Simple script I used to import entries from my old blog to tumblr

import_tumblr.py

"""My old blog -> tumblr importer

"""

import csv
from getpass import getpass
import json
import logging
import sys
import urllib

import httplib2

LOG = logging.getLogger("import")
HTTP = httplib2.Http()

class PostException(Exception):
    """Raised when there are problems posting an entry"""

if __name__ == "__main__":
    logging.basicConfig(level=logging.INFO)
    
    email = None  # Set this to something
    password = getpass()
    
    blogs = json.load(open("../../Backups/translucentcode.json"))
    
    # Read in any existing posts so we don't duplicate
    # Simple csv mapping original id -> tumblr id
    try:
        LOG.info("Read in posted entries.")
        posted = set(int(original) for original, tumblr in csv.reader(open("posted.csv")))
    except IOError:
        LOG.info("Creating fresh set of posted entries.")
        posted = set()
    
    entries = (entry for entry in blogs["entries"] 
        if (entry["blog"] == "mick") and (entry["pub_state"] == "published"))
    
    # Track what we've posted
    posted_fp = open("posted.csv", "ab")
    posted_writer = csv.writer(posted_fp)
    
    for entry in entries:
        original_id = entry["id"]
        if int(original_id) in posted:
            LOG.info("Already posted id %r", original_id)
            continue
        content = entry["content"]
        content_type = entry["content_type"]
        slug = entry["name"]
        published_date = entry["pub_date"]
        rendered = entry["rendered"]
        tags = entry["tags"]
        tags.append("imported from translucentcode.og")  # stick on a tag to find stuff later
        title = entry["title"]
        LOG.info("Posting %r - %r", original_id, title)
        
        params = {
            "email": email,
            "password": password,
            "type": "regular",
            "title": title.encode("utf-8"),
            "body": rendered.encode("utf-8"),
            "generator": "mick's importer",
            "date": published_date,
            "private": "0",
            "tags": ", ".join('"%s"' % tag for tag in tags),  # "tag1", "tag2", "tag3"
            "format": "html",
            "slug": slug.encode("utf-8"),
            "state": "published",
            "send-to-twitter": "no"
        }
        LOG.debug("Params: %r", params)
        data = urllib.urlencode(params)
        LOG.debug("Content: %r", data)
        response, content = HTTP.request("http://www.tumblr.com/api/write",
            method="POST", body=data, headers={"content-type": "application/x-www-form-urlencoded"})
        LOG.debug("response: %r content: %r", response, content)
        if response.status != 201:
            LOG.error("Problem posting: response: %r content: %r", response, content)
            sys.exit(1)
        LOG.info("Posted to %r", content)
        posted_writer.writerow((original_id, content))
        posted_fp.flush()  # just in case I hit ctrl+c :)