Pandoc filter to replace Link URLs with Wayback Machine URLs

wayback.py

#!/usr/local/bin/python
# -*- coding: utf-8 -*-

# Usage: pandoc --filter=wayback.py input
# Install pandocfilters and requests with pip before using
# Warning: may take a while to process input with lots of links
# Note: Links that can't be saved to WBM or already point to WBM are left as is

from pandocfilters import toJSONFilter, Link
import requests

base_url = 'http://web.archive.org'

def wayback(k, v, f, m):
    ''' Take a non-Wayback-Machine URL, save it to Wayback, replace with snapshot URL '''
    if k == 'Link':
        attrs = v[1]
        url = attrs[0]
        if base_url in url:
            return Link(v[0], attrs)
        else:
            r = requests.get(base_url + '/save/' + url)
            s = r.status_code
            new_url = base_url + r.headers['content-location'] if s == requests.codes.ok else url
            return Link(v[0], (new_url, attrs[1]))

if __name__ == "__main__":
    toJSONFilter(wayback)