Merge pull request #121 from scrapinghub/skip_empty_on_flattening_og

OpenGraph skipping empty for properties with values on flattening
scrapinghub · Jul 26, 2019 · f6ee940 · f6ee940
2 parents 6df8e19 + 5fa98cb
commit f6ee940
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 1 deletion.
diff --git a/extruct/uniform.py b/extruct/uniform.py
@@ -4,7 +4,13 @@
 def _uopengraph(extracted):
     out = []
     for obj in extracted:
-        flattened = dict(reversed(obj['properties']))
+        # In order of appearance in the page
+        properties = list(reversed(obj['properties']))
+        # Ensuring that never empty value is returned if there is a duplicated
+        # property with non empty value
+        non_empty_props = {k for k, v in properties if v and v.strip()}
+        flattened = {k: v for k, v in properties
+                     if k not in non_empty_props or (v and v.strip())}
         t = flattened.pop('og:type', None)
         if t:
             flattened['@type'] = t

diff --git a/tests/test_uniform.py b/tests/test_uniform.py
@@ -37,6 +37,28 @@ def test_uopengraph_duplicated_priorities(self):
         for k in range(5):
             assert data[0]['prop_{}'.format(k)] == 'value_0'
 
+        # Ensures that empty is not returned if a property contains any
+        # non empty value
+        data = _uopengraph([{'properties':
+                                 [('prop_empty', ' '),
+
+                                  ('prop_non_empty', ' '),
+                                  ('prop_non_empty', 'value!'),
+
+                                  ('prop_non_empty2', 'value!'),
+                                  ('prop_non_empty2', ' '),
+
+                                  ('prop_non_empty3', ' '),
+                                  ('prop_non_empty3', 'value!'),
+                                  ('prop_non_empty3', 'other value'),
+                                  ],
+                             'namespace': 'namespace'}])
+        assert data[0]['prop_empty'] == ' '
+        assert data[0]['prop_non_empty'] == 'value!'
+        assert data[0]['prop_non_empty2'] == 'value!'
+        assert data[0]['prop_non_empty3'] == 'value!'
+
+
     def test_umicroformat(self):
         expected = [ { '@context': 'http://microformats.org/wiki/',
                      '@type': ['h-hidden-phone', 'h-hidden-tablet'],