facebook-feed.py

# -*- coding: utf-8 -*-
 
import feedparser, time, sys, re, lxml
 
def strip_html(s):
    return re.sub('<[^>]+>', '', s)
 
def strip_entity(s):
    def unescape_entity(m):
        s = m.group(1).lower()
        transform_map = {
            'amp': '&',
            'gt': '>',
            'lt': '<',
            'quot': '"',
            'nbsp': ' '
        }
        if s[:2] == '#x':
            return unichr(int(s[2:], 16))
        elif s[0] == '#':
            return unichr(int(s[1:]))
        elif s == 'amp':
            return ' '
        else:
            try:
                return transform_map[s]
            except KeyError as e:
                print "Error while decode html entity '{}'".format(m.group(0))
    return re.sub('&([#\w]+);', unescape_entity, s)
 
def smart_break(s, indent = 0, max_width = 80):
    def char_width(s):
        if s == '\t':
            return 8
        elif ' ' <= s and s <= '~':
            return 1
        else:
            return 2
 
    out = ''
    last_line = ''
    w = indent
    for ch in s:
        if ch != '\n':
            w += char_width(ch)
            if w > max_width:
                out += '\n' + ' ' * indent + ch
                last_line = ch
                w = indent + char_width(ch)
            else:
                out += ch
                last_line += ch
        else:
            out += '\n' + ' ' * indent
            last_line = ''
            w = 0 
    return out
 
feed = feedparser.parse("https://www.facebook.com/feeds/notifications.php?id=10000014******3&viewer=0000014******3&key=A***********ZCiC&format=rss20")
entries = feed["entries"]
for entry in entries:
    title = entry['title']
    timestamp = time.localtime(time.mktime(entry['published_parsed']))
    url = entry['link']
 
    title = entry['summary_detail']['value']
    title = strip_html(title)
    title = strip_entity(title)
    title = smart_break(title, 22, 84)
    timestamp = time.strftime('%Y-%m-%d %H:%M:%S', timestamp)
    print u'{time} - {title}'.format(time = timestamp, title = title)