Sort the unsorted RDFa values in the tests

scrapinghub · Jul 16, 2019 · 3aaebd5 · 3aaebd5
1 parent 8e69411
commit 3aaebd5
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 9 deletions.
diff --git a/extruct/opengraph.py b/extruct/opengraph.py
@@ -30,8 +30,8 @@ def extract_items(self, document, base_url=None):
             namespaces.update(self.get_namespaces(head))
             props = []
             for el in head.xpath('meta[@property and @content]'):
-                prop = el.attrib['property']
-                val = el.attrib['content']
+                prop = el.attrib['property'].strip()
+                val = el.attrib['content'].strip()
                 if prop == '' or val == '':
                     continue
                 ns = prop.partition(':')[0]

diff --git a/tests/samples/songkick/elysianfields.json b/tests/samples/songkick/elysianfields.json
@@ -231,18 +231,18 @@
             ],
             "http://ogp.me/ns#description": [
                 {
-                    "@value": "Buy tickets for an upcoming Elysian Fields concert near you. List of all Elysian Fields tickets and tour dates for 2017."
+                    "@value": ""
                 },
                 {
-                    "@value": ""
+                    "@value": "Buy tickets for an upcoming Elysian Fields concert near you. List of all Elysian Fields tickets and tour dates for 2017."
                 }
             ],
             "http://ogp.me/ns#image": [
                 {
-                    "@value": "http://images.sk-static.com/images/media/img/col4/20100330-103600-169450.jpg"
+                    "@value": "http://images.sk-static.com/SECONDARY_IMAGE.jpg"
                 },
                 {
-                    "@value": "http://images.sk-static.com/SECONDARY_IMAGE.jpg"
+                    "@value": "http://images.sk-static.com/images/media/img/col4/20100330-103600-169450.jpg"
                 }
             ],
             "http://ogp.me/ns#site_name": [

diff --git a/tests/test_extruct.py b/tests/test_extruct.py
@@ -16,9 +16,10 @@ def test_all(self):
         body = get_testdata('songkick', 'elysianfields.html')
         expected = json.loads(get_testdata('songkick', 'elysianfields.json').decode('UTF-8'))
         data = extruct.extract(body, base_url='http://www.songkick.com/artists/236156-elysian-fields')
-        # See test_rdfa_not_preserving_order()
-        del data['rdfa'][0]['http://ogp.me/ns#image']
-        del expected['rdfa'][0]['http://ogp.me/ns#image']
+        for rdf in data['rdfa']:
+            for key, pairs in rdf.items():
+                if ':' in key and isinstance(pairs, list):
+                    rdf[key] = sorted(pairs, key=lambda e: e["@value"])
         self.assertEqual(jsonize_dict(data), expected)
 
     @pytest.mark.xfail