gistfile1.py

import re
import sys
import time

__author__ = 'altmind'

import MySQLdb

conn = MySQLdb.connect(host="localhost",
                                           user="root",
                                           passwd="",
                                           db="test")


def processReplacement(replacement, row, cursor):
        cursor.execute("UPDATE PROF SET INFO_PROCESSED=%s, PROCESSED=1 WHERE ID=%s",(replacement, row["ID"]))

def repl(m):
        return "?"

def processQuery(subject):
        subject = re.sub('"', "'", subject)
        subject = re.sub(r"\s+", " ", subject)
        subject = re.sub(r"(?i)(\s*in\s*)\([^)]*(?:\)|$)|(is)\s*(?:not)?\s*null|(?:_[a-z0-9]+)?'([^'\\]|\\.)*?'|(?:[^a-z_0-9])\d+", repl, subject)
        return re.sub(r"\?{2,}", "?", subject)



while(True):
        cursor = conn.cursor(MySQLdb.cursors.DictCursor)
        cursor.execute("SELECT * from PROF WHERE PROCESSED=0")
        result_set = cursor.fetchall()
        for row in result_set:
                replacement = processQuery(row["INFO"])
                processReplacement(replacement, row, cursor)
                print row["ID"]
                sys.stdout.flush()
        cursor.close()
        conn.commit()
        time.sleep(1)
conn.close()