解析一个912M(8e6行)的日志文本文件，同样的逻辑Python用了4'05''， C程序用了14''. 性能还是有差距的啊！

log_ana.c

#include <stdio.h>
#include <unistd.h>
#include <string.h>
#include <time.h>
#include <stdlib.h>

#define MAXLINE 1024
#define DATE_FORMAT "%Y-%m-%d %H:%M:%S"
#define GAP_IN_SEC 2

int main(int argc, char* argv[])
{
    struct tm tm;
    time_t ts0=0, ts1=0;
    char line[MAXLINE];
    unsigned long lineno=0;

    while (NULL!=fgets(line, MAXLINE, stdin)) {
        ++lineno;
        if (NULL!=strptime(line+5, DATE_FORMAT, &tm)) {
            ts1 = mktime(&tm);
            if (ts1 > ts0+GAP_IN_SEC && ts0>0) {
                printf("%6lu %s", lineno, line);
            }
            ts0 = ts1;
        } else {
            // printf("cant parse time\n");
        }
    }
    if (ferror(stdin)) {
        perror("fgets error");
        exit(1);
    }
    printf("\n");

    return 0;
}

log_ana.py

#-*- coding:utf8 -*-
from __future__ import print_function

import sys
import time


def filter(file):
    lineno = 0
    ts0 = 0
    doprint = False
    for line in file:
        lineno += 1
        ts = time.mktime(time.strptime(line[5:24], '%Y-%m-%d %H:%M:%S'))
        if ts > ts0+2 and ts0>0:
            print('%6d %s' % (lineno, line), end='')
        ts0 = ts


if __name__ == '__main__':
    filter(sys.stdin)

time 结果对比
Python:

real 3m42.473s
user 3m30.885s
sys 0m11.081s

C:

real 0m16.278s
user 0m10.465s
sys 0m5.744s