def parse(self, log: ProcessLogger, text, text_unit_id, _text_unit_lang, **kwargs) -> ParseResults:
found = [u.url for u in urls.get_url_annotations(text)]
if found:
unique = set(found)
return ParseResults({UrlUsage: [UrlUsage(text_unit_id=text_unit_id,
source_url=item,
count=found.count(item)) for item in unique]})
def test_ratios(self):
text = "I've been banned on www.google.com :("
ds = list(get_urls(text))
self.assertEqual(1, len(ds))
self.assertEqual('www.google.com', ds[0])
ants = list(get_url_annotations(text))
self.assertEqual(1, len(ds))
self.assertEqual('en', ants[0].locale)
self.assertEqual('www.google.com', ants[0].url)
