wantzel.py 19.7 KB
Newer Older
1 2
#encoding: utf-8
"""
Mindiell's avatar
Mindiell committed
3
Bot Wantzel from La Quadrature du Net.
4

Mindiell's avatar
Mindiell committed
5 6 7 8
License : AGPLv3
Doc     : https://wiki.laquadrature.net/Wantzel

TODO:
9 10 11 12 13
- Ajouter des commandes permettant de gérer le mediakit (moderators only)
    - Parser les urls et voir ce qu'on peut faire:
        - upload de vidéo
        - tag de vidéo
        - obtenir un lien vers une vidéo
14 15
"""

16
import feedparser
17
import HTMLParser
18
import importlib
19 20 21
from irc import IrcClientFactory
import MySQLdb
import re
22 23
import sqlite3
import time
24
from twisted.internet import reactor
25
from twitter import Twitter, OAuth
26 27 28 29 30
import urllib

import config
from messages import messages

Mindiell's avatar
Mindiell committed
31 32 33 34 35 36 37 38 39 40 41
LOG_FILE = "wantzel.log"
DEBUG = 3
WARNING = 2
INFO = 1
ERROR = 0
LOG_LEVEL = DEBUG

class Utils(object):
    @classmethod
    def log(cls, message):
        with open(LOG_FILE, 'a') as f:
42 43 44 45 46 47 48 49
            try:
                f.write("%s: %s\n" % (time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()), message.encode("utf-8")))
            except:
                f.write("%s: Erreur de log\n" % time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()))
                try:
                    f.write("%s: %s\n" % (time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()), message))
                except:
                    f.write("%s: Seconde erreur de log\n" % time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()))
Mindiell's avatar
Mindiell committed
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72

    @classmethod
    def debug(cls, message):
        if LOG_LEVEL>=DEBUG:
            cls.log("%s: %s" % ("DEBUG", message))

    @classmethod
    def warning(cls, message):
        if LOG_LEVEL>=WARNING:
            cls.log("%s: %s" % ("WARNING", message))

    @classmethod
    def info(cls, message):
        if LOG_LEVEL>=INFO:
            cls.log("%s: %s" % ("INFO", message))

    @classmethod
    def error(cls, message):
        if LOG_LEVEL>=ERROR:
            cls.log("%s: %s" % ("ERROR", message))



Mindiell's avatar
Mindiell committed
73 74
def get_cursor():
    """
75
    This function connects to a MySQL database and returns a usable cursor.
Mindiell's avatar
Mindiell committed
76
    """
77
    connection = MySQLdb.connect(
Mindiell's avatar
Mindiell committed
78 79 80 81 82
        host=config.dbserver,
        user=config.dbuser,
        passwd=config.dbpassword,
        db=config.dbname
    )
83 84
    if connection:
        return connection.cursor()
Mindiell's avatar
Mindiell committed
85 86
    return None

87
def get_url(message):
Mindiell's avatar
Mindiell committed
88
    """
89
    Retrieve the url in the message.
Mindiell's avatar
Mindiell committed
90
    """
91 92
    # Let's get the url
    result = re.search("(https?[^ ]+)", message)
Mindiell's avatar
Mindiell committed
93 94
    if not result:
        return "http"
95
    url = result.group(1)
Mindiell's avatar
Mindiell committed
96 97 98 99 100 101 102 103
    # 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

104 105 106
def get_title(message):
    title = ""
    website = ""
107 108 109 110
    exceptions = [
        "twitter.com",
        "github.com",
    ]
111
    try:
112 113 114 115
        url = get_url(message)
        for exception in exceptions:
            if exception in url:
                return ("", "")
116 117 118 119
        website = re.search("//([^/]*)", url).group(1)
        f = urllib.URLopener().open(url)
        content = f.read()
        title = re.search("<title>([^<]+)</title>", content).group(1)
120 121
        # Unescaping HTML entities and removing multiple lines
        title = HTMLParser.HTMLParser().unescape(re.sub("\n|\r", "", title))
122
    except:
123
        return ("", "")
124 125
    return (title, website)

126 127 128 129 130 131 132 133 134 135 136 137
def is_moderator(name):
    """
    This function verify if a user is a moderator.
    """
    connection = sqlite3.connect(config.sqlite_db)
    cursor = connection.cursor()
    cursor.execute("SELECT count(*) FROM moderator WHERE name=?", (name, ))
    if int(cursor.fetchone()[0])==1:
        return True
    return False


138 139 140 141 142 143 144 145
class Wantzel(object):
    """
    Wantzel bot.
    """
    def __init__(self):
        """
        Initialization of bot over IRC.
        """
146
        self.number = None
147 148 149 150 151 152 153 154 155
        # default last_entry_published
        self.last_entry_published = time.strptime("2000-01-01", "%Y-%m-%d")
        # See if there is something in the db
        connection = sqlite3.connect(config.sqlite_db)
        for row in connection.execute("SELECT last_entry_published FROM tweets"):
            self.last_entry_published = time.strptime(
                row[0].encode("utf-8"),
                "%Y-%m-%d %H:%M:%S %Z"
            )
156
        Utils.debug("Dernier tweet: %s" % self.last_entry_published)
157 158 159
        self.irc = IrcClientFactory(config)
        self.irc.set_privmsg = self.set_privmsg
        reactor.connectTCP(config.server, config.port, self.irc)
160 161 162 163 164 165 166
        # Prepare timer
        reactor.callLater(config.timer, self.timer)

    def timer(self):
        """
        This method launches function regularly (see config.timer).
        """
Mindiell's avatar
Mindiell committed
167
        Utils.debug("Timer called")
168 169 170 171 172
        self.rp_to_twitter("http://www.laquadrature.net/fr/revue-de-presse/feed")
        self.rp_to_twitter("http://www.laquadrature.net/en/press-review/feed")
        self.count_articles()
        # Recalling the timer
        reactor.callLater(config.timer, self.timer)
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188

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

    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):
        """
189 190
        Wantzel can understand a lot of commands. Commands followed by a (*)
        are accessible only to moderators:
191 192 193 194 195
        - 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)
Mindiell's avatar
Mindiell committed
196
            Add an article in the database
197
        - stats
Mindiell's avatar
Mindiell committed
198
            Show some statistics about the RP
199
        - kill (*)
Mindiell's avatar
Mindiell committed
200
            Kill an article by giving it a score of -100
201
        - admin list (*)
202
            List rights in private
203
        - admin add (*)
204
            Add a new moderator to list
205
        - admin remove (*)
206
            Remove a moderator from list
207
        - admin timer
208
            Relaunch a timer
209 210
        """
        # Cleaning user name
Mindiell's avatar
Mindiell committed
211
        user = re.search("([^!]*)!", user).group(1)
Mindiell's avatar
Mindiell committed
212
        Utils.debug("Message received: %s %s %s" % (user, channel, msg))
213
        # Whatever is done, get the title of an existing url in a message
214 215 216
        title = ""
        if "http" in msg:
            title, website = get_title(msg)
Mindiell's avatar
Mindiell committed
217 218 219 220 221 222 223
        # 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):
224
                self.help(user, channel, msg)
Mindiell's avatar
Mindiell committed
225
            # Find known command
226
            command = re.search("!(rp[acp]*|kill|help|stats|admin)", msg)
Mindiell's avatar
Mindiell committed
227
            Utils.debug("Command: %s" % command)
Mindiell's avatar
Mindiell committed
228
            if command:
Mindiell's avatar
Mindiell committed
229
                Utils.debug("group(0): %s" % command.group(0))
Mindiell's avatar
Mindiell committed
230
                command = command.group(1)
Mindiell's avatar
Mindiell committed
231
                Utils.debug("Command: %s" % command)
Mindiell's avatar
Mindiell committed
232
                if command.startswith("rp"):
Mindiell's avatar
Mindiell committed
233
                    Utils.debug("Calling self.rp")
234
                    self.rp(command, user, channel, msg)
Mindiell's avatar
Mindiell committed
235
                elif command=="help":
Mindiell's avatar
Mindiell committed
236
                    Utils.debug("Calling self.help")
Mindiell's avatar
Mindiell committed
237 238
                    self.help(user, channel, msg)
                elif command=="kill":
Mindiell's avatar
Mindiell committed
239
                    Utils.debug("Calling self.kill")
Mindiell's avatar
Mindiell committed
240 241
                    self.kill(user, channel, msg)
                elif command=="stats":
Mindiell's avatar
Mindiell committed
242
                    Utils.debug("Calling self.stats")
Mindiell's avatar
Mindiell committed
243
                    self.stats(user, channel, msg)
244
                elif command=="admin":
Mindiell's avatar
Mindiell committed
245
                    Utils.debug("Calling self.admin")
246
                    self.admin(user, channel, msg)
247 248 249
        # No more giving the title of an url
        #if title and website:
        #    self.send_message(channel, messages["title"] % (title, website))
250 251 252 253 254 255 256

    def help(self, user, channel, msg):
        """
        Show global help.
        If a known command is behind the !help command, an adequate message is
        returned.
        """
Mindiell's avatar
Mindiell committed
257
        Utils.debug("help command")
Mindiell's avatar
Mindiell committed
258
        # Searching for a command after help keyword
259
        command = re.search("!help (stats|rp|help|kill|admin)", msg)
Mindiell's avatar
Mindiell committed
260 261
        if command:
            command = command.group(1)
262
            self.send_message(channel, messages["help_"+command])            
Mindiell's avatar
Mindiell committed
263 264
        else:
            self.send_message(channel, messages["help"])
265

266
    def rp(self, command, user, channel, msg):
267 268 269
        """
        Adding the article in rp database.
        """
Mindiell's avatar
Mindiell committed
270 271 272 273
        Utils.debug("rp command : %s" % command)
        Utils.debug("rp user : %s" % user)
        Utils.debug("rp channel : %s" % channel)
        Utils.debug("rp msg : %s" % msg)
274
        cite = 0
275
        note = 1
276
        url = get_url(msg)
Mindiell's avatar
Mindiell committed
277
        Utils.debug("url: %s" % url)
Mindiell's avatar
Mindiell committed
278 279 280 281 282
        if url=="":
            return
        elif url=="http":
            self.send_message(channel, messages["rp_http"] % user)
            return
283

Mindiell's avatar
Mindiell committed
284 285
        # Looking for such an article in database
        cursor = get_cursor()
286 287 288 289 290 291 292 293 294 295 296
        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:
297
                note = -2
Mindiell's avatar
Mindiell committed
298
            Utils.debug("Adding an article by %s: %s" % (user, url))
299
            result = cursor.execute(
300
                """INSERT INTO presse SET 
301
                url=%s, provenance=%s, cite=%s, note=%s, datec=NOW(), title='',
302 303
                lang='', published=0, nid=0, screenshot=0, fetched=0, seemscite=0
                """,
304
                (url, user, cite, note)
305
            )
Mindiell's avatar
Mindiell committed
306
            self.send_message(channel, messages["rp_new_article"] % user)
307 308
        else:
            if rows[0][2]!=user:
Mindiell's avatar
Mindiell committed
309
                Utils.debug("Adding a point by %s on %s" % (user, rows[0][0]))
310 311 312 313 314
                result = cursor.execute(
                    "UPDATE presse SET note=note+1 WHERE id=%s",
                    (rows[0][0], )
                )
            if (rows[0][1]+1)<3:
Mindiell's avatar
Mindiell committed
315
                self.send_message(channel, messages["rp_known_article"] % user)
316
            else:
Mindiell's avatar
Mindiell committed
317
                self.send_message(channel, messages["rp_taken_article"] % user)
318 319
                # Update number of articles to do
                self.count_articles()
320 321 322

    def kill(self, user, channel, msg):
        """
Mindiell's avatar
Mindiell committed
323
        Kill an article by setting its score to -100.
324
        """
Mindiell's avatar
Mindiell committed
325
        Utils.debug("kill command")
326 327
        if is_moderator(user):
            url = get_url(msg)
Mindiell's avatar
Mindiell committed
328
            Utils.debug("url: %s" % url)
329 330 331 332 333 334 335 336 337 338 339 340 341 342
            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)
Mindiell's avatar
Mindiell committed
343
        else:
344
            self.send_message(channel, messages["not_moderator"])
345 346 347 348 349

    def stats(self, user, channel, msg):
        """
        Returns stats on articles in press review.
        """
Mindiell's avatar
Mindiell committed
350
        Utils.debug("stats command")
351
        cursor = get_cursor()
352 353 354 355 356 357 358
        periods = [1, 3, 7, 15]
        notes = [0, 3 ,4]
        notnull = 0
        somethingatall = 0
        result = ""
        for note in notes:
            notnull = 0
Mindiell's avatar
Mindiell committed
359
            result = result + "note>=%s: " % note
360 361 362 363 364 365 366 367 368 369 370 371
            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
Mindiell's avatar
Mindiell committed
372
                    somethingatall = 1
373 374
            if notnull:
                result = result[:-2] + "\n"
Mindiell's avatar
Mindiell committed
375 376
        if somethingatall==0:
            result = messages["stats_bravo"] % periods[-1]
377 378
        self.send_message(channel, result)

379 380 381 382 383
    def admin(self, user, channel, msg):
        """
        Manage moderation.
        A sub-command should be behind the !admin command.
        """
Mindiell's avatar
Mindiell committed
384
        Utils.debug("admin command")
385
        # Searching for a command after admin keyword
386
        command = re.search("!admin (list|add|del|timer)", msg)
387 388 389 390 391 392 393 394
        if command:
            command = command.group(1)
            if command=="list":
                self.admin_list(user, channel, msg)
            elif command=="add":
                self.admin_add(user, channel, msg)
            elif command=="del":
                self.admin_del(user, channel, msg)
395 396
            elif command=="del":
                self.admin_timer(user, channel, msg)
397 398 399 400 401

    def admin_list(self, user, channel, msg):
        """
        List actual moderators.
        """
Mindiell's avatar
Mindiell committed
402
        Utils.debug("admin_list command")
403 404 405 406 407 408 409 410 411 412 413 414 415
        if is_moderator(user):
            connection = sqlite3.connect(config.sqlite_db)
            names = []
            for row in connection.execute("SELECT name FROM moderator"):
                names.append(row[0].encode("utf-8"))
            self.send_message(channel, messages["admin_list"] % ", ".join(names))
        else:
            self.send_message(channel, messages["not_moderator"])

    def admin_add(self, user, channel, msg):
        """
        Add some new moderators if not existing yet.
        """
Mindiell's avatar
Mindiell committed
416
        Utils.debug("admin_add command")
417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
        if is_moderator(user):
            try:
                names = []
                connection = sqlite3.connect(config.sqlite_db)
                result = re.search("!admin add (([^,]+, ?)+)?(.*)", msg)
                if result.group(1):
                    names = [name for name in result.group(1).split(", ") if name!=""]
                names.append(result.group(3))
                # Do not add actual moderators
                moderators = []
                for row in connection.execute("SELECT name FROM moderator"):
                    moderators.append(row[0].encode("utf-8"))
                names = set([name for name in names if name not in moderators])
                if names:
                    # Converting set in list of tuples
                    values = [(name,) for name in names]
                    connection.executemany("INSERT INTO moderator (name) VALUES (?)", values)
                    connection.commit()
                    self.send_message(channel, messages["admin_add"] % ", ".join(names))
                else:
                    self.send_message(channel, messages["admin_add_empty"])
            except:
                pass
        else:
            self.send_message(channel, messages["not_moderator"])

    def admin_del(self, user, channel, msg):
        """
        Delete a moderator from list.
        """
Mindiell's avatar
Mindiell committed
447
        Utils.debug("admin_del command")
448 449 450 451 452 453 454 455
        if is_moderator(user):
            try:
                names = []
                result = re.search("!admin del (([^,]+, ?)+)?(.*)", msg)
                if result.group(1):
                    names = [name for name in result.group(1).split(", ") if name!=""]
                names.append(result.group(3))
                names = set(names)
Mindiell's avatar
Mindiell committed
456
                Utils.debug(names)
457 458 459 460 461 462 463 464 465 466
                connection = sqlite3.connect(config.sqlite_db)
                for name in names:
                    connection.execute("DELETE FROM moderator WHERE name=?", (name, ))
                connection.commit()
                self.send_message(channel, messages["admin_del"] % ", ".join(names))
            except:
                pass
        else:
            self.send_message(channel, messages["not_moderator"])

467 468 469 470 471 472 473 474 475 476 477 478 479 480
    def admin_timer(self, user, channel, msg):
        """
        Relaunch a timer.
        """
        Utils.debug("admin_timer command")
        if is_moderator(user):
            try:
                # Recalling the timer
                reactor.callLater(config.timer, self.timer)
            except:
                pass
        else:
            self.send_message(channel, messages["not_moderator"])

481 482 483 484 485
    def count_articles(self):
        """
        Count number of articles not done in RP and updates the topic of the
        press review channel if needed.
        """
Mindiell's avatar
Mindiell committed
486
        Utils.debug("count_articles method")
487 488 489 490 491 492 493
        cursor = get_cursor()
        cursor.execute("""SELECT COUNT(*) FROM presse
            WHERE DATE_SUB(NOW(), INTERVAL 2 MONTH)<datec
            AND note > 2
            AND nid = 0""")
        rows = cursor.fetchall()
        number = int(rows[0][0])
494
        Utils.debug("Found %s articles." % number)
495
        if self.number!=number:
Mindiell's avatar
Mindiell committed
496
            self.irc.client.topic("#lqdn-rp", messages["topic"] % number)
497 498 499 500 501 502 503
            pass
        self.number = number

    def rp_to_twitter(self, rss):
        """
        By parsing the RSS feed of the press-review, we know what to tweet.
        """
Mindiell's avatar
Mindiell committed
504
        Utils.debug("rp_to_twitter method")
505 506 507 508 509
        now = time.localtime()
        today = time.strptime("%s-%s-%s %s" % (
                now.tm_year,
                now.tm_mon,
                now.tm_mday,
Mindiell's avatar
Mindiell committed
510
                time.tzname[0]
511 512 513
            ),
            "%Y-%m-%d %Z"
        )
Mindiell's avatar
Mindiell committed
514 515 516
        language = "fr"
        if "/en/" in rss:
            language = "en"
517 518
        entries = feedparser.parse(rss)['entries']
        entries.reverse()
519
        Utils.debug(self.last_entry_published)
520 521 522 523 524
        for entry in entries:
            # if date of publication is greater than today, midnight, and
            # lesser than future
            if today < entry.published_parsed < now:
                if self.last_entry_published < entry.published_parsed:
Mindiell's avatar
Mindiell committed
525
                    self.tweet(messages["tweet_rp_%s" % language] % (
526 527 528
                        entry.title.encode("utf-8"),
                        entry.link.encode("utf-8")
                    ))
Mindiell's avatar
Mindiell committed
529
                    Utils.debug(entry.published_parsed)
Mindiell's avatar
Mindiell committed
530
                    Utils.debug(entry.title)
531 532 533 534 535 536 537 538 539 540 541 542 543 544
                    # Save last_entry_published
                    self.last_entry_published = entry.published_parsed
                    last_entry_published = time.strftime(
                        "%Y-%m-%d %H:%M:%S %Z",
                        self.last_entry_published
                    )
                    connection = sqlite3.connect(config.sqlite_db)
                    connection.execute(
                        "UPDATE tweets SET last_entry_published=?",
                        (last_entry_published,)
                    )
                    connection.commit()
                    # Tweet only one message in order not to spam
                    return
545 546 547
                else:
                    Utils.debug(entry.title)
                    Utils.debug(entry.published_parsed)
548 549 550 551 552

    def tweet(self, message):
        """
        Tweet message on specified account
        """
Mindiell's avatar
Mindiell committed
553
        Utils.debug("tweet method")
554 555 556 557 558 559 560 561
        auth = OAuth(
            config.TOKEN,
            config.TOKENSEC,
            config.CONSKEY,
            config.CONSSEC
        )
        twitter = Twitter(auth=auth)
        try:
Mindiell's avatar
Mindiell committed
562
            Utils.debug("Tweeting: %s" % message)
Mindiell's avatar
Mindiell committed
563
            twitter.statuses.update(status=message)
564 565 566
        except:
            pass

567 568 569 570

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