def save_user_counts():
"""This saves users and corresponding feature counts to a CSV file."""
statfile = csv.writer(open(GENERAL_STATS, "w"))
full_ones = user_coll.find(
spec={"username": {"$exists": True} },
fields=CONNS.values(),
timeout=False)
# Write CSV header
statfile.writerow(["username"] + CONNS.values() )
for each in full_ones:
row = []
row.append(each["_id"])
for field in CONNS:
row.append(each[CONNS[field]])
statfile.writerow(row)
def save_count_mismatch():
"""Finds and saves users for whom a feature's count does not match the
stored count value."""
field_names = CONNS.keys() + CONNS.values()
# Find users with complete documents
cursor = user_coll.find(
spec={"username": {"$exists": True} },
fields=field_names,
timeout=False)
statfile = csv.writer(open(MISMATCH_STATS, "w"))
# CSV header
statfile.writerow(["username", "connection", "real_count", "stored_count"])
for user in cursor:
for conn in CONNS:
if ( len(user[conn]) != user[CONNS[conn]] ):
row = []
row.append(user["_id"])
row.append(conn)
row.append(len(user[conn]))
row.append(user[CONNS[conn]])
statfile.writerow(row)
user_coll = MONGO_COLLECTION
def correct_counts(conn_list):
for conn in conn_list:
if conn not in CONNS:
print "ERROR: One of the metaconnections given is not valid:", conn
return
cursor = user_coll.find(
spec={"username": {"$exists": True} },
fields=[conn, CONNS[conn]],
timeout=False)
correct_counts = {}
for user in cursor:
real_count = len(user[conn])
if ( real_count != user[CONNS[conn]] ):
correct_counts[user["_id"]] = real_count
for each_id in correct_counts:
user_coll.update(
spec={"_id": each_id},
document={"$set": {CONNS[conn]: correct_counts[each_id]} }
)
if __name__ == "__main__":
import sys
correct_counts(CONNS.keys())
