from beautifulsoup.builder.html5lib_builder import HTML5TreeBuilder from test_lxml import ( TestLXMLBuilder, TestLXMLBuilderInvalidMarkup, ) class TestHTML5Builder(TestLXMLBuilder): """See `BuilderSmokeTest`.""" @property def default_builder(self): return HTML5TreeBuilder() 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

tag around the bare string. self.assertSoupEquals( "A bare string", "A bare string") def test_correctly_nested_tables(self): markup = ('' '' "') self.assertSoupEquals( markup, '
Here's another table:" '' '' '
foo
Here\'s another table:' '
foo
' '
') self.assertSoupEquals( "" "" "
Foo
Bar
Baz
") def test_collapsed_whitespace(self): """Whitespace is preserved even in tags that don't require it.""" self.assertSoupEquals("

Foo
Bar
Baz

") self.assertSoupEquals(" ") def test_cdata(self): print self.soup("

") class TestHTML5BuilderInvalidMarkup(TestLXMLBuilderInvalidMarkup): """See `BuilderInvalidMarkupSmokeTest`.""" @property def default_builder(self): return HTML5TreeBuilder() def test_unclosed_block_level_elements(self): # The unclosed tag is closed so that the block-level tag # can be closed, and another tag is inserted after the # next block-level tag begins. self.assertSoupEquals( '
Foo
Bar', '
Foo
Bar
') def test_table_containing_bare_markup(self): # Markup should be in table cells, not directly in the table. self.assertSoupEquals("
Foo
", "
Foo
") def test_incorrectly_nested_tables(self): self.assertSoupEquals( '
', ('
' '
')) def test_doctype_in_body(self): markup = "
onetwo
" self.assertSoupEquals(markup, "
onetwo
") def test_foo(self): isolatin = """Sacr\xe9 bleu!""" soup = self.soup(isolatin) utf8 = isolatin.replace("ISO-Latin-1".encode(), "utf-8".encode()) utf8 = utf8.replace("\xe9", "\xc3\xa9") #print soup