Added tests illustrating the different ways lxml and html5lib handle nested tables.

2 files changed, 72 insertions, 0 deletions

diff --git a/tests/test_html5lib.py b/tests/test_html5lib.py
index 3a4ee27..ef38f9f 100644
--- a/tests/test_html5lib.py
+++ b/tests/test_html5lib.py
@@ -19,6 +19,32 @@ class TestHTML5Builder(TestLXMLBuilder):
         self.assertSoupEquals(
             "A bare string", "A bare string")
 
+    def test_nested_tables(self):
+        # See TestLXMLBuilder for TABLE_MARKUP_1 and
+        # TABLE_MARKUP_2. They're both nested tables where the
+        # top-level <table> and <tr> aren't closed. In TABLE_MARKUP_1
+        # the second table is within a <td> tag. In
+        # TABLE_MARKUP_2, the second table is floating inside a <tr> tag.
+        #
+        # html5lib adds <tbody> tags to each table. It treats
+        # TABLE_MARKUP_1 as a nested table, and TABLE_MARKUP_2 as two
+        # different tables.
+        self.assertSoupEquals(
+            self.TABLE_MARKUP_1,
+            '<table id="1"><tbody>'
+            "<tr><td>Here's another table:"
+            '<table id="2"><tbody><tr><td>foo</td></tr></tbody></table>'
+            "</td></tr></tbody></table>"
+            )
+
+        self.assertSoupEquals(
+            self.TABLE_MARKUP_2,
+            '<table id="1"><tbody>'
+            "<tr><td>Here's another table:</td></tr>"
+            '</tbody></table>'
+            '<table id="2"><tbody><tr><td>foo</td></tr></tbody></table>'
+            )
+
     def test_collapsed_whitespace(self):
         """Whitespace is preserved even in tags that don't require it."""
         self.assertSoupEquals("<p>   </p>")
diff --git a/tests/test_lxml.py b/tests/test_lxml.py
index 455c953..76bcd32 100644
--- a/tests/test_lxml.py
+++ b/tests/test_lxml.py
@@ -1,5 +1,7 @@
 """Tests to ensure that the lxml tree builder generates good trees."""
 
+import re
+
 from beautifulsoup import BeautifulSoup
 from beautifulsoup.builder.lxml_builder import LXMLTreeBuilder
 from beautifulsoup.element import Comment
@@ -65,6 +67,50 @@ class TestLXMLBuilder(SoupTest):
         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>')
+
+    # This is the same as TABLE_MARKUP_1, but the nested table is
+    # floating freely rather than being inside a <td> cell.
+    TABLE_MARKUP_2 = ('<table id="1">'
+                     '<tr>'
+                     "<td>Here's another table:</td>"
+                     '<table id="2">'
+                     '<tr><td>foo</td></tr>'
+                     '</table></td>')
+
+
+    def test_nested_tables(self):
+        # lxml closes the <tr> and <table> tags that weren't closed by
+        # TABLE_MARKUP. Unlike html5lib, it treats both bits of markup
+        # as nested tables.
+        self.assertSoupEquals(
+            self.TABLE_MARKUP_1,
+            '<table id="1">'
+            '<tr>'
+            "<td>Here's another table:"
+            '<table id="2">'
+            '<tr><td>foo</td></tr>'
+            '</table>'
+            '</td></tr></table>')
+
+        self.assertSoupEquals(
+            self.TABLE_MARKUP_2,
+            '<table id="1">'
+            '<tr>'
+            "<td>Here's another table:</td>"
+            '<table id="2">'
+            '<tr><td>foo</td></tr>'
+            '</table>'
+            '</tr></table>')
+
+
     def test_collapsed_whitespace(self):
         """In most tags, whitespace is collapsed."""
         self.assertSoupEquals("<p>   </p>", "<p> </p>")