1 """

     2 usage: python exlog.py < rql.log

     3 

     4 will print out the following table

     5 

     6   total execution time || number of occurences || rql query

     7 

     8 sorted by descending total execution time

     9 

    10 chances are the lines at the top are the ones that will bring

    11 the higher benefit after optimisation. Start there.

    12 """

    13 import sys, re

    14 

    15 def run():

    16     requests = {}

    17     for line in sys.stdin:

    18         if not ' WHERE ' in line:

    19             continue

    20         #sys.stderr.write( line )

    21         rql, time = line.split('--')

    22         rql = re.sub("(\'\w+': \d*)", '', rql)

    23         req = requests.setdefault(rql, [])

    24         time.strip()

    25         chunks = time.split()

    26         cputime = float(chunks[-3])

    27         req.append( cputime )

    28 

    29     stat = []

    30     for rql, times in requests.items():

    31         stat.append( (sum(times), len(times), rql) )

    32 

    33     stat.sort()

    34     stat.reverse()

    35     for time, occ, rql in stat:

    36         print time, occ, rql

    37 

    38 if __name__ == '__main__':

    39     run()