Merge pull request #164 from sortafreel/fix_extract_regex_function

Add tests for `extract_regex` function.

tests/test_utils.py

@@ -1,4 +1,4 @@
-from parsel.utils import shorten
+from parsel.utils import shorten, extract_regex
 
 from pytest import mark, raises
 import six
@@ -24,3 +24,15 @@ def test_shorten(width, expected):
     else:
         with raises(expected):
             shorten(u'foobar', width)
+
+
+@mark.parametrize('regex, text, replace_entities, expected', (
+    [r'(?P<month>\w+)\s*(?P<day>\d+)\s*\,?\s*(?P<year>\d+)', 'October  25, 2019', True, ['October', '25', '2019']],
+    [r'(?P<month>\w+)\s*(?P<day>\d+)\s*\,?\s*(?P<year>\d+)', 'October  25 2019', True, ['October', '25', '2019']],
+    [r'(?P<extract>\w+)\s*(?P<day>\d+)\s*\,?\s*(?P<year>\d+)', 'October  25 2019', True, ['October']],
+    [r'\w+\s*\d+\s*\,?\s*\d+', 'October  25 2019', True, ['October  25 2019']],
+    [r'^.*$', '&quot;sometext&quot; &amp; &quot;moretext&quot;', True, ['"sometext" &amp; "moretext"']],
+    [r'^.*$', '&quot;sometext&quot; &amp; &quot;moretext&quot;', False, ['&quot;sometext&quot; &amp; &quot;moretext&quot;']],
+))
+def test_extract_regex(regex, text, replace_entities, expected):
+    assert extract_regex(regex, text, replace_entities) == expected