diff options
| -rw-r--r-- | bs4/tests/test_html5lib.py | 21 | ||||
| -rw-r--r-- | bs4/tests/test_lxml.py | 26 |
2 files changed, 47 insertions, 0 deletions
diff --git a/bs4/tests/test_html5lib.py b/bs4/tests/test_html5lib.py index 5b1d1e4..d257392 100644 --- a/bs4/tests/test_html5lib.py +++ b/bs4/tests/test_html5lib.py @@ -104,6 +104,27 @@ class TestHTML5BuilderInvalidMarkup(TestLXMLBuilderInvalidMarkup): self.assertSoupEquals("<table><div>Foo</div></table>", "<div>Foo</div><table></table>") + def test_unclosed_a_tag(self): + # n.b. the whitespace is important here. + markup = """<div id="1"> + <a href="foo"> +</div> +<div id="2"> + <div id="3"> + <a href="bar"></a> + </div> +</div>""" + + expect = """<div id="1"> + <a href="foo"> +</a></div><a href="foo"> +</a><div id="2"><a href="foo"> + </a><div id="3"><a href="foo"> + </a><a href="bar"></a> + </div> +</div>""" + self.assertSoupEquals(markup, expect) + def test_incorrectly_nested_tables(self): self.assertSoupEquals( '<table><tr><table><tr id="nested">', diff --git a/bs4/tests/test_lxml.py b/bs4/tests/test_lxml.py index 0adef20..e2cb2d2 100644 --- a/bs4/tests/test_lxml.py +++ b/bs4/tests/test_lxml.py @@ -332,6 +332,32 @@ class TestLXMLBuilderInvalidMarkup(SoupTest): '<tr><td>foo</td></tr>' '</table></td>') + + def test_unclosed_a_tag(self): + # <a> tags really ought to be closed at some point. + # + # We have all the <div> tags because HTML5 says to duplicate + # the <a> tag rather than closing it, and that's what html5lib + # does. + markup = """<div id="1"> + <a href="foo"> +</div> +<div id="2"> + <div id="3"> + <a href="bar"></a> + </div> +</div>""" + + expect = """<div id="1"> +<a href="foo"> +</a></div> +<div id="2"> +<div id="3"> +<a href="bar"></a> +</div> +</div>""" + self.assertSoupEquals(markup, expect) + def test_unclosed_block_level_elements(self): # Unclosed block-level elements should be closed. self.assertSoupEquals( |