plomlombot.py

   1 #!/usr/bin/python3
   2
   3 import argparse
   4 import socket
   5 import datetime
   6 import select
   7 import time
   8 import re
   9 import urllib.request
  10 import http.client
  11 import html
  12 import html.parser
  13
  14 # Defaults, may be overwritten by command line arguments.
  15 SERVER = "irc.freenode.net"
  16 PORT = 6667
  17 TIMEOUT = 240
  18 USERNAME = "plomlombot"
  19 NICKNAME = USERNAME
  20
  21
  22 class HTMLParser(html.parser.HTMLParser):
  23     def __init__(self, html, tag):
  24         super().__init__()
  25         self._tag_to_check = tag
  26         self._tag = ""
  27         self.data = ""
  28         self.feed(html)
  29     def handle_starttag(self, tag, attrs):
  30         if self.data == "" and tag == self._tag_to_check:
  31             self._tag = tag
  32     def handle_endtag(self, tag):
  33         self._tag = ""
  34     def handle_data(self, data):
  35         if self._tag != "":
  36             self.data = data
  37
  38
  39 class ExceptionForRestart(Exception):
  40     pass
  41
  42
  43 class IO:
  44
  45     def __init__(self, server, port, timeout):
  46         self.timeout = timeout
  47         self.socket = socket.socket()
  48         self.socket.connect((server, port))
  49         self.socket.setblocking(0)
  50         self.line_buffer = []
  51         self.rune_buffer = ""
  52         self.last_pong = time.time()
  53         self.servername = self.recv_line(send_ping=False).split(" ")[0][1:]
  54
  55     def _pingtest(self, send_ping=True):
  56         if self.last_pong + self.timeout < time.time():
  57             print("SERVER NOT ANSWERING")
  58             raise ExceptionForRestart
  59         if send_ping:
  60             self.send_line("PING " + self.servername)
  61
  62     def send_line(self, msg):
  63         msg = msg.replace("\r", " ")
  64         msg = msg.replace("\n", " ")
  65         if len(msg.encode("utf-8")) > 510:
  66             print("NOT SENT LINE TO SERVER (too long): " + msg)
  67         print("LINE TO SERVER: "
  68               + str(datetime.datetime.now()) + ": " + msg)
  69         msg = msg + "\r\n"
  70         msg_len = len(msg)
  71         total_sent_len = 0
  72         while total_sent_len < msg_len:
  73             sent_len = self.socket.send(bytes(msg[total_sent_len:], "UTF-8"))
  74             if sent_len == 0:
  75                 print("SOCKET CONNECTION BROKEN")
  76                 raise ExceptionForRestart
  77             total_sent_len += sent_len
  78
  79     def _recv_line_wrapped(self, send_ping=True):
  80         if len(self.line_buffer) > 0:
  81             return self.line_buffer.pop(0)
  82         while True:
  83             ready = select.select([self.socket], [], [], int(self.timeout / 2))
  84             if not ready[0]:
  85                 self._pingtest(send_ping)
  86                 return None
  87             self.last_pong = time.time()
  88             received_bytes = self.socket.recv(1024)
  89             try:
  90                 received_runes = received_bytes.decode("UTF-8")
  91             except UnicodeDecodeError:
  92                 received_runes = received_bytes.decode("latin1")
  93             if len(received_runes) == 0:
  94                 print("SOCKET CONNECTION BROKEN")
  95                 raise ExceptionForRestart
  96             self.rune_buffer += received_runes
  97             lines_split = str.split(self.rune_buffer, "\r\n")
  98             self.line_buffer += lines_split[:-1]
  99             self.rune_buffer = lines_split[-1]
 100             if len(self.line_buffer) > 0:
 101                 return self.line_buffer.pop(0)
 102
 103     def recv_line(self, send_ping=True):
 104         line = self._recv_line_wrapped(send_ping)
 105         if line:
 106             print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " +
 107                   line)
 108         return line
 109
 110
 111 def init_session(server, port, timeout, nickname, username, channel):
 112     print("CONNECTING TO " + server)
 113     io = IO(server, port, timeout)
 114     io.send_line("NICK " + nickname)
 115     io.send_line("USER " + username + " 0 * : ")
 116     io.send_line("JOIN " + channel)
 117     return io
 118
 119
 120 def lineparser_loop(io, nickname):
 121
 122     def act_on_privmsg(tokens):
 123
 124         def url_check(msg):
 125
 126             def notice(msg):
 127                 io.send_line("NOTICE " + target + " :" + msg)
 128
 129             matches = re.findall("(https?://[^\s>]+)", msg)
 130             for i in range(len(matches)):
 131                 url = matches[i]
 132                 request = urllib.request.Request(url, headers={
 133                     "User-Agent": "plomlombot"
 134                 })
 135                 try:
 136                     webpage = urllib.request.urlopen(request, timeout=15)
 137                 except (urllib.error.HTTPError, urllib.error.URLError,
 138                         UnicodeError, http.client.BadStatusLine) as error:
 139                     notice("TROUBLE FOLLOWING URL: " + str(error))
 140                     continue
 141                 charset = webpage.info().get_content_charset()
 142                 if not charset:
 143                     notice("TROUBLE READING PAGE TITLE: no charset in header")
 144                     continue
 145                 content_type = webpage.info().get_content_type()
 146                 if content_type not in ('text/html', 'text/xml',
 147                                         'application/xhtml+xml'):
 148                     notice("TROUBLE READING PAGE TITLE: bad content type "
 149                            + content_type)
 150                     continue
 151                 content = webpage.read().decode(charset)
 152                 title = HTMLParser(content, "title").data
 153                 title = html.unescape(title)
 154                 notice("PAGE TITLE FOR URL: " + title)
 155
 156         sender = ""
 157         for rune in tokens[0]:
 158             if rune == "!":
 159                 break
 160             if rune != ":":
 161                 sender += rune
 162         receiver = ""
 163         for rune in tokens[2]:
 164             if rune == "!":
 165                 break
 166             if rune != ":":
 167                 receiver += rune
 168         target = sender
 169         if receiver != nickname:
 170             target = receiver
 171         msg = str.join(" ", tokens[3:])[1:]
 172         url_check(msg)
 173
 174     while True:
 175         line = io.recv_line()
 176         if not line:
 177             continue
 178         tokens = line.split(" ")
 179         if len(tokens) > 1:
 180             if tokens[1] == "PRIVMSG":
 181                 act_on_privmsg(tokens)
 182             if tokens[0] == "PING":
 183                 io.send_line("PONG " + tokens[1])
 184
 185
 186 def parse_command_line_arguments():
 187     parser = argparse.ArgumentParser()
 188     parser.add_argument("-s, --server", action="store", dest="server",
 189                         default=SERVER,
 190                         help="server or server net to connect to (default: "
 191                         + SERVER + ")")
 192     parser.add_argument("-p, --port", action="store", dest="port", type=int,
 193                         default=PORT, help="port to connect to (default : "
 194                         + str(PORT) + ")")
 195     parser.add_argument("-t, --timeout", action="store", dest="timeout",
 196                         type=int, default=TIMEOUT,
 197                         help="timeout in seconds after which to attempt " +
 198                         "reconnect (default: " + str(TIMEOUT) + ")")
 199     parser.add_argument("-u, --username", action="store", dest="username",
 200                         default=USERNAME, help="username to use (default: "
 201                         + USERNAME + ")")
 202     parser.add_argument("-n, --nickname", action="store", dest="nickname",
 203                         default=NICKNAME, help="nickname to use (default: "
 204                         + NICKNAME + ")")
 205     parser.add_argument("CHANNEL", action="store", help="channel to join")
 206     opts, unknown = parser.parse_known_args()
 207     return opts
 208
 209 opts = parse_command_line_arguments()
 210 while True:
 211     try:
 212         io = init_session(opts.server, opts.port, opts.timeout, opts.nickname,
 213                           opts.username, opts.CHANNEL)
 214         lineparser_loop(io, opts.nickname)
 215     except ExceptionForRestart:
 216         io.socket.close()
 217         continue