def doc_op(self, doc, action='insert'):
if action == 'insert':
doc['_id'] = str(uuid.uuid1()).decode()
doc = self.integrity(doc, action)
path = self.gen_path(doc)
ext = ExtHandler.get_ext(path)
# file operation
self.fhdict[ext].operate(path, doc, action)
# index operation
with self.ix.writer() as writer:
self.index_doc(doc, writer, action)
def init(self, clean=False):
with self.ix.writer() as writer:
count = 0
for root, dirnames, filenames in os.walk(self.dir):
for filename in filenames:
ext = ExtHandler.get_ext(filename)
if ext not in self.ext:
continue
self.index_file(os.path.join(root, filename), writer)
count += 1
if count % 1000 == 0:
utils.logger.info('create index for %d files' % count)
writer.mergetype = writing.CLEAR # re-index without disturbing current users
def index_file(self, path, writer):
ext = ExtHandler.get_ext(path)
for doc in self.fhdict[ext].iter(path):
self.index_doc(doc, writer, 'insert')
