tune worldcat parser

2014-08-11 13:27:11 +02:00 · 2014-08-11 13:27:11 +02:00 · 2d3a0b2247
parent a8866ca3b3
commit 2d3a0b2247
1 changed files with 7 additions and 0 deletions
--- 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