This outputs tumblr posts to JSON and CSV. Requires beautifulsoup4, csvkit and requests.

tumblr_output_posts.py

#!/usr/bin/env python
"""
Code which generates tumblr post outputs -- in CSV and JSON.
"""
import json

from bs4 import BeautifulSoup
from csvkit.convert import convert
import requests

# Set up the tumblr urls and base params
base_url = 'http://api.tumblr.com/v2/blog/inauguration2013.tumblr.com/posts/photo'

# This should be your own app key.
key_param = '?api_key=Cxp2JzyA03QxmQixf7Fee0oIYaFtBTTHKzRA0AveHlh094bwDH'
limit_param = '&limit=20'
limit = 20
new_limit = limit

# Make the request.
r = requests.get(base_url + key_param)

# Get the total number of posts so we can calculate pages.
total_count = int(json.loads(r.content)['response']['total_posts'])
pages_count = (total_count / limit)

# Handle overflows. Modulus is fun.
pages_remainder = (total_count % limit)
if pages_remainder > 0:
    pages_count += 1

# Set up the list of pages.
pages = range(0, pages_count)

# Storage for the posts.
ALL_POSTS = []

# Loop!
for page in pages:

    # Handle pagination
    start_number = new_limit - limit
    end_number = new_limit

    if end_number > total_count:
        end_number = total_count

    new_limit = new_limit + limit

    page_param = '&offset=%s' % start_number
    page_url = base_url + key_param + limit_param + page_param

    # Request this page.
    r = requests.get(page_url)

    # Get the posts.
    posts = json.loads(r.content)

    # Loop!
    for post in posts['response']['posts']:

        # Intermediate storage for each post.
        post_dict = {}

        # Yeah, yeah, LXML is faster but this is easy.
        soup = BeautifulSoup(post['caption'])
        post_dict['location'] = soup.select('p.signature-name')[0].contents[2].split('from')[1]

        # Some posts error out. This is the WORST error checking.
        try:
            post_dict['message'] = soup.select('p.message')[0].contents[0].replace('<br/>', '')
        except:
            post_dict['message'] = ''

        # Only get tags related to vote type.
        for tag in post['tags']:
            if 'vot' in tag:
                post_dict['voted'] = tag

        # Append this post to the list of posts.
        ALL_POSTS.append(post_dict)

# Turn our post list into JSON.
all_posts_json = json.dumps(ALL_POSTS)

# Write our JSON to a file.
with open('posts.json', 'w') as f:
    f.write(all_posts_json)

# Pass the JSON file buffer to csvkit and write a csv.
with open('posts.json', 'r') as f:
    all_posts_csv = convert(f, format='json')
    with open('posts.csv', 'w') as f:
        f.write(all_posts_csv)