# encoding: utf-8
"""Helper classes for tests."""
# Use of this source code is governed by the MIT license.
__license__ = "MIT"
import pickle
import copy
import functools
import unittest
from unittest import TestCase
from bs4 import BeautifulSoup
from bs4.element import (
CharsetMetaAttributeValue,
Comment,
ContentMetaAttributeValue,
Doctype,
SoupStrainer,
Tag
)
from bs4.builder import HTMLParserTreeBuilder
default_builder = HTMLParserTreeBuilder
BAD_DOCUMENT = u"""A bare string
A tag
A tag that supposedly has contents.
AT&T
This numeric entity is missing the final semicolon:
"""
class SoupTest(unittest.TestCase):
@property
def default_builder(self):
return default_builder()
def soup(self, markup, **kwargs):
"""Build a Beautiful Soup object from markup."""
builder = kwargs.pop('builder', self.default_builder)
return BeautifulSoup(markup, builder=builder, **kwargs)
def document_for(self, markup):
"""Turn an HTML fragment into a document.
The details depend on the builder.
"""
return self.default_builder.test_fragment_to_document(markup)
def assertSoupEquals(self, to_parse, compare_parsed_to=None):
builder = self.default_builder
obj = BeautifulSoup(to_parse, builder=builder)
if compare_parsed_to is None:
compare_parsed_to = to_parse
self.assertEqual(obj.decode(), self.document_for(compare_parsed_to))
def assertConnectedness(self, element):
"""Ensure that next_element and previous_element are properly
set for all descendants of the given element.
"""
earlier = None
for e in element.descendants:
if earlier:
self.assertEqual(e, earlier.next_element)
self.assertEqual(earlier, e.previous_element)
earlier = e
def linkage_validator(self, el, _recursive_call=False):
"""Ensure proper linkage throughout the document."""
descendant = None
# Document element should have no previous element or previous sibling.
# It also shouldn't have a next sibling.
if el.parent is None:
assert el.previous_element is None,\
"Bad previous_element\nNODE: {}\nPREV: {}\nEXPECTED: {}".format(
el, el.previous_element, None
)
assert el.previous_sibling is None,\
"Bad previous_sibling\nNODE: {}\nPREV: {}\nEXPECTED: {}".format(
el, el.previous_sibling, None
)
assert el.next_sibling is None,\
"Bad next_sibling\nNODE: {}\nNEXT: {}\nEXPECTED: {}".format(
el, el.next_sibling, None
)
idx = 0
child = None
last_child = None
last_idx = len(el.contents) - 1
for child in el.contents:
descendant = None
# Parent should link next element to their first child
# That child should have no previous sibling
if idx == 0:
if el.parent is not None:
assert el.next_element is child,\
"Bad next_element\nNODE: {}\nNEXT: {}\nEXPECTED: {}".format(
el, el.next_element, child
)
assert child.previous_element is el,\
"Bad previous_element\nNODE: {}\nPREV: {}\nEXPECTED: {}".format(
child, child.previous_element, el
)
assert child.previous_sibling is None,\
"Bad previous_sibling\nNODE: {}\nPREV {}\nEXPECTED: {}".format(
child, child.previous_sibling, None
)
# If not the first child, previous index should link as sibling to this index
# Previous element should match the last index or the last bubbled up descendant
else:
assert child.previous_sibling is el.contents[idx - 1],\
"Bad previous_sibling\nNODE: {}\nPREV {}\nEXPECTED {}".format(
child, child.previous_sibling, el.contents[idx - 1]
)
assert el.contents[idx - 1].next_sibling is child,\
"Bad next_sibling\nNODE: {}\nNEXT {}\nEXPECTED {}".format(
el.contents[idx - 1], el.contents[idx - 1].next_sibling, child
)
if last_child is not None:
assert child.previous_element is last_child,\
"Bad previous_element\nNODE: {}\nPREV {}\nEXPECTED {}\nCONTENTS {}".format(
child, child.previous_element, last_child, child.parent.contents
)
assert last_child.next_element is child,\
"Bad next_element\nNODE: {}\nNEXT {}\nEXPECTED {}".format(
last_child, last_child.next_element, child
)
if isinstance(child, Tag) and child.contents:
descendant = self.linkage_validator(child, True)
# A bubbled up descendant should have no next siblings
assert descendant.next_sibling is None,\
"Bad next_sibling\nNODE: {}\nNEXT {}\nEXPECTED {}".format(
descendant, descendant.next_sibling, None
)
# Mark last child as either the bubbled up descendant or the current child
if descendant is not None:
last_child = descendant
else:
last_child = child
# If last child, there are non next siblings
if idx == last_idx:
assert child.next_sibling is None,\
"Bad next_sibling\nNODE: {}\nNEXT {}\nEXPECTED {}".format(
child, child.next_sibling, None
)
idx += 1
child = descendant if descendant is not None else child
if child is None:
child = el
if not _recursive_call and child is not None:
target = el
while True:
if target is None:
assert child.next_element is None, \
"Bad next_element\nNODE: {}\nNEXT {}\nEXPECTED {}".format(
child, child.next_element, None
)
break
elif target.next_sibling is not None:
assert child.next_element is target.next_sibling, \
"Bad next_element\nNODE: {}\nNEXT {}\nEXPECTED {}".format(
child, child.next_element, target.next_sibling
)
break
target = target.parent
# We are done, so nothing to return
return None
else:
# Return the child to the recursive caller
return child
class HTMLTreeBuilderSmokeTest(object):
"""A basic test of a treebuilder's competence.
Any HTML treebuilder, present or future, should be able to pass
these tests. With invalid markup, there's room for interpretation,
and different parsers can handle it differently. But with the
markup in these tests, there's not much room for interpretation.
"""
def test_empty_element_tags(self):
"""Verify that all HTML4 and HTML5 empty element (aka void element) tags
are handled correctly.
"""
for name in [
'area', 'base', 'br', 'col', 'embed', 'hr', 'img', 'input', 'keygen', 'link', 'menuitem', 'meta', 'param', 'source', 'track', 'wbr',
'spacer', 'frame'
]:
soup = self.soup("")
new_tag = soup.new_tag(name)
self.assertEqual(True, new_tag.is_empty_element)
def test_pickle_and_unpickle_identity(self):
# Pickling a tree, then unpickling it, yields a tree identical
# to the original.
tree = self.soup("foo")
dumped = pickle.dumps(tree, 2)
loaded = pickle.loads(dumped)
self.assertEqual(loaded.__class__, BeautifulSoup)
self.assertEqual(loaded.decode(), tree.decode())
def assertDoctypeHandled(self, doctype_fragment):
"""Assert that a given doctype string is handled correctly."""
doctype_str, soup = self._document_with_doctype(doctype_fragment)
# Make sure a Doctype object was created.
doctype = soup.contents[0]
self.assertEqual(doctype.__class__, Doctype)
self.assertEqual(doctype, doctype_fragment)
self.assertEqual(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.assertEqual(soup.p.contents[0], 'foo')
def _document_with_doctype(self, doctype_fragment):
"""Generate and parse a document with the given doctype."""
doctype = '' % doctype_fragment
markup = doctype + '\n
foo
'
soup = self.soup(markup)
return doctype, soup
def test_normal_doctypes(self):
"""Make sure normal, everyday HTML doctypes are handled correctly."""
self.assertDoctypeHandled("html")
self.assertDoctypeHandled(
'html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"')
def test_empty_doctype(self):
soup = self.soup("")
doctype = soup.contents[0]
self.assertEqual("", doctype.strip())
def test_public_doctype_with_url(self):
doctype = 'html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"'
self.assertDoctypeHandled(doctype)
def test_system_doctype(self):
self.assertDoctypeHandled('foo SYSTEM "http://www.example.com/"')
def test_namespaced_system_doctype(self):
# We can handle a namespaced doctype with a system ID.
self.assertDoctypeHandled('xsl:stylesheet SYSTEM "htmlent.dtd"')
def test_namespaced_public_doctype(self):
# Test a namespaced doctype with a public id.
self.assertDoctypeHandled('xsl:stylesheet PUBLIC "htmlent.dtd"')
def test_real_xhtml_document(self):
"""A real XHTML document should come out more or less the same as it went in."""
markup = b"""
Hello.
Goodbye.
"""
soup = self.soup(markup)
self.assertEqual(
soup.encode("utf-8").replace(b"\n", b""),
markup.replace(b"\n", b""))
def test_namespaced_html(self):
"""When a namespaced XML document is parsed as HTML it should
be treated as HTML with weird tag names.
"""
markup = b"""content"""
soup = self.soup(markup)
self.assertEqual(2, len(soup.find_all("ns1:foo")))
def test_processing_instruction(self):
# We test both Unicode and bytestring to verify that
# process_markup correctly sets processing_instruction_class
# even when the markup is already Unicode and there is no
# need to process anything.
markup = u""""""
soup = self.soup(markup)
self.assertEqual(markup, soup.decode())
markup = b""""""
soup = self.soup(markup)
self.assertEqual(markup, soup.encode("utf8"))
def test_deepcopy(self):
"""Make sure you can copy the tree builder.
This is important because the builder is part of a
BeautifulSoup object, and we want to be able to copy that.
"""
copy.deepcopy(self.default_builder)
def test_p_tag_is_never_empty_element(self):
"""A
tag is never designated as an empty-element tag.
Even if the markup shows it as an empty-element tag, it
shouldn't be presented that way.
"""
soup = self.soup("
")
self.assertFalse(soup.p.is_empty_element)
self.assertEqual(str(soup.p), "")
def test_unclosed_tags_get_closed(self):
"""A tag that's not closed by the end of the document should be closed.
This applies to all tags except empty-element tags.
"""
self.assertSoupEquals("
", "
")
self.assertSoupEquals("", "")
self.assertSoupEquals(" ", " ")
def test_br_is_always_empty_element_tag(self):
"""A tag is designated as an empty-element tag.
Some parsers treat as one tag, some parsers as
two tags, but it should always be an empty-element tag.
"""
soup = self.soup(" ")
self.assertTrue(soup.br.is_empty_element)
self.assertEqual(str(soup.br), " ")
def test_nested_formatting_elements(self):
self.assertSoupEquals("")
def test_double_head(self):
html = '''
Ordinary HEAD element test
Hello, world!
'''
soup = self.soup(html)
self.assertEqual("text/javascript", soup.find('script')['type'])
def test_comment(self):
# Comments are represented as Comment objects.
markup = "
foobaz
"
self.assertSoupEquals(markup)
soup = self.soup(markup)
comment = soup.find(text="foobar")
self.assertEqual(comment.__class__, Comment)
# The comment is properly integrated into the tree.
foo = soup.find(text="foo")
self.assertEqual(comment, foo.next_element)
baz = soup.find(text="baz")
self.assertEqual(comment, baz.previous_element)
def test_preserved_whitespace_in_pre_and_textarea(self):
"""Whitespace must be preserved in
and "
self.assertSoupEquals(pre_markup)
self.assertSoupEquals(textarea_markup)
soup = self.soup(pre_markup)
self.assertEqual(soup.pre.prettify(), pre_markup)
soup = self.soup(textarea_markup)
self.assertEqual(soup.textarea.prettify(), textarea_markup)
soup = self.soup("")
self.assertEqual(soup.textarea.prettify(), "")
def test_nested_inline_elements(self):
"""Inline elements can be nested indefinitely."""
b_tag = "Inside a B tag"
self.assertSoupEquals(b_tag)
nested_b_tag = "
"
self.assertSoupEquals(nested_b_tag)
def test_nested_block_level_elements(self):
"""Block elements can be nested."""
soup = self.soup('
Foo
')
blockquote = soup.blockquote
self.assertEqual(blockquote.p.b.string, 'Foo')
self.assertEqual(blockquote.b.string, 'Foo')
def test_correctly_nested_tables(self):
"""One table can go inside another one."""
markup = ('
'
'
'
"
Here's another table:"
'
'
'
foo
'
'
')
self.assertSoupEquals(
markup,
'
Here\'s another table:'
'
foo
'
'
')
self.assertSoupEquals(
"
Foo
"
"
Bar
"
"
Baz
")
def test_multivalued_attribute_with_whitespace(self):
# Whitespace separating the values of a multi-valued attribute
# should be ignored.
markup = '
'
soup = self.soup(markup)
self.assertEqual(['foo', 'bar'], soup.div['class'])
# If you search by the literal name of the class it's like the whitespace
# wasn't there.
self.assertEqual(soup.div, soup.find('div', class_="foo bar"))
def test_deeply_nested_multivalued_attribute(self):
# html5lib can set the attributes of the same tag many times
# as it rearranges the tree. This has caused problems with
# multivalued attributes.
markup = '
'
soup = self.soup(markup)
self.assertEqual(["css"], soup.div.div['class'])
def test_multivalued_attribute_on_html(self):
# html5lib uses a different API to set the attributes ot the
# tag. This has caused problems with multivalued
# attributes.
markup = ''
soup = self.soup(markup)
self.assertEqual(["a", "b"], soup.html['class'])
def test_angle_brackets_in_attribute_values_are_escaped(self):
self.assertSoupEquals('', '')
def test_strings_resembling_character_entity_references(self):
# "&T" and "&p" look like incomplete character entities, but they are
# not.
self.assertSoupEquals(
u"
• AT&T is in the s&p 500
",
u"
\u2022 AT&T is in the s&p 500
"
)
def test_entities_in_foreign_document_encoding(self):
# “ and ” are invalid numeric entities referencing
# Windows-1252 characters. - references a character common
# to Windows-1252 and Unicode, and ☃ references a
# character only found in Unicode.
#
# All of these entities should be converted to Unicode
# characters.
markup = "
')
def test_out_of_range_entity(self):
expect = u"\N{REPLACEMENT CHARACTER}"
self.assertSoupEquals("�", expect)
self.assertSoupEquals("�", expect)
self.assertSoupEquals("�", expect)
def test_multipart_strings(self):
"Mostly to prevent a recurrence of a bug in the html5lib treebuilder."
soup = self.soup("
\nfoo
")
self.assertEqual("p", soup.h2.string.next_element.name)
self.assertEqual("p", soup.p.name)
self.assertConnectedness(soup)
def test_empty_element_tags(self):
"""Verify consistent handling of empty-element tags,
no matter how they come in through the markup.
"""
self.assertSoupEquals('
', "
")
self.assertSoupEquals('
', "
")
def test_head_tag_between_head_and_body(self):
"Prevent recurrence of a bug in the html5lib treebuilder."
content = """
foo
"""
soup = self.soup(content)
self.assertNotEqual(None, soup.html.body)
self.assertConnectedness(soup)
def test_multiple_copies_of_a_tag(self):
"Prevent recurrence of a bug in the html5lib treebuilder."
content = """
"
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"
\x91Foo\x92
"
soup = self.soup(quote)
self.assertEqual(
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("")
self.assertEqual(soup.a.string, u"\N{NO-BREAK SPACE}" * 2)
def test_entities_converted_on_the_way_out(self):
text = "
<<sacré bleu!>>
"
expected = u"
<<sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!>>
".encode("utf-8")
soup = self.soup(text)
self.assertEqual(soup.p.encode("utf-8"), expected)
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'
Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!
'
# 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.assertEqual(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'
'
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'
')
unicode_html = shift_jis_html.decode("shift-jis")
soup = self.soup(unicode_html)
# Make sure the parse tree is correctly encoded to various
# encodings.
self.assertEqual(soup.encode("utf-8"), unicode_html.encode("utf-8"))
self.assertEqual(soup.encode("euc_jp"), unicode_html.encode("euc_jp"))
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.
hebrew_document = b'Hebrew (ISO 8859-8) in Visual Directionality
Hebrew (ISO 8859-8) in Visual Directionality
\xed\xe5\xec\xf9'
soup = self.soup(
hebrew_document, from_encoding="iso8859-8")
# Some tree builders call it iso8859-8, others call it iso-8859-9.
# That's not a difference we really care about.
assert soup.original_encoding in ('iso8859-8', 'iso-8859-8')
self.assertEqual(
soup.encode('utf-8'),
hebrew_document.decode("iso8859-8").encode("utf-8"))
def test_meta_tag_reflects_current_encoding(self):
# Here's the tag saying that a document is
# encoded in Shift-JIS.
meta_tag = ('')
# Here's a document incorporating that meta tag.
shift_jis_html = (
'\n%s\n'
''
'Shift-JIS markup goes here.') % meta_tag
soup = self.soup(shift_jis_html)
# Parse the document, and the charset is seemingly unaffected.
parsed_meta = soup.find('meta', {'http-equiv': 'Content-type'})
content = parsed_meta['content']
self.assertEqual('text/html; charset=x-sjis', content)
# But that value is actually a ContentMetaAttributeValue object.
self.assertTrue(isinstance(content, ContentMetaAttributeValue))
# And it will take on a value that reflects its current
# encoding.
self.assertEqual('text/html; charset=utf8', content.encode("utf8"))
# For the rest of the story, see TestSubstitutions in
# test_tree.py.
def test_html5_style_meta_tag_reflects_current_encoding(self):
# Here's the tag saying that a document is
# encoded in Shift-JIS.
meta_tag = ('')
# Here's a document incorporating that meta tag.
shift_jis_html = (
'\n%s\n'
''
'Shift-JIS markup goes here.') % meta_tag
soup = self.soup(shift_jis_html)
# Parse the document, and the charset is seemingly unaffected.
parsed_meta = soup.find('meta', id="encoding")
charset = parsed_meta['charset']
self.assertEqual('x-sjis', charset)
# But that value is actually a CharsetMetaAttributeValue object.
self.assertTrue(isinstance(charset, CharsetMetaAttributeValue))
# And it will take on a value that reflects its current
# encoding.
self.assertEqual('utf8', charset.encode("utf8"))
def test_tag_with_no_attributes_can_have_attributes_added(self):
data = self.soup("text")
data.a['foo'] = 'bar'
self.assertEqual('text', data.a.decode())
def test_worst_case(self):
"""Test the worst case (currently) for linking issues."""
soup = self.soup(BAD_DOCUMENT)
self.linkage_validator(soup)
class XMLTreeBuilderSmokeTest(object):
def test_pickle_and_unpickle_identity(self):
# Pickling a tree, then unpickling it, yields a tree identical
# to the original.
tree = self.soup("foo")
dumped = pickle.dumps(tree, 2)
loaded = pickle.loads(dumped)
self.assertEqual(loaded.__class__, BeautifulSoup)
self.assertEqual(loaded.decode(), tree.decode())
def test_docstring_generated(self):
soup = self.soup("")
self.assertEqual(
soup.encode(), b'\n')
def test_xml_declaration(self):
markup = b"""\n"""
soup = self.soup(markup)
self.assertEqual(markup, soup.encode("utf8"))
def test_processing_instruction(self):
markup = b"""\n"""
soup = self.soup(markup)
self.assertEqual(markup, soup.encode("utf8"))
def test_real_xhtml_document(self):
"""A real XHTML document should come out *exactly* the same as it went in."""
markup = b"""
Hello.
Goodbye.
"""
soup = self.soup(markup)
self.assertEqual(
soup.encode("utf-8"), markup)
def test_nested_namespaces(self):
doc = b"""
"""
soup = self.soup(doc)
self.assertEqual(doc, soup.encode())
def test_formatter_processes_script_tag_for_xml_documents(self):
doc = """
"""
soup = BeautifulSoup(doc, "lxml-xml")
# lxml would have stripped this while parsing, but we can add
# it later.
soup.script.string = 'console.log("< < hey > > ");'
encoded = soup.encode()
self.assertTrue(b"< < hey > >" in encoded)
def test_can_parse_unicode_document(self):
markup = u'Sacr\N{LATIN SMALL LETTER E WITH ACUTE} bleu!'
soup = self.soup(markup)
self.assertEqual(u'Sacr\xe9 bleu!', soup.root.string)
def test_popping_namespaced_tag(self):
markup = 'b2012-07-02T20:33:42Zcd'
soup = self.soup(markup)
self.assertEqual(
unicode(soup.rss), markup)
def test_docstring_includes_correct_encoding(self):
soup = self.soup("")
self.assertEqual(
soup.encode("latin1"),
b'\n')
def test_large_xml_document(self):
"""A large XML document should come out the same as it went in."""
markup = (b'\n'
+ b'0' * (2**12)
+ b'')
soup = self.soup(markup)
self.assertEqual(soup.encode("utf-8"), markup)
def test_tags_are_empty_element_if_and_only_if_they_are_empty(self):
self.assertSoupEquals("
", "
")
self.assertSoupEquals("
foo
")
def test_namespaces_are_preserved(self):
markup = 'This tag is in the a namespaceThis tag is in the b namespace'
soup = self.soup(markup)
root = soup.root
self.assertEqual("http://example.com/", root['xmlns:a'])
self.assertEqual("http://example.net/", root['xmlns:b'])
def test_closing_namespaced_tag(self):
markup = '