src/core/wp_utils.py
changeset 63 a7f4a418d5af
parent 61 0048668779c0
child 67 5d9223bb3aab
--- a/src/core/wp_utils.py	Wed Jul 10 17:34:14 2013 +0200
+++ b/src/core/wp_utils.py	Wed Jul 10 18:47:44 2013 +0200
@@ -1,10 +1,12 @@
 # -*- coding: utf-8 -*-
 from .models import Term, TermLabel, TERM_URL_STATUS_DICT
+from core.models.term import TERM_WK_LINK_SEMANTIC_LEVEL_DICT
 from django.conf import settings
 from django.utils.http import urlquote
+from urlparse import urlparse
 from wikitools import api, wiki
 import logging
-from core.models.term import TERM_WK_LINK_SEMANTIC_LEVEL_DICT
+import urllib2
 
 logger = logging.getLogger(__name__)
 site = None
@@ -172,6 +174,10 @@
         
     if not label:
         label = term.label
+    elif label.startswith(settings.WIKIPEDIA_PAGE_URL):
+        url_parts = urlparse(label)
+        label = urllib2.unquote(str(url_parts.path.split('/')[-1])).decode("utf-8")
+
     wp_res = query_wikipedia_title(site, label=label)
     new_label = wp_res['new_label']
     alternative_label= wp_res['alternative_label']