def update_database(self):
filenames = get_scanned_filenames(self.directory)
latest = self.get_latest()
for filename in filenames:
dt = datetime_from_pdf_filename(filename)
if latest is None or dt > latest.created:
q = self.session.query(ScannedDocument)
q = q.filter_by(name=filename)
try:
q.one()
except NoResultFound:
self.insert_scanned_file(filename)
def insert_scanned_file(self, filename):
fullname = os.path.join(self.directory, filename)
# file must fit in memory
content = file(fullname).read()
created = datetime_from_pdf_filename(filename)
with transaction.manager:
f = File()
f.content = content
self.session.add(f)
f = self.session.merge(f)
s = ScannedDocument()
s.created = created
s.name = filename
s.file_id = f.id
self.session.add(s)
s = self.session.merge(s)
return s