X-Git-Url: http://git.kaliko.me/?a=blobdiff_plain;f=sid%2Ffeeds.py;h=7fba4e25151b390d6c46370b26d28f05efbf8e24;hb=dc4ac4f37ac6b92014ec41f6233ada96c8484adb;hp=f06befa062b068eae46973aca1751df5278e8172;hpb=025ffd6ce2160db6d19849296a3651b2a700f8e0;p=sid.git diff --git a/sid/feeds.py b/sid/feeds.py index f06befa..7fba4e2 100644 --- a/sid/feeds.py +++ b/sid/feeds.py @@ -1,6 +1,6 @@ # -*- coding: utf-8 -*- -# Copyright (C) 2011, 2014 kaliko +# Copyright (C) 2011, 2014, 2020 kaliko # This program is free software: you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -47,84 +47,84 @@ class FeedMonitor(threading.Thread): threading.Thread.__init__(self) self.feeds_list = plugin.FEEDS self.tempo = plugin.TEMPO - self.bot = plugin.bot + self.plugin = plugin self.last_check = datetime.datetime.utcnow() self.seen = dict() self.thread_killed = False - def send(self, message): - """simple wrapper around bot send_message method""" - self.bot.send_message(mto=self.bot.room, - mbody=message[1], - mhtml=message[0], - mtype='groupchat') + def _update_cache(self, feed, parsed): + self.seen[feed].update({'ids': {p.id for p in parsed.entries} or {}}) + # Common HTTP caching + if parsed.get('etag', False): + self.seen[feed].update({'cache': {'etag': parsed.etag}}) + if parsed.get('modified', False): + self.seen[feed].update({'cache': {'modified': parsed.modified}}) def new_posts(self, feed): """Send new posts in feed""" - parsed_feed = feed_parse(feed) - + self.plugin.log.debug('feed: : "%s"', feed) + if self.seen.get(feed) and self.seen.get(feed).get('cache'): + parsed_feed = feed_parse(feed, **self.seen[feed]['cache']) + else: + if self.seen.get(feed): + self.plugin.log.debug('No cache headers set (etag/modified)') + parsed_feed = feed_parse(feed) # Cannot resolve address if 'status' not in parsed_feed: - self.bot.log.error('Error from "%s": %s.' % - (feed, parsed_feed.bozo_exception.__repr__())) + self.plugin.log.error('Error from "%s": %s.', + feed, parsed_feed.bozo_exception.__repr__()) + return + # http caching + if parsed_feed.status == 304: + self.plugin.log.debug('Got 304 not modified') return - # unusual return http code if parsed_feed.status != 200: - self.bot.log.error( - 'Got code %(status)d from "%(href)s" (please update).' % - parsed_feed) + self.plugin.log.warning( + 'Got code %(status)d from "%(href)s" (please update).', + parsed_feed) return - - feed_updated = parsed_feed.feed.get('updated_parsed', None) - - # Avoid looping over all posts if possible - if feed_updated and strtm_to_dtm(feed_updated) < self.last_check: - self.bot.log.debug('updated : %s' % strtm_to_dtm(feed_updated)) - self.bot.log.debug('last check: %s' % self.last_check) + if not self.seen.setdefault(feed): + # Fills with post id when first started (prevent from posting all + # entries at startup) + self.seen[feed] = {'cache': None} + self._update_cache(feed, parsed_feed) return - title = '"%s":' % parsed_feed.feed.get('title', 'n/a') xtitle = '%s:' % html_escape( parsed_feed.feed.get('title', 'n/a')) text = [title] xhtml = [xtitle] - feed_id = parsed_feed.feed.get('id', feed) - if not self.seen.setdefault(feed_id): - # Fills with post id when first started (prevent from posting all - # entries at startup) - self.seen[feed_id] = {p.id for p in parsed_feed.entries} - #return # Detecting new post entries = {p.id for p in parsed_feed.entries} + seen_ids = self.seen.get(feed).get('ids') new_entries = [p for p in parsed_feed.entries - if p.id in entries - self.seen.get(feed_id)] + if p.id in entries - seen_ids] for post in new_entries: - self.bot.log.info(post.title) - + self.plugin.log.info(post.title) body = '%(title)s %(link)s' % post text.append(body) - - xpost = dict(**post) - xpost['title'] = html_escape(xpost.get('title', 'n/a')) - xbody = '%(title)s' % xpost + xpost = {'title': html_escape(post.get('title', 'n/a'))} + xpost['link'] = html_escape(post.get('link',)) + xbody = '{title}'.format(**xpost) xhtml.append(xbody) - # Updating self.seen - self.seen[feed_id] = entries + # Updating self.seen, entries and cache headers + self._update_cache(feed, parsed_feed) if len(text) > 1: - self.bot.log.debug('
'.join(xhtml)) - self.send(('
'.join(xhtml), '\n'.join(text))) + self.plugin.send(self.plugin.bot.room, + {'mhtml': '
'.join(xhtml), 'mbody': '\n'.join(text)}, + mtype='groupchat') def run(self): while not self.thread_killed: - self.bot.log.debug('feeds check') + self.plugin.log.debug('feeds check') for feed in self.feeds_list: try: self.new_posts(feed) except Exception as err: - self.bot.log.error('feeds thread crashed: %s' % err) - self.bot.log.error(''.join(traceback.format_exc())) + self.plugin.log.error('feeds thread crashed: %s', err) + self.plugin.log.error(''.join(traceback.format_exc())) self.thread_killed = True self.last_check = datetime.datetime.utcnow() for _ in list(range(self.tempo)): @@ -134,22 +134,22 @@ class FeedMonitor(threading.Thread): class Feeds(Plugin): + """ + .. note:: + Feeds plugin depends on external module: **feedparser** + """ + + #: Time between feeds check TEMPO = 60 + #: Default feeds to monitor FEEDS = [ - # not working - # 'http://www.debian.org/security/dsa', - - # not working - # 'http://www.debian.org/News/news', - - # DPN in french - 'http://www.debian.org/News/weekly/dwn.fr.rdf', - + 'https://www.debian.org/security/dsa', + 'https://www.debian.org/News/news', + # Some packages + 'https://tracker.debian.org/pkg/prosody/rss', + 'https://tracker.debian.org/pkg/ejabberd/rss', # Misc - 'http://rss.gmane.org/topics/excerpts/gmane.linux.debian.devel.announce', - 'http://rss.gmane.org/gmane.linux.debian.user.security.announce', - 'http://planet-fr.debian.net/users/rss20.xml', - 'http://planet.debian.org/atom.xml', + 'https://planet.debian.org/atom.xml', ] def __init__(self, bot): @@ -162,10 +162,19 @@ class Feeds(Plugin): self.th_mon.thread_killed = True @botcmd - def feeds(self, message, args): + def feeds(self, rcv, args): """feeds monitors debian project related feeds. - !feeds : registred feeds list - !feeds last : last check time""" + + **commands**: + + * ``!feeds`` : registred feeds list + * ``!feeds last`` : last check time""" if 'last' in args: - return 'Last feeds check: %s' % self.th_mon.last_check - return '\n'.join(Feeds.FEEDS) + self.reply(rcv, 'Last feeds check: %s' % self.th_mon.last_check) + return + html = ['{1}'.format(html_escape(u), + html_escape(u[7:]) + ) for u in Feeds.FEEDS] + msg = {'mbody': 'Feeds:\n' + '\n'.join(Feeds.FEEDS), + 'mhtml': 'Feeds:
' + '
'.join(html)} + self.reply(rcv, msg)