Generate an RSS feed from your Tumblr dashboard and push it to S3.

tumblr2s3.py

"""
    Create an RSS feed from your Tumblr dashboard and write to S3.
    
    Run with -x command to write to stdout instead of pushing to S3.

    Requirements:
        boto
        python-s3file
        PyRSS2Gen
"""
from s3file import s3open
from urllib import urlencode
import datetime
import json
import PyRSS2Gen
import re
import sys
import urllib2

# Put your Tumblr credentials here...
TUMBLR_EMAIL = ""
TUMBLR_PASSWORD = ""

# and your S3 keys go here.
S3_KEY = ""
S3_SECRET = ""

# Full URL to file on S3. This one is an example... change it.
S3_URL = "http://mybucket.s3.amazonaws.com/feeds/tumblr-dashbord.rss"

# really bad tag stripper
TAG_STRIPPER = re.compile(r'<.*?>')

# Tumblr client

class Tumblr(object):
    
    def __init__(self, email=None, password=None):
        self.email = email
        self.password = password
    
    def _call(self, url, params=None, auth=True):
        params = params.copy() if params else {}
        if auth and self.email and self.password:
            params['email'] = self.email
            params['password'] = self.password
            resp = urllib2.urlopen(url, data=urlencode(params))
        else:
            resp = urllib2.urlopen(url)
        return resp.read()
    
    def dashboard(self, **kwargs):
        url = "http://www.tumblr.com/api/dashboard/json"
        resp = self._call(url, params=kwargs)
        return json.loads(resp.strip()[22:-1])

# utility methods

def parse_datetime(dt):
    return datetime.datetime.strptime(dt, "%Y-%m-%d %H:%M:%S %Z")

def titleify(t):
    t = TAG_STRIPPER.sub('', t)
    if len(t) > 60:
        t = t[:57] + "..."
    return t

# post handlers

def default_handler(post):
    return {
        "title": post['slug'],
        "description": "",
    }

def post_handler(post):
    return {
        "title": post['regular-title'],
        "description": post['regular-body'],
    }

def photo_handler(post):
    description = """<img src="%s">\n<p>%s</p>""" % (post['photo-url-400'], post['photo-caption'])
    return {
        "title": post['photo-caption'],
        "description": description,
    }

def quote_handler(post):
    description = """<blockquote>%s</blockquote>\n<p>%s</p>""" % (post['quote-text'], post['quote-source'])
    return {
        "title": post['quote-text'],
        "description": description,
    }

def link_handler(post):
    description = """<a href="%s">%s</a>\n\n%s""" % (post['link-url'], post['link-text'], post['link-description'])
    return {
        "link": post['link-url'],
        "title": post['link-text'],
        "description": description,
    }

def conversation_handler(post):
    pass

def video_handler(post):
    description = """%s\n<p>%s</p>""" % (post['video-player'], post['video-caption'])
    return {
        "title": post['video-caption'],
        "description": description,
    }

def audio_handler(post):
    pass

def answer_handler(post):
    return {
        "title": post['question'],
        "description": post['answer'],
    }

POST_HANDLERS = {
    'regular': post_handler,
    'photo': photo_handler,
    'quote': quote_handler,
    'link': link_handler,
    'conversation': conversation_handler,
    'video': video_handler,
    'audio': audio_handler,
    'answer': answer_handler,
}

# main jazz

if __name__ == "__main__":
    
    t = Tumblr(TUMBLR_EMAIL, TUMBLR_PASSWORD)
    dashboard = t.dashboard()
    
    items = []
    
    for post in dashboard['posts']:
        handler = POST_HANDLERS.get(post['type'], default_handler)
        params = handler(post)
        if params:
            if not 'link' in params:
                params['link'] = post['url']
            params['guid'] = PyRSS2Gen.Guid(post['url'])
            params['pubDate'] = parse_datetime(post['date-gmt'])
            if params['title']:
                params['title'] = "%s: %s" % (post['tumblelog']['name'], titleify(params['title']))
            else:
                params['title'] = post['tumblelog']['name']
            items.append(PyRSS2Gen.RSSItem(**params))
        else:
            print "NO HANDLER!!!!", post
    
    rss = PyRSS2Gen.RSS2(
        title="Tumblr Dashboard",
        link="http://www.tumblr.com/dashboard",
        description="Tumblr dashboard feed",
        lastBuildDate=datetime.datetime.utcnow(),
        items=items,
    )
    
    if "-x" in sys.argv:
    
        rss.write_xml(sys.stdout)
    
    else:
    
        s3 = s3open(S3_URL, S3_KEY, S3_SECRET)
        rss.write_xml(s3)
        s3.close()