home · contact · privacy
Use request library to simplify page retrieval code.
[plomlombot-irc.git] / 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 requests
10 import html
11 import html.parser
12
13 # Defaults, may be overwritten by command line arguments.
14 SERVER = "irc.freenode.net"
15 PORT = 6667
16 TIMEOUT = 240
17 USERNAME = "plomlombot"
18 NICKNAME = USERNAME
19
20
21 class HTMLParser(html.parser.HTMLParser):
22     def __init__(self, html, tag):
23         super().__init__()
24         self._tag_to_check = tag
25         self._tag = ""
26         self.data = ""
27         self.feed(html)
28     def handle_starttag(self, tag, attrs):
29         if self.data == "" and tag == self._tag_to_check:
30             self._tag = tag
31     def handle_endtag(self, tag):
32         self._tag = ""
33     def handle_data(self, data):
34         if self._tag != "":
35             self.data = data
36
37
38 class ExceptionForRestart(Exception):
39     pass
40
41
42 class IO:
43
44     def __init__(self, server, port, timeout):
45         self.timeout = timeout
46         self.socket = socket.socket()
47         self.socket.connect((server, port))
48         self.socket.setblocking(0)
49         self.line_buffer = []
50         self.rune_buffer = ""
51         self.last_pong = time.time()
52         self.servername = self.recv_line(send_ping=False).split(" ")[0][1:]
53
54     def _pingtest(self, send_ping=True):
55         if self.last_pong + self.timeout < time.time():
56             print("SERVER NOT ANSWERING")
57             raise ExceptionForRestart
58         if send_ping:
59             self.send_line("PING " + self.servername)
60
61     def send_line(self, msg):
62         msg = msg.replace("\r", " ")
63         msg = msg.replace("\n", " ")
64         if len(msg.encode("utf-8")) > 510:
65             print("NOT SENT LINE TO SERVER (too long): " + msg)
66         print("LINE TO SERVER: "
67               + str(datetime.datetime.now()) + ": " + msg)
68         msg = msg + "\r\n"
69         msg_len = len(msg)
70         total_sent_len = 0
71         while total_sent_len < msg_len:
72             sent_len = self.socket.send(bytes(msg[total_sent_len:], "UTF-8"))
73             if sent_len == 0:
74                 print("SOCKET CONNECTION BROKEN")
75                 raise ExceptionForRestart
76             total_sent_len += sent_len
77
78     def _recv_line_wrapped(self, send_ping=True):
79         if len(self.line_buffer) > 0:
80             return self.line_buffer.pop(0)
81         while True:
82             ready = select.select([self.socket], [], [], int(self.timeout / 2))
83             if not ready[0]:
84                 self._pingtest(send_ping)
85                 return None
86             self.last_pong = time.time()
87             received_bytes = self.socket.recv(1024)
88             try:
89                 received_runes = received_bytes.decode("UTF-8")
90             except UnicodeDecodeError:
91                 received_runes = received_bytes.decode("latin1")
92             if len(received_runes) == 0:
93                 print("SOCKET CONNECTION BROKEN")
94                 raise ExceptionForRestart
95             self.rune_buffer += received_runes
96             lines_split = str.split(self.rune_buffer, "\r\n")
97             self.line_buffer += lines_split[:-1]
98             self.rune_buffer = lines_split[-1]
99             if len(self.line_buffer) > 0:
100                 return self.line_buffer.pop(0)
101
102     def recv_line(self, send_ping=True):
103         line = self._recv_line_wrapped(send_ping)
104         if line:
105             print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " +
106                   line)
107         return line
108
109
110 def init_session(server, port, timeout, nickname, username, channel):
111     print("CONNECTING TO " + server)
112     io = IO(server, port, timeout)
113     io.send_line("NICK " + nickname)
114     io.send_line("USER " + username + " 0 * : ")
115     io.send_line("JOIN " + channel)
116     return io
117
118
119 def lineparser_loop(io, nickname):
120
121     def act_on_privmsg(tokens):
122
123         def url_check(msg):
124
125             def notice(msg):
126                 io.send_line("NOTICE " + target + " :" + msg)
127
128             matches = re.findall("(https?://[^\s>]+)", msg)
129             for i in range(len(matches)):
130                 url = matches[i]
131                 try:
132                     r = requests.get(url, timeout=15)
133                 except (requests.exceptions.TooManyRedirects,
134                         requests.exceptions.ConnectionError,
135                         requests.exceptions.InvalidSchema) as error:
136                     notice("TROUBLE FOLLOWING URL: " + str(error))
137                     continue
138                 content = r.text
139                 title = HTMLParser(content, "title").data
140                 title = html.unescape(title)
141                 notice("PAGE TITLE FOR URL: " + title)
142
143         sender = ""
144         for rune in tokens[0]:
145             if rune == "!":
146                 break
147             if rune != ":":
148                 sender += rune
149         receiver = ""
150         for rune in tokens[2]:
151             if rune == "!":
152                 break
153             if rune != ":":
154                 receiver += rune
155         target = sender
156         if receiver != nickname:
157             target = receiver
158         msg = str.join(" ", tokens[3:])[1:]
159         url_check(msg)
160
161     while True:
162         line = io.recv_line()
163         if not line:
164             continue
165         tokens = line.split(" ")
166         if len(tokens) > 1:
167             if tokens[1] == "PRIVMSG":
168                 act_on_privmsg(tokens)
169             if tokens[0] == "PING":
170                 io.send_line("PONG " + tokens[1])
171
172
173 def parse_command_line_arguments():
174     parser = argparse.ArgumentParser()
175     parser.add_argument("-s, --server", action="store", dest="server",
176                         default=SERVER,
177                         help="server or server net to connect to (default: "
178                         + SERVER + ")")
179     parser.add_argument("-p, --port", action="store", dest="port", type=int,
180                         default=PORT, help="port to connect to (default : "
181                         + str(PORT) + ")")
182     parser.add_argument("-t, --timeout", action="store", dest="timeout",
183                         type=int, default=TIMEOUT,
184                         help="timeout in seconds after which to attempt " +
185                         "reconnect (default: " + str(TIMEOUT) + ")")
186     parser.add_argument("-u, --username", action="store", dest="username",
187                         default=USERNAME, help="username to use (default: "
188                         + USERNAME + ")")
189     parser.add_argument("-n, --nickname", action="store", dest="nickname",
190                         default=NICKNAME, help="nickname to use (default: "
191                         + NICKNAME + ")")
192     parser.add_argument("CHANNEL", action="store", help="channel to join")
193     opts, unknown = parser.parse_known_args()
194     return opts
195
196 opts = parse_command_line_arguments()
197 while True:
198     try:
199         io = init_session(opts.server, opts.port, opts.timeout, opts.nickname,
200                           opts.username, opts.CHANNEL)
201         lineparser_loop(io, opts.nickname)
202     except ExceptionForRestart:
203         io.socket.close()
204         continue