def test_unquoted_attribute_value(self): + soup = self.soup('<a style={height:21px;}></a>') + self.assertEquals(soup.a['style'], '{height:21px;}') + + def test_attribute_value_with_embedded_brackets(self): + soup = self.soup('<a b="<a>">') + self.assertEquals(soup.a['b'], '<a>') + + def test_nonexistent_entity(self): + soup = self.soup("<p>foo&#bar;baz</p>") + self.assertEquals(soup.p.string, "foobar;baz") + + # Compare a real entity. + soup = self.soup("<p>foodbaz</p>") + self.assertEquals(soup.p.string, "foodbaz") + + # Also compare html5lib, which preserves the &# before the + # entity name. + + def test_entity_out_of_range(self): + # An entity that's out of range will be ignored. + soup = self.soup("<p>�</p>") + self.assertEquals(soup.p.string, None) + + soup = self.soup("<p>�</p>") + self.assertEquals(soup.p.string, None) + + + def test_entity_was_not_finished(self): + soup = self.soup("<p><Hello>") + # Compare html5lib, which completes the entity. + self.assertEquals(soup.p.string, "<Hello>") + + def test_document_ends_with_incomplete_declaration(self): + soup = self.soup('<p>a<!b') + # This becomes a string 'a'. The incomplete declaration is ignored. + # Compare html5lib, which turns it into a comment. + self.assertEquals(soup.p.contents, ['a']) + + def test_document_starts_with_bogus_declaration(self): + soup = self.soup('<! Foo ><p>a</p>') + # The declaration is ignored altogether. + self.assertEquals(soup.encode(), b"<html><body><p>a</p></body></html>") + + def test_tag_name_contains_unicode(self): + # Unicode characters in tag names are stripped. + tag_name = u"<our\N{SNOWMAN}>Joe</our\N{SNOWMAN}>" + self.assertSoupEquals("<our>Joe</our>") + +class TestLXMLBuilderEncodingConversion(SoupTest): + # Test Beautiful Soup's ability to decode and encode from various + # encodings. + + def setUp(self): + super(TestLXMLBuilderEncodingConversion, self).setUp() + self.unicode_data = u"<html><head></head><body><foo>Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!</foo></body></html>" + self.utf8_data = self.unicode_data.encode("utf-8") + # Just so you know what it looks like. + self.assertEqual( + self.utf8_data, + b"<html><head></head><body><foo>Sacr\xc3\xa9 bleu!</foo></body></html>") + + def test_ascii_in_unicode_out(self): + # ASCII input is converted to Unicode. The original_encoding + # attribute is set. + ascii = b"<foo>a</foo>" + soup_from_ascii = self.soup(ascii) + unicode_output = soup_from_ascii.decode() + self.assertTrue(isinstance(unicode_output, unicode)) + self.assertEquals(unicode_output, self.document_for(ascii.decode())) + self.assertEquals(soup_from_ascii.original_encoding, "ascii") + + 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.assertEquals(soup_from_unicode.decode(), self.unicode_data) + self.assertEquals(soup_from_unicode.foo.string, u'Sacr\xe9 bleu!') + self.assertEquals(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.assertEquals(soup_from_utf8.decode(), self.unicode_data) + self.assertEquals(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.assertEquals(soup_from_unicode.encode('utf-8'), self.utf8_data) + + HEBREW_DOCUMENT = b'<html><head><title>Hebrew (ISO 8859-8) in Visual Directionality</title></head><body><h1>Hebrew (ISO 8859-8) in Visual Directionality</h1>\xed\xe5\xec\xf9</body></html>' + + def test_real_hebrew_document(self): + # A real-world test to make sure we can convert ISO-8859-9 (a + # Hebrew encoding) to UTF-8. + soup = self.soup(self.HEBREW_DOCUMENT, + from_encoding="iso-8859-8") + self.assertEquals(soup.original_encoding, 'iso-8859-8') + self.assertEquals( + soup.encode('utf-8'), + self.HEBREW_DOCUMENT.decode("iso-8859-8").encode("utf-8")) + + +class TestLXMLXMLBuilder(SoupTest): + """Test XML-specific parsing behavior. + + Most of the tests use HTML as an example, since Beautiful Soup is + mainly an HTML parser. This test suite is a base for XML-specific + tree builders. + """ + + @property + def default_builder(self): + return LXMLTreeBuilderForXML() + + def test_mixed_case_tags(self): + # Mixed-case tags are *not* folded to lowercase, but the + # end tag is always the same case as the start tag. + self.assertSoupEquals( + "<a><B><Cd><EFG /></CD></b></A>", + "<a><B><Cd><EFG /></Cd></B></a>") + + + def test_cdata_becomes_text(self): + # LXML sends CData sections as 'data' events, so we can't + # create special CData objects for them. We have to use + # NavigableString. I would like to fix this, but it's not a + # very high priority. + markup = "<foo><![CDATA[iamcdata]]></foo>" + soup = self.soup(markup) + cdata = soup.foo.contents[0] + self.assertEquals(cdata.__class__.__name__, 'NavigableString') + + + def test_can_handle_invalid_xml(self): + self.assertSoupEquals("<a><b>", "<a><b /></a>") + + def test_empty_element_tag(self): + soup = self.soup("<p><iamselfclosing /></p>") + self.assertTrue(soup.iamselfclosing.is_empty_element) + + def test_self_empty_tag_treated_as_empty_element(self): + soup = self.soup("<p><iamclosed></iamclosed></p>") + self.assertTrue(soup.iamclosed.is_empty_element) + + def test_self_nonempty_tag_is_not_empty_element(self): + soup = self.soup("<p><ihavecontents>contents</ihavecontents></p>") + self.assertFalse(soup.ihavecontents.is_empty_element) + + def test_empty_tag_that_stops_being_empty_gets_a_closing_tag(self): + soup = self.soup("<bar />") + self.assertTrue(soup.bar.is_empty_element) + soup.bar.insert(1, "Contents") + self.assertFalse(soup.bar.is_empty_element) + self.assertEquals(str(soup), self.document_for("<bar>Contents</bar>")) + + def test_designated_empty_element_tag_has_no_closing_tag(self): + builder = LXMLTreeBuilderForXML(empty_element_tags=['bar']) + soup = BeautifulSoup(builder=builder, markup="<bar></bar>") + self.assertTrue(soup.bar.is_empty_element) + self.assertEquals(str(soup), self.document_for("<bar />")) + + def test_empty_tag_not_in_empty_element_tag_list_has_closing_tag(self): + builder = LXMLTreeBuilderForXML(empty_element_tags=['bar']) + + soup = BeautifulSoup(builder=builder, markup="<foo />") + self.assertFalse(soup.foo.is_empty_element) + self.assertEquals(str(soup), self.document_for("<foo></foo>")) + + def test_designated_empty_element_tag_does_not_change_parser_behavior(self): + # The designated list of empty-element tags only affects how + # empty tags are presented. It does not affect how tags are + # parsed--that's the parser's job. + builder = LXMLTreeBuilderForXML(empty_element_tags=['bar']) + soup = BeautifulSoup(builder=builder, markup="<bar>contents</bar>") + self.assertEquals(str(soup), self.document_for("<bar>contents</bar>")) diff --git a/bs4/tests/test_soup.py b/bs4/tests/test_soup.py new file mode 100644 index 0000000..3c5d742 --- /dev/null +++ b/bs4/tests/test_soup.py @@ -0,0 +1,130 @@ +# -*- coding: utf-8 -*- +"""Tests of Beautiful Soup as a whole.""" + +import unittest +from bs4.element import SoupStrainer +from bs4.dammit import EntitySubstitution, UnicodeDammit +from bs4.testing import SoupTest + + +class TestSelectiveParsing(SoupTest): + + def test_parse_with_soupstrainer(self): + markup = "No<b>Yes</b><a>No<b>Yes <c>Yes</c></b>" + strainer = SoupStrainer("b") + soup = self.soup(markup, parse_only=strainer) + self.assertEquals(soup.encode(), b"<b>Yes</b><b>Yes <c>Yes</c></b>") + + +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.assertEquals(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.assertEquals(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.assertEquals(self.sub.substitute_xml(s, False), s) + + def test_xml_attribute_quoting_normally_uses_double_quotes(self): + self.assertEquals(self.sub.substitute_xml("Welcome", True), + '"Welcome"') + self.assertEquals(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.assertEquals(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.assertEquals( + 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.assertEquals(self.sub.substitute_xml(quoted), quoted) + + def test_xml_quoting_handles_angle_brackets(self): + self.assertEquals( + self.sub.substitute_xml("foo<bar>"), + "foo<bar>") + + def test_xml_quoting_handles_ampersands(self): + self.assertEquals(self.sub.substitute_xml("AT&T"), "AT&T") + + def test_xml_quoting_ignores_ampersands_when_they_are_part_of_an_entity(self): + self.assertEquals( + self.sub.substitute_xml("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.assertEquals(self.sub.substitute_html(text), text) + +class TestUnicodeDammit(unittest.TestCase): + """Standalone tests of Unicode, Dammit.""" + + def test_smart_quotes_to_unicode(self): + markup = b"<foo>\x91\x92\x93\x94</foo>" + dammit = UnicodeDammit(markup) + self.assertEquals( + 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.assertEquals( + 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.assertEquals( + dammit.unicode_markup, "<foo>‘’“”</foo>") + + def test_detect_utf8(self): + utf8 = b"\xc3\xa9" + dammit = UnicodeDammit(utf8) + self.assertEquals(dammit.unicode_markup, u'\xe9') + self.assertEquals(dammit.original_encoding, 'utf-8') + + def test_convert_hebrew(self): + hebrew = b"\xed\xe5\xec\xf9" + dammit = UnicodeDammit(hebrew, ["iso-8859-8"]) + self.assertEquals(dammit.original_encoding, 'iso-8859-8') + self.assertEquals(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.assertEquals(dammit.original_encoding, 'utf-8') + self.assertEquals(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.assertEquals(dammit.original_encoding, '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.assertEquals(dammit.original_encoding, 'utf-8') diff --git a/bs4/tests/test_tree.py b/bs4/tests/test_tree.py new file mode 100644 index 0000000..e11b1b9 --- /dev/null +++ b/bs4/tests/test_tree.py @@ -0,0 +1,982 @@ +# -*- coding: utf-8 -*- +"""Tests for Beautiful Soup's tree traversal methods. + +The tree traversal methods are the main advantage of using Beautiful +Soup over other parsers. + +Different parsers will build different Beautiful Soup trees given the +same markup, but all Beautiful Soup trees can be traversed with the +methods tested here. +""" + +import copy +import pickle +import re +from bs4 import BeautifulSoup +from bs4.builder import builder_registry +from bs4.element import CData, SoupStrainer, Tag +from bs4.testing import SoupTest + +class TreeTest(SoupTest): + + def assertSelects(self, tags, should_match): + """Make sure that the given tags have the correct text. + + This is used in tests that define a bunch of tags, each + containing a single string, and then select certain strings by + some mechanism. + """ + self.assertEqual([tag.string for tag in tags], should_match) + + def assertSelectsIDs(self, tags, should_match): + """Make sure that the given tags have the correct IDs. + + This is used in tests that define a bunch of tags, each + containing a single string, and then select certain strings by + some mechanism. + """ + self.assertEqual([tag['id'] for tag in tags], should_match) + + +class TestFind(TreeTest): + """Basic tests of the find() method. + + find() just calls find_all() with limit=1, so it's not tested all + that thouroughly here. + """ + + def test_find_tag(self): + soup = self.soup("<a>1</a><b>2</b><a>3</a><b>4</b>") + self.assertEqual(soup.find("b").string, "2") + + def test_unicode_text_find(self): + soup = self.soup(u'<h1>Räksmörgås</h1>') + self.assertEqual(soup.find(text=u'Räksmörgås'), u'Räksmörgås') + + +class TestFindAll(TreeTest): + """Basic tests of the find_all() method.""" + + def test_find_all_text_nodes(self): + """You can search the tree for text nodes.""" + soup = self.soup("<html>Foo<b>bar</b>\xbb</html>") + # Exact match. + self.assertEqual(soup.find_all(text="bar"), [u"bar"]) + # Match any of a number of strings. + self.assertEqual( + soup.find_all(text=["Foo", "bar"]), [u"Foo", u"bar"]) + # Match a regular expression. + self.assertEqual(soup.find_all(text=re.compile('.*')), + [u"Foo", u"bar", u'\xbb']) + # Match anything. + self.assertEqual(soup.find_all(text=True), + [u"Foo", u"bar", u'\xbb']) + + def test_find_all_limit(self): + """You can limit the number of items returned by find_all.""" + soup = self.soup("<a>1</a><a>2</a><a>3</a><a>4</a><a>5</a>") + self.assertSelects(soup.find_all('a', limit=3), ["1", "2", "3"]) + self.assertSelects(soup.find_all('a', limit=1), ["1"]) + self.assertSelects( + soup.find_all('a', limit=10), ["1", "2", "3", "4", "5"]) + + # A limit of 0 means no limit. + self.assertSelects( + soup.find_all('a', limit=0), ["1", "2", "3", "4", "5"]) + +class TestFindAllByName(TreeTest): + """Test ways of finding tags by tag name.""" + + def setUp(self): + super(TreeTest, self).setUp() + self.tree = self.soup("""<a>First tag.</a> + <b>Second tag.</b> + <c>Third <a>Nested tag.</a> tag.</c>""") + + def test_find_all_by_tag_name(self): + # Find all the <a> tags. + self.assertSelects( + self.tree.find_all('a'), ['First tag.', 'Nested tag.']) + + def test_find_all_on_non_root_element(self): + # You can call find_all on any node, not just the root. + self.assertSelects(self.tree.c.find_all('a'), ['Nested tag.']) + + def test_calling_element_invokes_find_all(self): + self.assertSelects(self.tree('a'), ['First tag.', 'Nested tag.']) + + def test_find_all_by_tag_strainer(self): + self.assertSelects( + self.tree.find_all(SoupStrainer('a')), + ['First tag.', 'Nested tag.']) + + def test_find_all_by_tag_names(self): + self.assertSelects( + self.tree.find_all(['a', 'b']), + ['First tag.', 'Second tag.', 'Nested tag.']) + + def test_find_all_by_tag_dict(self): + self.assertSelects( + self.tree.find_all({'a' : True, 'b' : True}), + ['First tag.', 'Second tag.', 'Nested tag.']) + + def test_find_all_by_tag_re(self): + self.assertSelects( + self.tree.find_all(re.compile('^[ab]$')), + ['First tag.', 'Second tag.', 'Nested tag.']) + + def test_find_all_with_tags_matching_method(self): + # You can define an oracle method that determines whether + # a tag matches the search. + def id_matches_name(tag): + return tag.name == tag.get('id') + + tree = self.soup("""<a id="a">Match 1.</a> + <a id="1">Does not match.</a> + <b id="b">Match 2.</a>""") + + self.assertSelects( + tree.find_all(id_matches_name), ["Match 1.", "Match 2."]) + + +class TestFindAllByAttribute(TreeTest): + + def test_find_all_by_attribute_name(self): + # You can pass in keyword arguments to find_all to search by + # attribute. + tree = self.soup(""" + <a id="first">Matching a.</a> + <a id="second"> + Non-matching <b id="first">Matching b.</b>a. + </a>""") + self.assertSelects(tree.find_all(id='first'), + ["Matching a.", "Matching b."]) + + def test_find_all_by_attribute_dict(self): + # You can pass in a dictionary as the argument 'attrs'. This + # lets you search for attributes like 'name' (a fixed argument + # to find_all) and 'class' (a reserved word in Python.) + tree = self.soup(""" + <a name="name1" class="class1">Name match.</a> + <a name="name2" class="class2">Class match.</a> + <a name="name3" class="class3">Non-match.</a> + <name1>A tag called 'name1'.</name1> + """) + + # This doesn't do what you want. + self.assertSelects(tree.find_all(name='name1'), + ["A tag called 'name1'."]) + # This does what you want. + self.assertSelects(tree.find_all(attrs={'name' : 'name1'}), + ["Name match."]) + + # Passing class='class2' would cause a syntax error. + self.assertSelects(tree.find_all(attrs={'class' : 'class2'}), + ["Class match."]) + + def test_find_all_by_class(self): + # Passing in a string to 'attrs' will search the CSS class. + tree = self.soup(""" + <a class="1">Class 1.</a> + <a class="2">Class 2.</a> + <b class="1">Class 1.</b> + <c class="3 4">Class 3 and 4.</c> + """) + self.assertSelects(tree.find_all('a', '1'), ['Class 1.']) + self.assertSelects(tree.find_all(attrs='1'), ['Class 1.', 'Class 1.']) + self.assertSelects(tree.find_all('c', '3'), ['Class 3 and 4.']) + self.assertSelects(tree.find_all('c', '4'), ['Class 3 and 4.']) + + def test_find_all_by_attribute_soupstrainer(self): + tree = self.soup(""" + <a id="first">Match.</a> + <a id="second">Non-match.</a>""") + + strainer = SoupStrainer(attrs={'id' : 'first'}) + self.assertSelects(tree.find_all(strainer), ['Match.']) + + def test_find_all_with_missing_atribute(self): + # You can pass in None as the value of an attribute to find_all. + # This will match tags that do not have that attribute set. + tree = self.soup("""<a id="1">ID present.</a> + <a>No ID present.</a> + <a id="">ID is empty.</a>""") + self.assertSelects(tree.find_all('a', id=None), ["No ID present."]) + + def test_find_all_with_defined_attribute(self): + # You can pass in None as the value of an attribute to find_all. + # This will match tags that have that attribute set to any value. + tree = self.soup("""<a id="1">ID present.</a> + <a>No ID present.</a> + <a id="">ID is empty.</a>""") + self.assertSelects( + tree.find_all(id=True), ["ID present.", "ID is empty."]) + + def test_find_all_with_numeric_attribute(self): + # If you search for a number, it's treated as a string. + tree = self.soup("""<a id=1>Unquoted attribute.</a> + <a id="1">Quoted attribute.</a>""") + + expected = ["Unquoted attribute.", "Quoted attribute."] + self.assertSelects(tree.find_all(id=1), expected) + self.assertSelects(tree.find_all(id="1"), expected) + + def test_find_all_with_list_attribute_values(self): + # You can pass a list of attribute values instead of just one, + # and you'll get tags that match any of the values. + tree = self.soup("""<a id="1">1</a> + <a id="2">2</a> + <a id="3">3</a> + <a>No ID.</a>""") + self.assertSelects(tree.find_all(id=["1", "3", "4"]), + ["1", "3"]) + + def test_find_all_with_regular_expression_attribute_value(self): + # You can pass a regular expression as an attribute value, and + # you'll get tags whose values for that attribute match the + # regular expression. + tree = self.soup("""<a id="a">One a.</a> + <a id="aa">Two as.</a> + <a id="ab">Mixed as and bs.</a> + <a id="b">One b.</a> + <a>No ID.</a>""") + + self.assertSelects(tree.find_all(id=re.compile("^a+$")), + ["One a.", "Two as."]) + + +class TestIndex(TreeTest): + """Test Tag.index""" + def test_index(self): + tree = self.soup("""<wrap> + <a>Identical</a> + <b>Not identical</b> + <a>Identical</a> + + <c><d>Identical with child</d></c> + <b>Also not identical</b> + <c><d>Identical with child</d></c> + </wrap>""") + wrap = tree.wrap + for i, element in enumerate(wrap.contents): + self.assertEqual(i, wrap.index(element)) + self.assertRaises(ValueError, tree.index, 1) + + +class TestParentOperations(TreeTest): + """Test navigation and searching through an element's parents.""" + + def setUp(self): + super(TestParentOperations, self).setUp() + self.tree = self.soup('''<ul id="empty"></ul> + <ul id="top"> + <ul id="middle"> + <ul id="bottom"> + <b>Start here</b> + </ul> + </ul>''') + self.start = self.tree.b + + + def test_parent(self): + self.assertEquals(self.start.parent['id'], 'bottom') + self.assertEquals(self.start.parent.parent['id'], 'middle') + self.assertEquals(self.start.parent.parent.parent['id'], 'top') + + def test_parent_of_top_tag_is_soup_object(self): + top_tag = self.tree.contents[0] + self.assertEquals(top_tag.parent, self.tree) + + def test_soup_object_has_no_parent(self): + self.assertEquals(None, self.tree.parent) + + def test_find_parents(self): + self.assertSelectsIDs( + self.start.find_parents('ul'), ['bottom', 'middle', 'top']) + self.assertSelectsIDs( + self.start.find_parents('ul', id="middle"), ['middle']) + + def test_find_parent(self): + self.assertEquals(self.start.find_parent('ul')['id'], 'bottom') + + def test_parent_of_text_element(self): + text = self.tree.find(text="Start here") + self.assertEquals(text.parent.name, 'b') + + def test_text_element_find_parent(self): + text = self.tree.find(text="Start here") + self.assertEquals(text.find_parent('ul')['id'], 'bottom') + + def test_parent_generator(self): + parents = [parent['id'] for parent in self.start.parents + if parent is not None and 'id' in parent.attrs] + self.assertEquals(parents, ['bottom', 'middle', 'top']) + + +class ProximityTest(TreeTest): + + def setUp(self): + super(TreeTest, self).setUp() + self.tree = self.soup( + '<html id="start"><head></head><body><b id="1">One</b><b id="2">Two</b><b id="3">Three</b></body></html>') + + +class TestNextOperations(ProximityTest): + + def setUp(self): + super(TestNextOperations, self).setUp() + self.start = self.tree.b + + def test_next(self): + self.assertEquals(self.start.next_element, "One") + self.assertEquals(self.start.next_element.next_element['id'], "2") + + def test_next_of_last_item_is_none(self): + last = self.tree.find(text="Three") + self.assertEquals(last.next_element, None) + + def test_next_of_root_is_none(self): + # The document root is outside the next/previous chain. + self.assertEquals(self.tree.next_element, None) + + def test_find_all_next(self): + self.assertSelects(self.start.find_all_next('b'), ["Two", "Three"]) + self.start.find_all_next(id=3) + self.assertSelects(self.start.find_all_next(id=3), ["Three"]) + + def test_find_next(self): + self.assertEquals(self.start.find_next('b')['id'], '2') + self.assertEquals(self.start.find_next(text="Three"), "Three") + + def test_find_next_for_text_element(self): + text = self.tree.find(text="One") + self.assertEquals(text.find_next("b").string, "Two") + self.assertSelects(text.find_all_next("b"), ["Two", "Three"]) + + def test_next_generator(self): + start = self.tree.find(text="Two") + successors = [node for node in start.next_elements] + # There are two successors: the final <b> tag and its text contents. + # Then we go off the end. + tag, contents, none = successors + self.assertEquals(tag['id'], '3') + self.assertEquals(contents, "Three") + self.assertEquals(none, None) + + # XXX Should next_elements really return None? Seems like it + # should just stop. + + +class TestPreviousOperations(ProximityTest): + + def setUp(self): + super(TestPreviousOperations, self).setUp() + self.end = self.tree.find(text="Three") + + def test_previous(self): + self.assertEquals(self.end.previous_element['id'], "3") + self.assertEquals(self.end.previous_element.previous_element, "Two") + + def test_previous_of_first_item_is_none(self): + first = self.tree.find('html') + self.assertEquals(first.previous_element, None) + + def test_previous_of_root_is_none(self): + # The document root is outside the next/previous chain. + # XXX This is broken! + #self.assertEquals(self.tree.previous_element, None) + pass + + def test_find_all_previous(self): + # The <b> tag containing the "Three" node is the predecessor + # of the "Three" node itself, which is why "Three" shows up + # here. + self.assertSelects( + self.end.find_all_previous('b'), ["Three", "Two", "One"]) + self.assertSelects(self.end.find_all_previous(id=1), ["One"]) + + def test_find_previous(self): + self.assertEquals(self.end.find_previous('b')['id'], '3') + self.assertEquals(self.end.find_previous(text="One"), "One") + + def test_find_previous_for_text_element(self): + text = self.tree.find(text="Three") + self.assertEquals(text.find_previous("b").string, "Three") + self.assertSelects( + text.find_all_previous("b"), ["Three", "Two", "One"]) + + def test_previous_generator(self): + start = self.tree.find(text="One") + predecessors = [node for node in start.previous_elements] + + # There are four predecessors: the <b> tag containing "One" + # the <body> tag, the <head> tag, and the <html> tag. Then we + # go off the end. + b, body, head, html, none = predecessors + self.assertEquals(b['id'], '1') + self.assertEquals(body.name, "body") + self.assertEquals(head.name, "head") + self.assertEquals(html.name, "html") + self.assertEquals(none, None) + + # Again, we shouldn't be returning None. + + +class SiblingTest(TreeTest): + + def setUp(self): + super(SiblingTest, self).setUp() + markup = '''<html> + <span id="1"> + <span id="1.1"></span> + </span> + <span id="2"> + <span id="2.1"></span> + </span> + <span id="3"> + <span id="3.1"></span> + </span> + <span id="4"></span> + </html>''' + # All that whitespace looks good but makes the tests more + # difficult. Get rid of it. + markup = re.compile("\n\s*").sub("", markup) + self.tree = self.soup(markup) + + +class TestNextSibling(SiblingTest): + + def setUp(self): + super(TestNextSibling, self).setUp() + self.start = self.tree.find(id="1") + + def test_next_sibling_of_root_is_none(self): + self.assertEquals(self.tree.next_sibling, None) + + def test_next_sibling(self): + self.assertEquals(self.start.next_sibling['id'], '2') + self.assertEquals(self.start.next_sibling.next_sibling['id'], '3') + + # Note the difference between next_sibling and next_element. + self.assertEquals(self.start.next_element['id'], '1.1') + + def test_next_sibling_may_not_exist(self): + self.assertEquals(self.tree.html.next_sibling, None) + + nested_span = self.tree.find(id="1.1") + self.assertEquals(nested_span.next_sibling, None) + + last_span = self.tree.find(id="4") + self.assertEquals(last_span.next_sibling, None) + + def test_find_next_sibling(self): + self.assertEquals(self.start.find_next_sibling('span')['id'], '2') + + def test_next_siblings(self): + self.assertSelectsIDs(self.start.find_next_siblings("span"), + ['2', '3', '4']) + + self.assertSelectsIDs(self.start.find_next_siblings(id='3'), ['3']) + + def test_next_sibling_for_text_element(self): + soup = self.soup("Foo<b>bar</b>baz") + start = soup.find(text="Foo") + self.assertEquals(start.next_sibling.name, 'b') + self.assertEquals(start.next_sibling.next_sibling, 'baz') + + self.assertSelects(start.find_next_siblings('b'), ['bar']) + self.assertEquals(start.find_next_sibling(text="baz"), "baz") + self.assertEquals(start.find_next_sibling(text="nonesuch"), None) + + +class TestPreviousSibling(SiblingTest): + + def setUp(self): + super(TestPreviousSibling, self).setUp() + self.end = self.tree.find(id="4") + + def test_previous_sibling_of_root_is_none(self): + self.assertEquals(self.tree.previous_sibling, None) + + def test_previous_sibling(self): + self.assertEquals(self.end.previous_sibling['id'], '3') + self.assertEquals(self.end.previous_sibling.previous_sibling['id'], '2') + + # Note the difference between previous_sibling and previous_element. + self.assertEquals(self.end.previous_element['id'], '3.1') + + def test_previous_sibling_may_not_exist(self): + self.assertEquals(self.tree.html.previous_sibling, None) + + nested_span = self.tree.find(id="1.1") + self.assertEquals(nested_span.previous_sibling, None) + + first_span = self.tree.find(id="1") + self.assertEquals(first_span.previous_sibling, None) + + def test_find_previous_sibling(self): + self.assertEquals(self.end.find_previous_sibling('span')['id'], '3') + + def test_previous_siblings(self): + self.assertSelectsIDs(self.end.find_previous_siblings("span"), + ['3', '2', '1']) + + self.assertSelectsIDs(self.end.find_previous_siblings(id='1'), ['1']) + + def test_previous_sibling_for_text_element(self): + soup = self.soup("Foo<b>bar</b>baz") + start = soup.find(text="baz") + self.assertEquals(start.previous_sibling.name, 'b') + self.assertEquals(start.previous_sibling.previous_sibling, 'Foo') + + self.assertSelects(start.find_previous_siblings('b'), ['bar']) + self.assertEquals(start.find_previous_sibling(text="Foo"), "Foo") + self.assertEquals(start.find_previous_sibling(text="nonesuch"), None) + + +class TestTreeModification(SoupTest): + + def test_attribute_modification(self): + soup = self.soup('<a id="1"></a>') + soup.a['id'] = 2 + self.assertEqual(soup.decode(), self.document_for('<a id="2"></a>')) + del(soup.a['id']) + self.assertEqual(soup.decode(), self.document_for('<a></a>')) + soup.a['id2'] = 'foo' + self.assertEqual(soup.decode(), self.document_for('<a id2="foo"></a>')) + + def test_new_tag_creation(self): + builder = builder_registry.lookup('html5lib')() + soup = self.soup("<body></body>", builder=builder) + a = Tag(soup, builder, 'a') + ol = Tag(soup, builder, 'ol') + a['href'] = 'http://foo.com/' + soup.body.insert(0, a) + soup.body.insert(1, ol) + self.assertEqual( + soup.body.encode(), + '<body><a href="http://foo.com/"></a><ol></ol></body>') + + def test_append_to_contents_moves_tag(self): + doc = """<p id="1">Don't leave me <b>here</b>.</p> + <p id="2">Don\'t leave!</p>""" + soup = self.soup(doc) + second_para = soup.find(id='2') + bold = soup.b + + # Move the <b> tag to the end of the second paragraph. + soup.find(id='2').append(soup.b) + + # The <b> tag is now a child of the second paragraph. + self.assertEqual(bold.parent, second_para) + + self.assertEqual( + soup.decode(), self.document_for( + '<p id="1">Don\'t leave me .</p>\n' + '<p id="2">Don\'t leave!<b>here</b></p>')) + + def test_replace_tag_with_itself(self): + text = "<a><b></b><c>Foo<d></d></c></a><a><e></e></a>" + soup = self.soup(text) + c = soup.c + soup.c.replace_with(c) + self.assertEquals(soup.decode(), self.document_for(text)) + + def test_replace_final_node(self): + soup = self.soup("<b>Argh!</b>") + soup.find(text="Argh!").replace_with("Hooray!") + new_text = soup.find(text="Hooray!") + b = soup.b + self.assertEqual(new_text.previous_element, b) + self.assertEqual(new_text.parent, b) + self.assertEqual(new_text.previous_element.next_element, new_text) + self.assertEqual(new_text.next_element, None) + + def test_consecutive_text_nodes(self): + # A builder should never create two consecutive text nodes, + # but if you insert one next to another, Beautiful Soup will + # handle it correctly. + soup = self.soup("<a><b>Argh!</b><c></c></a>") + soup.b.insert(1, "Hooray!") + + self.assertEqual( + soup.decode(), self.document_for( + "<a><b>Argh!Hooray!</b><c></c></a>")) + + new_text = soup.find(text="Hooray!") + self.assertEqual(new_text.previous_element, "Argh!") + self.assertEqual(new_text.previous_element.next_element, new_text) + + self.assertEqual(new_text.previous_sibling, "Argh!") + self.assertEqual(new_text.previous_sibling.next_sibling, new_text) + + self.assertEqual(new_text.next_sibling, None) + self.assertEqual(new_text.next_element, soup.c) + + def test_insert_tag(self): + builder = self.default_builder + soup = self.soup( + "<a><b>Find</b><c>lady!</c><d></d></a>", builder=builder) + magic_tag = Tag(soup, builder, 'magictag') + magic_tag.insert(0, "the") + soup.a.insert(1, magic_tag) + + self.assertEqual( + soup.decode(), self.document_for( + "<a><b>Find</b><magictag>the</magictag><c>lady!</c><d></d></a>")) + + # Make sure all the relationships are hooked up correctly. + b_tag = soup.b + self.assertEqual(b_tag.next_sibling, magic_tag) + self.assertEqual(magic_tag.previous_sibling, b_tag) + + find = b_tag.find(text="Find") + self.assertEqual(find.next_element, magic_tag) + self.assertEqual(magic_tag.previous_element, find) + + c_tag = soup.c + self.assertEqual(magic_tag.next_sibling, c_tag) + self.assertEqual(c_tag.previous_sibling, magic_tag) + + the = magic_tag.find(text="the") + self.assertEqual(the.parent, magic_tag) + self.assertEqual(the.next_element, c_tag) + self.assertEqual(c_tag.previous_element, the) + + def test_insert_works_on_empty_element_tag(self): + # This is a little strange, since most HTML parsers don't allow + # markup like this to come through. But in general, we don't + # know what the parser would or wouldn't have allowed, so + # I'm letting this succeed for now. + soup = self.soup("<br />") + soup.br.insert(1, "Contents") + self.assertEquals(str(soup.br), "<br>Contents</br>") + + def test_replace_with(self): + soup = self.soup( + "<p>There's <b>no</b> business like <b>show</b> business</p>") + no, show = soup.find_all('b') + show.replace_with(no) + self.assertEquals( + soup.decode(), + self.document_for( + "<p>There's business like <b>no</b> business</p>")) + + self.assertEquals(show.parent, None) + self.assertEquals(no.parent, soup.p) + self.assertEquals(no.next_element, "no") + self.assertEquals(no.next_sibling, " business") + + def test_nested_tag_replace_with(self): + soup = self.soup( + """<a>We<b>reserve<c>the</c><d>right</d></b></a><e>to<f>refuse</f><g>service</g></e>""") + + # Replace the entire <b> tag and its contents ("reserve the + # right") with the <f> tag ("refuse"). + remove_tag = soup.b + move_tag = soup.f + remove_tag.replace_with(move_tag) + + self.assertEqual( + soup.decode(), self.document_for( + "<a>We<f>refuse</f></a><e>to<g>service</g></e>")) + + # The <b> tag is now an orphan. + self.assertEqual(remove_tag.parent, None) + self.assertEqual(remove_tag.find(text="right").next_element, None) + self.assertEqual(remove_tag.previous_element, None) + self.assertEqual(remove_tag.next_sibling, None) + self.assertEqual(remove_tag.previous_sibling, None) + + # The <f> tag is now connected to the <a> tag. + self.assertEqual(move_tag.parent, soup.a) + self.assertEqual(move_tag.previous_element, "We") + self.assertEqual(move_tag.next_element.next_element, soup.e) + self.assertEqual(move_tag.next_sibling, None) + + # The gap where the <f> tag used to be has been mended, and + # the word "to" is now connected to the <g> tag. + to_text = soup.find(text="to") + g_tag = soup.g + self.assertEqual(to_text.next_element, g_tag) + self.assertEqual(to_text.next_sibling, g_tag) + self.assertEqual(g_tag.previous_element, to_text) + self.assertEqual(g_tag.previous_sibling, to_text) + + def test_replace_with_children(self): + tree = self.soup(""" + <p>Unneeded <em>formatting</em> is unneeded</p> + """) + tree.em.replace_with_children() + self.assertEqual(tree.em, None) + self.assertEqual(tree.p.text, "Unneeded formatting is unneeded") + + def test_extract(self): + soup = self.soup( + '<html><body>Some content. <div id="nav">Nav crap</div> More content.</body></html>') + + self.assertEqual(len(soup.body.contents), 3) + extracted = soup.find(id="nav").extract() + + self.assertEqual( + soup.decode(), "<html><body>Some content. More content.</body></html>") + self.assertEqual(extracted.decode(), '<div id="nav">Nav crap</div>') + + # The extracted tag is now an orphan. + self.assertEqual(len(soup.body.contents), 2) + self.assertEqual(extracted.parent, None) + self.assertEqual(extracted.previous_element, None) + self.assertEqual(extracted.next_element.next_element, None) + + # The gap where the extracted tag used to be has been mended. + content_1 = soup.find(text="Some content. ") + content_2 = soup.find(text=" More content.") + self.assertEquals(content_1.next_element, content_2) + self.assertEquals(content_1.next_sibling, content_2) + self.assertEquals(content_2.previous_element, content_1) + self.assertEquals(content_2.previous_sibling, content_1) + + def test_clear(self): + """Tag.clear()""" + soup = self.soup("<p><a>String <em>Italicized</em></a> and another</p>") + # clear using extract() + a = soup.a + soup.p.clear() + self.assertEqual(len(soup.p.contents), 0) + self.assertTrue(hasattr(a, "contents")) + + # clear using decompose() + em = a.em + a.clear(decompose=True) + self.assertFalse(hasattr(em, "contents")) + + def test_string_set(self): + """Tag.string = 'string'""" + soup = self.soup("<a></a> <b><c></c></b>") + soup.a.string = "foo" + self.assertEqual(soup.a.contents, ["foo"]) + soup.b.string = "bar" + self.assertEqual(soup.b.contents, ["bar"]) + + +class TestElementObjects(SoupTest): + """Test various features of element objects.""" + + def test_len(self): + """The length of an element is its number of children.""" + soup = self.soup("<top>1<b>2</b>3</top>") + + # The BeautifulSoup object itself contains one element: the + # <top> tag. + self.assertEquals(len(soup.contents), 1) + self.assertEquals(len(soup), 1) + + # The <top> tag contains three elements: the text node "1", the + # <b> tag, and the text node "3". + self.assertEquals(len(soup.top), 3) + self.assertEquals(len(soup.top.contents), 3) + + def test_member_access_invokes_find(self): + """Accessing a Python member .foo or .fooTag invokes find('foo')""" + soup = self.soup('<b><i></i></b>') + self.assertEqual(soup.b, soup.find('b')) + self.assertEqual(soup.bTag, soup.find('b')) + self.assertEqual(soup.b.i, soup.find('b').find('i')) + self.assertEqual(soup.bTag.iTag, soup.find('b').find('i')) + self.assertEqual(soup.a, None) + self.assertEqual(soup.aTag, None) + + def test_has_attr(self): + """has_attr() checks for the presence of an attribute. + + Please note note: has_attr() is different from + __in__. has_attr() checks the tag's attributes and __in__ + checks the tag's chidlren. + """ + soup = self.soup("<foo attr='bar'>") + self.assertTrue(soup.foo.has_attr('attr')) + self.assertFalse(soup.foo.has_attr('attr2')) + + + def test_attributes_come_out_in_alphabetical_order(self): + markup = '<b a="1" z="5" m="3" f="2" y="4"></b>' + self.assertSoupEquals(markup, '<b a="1" f="2" m="3" y="4" z="5"></b>') + + def test_multiple_values_for_the_same_attribute_are_collapsed(self): + markup = '<b b="20" a="1" b="10" a="2" a="3" a="4"></b>' + self.assertSoupEquals(markup, '<b a="1" b="20"></b>') + + def test_string(self): + # A tag that contains only a text node makes that node + # available as .string. + soup = self.soup("<b>foo</b>") + self.assertEquals(soup.b.string, 'foo') + + def test_empty_tag_has_no_string(self): + # A tag with no children has no .stirng. + soup = self.soup("<b></b>") + self.assertEqual(soup.b.string, None) + + def test_tag_with_multiple_children_has_no_string(self): + # A tag with no children has no .string. + soup = self.soup("<a>foo<b></b><b></b></b>") + self.assertEqual(soup.b.string, None) + + soup = self.soup("<a>foo<b></b>bar</b>") + self.assertEqual(soup.b.string, None) + + # Even if all the children are strings, due to trickery, + # it won't work--but this would be a good optimization. + soup = self.soup("<a>foo</b>") + soup.a.insert(1, "bar") + self.assertEqual(soup.a.string, None) + + def test_tag_with_recursive_string_has_string(self): + # A tag with a single child which has a .string inherits that + # .string. + soup = self.soup("<a><b>foo</b></a>") + self.assertEqual(soup.a.string, "foo") + self.assertEqual(soup.string, "foo") + + def test_lack_of_string(self): + """Only a tag containing a single text node has a .string.""" + soup = self.soup("<b>f<i>e</i>o</b>") + self.assertFalse(soup.b.string) + + soup = self.soup("<b></b>") + self.assertFalse(soup.b.string) + + def test_all_text(self): + """Tag.text and Tag.get_text(sep=u"") -> all child text, concatenated""" + soup = self.soup("<a>a<b>r</b> <r> t </r></a>") + self.assertEqual(soup.a.text, "ar t ") + self.assertEqual(soup.a.get_text(strip=True), "art") + self.assertEqual(soup.a.get_text(","), "a,r, , t ") + self.assertEqual(soup.a.get_text(",", strip=True), "a,r,t") + + +class TestPersistence(SoupTest): + "Testing features like pickle and deepcopy." + + def setUp(self): + super(TestPersistence, self).setUp() + self.page = """<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" +"http://www.w3.org/TR/REC-html40/transitional.dtd"> +<html> +<head> +<meta http-equiv="Content-Type" content="text/html; charset=utf-8"> +<title>Beautiful Soup: We called him Tortoise because he taught us.</title> +<link rev="made" href="mailto:leonardr@segfault.org"> +<meta name="Description" content="Beautiful Soup: an HTML parser optimized for screen-scraping."> +<meta name="generator" content="Markov Approximation 1.4 (module: leonardr)"> +<meta name="author" content="Leonard Richardson"> +</head> +<body> +<a href="foo">foo</a> +<a href="foo"><b>bar</b></a> +</body> +</html>""" + self.tree = self.soup(self.page) + + def test_pickle_and_unpickle_identity(self): + # Pickling a tree, then unpickling it, yields a tree identical + # to the original. + dumped = pickle.dumps(self.tree, 2) + loaded = pickle.loads(dumped) + self.assertEqual(loaded.__class__, BeautifulSoup) + self.assertEqual(loaded.decode(), self.tree.decode()) + + def test_deepcopy_identity(self): + # Making a deepcopy of a tree yields an identical tree. + copied = copy.deepcopy(self.tree) + self.assertEqual(copied.decode(), self.tree.decode()) + + def test_unicode_pickle(self): + # A tree containing Unicode characters can be pickled. + html = u"<b>\N{SNOWMAN}</b>" + soup = self.soup(html) + dumped = pickle.dumps(soup, pickle.HIGHEST_PROTOCOL) + loaded = pickle.loads(dumped) + self.assertEqual(loaded.decode(), soup.decode()) + + +class TestSubstitutions(SoupTest): + + def test_html_entity_substitution(self): + soup = self.soup( + u"<b>Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!</b>") + decoded = soup.decode(substitute_html_entities=True) + self.assertEquals(decoded, + self.document_for("<b>Sacré bleu!</b>")) + + def test_html_entity_substitution_off_by_default(self): + markup = u"<b>Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!</b>" + soup = self.soup(markup) + encoded = soup.b.encode("utf-8") + self.assertEquals(encoded, markup.encode('utf-8')) + + def test_encoding_substitution(self): + # Here's the <meta> tag saying that a document is + # encoded in Shift-JIS. + meta_tag = ('<meta content="text/html; charset=x-sjis" ' + 'http-equiv="Content-type" />') + soup = self.soup(meta_tag) + + # Parse the document, and the charset is replaced with a + # generic value. + self.assertEquals(soup.meta['content'], + 'text/html; charset=%SOUP-ENCODING%') + + # Encode the document into some encoding, and the encoding is + # substituted into the meta tag. + utf_8 = soup.encode("utf-8") + self.assertTrue(b"charset=utf-8" in utf_8) + + euc_jp = soup.encode("euc_jp") + self.assertTrue(b"charset=euc_jp" in euc_jp) + + shift_jis = soup.encode("shift-jis") + self.assertTrue(b"charset=shift-jis" in shift_jis) + + utf_16_u = soup.encode("utf-16").decode("utf-16") + self.assertTrue("charset=utf-16" in utf_16_u) + + def test_encoding_substitution_doesnt_happen_if_tag_is_strained(self): + markup = ('<head><meta content="text/html; charset=x-sjis" ' + 'http-equiv="Content-type" /></head><pre>foo</pre>') + + # Beautiful Soup used to try to rewrite the meta tag even if the + # meta tag got filtered out by the strainer. This test makes + # sure that doesn't happen. + strainer = SoupStrainer('pre') + soup = self.soup(markup, parse_only=strainer) + self.assertEquals(soup.contents[0].name, 'pre') + + +class TestEncoding(SoupTest): + """Test the ability to encode objects into strings.""" + + def test_unicode_string_can_be_encoded(self): + html = u"<b>\N{SNOWMAN}</b>" + soup = self.soup(html) + self.assertEquals(soup.b.string.encode("utf-8"), + u"\N{SNOWMAN}".encode("utf-8")) + + def test_tag_containing_unicode_string_can_be_encoded(self): + html = u"<b>\N{SNOWMAN}</b>" + soup = self.soup(html) + self.assertEquals( + soup.b.encode("utf-8"), html.encode("utf-8")) + + +class TestNavigableStringSubclasses(SoupTest): + + def test_cdata(self): + # None of the current builders turn CDATA sections into CData + # objects, but you can create them manually. + soup = self.soup("") + cdata = CData("foo") + soup.insert(1, cdata) + self.assertEquals(str(soup), "<![CDATA[foo]]>") + self.assertEquals(soup.find(text="foo"), "foo") + self.assertEquals(soup.contents[0], "foo") |