Merge branch 'master' into re-replace-entities-switch

Author: dangra

parsel/utils.py

@@ -65,12 +65,19 @@ def extract_regex(regex, text, replace_entities=True):
     if isinstance(regex, six.string_types):
         regex = re.compile(regex, re.UNICODE)
 
-    try:
-        strings = [regex.search(text).group('extract')]   # named group
-    except:
-        strings = regex.findall(text)    # full regex or numbered groups
+    if 'extract' in regex.groupindex:
+        # named group
+        try:
+            extracted = regex.search(text).group('extract')
+        except AttributeError:
+            strings = []
+        else:
+            strings = [extracted] if extracted is not None else []
+    else:
+        # full regex or numbered groups
+        strings = regex.findall(text)
+
     strings = flatten(strings)
     if not replace_entities:
         return strings
-    return [w3lib_replace_entities(s, keep=['lt', 'amp']) for s in strings]
-
+    return [w3lib_replace_entities(s, keep=['lt', 'amp']) for s in strings]

tests/test_selector.py

@@ -455,11 +455,20 @@ def test_re(self):
                          ["John", "Paul"])
         self.assertEqual(x.xpath("//ul/li").re("Age: (\d+)"),
                          ["10", "20"])
+        
+        # Test named group, hit and miss
+        x = self.sscls(text=u'foobar')
+        self.assertEqual(x.re('(?P<extract>foo)'), ['foo'])
+        self.assertEqual(x.re('(?P<extract>baz)'), [])
+
+        # A purposely constructed test for an edge case
+        x = self.sscls(text=u'baz')
+        self.assertEqual(x.re('(?P<extract>foo)|(?P<bar>baz)'), [])
 
     def test_re_replace_entities(self):
         body = u"""<script>{"foo":"bar &amp; &quot;baz&quot;"}</script>"""
         x = self.sscls(text=body)
-
+        
         name_re = re.compile('{"foo":(.*)}')
 
         # by default, only &amp; and &lt; are preserved ;