home · contact · privacy
Use proper HTML parsing for page title retrieval.
[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 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 = ""
26         self.data = ""
27         self.feed(html)
28     def handle_starttag(self, tag, attrs):
29         if self.data == "":
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                 request = urllib.request.Request(url, headers={
132                     "User-Agent": "plomlombot"
133                 })
134                 try:
135                     webpage = urllib.request.urlopen(request, timeout=15)
136                 except (urllib.error.HTTPError, urllib.error.URLError,
137                         UnicodeError, http.client.BadStatusLine) as error:
138                     notice("TROUBLE FOLLOWING URL: " + str(error))
139                     continue
140                 charset = webpage.info().get_content_charset()
141                 if not charset:
142                     notice("TROUBLE READING PAGE TITLE: no charset in header")
143                     continue
144                 content_type = webpage.info().get_content_type()
145                 if content_type not in ('text/html', 'text/xml',
146                                         'application/xhtml+xml'):
147                     notice("TROUBLE READING PAGE TITLE: bad content type "
148                            + content_type)
149                     continue
150                 content = webpage.read().decode(charset)
151                 title = HTMLParser(content, "title").data
152                 title = html.unescape(title)
153                 notice("PAGE TITLE FOR URL: " + title)
154
155         sender = ""
156         for rune in tokens[0]:
157             if rune == "!":
158                 break
159             if rune != ":":
160                 sender += rune
161         receiver = ""
162         for rune in tokens[2]:
163             if rune == "!":
164                 break
165             if rune != ":":
166                 receiver += rune
167         target = sender
168         if receiver != nickname:
169             target = receiver
170         msg = str.join(" ", tokens[3:])[1:]
171         url_check(msg)
172
173     while True:
174         line = io.recv_line()
175         if not line:
176             continue
177         tokens = line.split(" ")
178         if len(tokens) > 1:
179             if tokens[1] == "PRIVMSG":
180                 act_on_privmsg(tokens)
181             if tokens[0] == "PING":
182                 io.send_line("PONG " + tokens[1])
183
184
185 def parse_command_line_arguments():
186     parser = argparse.ArgumentParser()
187     parser.add_argument("-s, --server", action="store", dest="server",
188                         default=SERVER,
189                         help="server or server net to connect to (default: "
190                         + SERVER + ")")
191     parser.add_argument("-p, --port", action="store", dest="port", type=int,
192                         default=PORT, help="port to connect to (default : "
193                         + str(PORT) + ")")
194     parser.add_argument("-t, --timeout", action="store", dest="timeout",
195                         type=int, default=TIMEOUT,
196                         help="timeout in seconds after which to attempt " +
197                         "reconnect (default: " + str(TIMEOUT) + ")")
198     parser.add_argument("-u, --username", action="store", dest="username",
199                         default=USERNAME, help="username to use (default: "
200                         + USERNAME + ")")
201     parser.add_argument("-n, --nickname", action="store", dest="nickname",
202                         default=NICKNAME, help="nickname to use (default: "
203                         + NICKNAME + ")")
204     parser.add_argument("CHANNEL", action="store", help="channel to join")
205     opts, unknown = parser.parse_known_args()
206     return opts
207
208 opts = parse_command_line_arguments()
209 while True:
210     try:
211         io = init_session(opts.server, opts.port, opts.timeout, opts.nickname,
212                           opts.username, opts.CHANNEL)
213         lineparser_loop(io, opts.nickname)
214     except ExceptionForRestart:
215         io.socket.close()
216         continue