Bing美图自动抓取源码

本来想做成一个Bing美图的主题站的，后来发现网上已经有成熟的了。于是把自己写的自动抓取Bing美图的Python脚本放出

支持Bing中国版和Bing国际版
~~支持采集国际版的动态背景~~(已死，原因不明)
图像，介绍都可传到七牛

图像简介，搜索链接可存入MySQL数据库
使用前请先安装七牛sdk：pip install qiniu和MySQLdb
中国版采集：

#!/usr/bin/python
import urllib2
import json
import time
import MySQLdb
from qiniu import Auth, put_file, etag, urlsafe_base64_encode, BucketManager
import qiniu.config
def imgDownload():

print "Loaing json data from bing..."
jsonContent = urllib2.urlopen('http://cn.bing.com/HPImageArchive.aspx?format=js&amp;idx=0&amp;n=1&amp;&amp;pid=hp')
jsonData = jsonContent.read()
s = json.loads(jsonData)
imageUrl = s['images'][0]['url']
print "Downloading image..."
imgData = urllib2.urlopen(imageUrl)
imgWrite = imgData.read() 
fileName= time.strftime("%Y%m%d", time.localtime())
with open("/home/bing/bing.jpg", "wb") as code:     
       code.write(imgWrite)
   with open("/home/bing/info.json", "wb") as code:     
       code.write(jsonData)
   conn=MySQLdb.connect(host='127.0.0.1',user='root',passwd='root',port=3306,read_default_file="/tmp/mysql.sock")
cur=conn.cursor()     
conn.select_db('hdlcom_diao')
cur.execute('SET NAMES utf8;')
dis = s['images'][0]['copyright'].encode('utf-8')
se =s['images'][0]['copyrightlink'].encode('utf-8')
videoUrl = ''
value=[fileName,'cn',dis,se,imageUrl,videoUrl,'false']
cur.execute('insert into bing(`date`, `area`, `info`, `search`, `img`, `vid`, `video`) values(%s,%s,%s,%s,%s,%s,%s)',value)
conn.commit()
cur.close()
conn.close()

def upToQiniu():

print "Uploaing image to qiniu..."
access_key = ''
secret_key = ''
q = Auth(access_key, secret_key)
bucket_name = '空间名'
key = 'bing.jpg';
bucket = BucketManager(q)
print "Deleting old image..."
ret, info = bucket.delete(bucket_name, key)
assert ret == {}
print "Uploading images to qiniu..."
fileName= time.strftime("%Y%m%d", time.localtime())
token = q.upload_token(bucket_name, key, 3600)
localfile = '/home/bing/bing.jpg'
ret, info = put_file(token, key, localfile)
token1 = q.upload_token(bucket_name, fileName+"_cn.jpg", 3600)
localfile1 = '/home/bing/bing.jpg'
ret1, info1 = put_file(token1, fileName+"_cn.jpg", localfile1)
token2 = q.upload_token(bucket_name, fileName+"_cn.json", 3600)
localfile2 = '/home/bing/info.json'
ret1, info1 = put_file(token2, fileName+"_cn.json", localfile2)
assert ret['key'] == key
assert ret['hash'] == etag(localfile)

imgDownload()
upToQiniu()
</pre> 国际版采集： <pre class="lang:python decode:true" title="国际版 ">#!/usr/bin/python

import urllib2
import json
import time
import MySQLdb
from qiniu import Auth, put_file, etag, urlsafe_base64_encode, BucketManager
import qiniu.config
def imgDownload():

print "Loaing json data from bing..."
jsonContent = urllib2.urlopen('http://global.bing.com/HPImageArchive.aspx?format=js&amp;idx=-1&amp;n=1&amp;pid=hp&amp;FORM=HPCNEN&amp;setmkt=en-us&amp;setlang=en-us&amp;video=1')
jsonData = jsonContent.read()
s = json.loads(jsonData)
try:
    videoUrl =  "http:"+s['images'][0]['vid']['sources'][1][2]
    video = 'true'
except:
    video = 'false'
finally:
    videoUrl=''
    imageUrl =  "http://global.bing.com/"+s['images'][0]['url']
print "Downloading image..."
imgData = urllib2.urlopen(imageUrl)
imgWrite = imgData.read() 
with open("/home/bing/bing.jpg", "wb") as code:     
       code.write(imgWrite)
   with open("/home/bing/info.json", "wb") as code:     
       code.write(jsonData)
   if (video == 'true'):
       print videoUrl
       vidData = urllib2.urlopen(videoUrl)
    vidWrite = vidData.read()
    with open("/home/bing/bing.mp4", "wb") as code:     
           code.write(vidWrite)
print "Uploaing image to qiniu..."
access_key = 'AK'
secret_key = 'SK-'
q = Auth(access_key, secret_key)
bucket_name = '空间名'
key = '/home/bing/bing_en.jpg';
bucket = BucketManager(q)
print "Uploading images to qiniu..."
fileName= time.strftime("%Y%m%d", time.localtime())
token1 = q.upload_token(bucket_name, fileName+"_en.jpg", 3600)
localfile1 = '/home/bing/bing.jpg'
ret1, info1 = put_file(token1, fileName+"_en.jpg", localfile1)
token2 = q.upload_token(bucket_name, fileName+"_en.json", 3600)
localfile2 = '/home/bing/info.json'
ret2, info2 = put_file(token2, fileName+"_en.json", localfile2)
if(video == 'true'):
    token3 = q.upload_token(bucket_name, fileName+"_en.mp4", 3600)
    localfile3 = '/home/bing/bing.mp4'
    ret2, info2 = put_file(token3, fileName+"_en.mp4", localfile3)
conn=MySQLdb.connect(host='127.0.0.1',user='root',passwd='root',port=3306,read_default_file="/tmp/mysql.sock")
cur=conn.cursor()     
conn.select_db('hdlcom_diao')
dis = s['images'][0]['copyright']
se =s['images'][0]['copyrightlink']
value=[fileName,'en',dis,se,imageUrl,videoUrl,video]
cur.execute('insert into bing(`date`, `area`, `info`, `search`, `img`, `vid`, `video`) values(%s,%s,%s,%s,%s,%s,%s)',value)
conn.commit()
cur.close()
conn.close()

imgDownload()

注意：使用前请先更改代码中的七牛ak，sk，空间名，MySQL帐号,图像储存的绝对路径
数据库结构如下

`
CREATE TABLE IF NOT EXISTS `bing` (
  `id` int(11) NOT NULL,
  `date` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL,
  `area` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL,
  `info` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL,
  `search` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL,
  `img` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL,
  `vid` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL,
  `video` text CHARACTER SET utf8 COLLATE utf8_bin NOT NULL
) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4;

--
-- Indexes for dumped tables
--

--
-- Indexes for table `bing`
--
ALTER TABLE `bing`
  ADD PRIMARY KEY (`id`);

--
-- AUTO_INCREMENT for dumped tables
--

--
-- AUTO_INCREMENT for table `bing`
--
ALTER TABLE `bing`
  MODIFY `id` int(11) NOT NULL AUTO_INCREMENT;
`

Bing美图自动抓取源码

感谢投食~