tumblr_backup.py

# -*- coding: utf8 -*-

from google.appengine.ext import webapp
from google.appengine.ext.webapp.util import run_wsgi_app
from google.appengine.api import urlfetch
from google.appengine.ext import db
from google.appengine.api import mail
from google.appengine.api.labs import taskqueue
from BeautifulSoup import BeautifulStoneSoup
import string
import random
import re

tumblr_account = ''
from_mailaddress = ''
to_mailaddress = ''

class TumblrPhoto(db.Model):
    url = db.StringProperty()
    queued = db.BooleanProperty()
    mailsent = db.BooleanProperty()
    content = db.BlobProperty()

class MailHandler(webapp.RequestHandler):
    def get(self, id=''):
        if not id:
            return
        
        photo = TumblrPhoto.get_by_key_name(id)
        if photo is None:
            return
        
        filename = photo.url
        if not re.compile('^.*\.jpe?g$').search(filename):
            alphabets = string.digits + string.letters
            filename = ''.join(random.choice(alphabets) for i in xrange(20)) + '.jpg'
        
        mail.send_mail(
            sender = from_mailaddress,
            to = to_mailaddress,
            subject = filename,
            body = filename,
            attachments = [(filename, photo.content)]
        )
        
        photo.content = None
        photo.mailsent = True
        photo.put()

class FetchHandler(webapp.RequestHandler):
    def get(self, id=''):
        if not id:
            return
        
        photo = TumblrPhoto.get_by_key_name(id)
        if photo is None:
            return
        
        result = urlfetch.fetch(photo.url)
        if result.status_code != 200:
            taskqueue.add(url='/tumblr/fetch/'+id, method='GET')
            return
        
        photo.content = result.content
        photo.put()
        taskqueue.add(url='/tumblr/mail/'+id, method='GET')
        

class CronHandler(webapp.RequestHandler):
    def get(self):
        photos = db.Query(TumblrPhoto).filter('queued =', False).fetch(10)
        for photo in photos:
            taskqueue.add(url='/tumblr/fetch/'+photo.key().name(), method='GET')
            photo.queued = True
            photo.put()

class CheckHandler(webapp.RequestHandler):
    def get(self, start='', end=''):
        if not start or not end:
            return
        
        url = 'http://' + tumblr_account + '.tumblr.com/api/read?num=50&type=photo&start=' + start
        result = urlfetch.fetch(url)
        if result.status_code == 200:
            soup = BeautifulStoneSoup(result.content)
            for (post, photo_url) in zip(soup('post'), soup('photo-url', {'max-width': '1280'})):
                self.response.out.write(post['id']+','+photo_url.string+'<br>')
                
                p = TumblrPhoto.get_by_key_name(post['id'])
                if p is None:
                    p = TumblrPhoto(key_name=post['id'], url=photo_url.string+'', queued=False, mailsent=False)
                    p.put()
            
            if int(start)+50 < int(end):
                start = str(int(start)+50)
                taskqueue.add(url='/tumblr/check/'+start+'/'+end, method='GET')
        else:
            taskqueue.add(url='/tumblr/check/'+start+'/'+end, method='GET')

class StartHandler(webapp.RequestHandler):
    def get(self):
        url = 'http://' + tumblr_account + '.tumblr.com/api/read?type=photo'
        result = urlfetch.fetch(url)
        if result.status_code == 200:
            soup = BeautifulStoneSoup(result.content)
            posts = soup('posts')
            end = posts[0]['total']
            taskqueue.add(url='/tumblr/check/0/'+end, method='GET')
            self.response.out.write('Task start.')
        else:
            self.response.out.write('error')

application = webapp.WSGIApplication(
    [('/tumblr/mail/(.+)', MailHandler),
     ('/tumblr/fetch/(.+)', FetchHandler),
     ('/tumblr/cron', CronHandler),
     ('/tumblr/check/(.+)/(.+)', CheckHandler),
     ('/tumblr/start', StartHandler)],
    debug=True)

def main():
    run_wsgi_app(application)

if __name__ == '__main__':
    main()