Fix liberal XHTML support for <br/>

rubys · rubys · commit 9b21b2ac3b94 · 2007-06-22T01:35:12.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40802
diff --git a/src/liberalxmlparser.py b/src/liberalxmlparser.py
@@ -19,6 +19,8 @@
 import gettext
 _ = gettext.gettext
 
+from xml.dom import XHTML_NAMESPACE
+
 class XMLParser(html5parser.HTMLParser):
     """ liberal XML parser """
 
@@ -66,16 +68,21 @@ def normalizeToken(self, token):
 
         # ensure that non-void XHTML elements have content so that separate
         # open and close tags are emitted
-        if token["type"]  == "EndTag" and \
-            token["name"] not in voidElements and \
-            token["name"] == self.tree.openElements[-1].name and \
-            not self.tree.openElements[-1].hasContent():
-            for e in self.tree.openElements:
-                if 'xmlns' in e.attributes.keys():
-                    if e.attributes['xmlns'] <> 'http://www.w3.org/1999/xhtml':
-                        break
+        if token["type"]  == "EndTag":
+            if token["name"] in voidElements:
+                if not self.tree.openElements or \
+                  self.tree.openElements[-1].name != token["name"]:
+                    token["type"] = "EmptyTag"
+                    if not token.has_key("data"): token["data"] = {}
             else:
-                self.tree.insertText('')
+                if token["name"] == self.tree.openElements[-1].name and \
+                  not self.tree.openElements[-1].hasContent():
+                    for e in self.tree.openElements:
+                        if 'xmlns' in e.attributes.keys():
+                            if e.attributes['xmlns'] != XHTML_NAMESPACE:
+                                break
+                    else:
+                        self.tree.insertText('')
 
         return token
 
diff --git a/tests/test_lxp.py b/tests/test_lxp.py
@@ -196,11 +196,6 @@ def test_br(self):
 <head><title>XLINK</title></head>
 <body>
 <br/>
-</body></html>""",
-"""<html xmlns="http://www.w3.org/1999/xhtml">
-<head><title>XLINK</title></head>
-<body>
-<br/><br/>
 </body></html>""")
 
   def test_strong(self):
