Changeset 3189 in products for quintagroup.seoptimizer/trunk/quintagroup/seoptimizer/util.py

Timestamp:

Apr 28, 2011 1:52:45 PM (13 years ago)

Author:

vmaksymiv

Message:

added unescaping functionality

File:

• quintagroup.seoptimizer/trunk/quintagroup/seoptimizer/util.py (modified) (2 diffs)

quintagroup.seoptimizer/trunk/quintagroup/seoptimizer/util.py

@@ -1 +1 @@
 from AccessControl import ClassSecurityInfo
+from htmlentitydefs import entitydefs
+import re
 
 try:
@@ -45 +47 @@
 except:
     pass
+
+
+def _group_unescape(m):
+    if m.group("ent"):
+        try:
+            return entitydefs[m.group("ent")]
+        except KeyError:
+            return m.group(0)
+    if m.group("dec"):
+        return unichr(int(m.group("dec")))
+    if m.group("hex"):
+        return unichr(int(m.group("hex"), 16))
+
+expr = re.compile(r'&(?:(?P<ent>\w+?)|'\
+                   '#(?P<dec>\d{1,10})|'\
+                   '#x(?P<hex>[0-9a-fA-F]{1,8}));')
+
+
+def unescape(s):
+    result = expr.sub(_group_unescape, s)
+
+    if isinstance(s, unicode):
+        return result
+    else:
+        return unicode(result)