Takes a log file with multipe SELECT query that was split across multiple lines, and extracts the SELECT query and prints it out on stdout.

select_query_extractor.py

"""
******************************************************************
WARNING: THIS DOES NOT TAKE INTO ACCOUNT NESTED SELECT QUERIES.
******************************************************************

Select Query query_extractor
============================

Takes a log file with multipe SELECT query that was split across
multiple lines, and extracts the SELECT query and prints it out on
stdout.

Usage:

    select_query_extractor.py <filename>


******************************************************************
WARNING: THIS DOES NOT TAKE INTO ACCOUNT NESTED SELECT QUERIES.
******************************************************************

"""

import os
import sys
from argparse import ArgumentParser

query_keywords = ("SELECT", "FROM", "WHERE", "LIMIT", "ORDER BY" )

def query_extractor(line):

    is_query = False
    for keyword in query_keywords:
        if keyword in line:
            is_query = True
            line = line[line.find(keyword):].strip()

    return line if is_query else ''

def runner(filepath):

    with open(filepath) as f:
        
        lines = f.readlines()
        extracted = map(query_extractor, lines)
        for line in extracted:
            if line:

                if query_keywords[0] in line:
                    print "\n===================="
                
                sys.stdout.write(" %s " % line)

    print "\n===================="



if __name__ == "__main__":
    
    parser = ArgumentParser(description='Takes a log file and returns out SELECT query information')
    parser.add_argument('filepath', type=str)
    args = parser.parse_args()

    filepath = os.path.abspath(args.filepath)
    runner(filepath)