comparison between cffi_re2 and builtin re module

test_cffi_re2.py

#!/usr/bin/env python


import sys
import time


def test_re():
    print 'test python re'
    import re
    arr = []
    with open('regex.txt') as f:
        for line in f:
            s = line.rstrip('\r\n')
            s = s.decode('utf-8')
            arr.append(re.compile(s))

    t0 = time.time()
    count_match = 0
    with open('test_str.txt') as f:
        for line in f:
            line = line.decode('utf-8')

            for sobj in arr:
                m = sobj.search(line)
                if m:
                    count_match += 1


    eltime = time.time() - t0
    print 'python re', eltime
    print 'count match', count_match
    return eltime

def test_cffi():
    print 'test cffi'
    import cffi_re2
    arr = []
    with open('regex.txt') as f:
        for line in f:
            s = line.rstrip('\r\n')
            arr.append(cffi_re2.compile(s))

    t0 = time.time()
    count_match = 0
    with open('test_str.txt') as f:
        for line in f:
            for sobj in arr:
                m = sobj.search(line)
                if m:
                    count_match += 1


    eltime = time.time() - t0
    print 'cffi re2', eltime
    print 'count match', count_match
    return eltime



if __name__ == '__main__':
    repeat = 3
    if len(sys.argv) >= 2 and sys.argv[1] == 'cffi':
        test_cffi()
        total = 0
        for i in xrange(repeat):
            total += test_cffi()
        print 'avg time', total * 1.0 / repeat
    else:
        test_re()
        total = 0
        for i in xrange(repeat):
            total += test_re()
        print 'avg time', total * 1.0 / repeat