Python脚本实现下载合并SAE日志

由于一些原因，需要SAE上站点的日志文件，从SAE上只能按天下载，下载下来手动处理比较蛋疼，尤其是数量很大的时候。还好SAE提供了API可以批量获得日志文件下载地址，刚刚写了python脚本自动下载和合并这些文件

调用API获得下载地址

文档位置在这里

设置自己的应用和下载参数

请求中需要设置的变量如下

api_url = 'http://dloadcenter.sae.sina.com.cn/interapi.php?'

appname = 'xxxxx'

from_date = '20140101'

to_date = '20140116'

url_type = 'http' # http|taskqueue|cron|mail|rdc

url_type2 = 'access' # only when type=http  access|debug|error|warning|notice|resources

secret_key = 'xxxxx'

生成请求地址

请求地址生成方式可以看一下官网的要求：

1.将参数排序
2.生成请求字符串，去掉&
3.附加access_key
4.请求字符串求md5，形成sign
5.把sign增加到请求字符串中

具体实现代码如下

params = dict()

params['act'] = 'log'

params['appname'] = appname

params['from'] = from_date

params['to'] = to_date

params['type'] = url_type
if url_type == 'http':

    params['type2'] = url_type2
params = collections.OrderedDict(sorted(params.items()))
request = ''

for k,v in params.iteritems():

    request += k+'='+v+'&'
sign = request.replace('&','')

sign += secret_key
md5 = hashlib.md5()

md5.update(sign)

sign = md5.hexdigest()
request = api_url + request + 'sign=' + sign
if response['errno'] != 0:

    print '[!] '+response['errmsg']

    exit()
print '[#] request success'

下载日志文件

SAE将每天的日志文件都打包成tar.gz的格式，下载保存下来即可，文件名以日期.tar.gz命名

log_files = list()
for down_url in response['data']:    

    file_name = re.compile(r'\d{4}-\d{2}-\d{2}').findall(down_url)[0] + '.tar.gz'

    log_files.append(file_name)

    data = urllib2.urlopen(down_url).read()

    with open(file_name, "wb") as file:

        file.write(data)
print '[#] you got %d log files' % len(log_files)

合并文件

合并文件方式用trafile库解压缩每个文件，然后把文件内容附加到access_log下就可以了

# compress these files to access_log

access_log = open('access_log','w');
for log_file in log_files:

    tar = tarfile.open(log_file)

    log_name = tar.getnames()[0]

    tar.extract(log_name)

    # save to access_log

    data = open(log_name).read()

    access_log.write(data)

    os.remove(log_name)
print '[#] all file has writen to access_log'

完整代码

#!/usr/bin/env python

# -*- coding: utf-8 -*-

# @Author: Su Yan <http://yansu.org>

# @Date:   2014-01-17 12:05:19

# @Last Modified by:   Su Yan

# @Last Modified time: 2014-01-17 14:15:41
import os

import collections

import hashlib

import urllib2

import json

import re

import tarfile
# settings

# documents http://sae.sina.com.cn/?m=devcenter&catId=281

api_url = 'http://dloadcenter.sae.sina.com.cn/interapi.php?'

appname = 'yansublog'

from_date = '20140101'

to_date = '20140116'

url_type = 'http' # http|taskqueue|cron|mail|rdc

url_type2 = 'access' # only when type=http  access|debug|error|warning|notice|resources

secret_key = 'zwzim4zhk35i50003kz2lh3hyilz01m03515j0i5'
# encode request

params = dict()

params['act'] = 'log'

params['appname'] = appname

params['from'] = from_date

params['to'] = to_date

params['type'] = url_type
if url_type == 'http':

    params['type2'] = url_type2
params = collections.OrderedDict(sorted(params.items()))
request = ''

for k,v in params.iteritems():

    request += k+'='+v+'&'
sign = request.replace('&','')

sign += secret_key
md5 = hashlib.md5()

md5.update(sign)

sign = md5.hexdigest()
request = api_url + request + 'sign=' + sign
# request api

response = urllib2.urlopen(request).read()

response = json.loads(response)
if response['errno'] != 0:

    print '[!] '+response['errmsg']

    exit()
print '[#] request success'
# download and save files

log_files = list()
for down_url in response['data']:    

    file_name = re.compile(r'\d{4}-\d{2}-\d{2}').findall(down_url)[0] + '.tar.gz'

    log_files.append(file_name)

    data = urllib2.urlopen(down_url).read()

    with open(file_name, "wb") as file:

        file.write(data)
print '[#] you got %d log files' % len(log_files)
# compress these files to access_log

access_log = open('access_log','w');
for log_file in log_files:

    tar = tarfile.open(log_file)

    log_name = tar.getnames()[0]

    tar.extract(log_name)

    # save to access_log

    data = open(log_name).read()

    access_log.write(data)

    os.remove(log_name)
print '[#] all file has writen to access_log'