+
+ # For each snippet of select_length, use markov() to find continuation
+ # token from selections. Replace present users' names with malkovich.
+ # Start snippets with the beginning of a sentence, if possible.
+ for i in range(len(tokens) - select_length):
+ token_list = []
+ for j in range(select_length + 1):
+ token_list += [tokens[i + j]]
+ selections += [token_list]
+ snippet = []
+ for i in range(select_length):
+ snippet += [""]
+ shuffle(selections)
+ for i in range(len(selections)):
+ if selections[i][0][-1] in sentence_end_markers:
+ for i in range(select_length):
+ snippet[i] = selections[i][i + 1]
+ break
+ msg = ""
+ malkovich = "malkovich"
+ while 1:
+ new_end = markov(snippet)
+ for name in session.users_in_chan:
+ if new_end[:len(name)] == name.lower():
+ new_end = malkovich + new_end[len(name):]
+ break
+ if len(msg) + len(new_end) > 200:
+ break
+ msg += new_end + " "
+ for i in range(select_length - 1):
+ snippet[i] = snippet[i + 1]
+ snippet[select_length - 1] = new_end
+
+ # Replace occurences of url escape string with random choice from urls.
+ while True:
+ index = msg.find(url_escape)
+ if index < 0:
+ break
+ msg = msg.replace(url_escape, choice(urls), 1)
+
+ # More meaningful ways to randomly end sentences.
+ notice(msg + malkovich + ".")
+
+ def twt():
+ def try_open(mode):
+ try:
+ twtfile = open(session.twtfile, mode)
+ except (PermissionError, FileNotFoundError) as err:
+ notice("CAN'T ACCESS OR CREATE TWT FILE: " + str(err))
+ return None
+ return twtfile
+
+ from datetime import datetime
+ if not os.access(session.twtfile, os.F_OK):
+ twtfile = try_open("w")
+ if None == twtfile:
+ return
+ twtfile.close()
+ twtfile = try_open("a")
+ if None == twtfile:
+ return
+ twtfile.write(datetime.utcnow().isoformat() + "\t" + argument + "\n")
+ twtfile.close()
+ notice("WROTE TWT.")
+
+ if "addquote" == command:
+ addquote()
+ elif "quote" == command:
+ quote()
+ elif "markov" == command:
+ markov()
+ elif "twt" == command:
+ twt()
+
+
+def handle_url(url, notice, show_url=False):
+
+ def mobile_twitter_hack(url):
+ re1 = 'https?://(mobile.twitter.com/)[^/]+(/status/)'
+ re2 = 'https?://mobile.twitter.com/([^/]+)/status/([^\?/]+)'
+ m = re.search(re1, url)
+ if m and m.group(1) == 'mobile.twitter.com/' \
+ and m.group(2) == '/status/':
+ m = re.search(re2, url)
+ url = 'https://twitter.com/' + m.group(1) + '/status/' + m.group(2)
+ handle_url(url, notice, True)
+ return True
+
+ try:
+ r = requests.get(url, timeout=15)
+ except (requests.exceptions.TooManyRedirects,
+ requests.exceptions.ConnectionError,
+ requests.exceptions.InvalidURL,
+ UnicodeError,
+ requests.exceptions.InvalidSchema) as error:
+ notice("TROUBLE FOLLOWING URL: " + str(error))
+ return
+ if mobile_twitter_hack(url):
+ return
+ title = bs4.BeautifulSoup(r.text, "html5lib").title
+ if title and title.string:
+ prefix = "PAGE TITLE: "
+ if show_url:
+ prefix = "PAGE TITLE FOR <" + url + ">: "
+ notice(prefix + title.string.strip())
+ else:
+ notice("PAGE HAS NO TITLE TAG")
+
+
+class Session:
+
+ def __init__(self, io, username, nickname, channel, twtfile, dbdir, rmlogs):
+ self.io = io
+ self.nickname = nickname
+ self.username = username
+ self.channel = channel
+ self.users_in_chan = []
+ self.twtfile = twtfile
+ self.dbdir = dbdir
+ self.rmlogs = rmlogs
+ self.io.send_line("NICK " + self.nickname)
+ self.io.send_line("USER " + self.username + " 0 * : ")
+ self.io.send_line("JOIN " + self.channel)
+ hash_channel = hashlib.md5(self.channel.encode("utf-8")).hexdigest()
+ self.chandir = self.dbdir + "/" + hash_channel + "/"
+ self.rawlogdir = self.chandir + "raw_logs/"
+ self.logdir = self.chandir + "logs/"
+ if not os.path.exists(self.logdir):
+ os.makedirs(self.logdir)
+ if not os.path.exists(self.rawlogdir):
+ os.makedirs(self.rawlogdir)
+ self.markovfile = self.chandir + "markovfeed"
+ self.quotesfile = self.chandir + "quotes"
+
+ def loop(self):
+
+ def log(line):
+ if type(line) == str:
+ line = Line(":" + self.nickname + "!~" + self.username +
+ "@localhost" + " " + line)
+ now = datetime.datetime.utcnow()
+ form = "%Y-%m-%d %H:%M:%S UTC\t"
+ write_to_file(self.rawlogdir + now.strftime("%Y-%m-%d") + ".txt",
+ "a", now.strftime(form) + " " + line.line + "\n")
+ to_log = irclog.format_logline(line, self.channel)
+ if to_log != None:
+ write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt",
+ "a", now.strftime(form) + " " + to_log + "\n")
+
+ def handle_privmsg(line):
+
+ def notice(msg):
+ line = "NOTICE " + target + " :" + msg
+ self.io.send_line(line)
+ log(line)
+
+ target = line.sender
+ if line.receiver != self.nickname:
+ target = line.receiver
+ msg = str.join(" ", line.tokens[3:])[1:]
+ matches = re.findall("(https?://[^\s>]+)", msg)
+ for i in range(len(matches)):
+ handle_url(matches[i], notice)
+ if "!" == msg[0]:
+ tokens = msg[1:].split()
+ argument = str.join(" ", tokens[1:])
+ handle_command(tokens[0], argument, notice, target, self)
+ return
+ write_to_file(self.markovfile, "a", msg + "\n")
+
+ now = datetime.datetime.utcnow()
+ write_to_file(self.logdir + now.strftime("%Y-%m-%d") + ".txt", "a",
+ "-----------------------\n")
+ while True:
+ if self.rmlogs > 0:
+ for f in os.listdir(self.logdir):
+ f = os.path.join(self.logdir, f)
+ if os.path.isfile(f) and \
+ os.stat(f).st_mtime < time.time() - self.rmlogs:
+ os.remove(f)
+ line = self.io.recv_line()
+ if not line:
+ continue
+ line = Line(line)
+ log(line)
+ if len(line.tokens) > 1:
+ if line.tokens[0] == "PING":
+ self.io.send_line("PONG " + line.tokens[1])
+ elif line.tokens[1] == "PRIVMSG":
+ handle_privmsg(line)
+ elif line.tokens[1] == "353":
+ names = line.tokens[5:]
+ names[0] = names[0][1:]
+ for i in range(len(names)):
+ names[i] = names[i].replace("@", "").replace("+", "")
+ self.users_in_chan += names
+ elif line.tokens[1] == "JOIN" and line.sender != self.nickname:
+ self.users_in_chan += [line.sender]
+ elif line.tokens[1] == "PART":
+ del(self.users_in_chan[self.users_in_chan.index(line.sender)])
+ elif line.tokens[1] == "NICK":
+ del(self.users_in_chan[self.users_in_chan.index(line.sender)])
+ self.users_in_chan += [line.receiver]
+
+
+def parse_command_line_arguments():
+ parser = argparse.ArgumentParser()
+ parser.add_argument("-s, --server", action="store", dest="server",
+ default=SERVER,
+ help="server or server net to connect to (default: "
+ + SERVER + ")")
+ parser.add_argument("-p, --port", action="store", dest="port", type=int,
+ default=PORT, help="port to connect to (default : "
+ + str(PORT) + ")")
+ parser.add_argument("-w, --wait", action="store", dest="timeout",
+ type=int, default=TIMEOUT,
+ help="timeout in seconds after which to attempt "
+ "reconnect (default: " + str(TIMEOUT) + ")")
+ parser.add_argument("-u, --username", action="store", dest="username",
+ default=USERNAME, help="username to use (default: "
+ + USERNAME + ")")
+ parser.add_argument("-n, --nickname", action="store", dest="nickname",
+ default=NICKNAME, help="nickname to use (default: "
+ + NICKNAME + ")")
+ parser.add_argument("-t, --twtxtfile", action="store", dest="twtfile",
+ default=TWTFILE, help="twtxt file to use (default: "
+ + TWTFILE + ")")
+ parser.add_argument("-d, --dbdir", action="store", dest="dbdir",
+ default=DBDIR, help="directory to store DB files in")
+ parser.add_argument("-r, --rmlogs", action="store", dest="rmlogs",
+ type=int, default=0,
+ help="maximum age in seconds for logfiles in logs/ "
+ "(0 means: never delete, and is default)")
+ parser.add_argument("CHANNEL", action="store", help="channel to join")
+ opts, unknown = parser.parse_known_args()
+ return opts
+
+
+opts = parse_command_line_arguments()
+while True:
+ try:
+ io = IO(opts.server, opts.port, opts.timeout)
+ hash_server = hashlib.md5(opts.server.encode("utf-8")).hexdigest()
+ dbdir = opts.dbdir + "/" + hash_server
+ session = Session(io, opts.username, opts.nickname, opts.CHANNEL,
+ opts.twtfile, dbdir, opts.rmlogs)
+ session.loop()
+ except ExceptionForRestart:
+ io.socket.close()
+ continue