Some refactorings, also tabs replaced to spaces

This commit is contained in:
2016-02-28 02:42:23 +03:00
parent 30dbc8e16c
commit 614bf2c0d4

View File

@@ -23,36 +23,36 @@ ids = set()
idsFile = sys.argv[3] idsFile = sys.argv[3]
if not os.path.isfile(idsFile): if not os.path.isfile(idsFile):
idsRegex = re.compile('\\"_id\\": (.+?),') idsRegex = re.compile('\\"_id\\": (.+?),')
with open(usersJsonFile, "r") as file_name: with open(usersJsonFile, "r") as file_name:
for line in file_name: for line in file_name:
groups = idsRegex.search(line) groups = idsRegex.search(line)
uid = int(groups.group(1)) uid = int(groups.group(1))
ids.add(uid) ids.add(uid)
with open(idsFile, "w") as file_name: with open(idsFile, "w") as file_name:
for uid in ids: for uid in ids:
file_name.write(str(uid) + "\n") file_name.write(str(uid) + "\n")
else: else:
with open(idsFile, "r") as f_ids: with open(idsFile, "r") as f_ids:
for line in f_ids: for line in f_ids:
f_ids.add(int(line)) f_ids.add(int(line))
for file_name in os.listdir(dirWithIds): for file_name in os.listdir(dirWithIds):
print("parsing {}".format(file_name)) print("parsing {}".format(file_name))
with open(os.path.join(dirWithIds, file_name), "r") as file: with open(os.path.join(dirWithIds, file_name), "r") as f_out:
for line in file: for line in f_out:
id = int(line) uid = int(line)
if id in ids: if uid in ids:
pazan = pazansCollection.find_one(id) pazan = pazansCollection.find_one(uid)
if pazan is None: if pazan is None:
pazansCollection.insert_one({"_id": id, "groups": [file_name]}) pazansCollection.insert_one({"_id": uid, "groups": [file_name]})
elif file_name not in pazan["groups"]: elif file_name not in pazan["groups"]:
pazan["groups"].append(file_name) pazan["groups"].append(file_name)
pazansCollection.update_one( pazansCollection.update_one(
{"_id": pazan["_id"]}, {"_id": pazan["_id"]},
{"$set": {"groups": pazan["groups"]}} {"$set": {"groups": pazan["groups"]}}
) )
print("- done") print("- done")