diff --git a/main.py b/main.py index da56b90..2289726 100644 --- a/main.py +++ b/main.py @@ -90,7 +90,7 @@ def get_date(text: str): return res res = re.search( - r"([j|J]anuary|[f|F]ebruary|[m|M]arch|[A|a]pril|[M|m]ay|[J|j]une|[J|j]uly|[A|a]ugust|[S|s]eptember|[O|o]ctober|[n|N]ovember|[d|D]ecember)[,\s|\s]+(0[1-9]|[1-2][0-9]|3[0-1])[,\s|\s|,]+(19[0-9][0-9]|20[0-9][0-9])", text) + r"([j|J]anuary|[f|F]ebruary|[m|M]arch|[A|a]pril|[M|m]ay|[J|j]une|[J|j]uly|[A|a]ugust|[S|s]eptember|[O|o]ctober|[n|N]ovember|[d|D]ecember),?\s*(0[1-9]|[1-2][0-9]|3[0-1]),?\s*(19[0-9][0-9]|20[0-9][0-9])", text) return res diff --git a/train/out.tsv b/train/out.tsv index 632c1bb..02aba1d 100644 --- a/train/out.tsv +++ b/train/out.tsv @@ -135,7 +135,7 @@ jurisdiction=Wisconsin jurisdiction=Washington effective_date=2006-04-13 party=Burlington_Coat_Factory_Holdings_Inc. - effective_date=2015-05-31 jurisdiction=California + effective_date=2014-07-24 jurisdiction=California effective_date=2013-12-20 jurisdiction=Delaware jurisdiction=Massachusetts