wantzel.py

#encoding: utf-8
"""
Bot Wantzel from La Quadrature du Net.

License : AGPLv3
Doc     : https://wiki.laquadrature.net/Wantzel

TODO:
- Ajouter la gestion des droits pour certaines commandes
- Mettre une valeur par défaut pour les champs concernés
- Afficher les titres des urls fournies sur le canal
"""

from irc import IrcClientFactory
import MySQLdb
import re
from twisted.internet import reactor
import urllib

import config
from messages import messages

def get_cursor():
    """
    This function connects to a database and returns a usable cursor.
    """
    db = MySQLdb.connect(
        host=config.dbserver,
        user=config.dbuser,
        passwd=config.dbpassword,
        db=config.dbname
    )
    if db:
        return db.cursor()
    return None

def get_url(message):
    """
    Retrieve the url in the message.
    """
    # Let's get the url
    result = re.search("(https?[^ ]+)", message)
    if not result:
        return "http"
    url = result.group(1)
    # Removing anchor if needed
    result = re.search("^([^#]*)", url)
    if result:
        url = result.group(1)
    # Removing trackers
    url = re.sub("[?&](utm_medium|utm_source|utm_campaign|xtor)=[^&]*", "", url)
    return url

def get_title(message):
    title = ""
    website = ""
    try:
        url = get_url(message)#re.search("(http[^ ]*)", msg).group(1)
        website = re.search("//([^/]*)", url).group(1)
        f = urllib.URLopener().open(url)
        content = f.read()
        title = re.search("<title>([^<]+)</title>", content).group(1)
    except:
        pass
    # Unescaping HTML entities
    if title:
        title = re.sub("&gt;", ">", title)
        title = re.sub("&lt;", "<", title)
        title = re.sub("&quot;", '"', title)
        title = re.sub("&amp;", "&", title)
    return (title, website)

class Wantzel(object):
    """
    Wantzel bot.
    """
    def __init__(self):
        """
        Initialization of bot over IRC.
        """
        self.irc = IrcClientFactory(config)
        self.irc.set_privmsg = self.set_privmsg
        reactor.connectTCP(config.server, config.port, self.irc)

    def set_privmsg(self):
        """
        This method set the methods to call for each callback received from IRC.
        """
        self.irc.client.privmsg = self.on_privmsg
        self.irc.client.joined = self.on_joined

    def send_message(self, channel, messages):
        """
        Sends a message on specified channel, cutting each line in a new message
        """
        for message in messages.splitlines():
            self.irc.client.msg(channel, message)

    def on_privmsg(self, user, channel, msg):
        """
        Wantzel can understand some commands :
        - help
            Returns a message about how to use the bot.
            If a command is passed after help, the message explains how to use
            the command. 
        - rp(acp)
            Add an article in the database
        - stats
            Show some statistics about the RP
        - kill
            Kill an article by giving it a score of -100
        """
        # Cleaning user name
        user = re.search("([^!]*)!", user).group(1)
        print("Message received: %s %s %s" % (user, channel, msg))
        # Whatever is done, get the title of an existing url in a message
        title = ""
        if "http" in msg:
            title, website = get_title(msg)
        # Never answer to botself
        if user!=config.nickname:
            # If it's a query, bot should answer to the user as the channel
            if "#" not in channel:
                channel = user
            # Help command, specific
            if "wantzel" in msg and ("help" in msg or "aide" in msg):
                self.help(user, channel, msg)
            # Find known command
            command = re.search("!(rp[acp]*|kill|help|stats)", msg)
            if command:
                command = command.group(1)
                print("Command: %s" % command)
                if command.startswith("rp"):
                    self.rp(command, user, channel, msg, title)
                elif command=="help":
                    self.help(user, channel, msg)
                elif command=="kill":
                    self.kill(user, channel, msg)
                elif command=="stats":
                    self.stats(user, channel, msg)
        if title and website:
            self.send_message(channel, messages["title"] % (title, website))

    def on_joined(self, channel):
        """
        Say hello to everyone.
        """
        print("Joined channel %s" % channel)
        # Specific message for specific channel
        if "hello_"+channel[1:] in messages:
            self.send_message(channel, messages["hello_"+channel[1:]])
        else:
            self.send_message(channel, messages["hello"])

    def help(self, user, channel, msg):
        """
        Show global help.
        If a known command is behind the !help command, an adequate message is
        returned.
        """
        print("help command")
        # Searching for a command after help keyword
        command = re.search("!help (stats|rp|help|kill)", msg)
        if command:
            command = command.group(1)
            self.send_message(user, messages["help_"+command])            
        else:
            self.send_message(channel, messages["help"])

    def rp(self, command, user, channel, msg, title=""):
        """
        Adding the article in rp database.
        """
        print("rp command %s" % command)
        cite = 0
        note = 0
        url = get_url(msg)
        print("url: %s" % url)
        if url=="":
            return
        elif url=="http":
            self.send_message(channel, messages["rp_http"] % user)
            return

        # Looking for such an article in database
        cursor = get_cursor()
        cursor.execute("SELECT id, note, provenance FROM presse WHERE url = %s", (url, ))
        rows = cursor.fetchall()
        if not rows:
            # LQdN is quoted
            if "c" in command:
                cite += 2
            # the article speak about LQdN
            if command.count("p")>1:
                cite += 2
            # Archive this article
            if "a" in command:
                note -= 2
            #TODO: Gérer les autres champs qui n'ont pas de valeur par défaut
            # lang, published, nid, screenshot, title, fetched, seemscite
            print("Adding an article by %s: %s" % (user, url))
            result = cursor.execute(
                """INSERT INTO presse SET 
                url=%s, provenance=%s, cite=%s, note=%s, datec=NOW(), title=%s,
                lang='', published=0, nid=0, screenshot=0, fetched=0, seemscite=0
                """,
                (url, user, cite, note, title)
            )
            self.send_message(channel, messages["rp_new_article"] % user)
        else:
            if rows[0][2]!=user:
                print("Adding a point by %s on %s" % (user, rows[0][0]))
                result = cursor.execute(
                    "UPDATE presse SET note=note+1 WHERE id=%s",
                    (rows[0][0], )
                )
            if (rows[0][1]+1)<3:
                self.send_message(channel, messages["rp_known_article"] % user)
            else:
                self.send_message(channel, messages["rp_taken_article"] % user)

    def kill(self, user, channel, msg):
        """
        Kill an article by setting its score to -100.
        """
        #TODO: Gérer les droits de cette commande
        print("kill command")
        url = get_url(msg)
        print("url: %s" % url)
        if url=="":
            return
        elif url=="http":
            self.send_message(channel, messages["rp_http"] % user)
            return
        # Looking for such an article in database
        cursor = get_cursor()
        cursor.execute("SELECT id, note FROM presse WHERE url=%s", (url, ))
        rows = cursor.fetchall()
        if not rows:
            self.send_message(channel, messages["kill_none"] % url)
        else:
            cursor.execute("UPDATE presse SET note=-100 WHERE id=%s", (rows[0][0], ))
            self.send_message(channel, messages["kill_done"] % url)

    def stats(self, user, channel, msg):
        """
        Returns stats on articles in press review.
        """
        print("stats command")
        db = MySQLdb.connect(host="localhost", user="root", passwd="root", db="site")
        cursor = db.cursor()
        periods = [1, 3, 7, 15]
        notes = [0, 3 ,4]
        notnull = 0
        somethingatall = 0
        result = ""
        for note in notes:
            notnull = 0
            result = result + "note>=%s: " % note
            for period in periods:
                cursor.execute(
                    """SELECT COUNT(id) AS cid FROM presse 
                    WHERE nid=0
                    AND datec>(NOW()-INTERVAL %s DAY)
                    AND note>=%s""",
                    (period, note)
                )
                rows = cursor.fetchall()
                if rows[0][0]>0:
                    result = result + "%sj:%s, " % (period, rows[0][0])
                    notnull = 1
                    somethingatall = 1
            if notnull:
                result = result[:-2] + "\n"
        if somethingatall==0:
            result = messages["stats_bravo"] % periods[-1]
        self.send_message(channel, result)


if __name__ == '__main__':
    wantzel = Wantzel()
    reactor.run()