path: root/unfoggedbot.py

#! /usr/bin/python
#from twisted.words.protocols import oscar
import oscar
## local copy taken from the py-aimt project.
## some API differences, but it automagically does rate-limiting stuff.
from twisted.internet import protocol, reactor
from twisted.enterprise.adbapi import ConnectionPool
from twisted.python import log
import re, feedparser, socket, sets, os
#socket.setdefaulttimeout(60)


DB = "ubotdb"
RC = ".unfoggedbotrc" 
host, port = ('login.oscar.aol.com', 5190)

## longer SQL.
GETCOMMENT = "SELECT comment_id,link,thread_url,author,content,title FROM rss_items WHERE processed = 0 ORDER BY comment_id ASC LIMIT 1"
GETPOST = "SELECT post_id, post_url, title FROM posts WHERE processed = 0 ORDER BY post_id ASC LIMIT 1"
GETSUBSCRIBERS = "SELECT screen_name FROM subscriptions WHERE thread_url = ?"
INSERTCOMMENTS = "INSERT OR IGNORE INTO rss_items (link, processed, thread_url, author, content, title) VALUES (?, ?, ?, ?, ?, ?)"
INSERTPOSTS = "INSERT OR IGNORE INTO posts (post_url, title, processed) VALUES (?, ?, ?)"

db = ConnectionPool("pysqlite2.dbapi2", DB)

permalink = re.compile("http://(www\.)?unfogged\.com/archives/week_...._.._..\.html#(\d+)")
commentlink = re.compile("http://(www\.)?unfogged\.com/archives/comments_(\d+).html")

ignorefirst = lambda f: lambda *a: f(*a[1:])
mkmsg = lambda txt: [[txt.decode('utf-8').encode('utf-16-be'), 'unicode']]

class UnfoggedBot(oscar.BOSConnection):
    def __init__(self, username, cookie):
        oscar.BOSConnection.__init__(self, username, cookie)
        self.offlines = sets.Set()
        self.lost = False
        db.runOperation("UPDATE posts SET processed = 1")\
                       .addCallback(ignorefirst(db.runOperation),
                                    "UPDATE rss_items SET processed = 1")\
                       .addCallback(ignorefirst(self.getnew))

    def connectionMade(self):
        oscar.BOSConnection.connectionMade(self)
        self.lost = False

    def connectionLost(self, reason):
        oscar.BOSConnection.connectionLost(self, reason)
        self.lost = True
        
    def initDone(self):
        log.msg("%s: in initDone" % self)
        self.requestSelfInfo()
        self.requestSSI().addCallback(self.gotBuddylist)

    def gotBuddylist(self, l):
        self.activateSSI()
        self.clientReady()

    def offlineBuddy(self, user):
        log.msg("User %s went offline" % user)
        self.offlines.add(user.name)

    def updateBuddy(self, user):
        msg = "User %s status change" % user.name
        if user.name in self.offlines:
            msg += "; removing from offlines set"
            self.offlines.remove(user.name)
        log.msg(msg)
        

    def _getnew(self, txn):
        if self.lost: return
        comment = txn.execute(GETCOMMENT).fetchall()
        if comment:
            turl = comment[0][2]
            crecipients = txn.execute(GETSUBSCRIBERS, (turl,)).fetchall()
        else: crecipients = [[[]]]
        post = txn.execute(GETPOST).fetchall()
        precipients = txn.execute("SELECT screen_name FROM subscriptions WHERE thread_url = ?", ("posts",)).fetchall()
        return (comment,
                post,
                [cr[0] for cr in crecipients],
                [pr[0] for pr in precipients],
                )

    def getnew(self):
        if self.lost: return
        reactor.callLater(3, self.getnew)
        db.runInteraction(self._getnew).addCallback(self.sendnew)

    def markread(self, txn, cid, pid):
        if self.lost: return
        if cid:
            txn.execute("UPDATE rss_items SET processed = 1 WHERE comment_id = ?", (cid,))
        if pid:
            txn.execute("UPDATE posts SET processed = 1 WHERE post_id = ?", (pid,))

    def sendnew(self, (comment, post, crecs, precs)):
        if comment:
            cid, link, turl, auth, cnt, title = comment[0]
            if len(cnt) >= 500:
                cnt = cnt[:497]+"..." #possibly cutting off tags
            msg = mkmsg('<b>%s</b> on <b><a href="%s">%s</a></b>: %s' %\
                        (auth, link, title, cnt))
            crecs = [c for c in crecs if c not in self.offlines]
            if crecs: log.msg("Sending message to %s" % crecs)
            for crec in crecs:
                self.sendMessage(crec, msg)
        else: cid = None
        if post:
            pid, link, title = post[0]
            msg = mkmsg('New post!\n<a href="%s">%s</a>' % (link, title))
            precs = [p for p in precs if p not in self.offlines]
            if precs: log.msg("Sending new post alert to %s" % precs)
            for prec in precs:
                self.sendMessage(prec, msg)
        else: pid = None
        db.runInteraction(self.markread, cid, pid)

    def receiveMessage(self, user, multiparts, flags):
        try:
            msg = str(multiparts[0][0])
        except IndexError:
            log.msg("Index error on msg: \"%s\"?" % multiparts)
            return
        if self.trycommentsub(user, msg): return
        if re.search("(un)?subscribe posts", msg):
            db.runQuery("SELECT screen_name FROM subscriptions WHERE thread_url = ?", ("posts",)).addCallback(self.addremovepostsub, user.name)
                
    def trycommentsub(self, user, msg):
        for pat in (permalink, commentlink):
            m = pat.search(msg)
            if m:
                thread_id = re.sub('^0+', '', m.group(2))
                thread_url = "http://www.unfogged.com/archives/comments_%s.html" % thread_id
                break
        else:
            return
        return db.runInteraction(self._subinfo, thread_url).addCallback(self.addremovecommentsub, thread_url, user.name)


    def _subinfo(self, txn, turl):
        subscribers = txn.execute(GETSUBSCRIBERS, (turl,)).fetchall()
        try:
            threadname = txn.execute("SELECT title FROM posts WHERE post_url = ?", (turl,)).fetchall()[0][0]
        except IndexError:
            threadname = turl
        return ((str(s[0]) for s in subscribers), str(threadname))

    def addremovecommentsub(self, (subscribers, threadname), threadurl, uname):
        def sendfromcb(_, msg): self.sendMessage(uname, mkmsg(msg))
        if uname in subscribers:
            log.msg("unsubscribe %s from %s" % (uname, threadname))
            db.runOperation("DELETE FROM subscriptions WHERE screen_name = ? AND thread_url = ?", (uname, threadurl)).addCallback(sendfromcb, 'subscription to "%s" disabled.' % threadname)
        else:
            log.msg("subscribe %s to %s" % (uname, threadname))
            db.runOperation("INSERT INTO subscriptions (screen_name, thread_url) VALUES (?, ?)", (uname, threadurl)).addCallback(sendfromcb, 'subscription to "%s" enabled.' % threadname)

    def addremovepostsub(self, subscribers, uname):
        def sendfromcb(_, msg): self.sendMessage(uname, mkmsg(msg))
        if uname in (str(s[0]) for s in subscribers):
            log.msg("unsubscribe %s from posts" % uname)
            db.runOperation("DELETE FROM subscriptions WHERE screen_name = ? AND thread_url = ?", (uname, "posts")).addCallback(sendfromcb, "subscription to posts disabled.")
        else:
            log.msg("subscribe %s to posts" % uname)
            db.runOperation("INSERT INTO subscriptions (screen_name, thread_url) VALUES (?, ?)", (uname, "posts")).addCallback(sendfromcb, "subscription to posts enabled.")

class ReconnectingOSCARFactory(protocol.ClientFactory):
    delay = 10.0
    protocol = UnfoggedBot
    def __init__(self, un, cookie):
        self.un = un
        self.cookie = cookie
    def buildProtocol(self, addr):
        p = self.protocol(self.un, self.cookie)
        p.factory = self
        return p
    def clientConnectionLost(self, connector, reason):
        reactor.callLater(self.delay, self._reconnect)
    def _reconnect(self):
        f = ReconnectingOSCARLoginFactory(sn, pass_)
        return reactor.connectTCP(host, port, f)

class OA(oscar.OscarAuthenticator):
    BOSClass = UnfoggedBot
    connectfactory = ReconnectingOSCARFactory
    def connectToBOS(self, server, port):
        if not self.connectfactory:
            c = protocol.ClientCreator(reactor, self.BOSClass, self.username, self.cookie)
            return c.connectTCP(server, int(port))
        else:
            f = self.connectfactory(self.username, self.cookie)
            return reactor.connectTCP(server, int(port), f)

class ReconnectingOSCARLoginFactory(protocol.ReconnectingClientFactory):
    protocol = OA
    def __init__(self, sn, pass_):
        self.sn = sn
        self.pass_ = pass_
    def buildProtocol(self, addr):
        p = self.protocol(self.sn, self.pass_, icq=0)
        p.factory = self
        return p
    ## only reconnect on *failures*
    def clientConnectionLost(self, con, reason):
        pass

def getrss():
##on the assumption that this can be done w/o using too much time,
##I have taken this out of a separate thread.
    comments = feedparser.parse("/home/unfogged/unfogged/html/bridgeplate.rdf")
    posts = feedparser.parse("/home/unfogged/unfogged/html/index.xml")
    updaterss(comments, posts)
    reactor.callLater(15, getrss)

def updaterss(comments, posts):
    def doinsertions(txn, commentdata, postdata):
        txn.executemany(INSERTCOMMENTS, commentdata)
        txn.executemany(INSERTPOSTS, postdata)
    commentdata, postdata = [], []
    for e in comments.entries:
        title = e.title.split("comments on ")[1][1:-1]
        thread_url = e.link.split('#')[0]
        commentdata.append((e.link, 0, thread_url, e.contributors[0]['name'],
                            e.description, title))
    postdata  = []
    for e in posts.entries:
            postdata.append((e.link, e.title, 0))
    db.runInteraction(doinsertions, commentdata, postdata)


if __name__ == '__main__':
    execfile(RC) # values of "sn" and "pass_"
    import sys
    sys.stderr = sys.stdout = open("ubot.log", "a+")
    log.startLogging(sys.stdout)
    f = ReconnectingOSCARLoginFactory(sn, pass_)
    reactor.connectTCP(host, port, f)

    getrss()
    reactor.run()