from beautifulsoup.builder.html5lib_builder import HTML5TreeBuilder from beautifulsoup.testing import ( BuilderInvalidMarkupSmokeTest, BuilderSmokeTest, ) class TestHTML5Builder(BuilderSmokeTest): """See `BuilderSmokeTest`.""" def setUp(self): self.default_builder = HTML5TreeBuilder() def test_collapsed_whitespace(self): """Whitespace is preserved even in tags that don't require it.""" self.assertSoupEquals("

") self.assertSoupEquals(" ") class TestHTML5BuilderInvalidMarkup(BuilderInvalidMarkupSmokeTest): """See `BuilderInvalidMarkupSmokeTest`.""" def setUp(self): self.default_builder = 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_incorrectly_nested_tables(self): self.assertSoupEquals( '
', ('
' '
')) 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