MySQL 数据库慢查询日志分析脚本

# coding = gbk
__author__ = ‘T_two‘


import datetime
import os


IP = ‘111‘
dirname = os.path.dirname(os.path.abspath(__file__))
# 解析后的目录名
slow_query= os.path.join(dirname, ‘slow_query‘)
# pt-query-digest前的目录的
mysql_data = os.path.join(os.path.join(dirname, ‘mysql_data‘), ‘log‘)
# pt-query-digest后的目录的
tmp = os.path.join(mysql_data, ‘tmp‘)


def getYesterday():
    today=datetime.date.today()
    yesterday = str(today - datetime.timedelta(days=1))
    return yesterday

def getLog(yes_time, slow_query):
    # 对日志进行pt-query-digest分析
    before_name = yes_time.replace(‘-‘, ‘‘) + ‘-‘ + ‘slow-query.log‘
    # pt-query-digest之前的日志 b_filename
    b_filename = os.path.join(mysql_data, before_name)
    # print(b_filename)
    # pt-query-digest之后的日志 a_filename
    after_name = yes_time.replace(‘-‘, ‘‘) + ‘-‘ + IP + ‘-‘ + ‘slow-query.log‘
    a_filename = os.path.join(tmp, after_name)
    # print(a_filename)
    # 最终格式化的日志 e_filename
    end_name = IP + ‘-slow-log-‘ + yes_time + ‘.txt‘
    e_filename = os.path.join(slow_query, end_name)
    #print(e_filename)
    return b_filename,a_filename,e_filename

def getSlowquery(b_filename,a_filename,e_filename):
    print(‘File format starting...‘)
    #os.system(‘pt-query-digest ‘+ b_filename + ‘>‘ + a_filename)

    a_slow_query = open(a_filename, ‘r‘, encoding = ‘utf8‘)
    e_slow_query = open(e_filename, ‘w‘, encoding = ‘utf8‘)
    _line = ‘‘
    line = a_slow_query.readlines()[20:]            # 对文件切片，去除不需要的前20行。
    for line in line:
        line = line.strip()
        # 提取需要的行
        if  line.startswith(‘#‘)   and ‘# Hosts‘ not in line and ‘# Users‘ not in line and ‘# Databases‘ not in line  and  ‘byte‘ not in line                 and ‘# Count‘ not in line  and ‘# Exec time‘ not in line :
            pass
        elif line == ‘‘:
            pass
        else:
            # 序列号
            if  ‘# Query‘ in line: line = (‘\nNO.%s‘ % line.split()[2])
            # 执行次数
            elif ‘# Count‘ in line: line = (‘执行次数: %s‘ % line.split()[3])
            #执行时间
            elif ‘# Exec time‘ in line: line = (‘执行时间 Total: %s  min: %s  max: %s‘ % (line.split()[4],line.split()[5],line.split()[6],))
            # DB
            elif ‘# Databases‘ in line: line = (‘库名: %s‘ % line.split()[2])
            # 源IP
            elif ‘# Host‘ in line:line = (‘源IP: %s‘ % line.split()[2])
            # 用户名
            elif ‘# User‘ in line: line = (‘用户名: %s‘ % line.split()[2])

            _line = _line + line + ‘\n‘

    e_slow_query.write(_line)
    a_slow_query.close()
    e_slow_query.close()
    # 将文件拷贝到web目录下
    os.system(‘cp ‘ + e_filename + ‘ ‘ + web_dir)
    # 删除10天之前的数据
    os.system(‘find ‘ + str(slow_query) + ‘ -mtime +10 | xargs rm -rf ‘)
    os.system(‘find ‘ + mysql_data + ‘ -mtime +10 | xargs rm -rf ‘)
    os.system(‘find ‘ + tmp + ‘ -mtime +10 | xargs rm -rf ‘)


    print (‘File format end...‘)

if __name__ == ‘__main__‘:
    yes_time = getYesterday()
    b_filename,a_filename, e_filename = getLog(yes_time, slow_query)
    getSlowquery(b_filename,a_filename,e_filename)