def main(src_dir, dst_fp):
engine, session = get_engine_and_session(dst_fp)
objectIDs = get_objectIDs(session)
print('get_objectIDs done')
filelist = list(yield_filelist(src_dir))
total = len(filelist)
for n, fp in enumerate(filelist):
j = load_json(fp)
hits = j['hits']
insert_hits(hits, objectIDs, engine)
print('{0}/{1}'.format(n + 1, total))
commit(session)
def main(src_dir, dst_fp):
engine, session = get_engine_and_session(dst_fp)
objectIDs = get_objectIDs(session)
print('get_objectIDs done')
filelist = list(yield_filelist(src_dir))
total = len(filelist)
for n, fp in enumerate(filelist):
j = load_json(fp)
hits = j['hits']
is_ok_to_delete = is_this_json_in_db(hits, objectIDs)
if is_ok_to_delete:
os.remove(fp)
print('{0}/{1}'.format(n, total))