| Index: trunk/tools/wsor/scripts/fix_reg_date.dumb.py |
| — | — | @@ -1,106 +0,0 @@ |
| 2 | | -import sys, MySQLdb, MySQLdb.cursors, argparse, os, logging, types |
| 3 | | -import wmf |
| 4 | | - |
| 5 | | -def encode(v): |
| 6 | | - if v == None: return "\N" |
| 7 | | - |
| 8 | | - if type(v) == types.LongType: v = int(v) |
| 9 | | - elif type(v) == types.UnicodeType: v = v.encode('utf-8') |
| 10 | | - |
| 11 | | - return str(v).encode("string-escape") |
| 12 | | - |
| 13 | | - |
| 14 | | -def main(): |
| 15 | | - parser = argparse.ArgumentParser( |
| 16 | | - description='Gathers editor data for first and last session' |
| 17 | | - ) |
| 18 | | - parser.add_argument( |
| 19 | | - 'date', |
| 20 | | - type=str, |
| 21 | | - help='the date to start querying for users with dumb registration dates' |
| 22 | | - ) |
| 23 | | - parser.add_argument( |
| 24 | | - '-c', '--cnf', |
| 25 | | - metavar="<path>", |
| 26 | | - type=str, |
| 27 | | - help='the path to MySQL config info (defaults to ~/.my.cnf)', |
| 28 | | - default=os.path.expanduser("~/.my.cnf") |
| 29 | | - ) |
| 30 | | - parser.add_argument( |
| 31 | | - '-s', '--host', |
| 32 | | - type=str, |
| 33 | | - help='the database host to connect to (defaults to localhost)', |
| 34 | | - default="localhost" |
| 35 | | - ) |
| 36 | | - parser.add_argument( |
| 37 | | - '-d', '--db', |
| 38 | | - type=str, |
| 39 | | - help='the language db to run the query in (defaults to enwiki)', |
| 40 | | - default="enwiki" |
| 41 | | - ) |
| 42 | | - args = parser.parse_args() |
| 43 | | - |
| 44 | | - LOGGING_STREAM = sys.stderr |
| 45 | | - logging.basicConfig( |
| 46 | | - level=logging.DEBUG, |
| 47 | | - stream=LOGGING_STREAM, |
| 48 | | - format='%(asctime)s %(levelname)-8s %(message)s', |
| 49 | | - datefmt='%b-%d %H:%M:%S' |
| 50 | | - ) |
| 51 | | - |
| 52 | | - logging.info("Connecting to %s:%s using %s." % (args.host, args.db, args.cnf)) |
| 53 | | - db = Database( |
| 54 | | - host=args.host, |
| 55 | | - db=args.db, |
| 56 | | - read_default_file=args.cnf |
| 57 | | - ) |
| 58 | | - headers = [ |
| 59 | | - 'user_id', |
| 60 | | - 'user_registration' |
| 61 | | - ] |
| 62 | | - |
| 63 | | - lowestDate = args.date |
| 64 | | - logging.info("foo") |
| 65 | | - for user in db.getUsersBefore(args.date): |
| 66 | | - if user['user_registration'] == None: |
| 67 | | - LOGGING_STREAM.write("!") |
| 68 | | - user['user_registration'] = lowestDate |
| 69 | | - print("\t".join(str(user[h]) for h in headers)) |
| 70 | | - else: |
| 71 | | - LOGGING_STREAM.write(".") |
| 72 | | - |
| 73 | | - lowestDate = min(user['user_registration'], lowestDate) |
| 74 | | - |
| 75 | | - LOGGING_STREAM.write("\n") |
| 76 | | - |
| 77 | | - |
| 78 | | - |
| 79 | | - |
| 80 | | - |
| 81 | | -class Database: |
| 82 | | - |
| 83 | | - def __init__(self, *args, **kwargs): |
| 84 | | - self.args = args |
| 85 | | - self.kwargs = kwargs |
| 86 | | - self.usersConn = MySQLdb.connect(*args, **kwargs) |
| 87 | | - |
| 88 | | - def getUsersBefore(self, date): |
| 89 | | - cursor = self.usersConn.cursor(MySQLdb.cursors.SSDictCursor) |
| 90 | | - cursor.execute( |
| 91 | | - """ |
| 92 | | - SELECT |
| 93 | | - user_id, |
| 94 | | - user_registration |
| 95 | | - FROM user |
| 96 | | - WHERE user_registration <= %(date)s |
| 97 | | - OR user_registration IS NULL |
| 98 | | - ORDER BY user_id DESC |
| 99 | | - """, |
| 100 | | - { |
| 101 | | - 'date': date |
| 102 | | - } |
| 103 | | - ) |
| 104 | | - for row in cursor: |
| 105 | | - yield row |
| 106 | | - |
| 107 | | -if __name__ == "__main__": main() |
| Index: trunk/tools/wsor/scripts/user_approx_registration.py |
| — | — | @@ -0,0 +1,106 @@ |
| | 2 | +import sys, MySQLdb, MySQLdb.cursors, argparse, os, logging, types |
| | 3 | +import wmf |
| | 4 | + |
| | 5 | +def encode(v): |
| | 6 | + if v == None: return "\N" |
| | 7 | + |
| | 8 | + if type(v) == types.LongType: v = int(v) |
| | 9 | + elif type(v) == types.UnicodeType: v = v.encode('utf-8') |
| | 10 | + |
| | 11 | + return str(v).encode("string-escape") |
| | 12 | + |
| | 13 | + |
| | 14 | +def main(): |
| | 15 | + parser = argparse.ArgumentParser( |
| | 16 | + description='Gathers approximate registration date by walking ' + |
| | 17 | + 'backwards through the user table and guessing at registration ' + |
| | 18 | + 'dates based on user_id. Assumes user_id is ordered.' |
| | 19 | + ) |
| | 20 | + parser.add_argument( |
| | 21 | + 'date', |
| | 22 | + type=str, |
| | 23 | + help='the date to start querying for users with dumb registration dates' |
| | 24 | + ) |
| | 25 | + parser.add_argument( |
| | 26 | + '-c', '--cnf', |
| | 27 | + metavar="<path>", |
| | 28 | + type=str, |
| | 29 | + help='the path to MySQL config info (defaults to ~/.my.cnf)', |
| | 30 | + default=os.path.expanduser("~/.my.cnf") |
| | 31 | + ) |
| | 32 | + parser.add_argument( |
| | 33 | + '-s', '--host', |
| | 34 | + type=str, |
| | 35 | + help='the database host to connect to (defaults to localhost)', |
| | 36 | + default="localhost" |
| | 37 | + ) |
| | 38 | + parser.add_argument( |
| | 39 | + '-d', '--db', |
| | 40 | + type=str, |
| | 41 | + help='the language db to run the query in (defaults to enwiki)', |
| | 42 | + default="enwiki" |
| | 43 | + ) |
| | 44 | + args = parser.parse_args() |
| | 45 | + |
| | 46 | + LOGGING_STREAM = sys.stderr |
| | 47 | + logging.basicConfig( |
| | 48 | + level=logging.DEBUG, |
| | 49 | + stream=LOGGING_STREAM, |
| | 50 | + format='%(asctime)s %(levelname)-8s %(message)s', |
| | 51 | + datefmt='%b-%d %H:%M:%S' |
| | 52 | + ) |
| | 53 | + |
| | 54 | + logging.info("Connecting to %s:%s using %s." % (args.host, args.db, args.cnf)) |
| | 55 | + db = Database( |
| | 56 | + host=args.host, |
| | 57 | + db=args.db, |
| | 58 | + read_default_file=args.cnf |
| | 59 | + ) |
| | 60 | + headers = [ |
| | 61 | + 'user_id', |
| | 62 | + 'user_registration' |
| | 63 | + ] |
| | 64 | + |
| | 65 | + lowestDate = args.date |
| | 66 | + for user in db.getUsersBefore(args.date): |
| | 67 | + if user['user_registration'] == None: |
| | 68 | + LOGGING_STREAM.write("!") |
| | 69 | + user['user_registration'] = lowestDate |
| | 70 | + print("\t".join(str(user[h]) for h in headers)) |
| | 71 | + else: |
| | 72 | + LOGGING_STREAM.write(".") |
| | 73 | + |
| | 74 | + lowestDate = min(user['user_registration'], lowestDate) |
| | 75 | + |
| | 76 | + LOGGING_STREAM.write("\n") |
| | 77 | + |
| | 78 | + |
| | 79 | + |
| | 80 | + |
| | 81 | +class Database: |
| | 82 | + |
| | 83 | + def __init__(self, *args, **kwargs): |
| | 84 | + self.args = args |
| | 85 | + self.kwargs = kwargs |
| | 86 | + self.usersConn = MySQLdb.connect(*args, **kwargs) |
| | 87 | + |
| | 88 | + def getUsersBefore(self, date): |
| | 89 | + cursor = self.usersConn.cursor(MySQLdb.cursors.SSDictCursor) |
| | 90 | + cursor.execute( |
| | 91 | + """ |
| | 92 | + SELECT |
| | 93 | + user_id, |
| | 94 | + user_registration |
| | 95 | + FROM user |
| | 96 | + WHERE user_registration <= %(date)s |
| | 97 | + OR user_registration IS NULL |
| | 98 | + ORDER BY user_id DESC |
| | 99 | + """, |
| | 100 | + { |
| | 101 | + 'date': date |
| | 102 | + } |
| | 103 | + ) |
| | 104 | + for row in cursor: |
| | 105 | + yield row |
| | 106 | + |
| | 107 | +if __name__ == "__main__": main() |