Google +1 feed generator

plusone2rss.py

# -*- coding: utf-8 -*-

"""
Google Plus has no feeds for my +1 and I want to share them in my blog.
This is an incredible simple scrapper to hack that stupidly wrong policy.
My contents are mine!

author: Martín Gaitán <[email protected]>

How to use?
-----------

set your GOOGLE_ID and run it. you'll get a plusone.xml file

Dependencies: pyquery and PyRSS2Gen (both are pip-installables)
"""


import datetime
from pyquery import PyQuery
import PyRSS2Gen

######
GOOGLE_ID = '102449284377784435533'
######

def _(text):
    """latin1 to utf8 transcode helper"""
    try:
        return text.encode('latin1').decode('utf8')
    except UnicodeError:
        return text

class PlusOne2Rss(object):

    def __init__(self, google_id):
        self.url = 'https://plus.google.com/%s/plusones' % google_id
        self.pq = PyQuery(url=self.url)
        self.user_name = self.pq('a.a-n[href="%s"]' % google_id).text()

    def _title(self, table):
        return _(self.pq('div.zCa a.QC', table).text())

    def _link(self, table):
        return _(self.pq('div.zCa a.QC', table).attr('href'))

    def _description(self, table):
        return _(self.pq('div.zCa p.GCa', table).text())

    def _image_url(pq, table):
        return _(self.pq('div.BT a', table).attr('title'))

    def write_feed(self, filename='plusone.xml'):
        rss_items = []
        tables = self.pq('table.KX')
        for table in tables:
            item = PyRSS2Gen.RSSItem(
                    title = self._title(table),
                    link = self._link(table),
                    description = self._description(table),
                    guid = PyRSS2Gen.Guid(self._link(table)),
                    pubDate = datetime.datetime.now()
                )
            rss_items.append(item)

        rss = PyRSS2Gen.RSS2(
                title = _("%s's +1s" % self.user_name),
                link = self.url,
                generator= u'plusone2rss',
                description = _("The latest %s's +1s" % self.user_name.split(' ')[0]),
                lastBuildDate = datetime.datetime.utcnow(),
                items = rss_items
            )
        rss.write_xml(open(filename, "w"), encoding='utf-8')

if __name__ == '__main__':
    PlusOne2Rss(GOOGLE_ID).write_feed()