undo cdata processing in liberal xml parsing

rubys · rubys · commit e30e89a0d200 · 2007-06-22T13:03:55.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40806
diff --git a/src/liberalxmlparser.py b/src/liberalxmlparser.py
@@ -15,11 +15,12 @@
 """
 
 import html5parser
-from constants import voidElements
+from constants import voidElements, contentModelFlags
 import gettext
 _ = gettext.gettext
 
 from xml.dom import XHTML_NAMESPACE
+from xml.sax.saxutils import unescape
 
 class XMLParser(html5parser.HTMLParser):
     """ liberal XML parser """
@@ -47,6 +48,11 @@ def normalizeToken(self, token):
             if token["data"]:
                self.parseError(_("End tag contains unexpected attributes."))
 
+        elif token["type"] == "Characters":
+            # un-escape rcdataElements (e.g. style, script)
+            if self.tokenizer.contentModelFlag == contentModelFlags["CDATA"]:
+                token["data"] = unescape(token["data"])
+
         elif token["type"] == "Comment":
             # Rescue CDATA from the comments
             if (token["data"].startswith("[CDATA[") and
diff --git a/tests/test_lxp.py b/tests/test_lxp.py
@@ -193,16 +193,29 @@ def test_xlink(self):
 
   def test_br(self):
     self.assertXhtmlEquals("""<html xmlns="http://www.w3.org/1999/xhtml">
-<head><title>XLINK</title></head>
+<head><title>BR</title></head>
 <body>
 <br/>
 </body></html>""")
 
   def test_strong(self):
     self.assertXhtmlEquals("""<html xmlns="http://www.w3.org/1999/xhtml">
-<head><title>XLINK</title></head>
+<head><title>STRONG</title></head>
 <body>
 <strong></strong>
+</body></html>""")
+
+  def test_script(self):
+    self.assertXhtmlEquals("""<html xmlns="http://www.w3.org/1999/xhtml">
+<head><title>SCRIPT</title></head>
+<body>
+<script>1 &lt; 2 &amp; 3</script>
+</body></html>""")
+
+  def test_title(self):
+    self.assertXhtmlEquals("""<html xmlns="http://www.w3.org/1999/xhtml">
+<head><title>1 &lt; 2 &amp; 3</title></head>
+<body>
 </body></html>""")
 
 def buildTestSuite():
