citeurl.py

#!/usr/bin/python
from __future__ import unicode_literals
import urllib
import BeautifulSoup
import sys
import string
import codecs

soup = BeautifulSoup.BeautifulSoup(urllib.urlopen(sys.argv[1]))

try:
    author = soup.findAll(attrs={"name": "author"})[0].content.string
except:
    author = "FIXME"
slug = sys.argv[1][7:20].translate(string.maketrans("", ""),
                                       string.punctuation)
with codecs.open("mybib.bib", "a", "utf-8") as f:
    f.write('''
@misc{%s,
    author = {%s},
    title = {%s},
    howpublished = {\\url{%s}}
}''' % (slug, author, soup.title.string, sys.argv[1]))
print "\cite{%s}" % (slug)