Fix lxml walker for empty fragments

Author: shkyyy18

AUTHORS.rst

@@ -63,4 +63,5 @@ Patches and suggestions
 - Ville Skyttä
 - Hugo van Kemenade
 - Mark Vasilkov
+- shkyyy18
 

CHANGES.rst

@@ -18,6 +18,8 @@ Features:
 
 Bug fixes:
 
+* Allow the lxml treewalker to serialize empty fragments without raising
+  ``IndexError``. (#343)
 * The sanitizer now permits ``<summary>`` tags. It used to allow ``<details>``
   already. (#423)
 

html5lib/tests/test_treewalkers.py

@@ -15,6 +15,7 @@
 
 from html5lib import html5parser, treewalkers
 from html5lib.filters.lint import Filter as Lint
+from html5lib.serializer import HTMLSerializer
 
 import re
 attrlist = re.compile(r"^(\s+)\w+=.*(\n\1\w+=.*)+", re.M)
@@ -123,6 +124,28 @@ def test_fragment_single_char(tree, char):
     assert list(output) == expected
 
 
+@pytest.mark.parametrize("tree,intext",
+                         itertools.product(sorted(treeTypes.items()),
+                                           ["", "<!doctype html>"]))
+def test_fragment_empty(tree, intext):
+    treeName, treeClass = tree
+    if treeClass is None:
+        pytest.skip("Treebuilder not loaded")
+
+    parser = html5parser.HTMLParser(tree=treeClass["builder"])
+    document = parser.parseFragment(intext)
+    document = treeClass.get("adapter", lambda x: x)(document)
+
+    assert list(Lint(treeClass["walker"](document))) == []
+    assert "".join(HTMLSerializer().serialize(treeClass["walker"](document))) == ""
+
+
+@pytest.mark.skipif(treeTypes["lxml"] is None, reason="lxml not importable")
+def test_lxml_empty_list():
+    walker = treewalkers.getTreeWalker('lxml')
+    assert list(Lint(walker([]))) == []
+
+
 @pytest.mark.skipif(treeTypes["lxml"] is None, reason="lxml not importable")
 def test_lxml_xml():
     expected = [

html5lib/treewalkers/etree_lxml.py

@@ -55,7 +55,7 @@ def getnext(self):
         return None
 
     def __len__(self):
-        return 1
+        return len(self.children)
 
 
 class Doctype(object):
@@ -180,11 +180,12 @@ def getNodeDetails(self, node):
     def getFirstChild(self, node):
         assert not isinstance(node, tuple), "Text nodes have no children"
 
-        assert len(node) or node.text, "Node has no children"
         if node.text:
             return (node, "text")
-        else:
+        elif len(node):
             return node[0]
+        else:
+            return None
 
     def getNextSibling(self, node):
         if isinstance(node, tuple):  # Text node