def make_doc(s, api_root, docsetid, year, groupid, docfname, url, srcurl, fsize, pagecount, note):
m = YEAR_RE.match(year)
if not m:
raise ValueError("Year %s in metadata for %s doesn't look like a year." % (year, docfname))
year = year_to_western(*m.groups())
obj = {'docset_id': docsetid, 'year': year, 'group_id': groupid, 'filename': docfname,
'url': url, 'srcurl': srcurl, 'size': fsize, 'pages': pagecount, 'note': note}
result = s.post(api_root + 'document', data=json.dumps(obj), verify=False).json()
if 'id' not in result:
raise ValueError('Document add: got back %s' % result)
return result
def title_date(title):
result = DATE_RE.search(title)
if result is None:
import pdb; pdb.set_trace()
raise ValueError("Title didn't match expected format: %s" % title)
(emp, year, month, day) = result.groups()[:4]
year = int(year.strip())
month = int(month.strip())
day = int(day.strip())
year = year_to_western(emp, year)
return date(year, month, day)