summaryrefslogtreecommitdiff
path: root/bs4/tests/test_tree.py
diff options
context:
space:
mode:
Diffstat (limited to 'bs4/tests/test_tree.py')
-rw-r--r--bs4/tests/test_tree.py42
1 files changed, 42 insertions, 0 deletions
diff --git a/bs4/tests/test_tree.py b/bs4/tests/test_tree.py
index 6510f85..e655dcc 100644
--- a/bs4/tests/test_tree.py
+++ b/bs4/tests/test_tree.py
@@ -417,6 +417,48 @@ class TestFindAllByAttribute(TreeTest):
self.assertEqual([], soup.find_all(id=1, text="bar"))
+class TestSmooth(TreeTest):
+ """Test Tag.smooth."""
+
+ def test_smooth(self):
+ soup = self.soup("<div>a</div>")
+ div = soup.div
+ div.append("b")
+ div.append("c")
+ div.append(Comment("Comment 1"))
+ div.append(Comment("Comment 2"))
+ div.append("d")
+ builder = self.default_builder()
+ span = Tag(soup, builder, 'span')
+ span.append('1')
+ span.append('2')
+ div.append(span)
+
+ # At this point the tree has a bunch of adjacent
+ # NavigableStrings. This is normal, but it has no meaning in
+ # terms of HTML, so we may want to smooth things out for
+ # output.
+
+ # Since the <span> tag has two children, its .string is None.
+ self.assertEquals(None, div.span.string)
+
+ self.assertEqual(7, len(div.contents))
+ div.smooth()
+ self.assertEqual(5, len(div.contents))
+
+ # The three strings at the beginning of div.contents have been
+ # merged into on string.
+ #
+ self.assertEqual('abc', div.contents[0])
+
+ # The call is recursive -- the <span> tag was also smoothed.
+ self.assertEqual('12', div.span.string)
+
+ # The two comments have _not_ been merged, even though
+ # comments are strings. Merging comments would change the
+ # meaning of the HTML.
+ self.assertEqual('Comment 1', div.contents[1])
+ self.assertEqual('Comment 2', div.contents[2])
class TestIndex(TreeTest):