attachment:wordpress2blogger.py of MicroProj/2007-02-01

   1 # blogger wordpress api libraries
   2 # Author: Yichao Zhang
   3 # Email: [email protected]
   4 
   5 
   6 import httplib
   7 import urllib
   8 import xmlrpclib
   9 import time
  10 from xmlrpclib import Error
  11 
  12 # set debug mode        
  13 DEBUG = False
  14 myencoding = 'utf-8' 
  15 
  16 # set your wordpress username, password, xmlrpc url     
  17 wp_username = 'admin'   
  18 wp_passwd = 'mypwd'
  19 wp_url = 'http://localhost/wordpress/xmlrpc.php'
  20                         
  21 # set your blogger email, password, blogid
  22 # blogid is in your blogspot Dashboard url, for example:
  23 # http://www2.blogger.com/posts.g?blogID=18083698, "18083698" is your blogid.
  24 blogger_email = '[email protected]'
  25 blogger_passwd = 'mypwd'
  26 blogger_blogid = '18083698'
  27 
  28 # set how many recent posts do you want to import from wordpress to blogger
  29 wp_recent_posts_num = 5  
  30 # set how many times do you want to try if it fails to post to blogger
  31 tryCounter = 3
  32 
  33 class SyncWorker:
  34     def main(self):
  35         wp = WordpressClient(wp_username, wp_passwd, wp_url)
  36         blogs = wp.getRecentPosts(wp_recent_posts_num)
  37         
  38         clientLogin = GoogleClientLogin(blogger_email, blogger_passwd, 'blogger', 'easter.blogsync')
  39         auth = clientLogin.getAuth()
  40         bg = BloggerClient(auth, blogger_blogid)  
  41         
  42         failedList = []
  43         for blog in blogs:
  44             title = blog['title'].encode(myencoding)
  45             content = blog['description'].encode(myencoding)
  46             labels = ','.join([c for c in blog['categories']]).encode(myencoding)
  47             publishedDate = blog['dateCreated'] 
  48             postid = blog['postid']
  49             entry = Entry(title, content, labels, publishedDate)   
  50             print '>>>> starting postid == %s' % postid
  51             print '>>>> ........ title  == %s' % title  
  52             isSuccessfull = bg.insertEntry(entry)  
  53             counter = tryCounter;
  54             while counter > 0 and not isSuccessfull:  
  55                 print 'post failed, try again!'
  56                 isSuccessfull = bg.insertEntry(entry) 
  57                 counter -= 1  
  58             if isSuccessfull:
  59                 print '>>>> post successfully\n'
  60             else: 
  61                 failedList.append(postid)
  62                 print '>>>> post unsuccessfully\n'
  63         if failedList:  
  64             print '>>>> these posts not added:'
  65             print failedList
  66             
  67 class GoogleClientLogin:
  68     def __init__(self, email, passwd, service, source):
  69         params = urllib.urlencode({'Email': email, 'Passwd': passwd, 'service': service, 'source': source})
  70         headers = {"Content-type": "application/x-www-form-urlencoded"}
  71         conn = httplib.HTTPSConnection("www.google.com")
  72         conn.request("POST", "/accounts/ClientLogin", params, headers)
  73         response = conn.getresponse()
  74         data = response.read()
  75         # failed
  76         if response.status == 403:
  77             self.auth = None
  78         else:
  79             self.auth = data.splitlines()[2].split('=')[1]
  80         conn.close()
  81 
  82     def getAuth(self):
  83         return self.auth
  84 
  85 class BloggerClient:
  86     def __init__(self, auth, blogid):
  87         self.auth = auth
  88         self.blogid = blogid
  89         self.host = 'www.blogger.com'
  90         self.path = '/feeds/%s/posts/default' % (self.blogid)
  91 
  92     def insertEntry(self, entry):   
  93         return self.insertEntryXML(entry.getEntryXML(), self.host, self.path)
  94     
  95     def insertEntryXML(self, entryXML, host, path):
  96         headers = {"Authorization": "GoogleLogin auth=%s" % self.auth, "Content-type": "application/atom+xml"}
  97         conn = httplib.HTTPConnection(self.host)
  98         conn.request("POST", self.path, entryXML, headers)
  99         data = conn.getresponse()  
 100         if DEBUG:  
 101             print '>>>> response.status == %s' % data.status
 102             print '>>>> blogger server response:'
 103             print data.read()
 104         if data.status == 302:
 105             location = data.getheader('location')[7:].split('/', 1)
 106             host = location[0]
 107             path = '/' + location[1]
 108             self.insertEntryXML(entryXML, host, path)
 109         elif data.status == 201:
 110             return True
 111         else:
 112             return False
 113 
 114 class Entry:
 115     def __init__(self, title, content, labels, publishedDate):
 116         self.title = title
 117         self.content = content
 118         self.labels = labels
 119         self.publishedDate = publishedDate
 120 
 121     def getLabelXML(self, label):
 122         return '<category scheme="http://www.blogger.com/atom/ns#" term="%s"/>' % label
 123 
 124     def __getLabelsXML(self):
 125         return ''.join([self.getLabelXML(label) for label in self.labels.split(',')])
 126 
 127     def __getPublishedDate(self):
 128         date = str(self.publishedDate)
 129         timezone = time.timezone
 130         if timezone > 0:
 131             sign = '-'
 132         else:
 133             sign = '+'
 134             timezone = -timezone
 135         tzHours = timezone/3600
 136         tzMinutes = timezone % 3600 / 60
 137         return '%(year)s-%(month)s-%(other)s%(sign)s%(hours)02d:%(minutes)02d' % \
 138 	        {'year':date[0:4], 'month':date[4:6], 'other':date[6:], 'sign':sign, 'hours':tzHours, 'minutes':tzMinutes}
 139     
 140     def getEntryXML(self):
 141         entryXML = \
 142             '''
 143             <entry xmlns='http://www.w3.org/2005/Atom'>
 144               <title type='text'><![CDATA[%s]]></title>
 145               <published>%s</published>
 146               %s
 147               <content type="text"><![CDATA[%s]]></content>
 148             </entry>
 149             ''' % (self.title, self.__getPublishedDate(), self.__getLabelsXML(), self.content)
 150         return entryXML  
 151 
 152 class WordpressClient:
 153     def __init__(self, username, passwd, url):
 154         self.username = username
 155         self.passwd = passwd
 156         self.url = url
 157         self.wp = xmlrpclib.ServerProxy(url)
 158     def getRecentPosts(self, num):  
 159         try:
 160             blogs = self.wp.metaWeblog.getRecentPosts('', self.username, self.passwd, num)    
 161         except Error, v:  
 162             print v
 163         return blogs
 164 
 165 if __name__ == "__main__":
 166     SyncWorker().main()