tumblr.py

#!/usr/bin/env python
'Downloads tumblr images from dashboard'
import urllib
import os
from time import sleep
import simplejson as json
from threading import Thread
from Queue import Queue

MY_QUEUE = Queue()

PHOTOS_FOLDER = os.path.join(os.path.dirname(__file__), 'photos')

BASEURL = 'https://www.tumblr.com/api/dashboard/json'

PARAMS = {
    'type' : 'photo',
    'num'  : 50,
    'debug': 1,
    'email': '[email protected]',
    'password': '********************'
}

def work(worker_name, item):
    'downloader'
    filename = os.path.basename(item)
    savename = os.path.join(PHOTOS_FOLDER, filename)
    exists   = os.path.exists(savename)
    print '\n%(worker_name)s : %(filename)s %(exists)s' % {
        'worker_name' : worker_name.rjust(10),
           'filename' : filename,
             'exists' : ['',' (exists)'][exists]
    },
    if not exists:
        try:
            fhan = urllib.urlopen(item)
            phan = open(savename, 'wb')
            phan.write(fhan.read())
            fhan.close()
            phan.close()
        except IOError:
            print 'Could not download: %s' % filename
    else:
        pass # exists

def worker(worker_name):
    'thread'
    while True:
        item = MY_QUEUE.get()
        work(worker_name, item)
        MY_QUEUE.task_done()

if not os.path.exists(PHOTOS_FOLDER):
    os.mkdir(PHOTOS_FOLDER)

for i in ['essa', 'tessa', 'kossa']:
    t = Thread(target = worker, args = (i,))
    t.daemon = True
    t.start()

for start in range(0, 251, 50):
    PARAMS.update({'start' : start})
    TUMBLR_API_URL = '%s?%s' % (BASEURL, urllib.urlencode(PARAMS))

    handler = urllib.urlopen(TUMBLR_API_URL)
    data = json.load(handler)
    handler.close()
    if not data:
        data = {}
    for post in data.get('posts', []):
        photos = []
        for key in post.keys():
            if key[:10] == 'photo-url-':
                photos.append(int(key[10:]))
        if len(photos):
            photos.sort()
            largephoto = 'photo-url-%s' % photos[-1]
            MY_QUEUE.put(post[largephoto])
    sleep(15)

MY_QUEUE.join()