From 2d3a0b22471aefc22da0517f86cc224f61e88c7d Mon Sep 17 00:00:00 2001 From: j Date: Mon, 11 Aug 2014 13:27:11 +0200 Subject: [PATCH] tune worldcat parser --- oml/meta/worldcat.py | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/oml/meta/worldcat.py b/oml/meta/worldcat.py index 1e0fe8c..8940165 100644 --- a/oml/meta/worldcat.py +++ b/oml/meta/worldcat.py @@ -85,6 +85,13 @@ def lookup(id): data['publisher'] = publisher data['date'] = date data['place'] = [place] + elif ':' in data['publisher']: + place, publisher = data['publisher'].split(':', 2) + data['place'] = place.strip() + data['publisher'] = publisher.split(',')[0].strip() + m = re.compile('\d{4}').findall(publisher) + if m: + data['date'] = m[0] logger.debug('lookup %s => %s', id, data.keys()) return data