home · contact · privacy
Handle mobile twitter status URLs properly.
[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 bs4
11 import random
12 import hashlib
13 import os
14
15 # Defaults, may be overwritten by command line arguments.
16 SERVER = "irc.freenode.net"
17 PORT = 6667
18 TIMEOUT = 240
19 USERNAME = "plomlombot"
20 NICKNAME = USERNAME
21
22
23 class ExceptionForRestart(Exception):
24     pass
25
26
27 class IO:
28
29     def __init__(self, server, port, timeout):
30         self.timeout = timeout
31         self.socket = socket.socket()
32         self.socket.connect((server, port))
33         self.socket.setblocking(0)
34         self.line_buffer = []
35         self.rune_buffer = ""
36         self.last_pong = time.time()
37         self.servername = self.recv_line(send_ping=False).split(" ")[0][1:]
38
39     def _pingtest(self, send_ping=True):
40         if self.last_pong + self.timeout < time.time():
41             print("SERVER NOT ANSWERING")
42             raise ExceptionForRestart
43         if send_ping:
44             self.send_line("PING " + self.servername)
45
46     def send_line(self, msg):
47         msg = msg.replace("\r", " ")
48         msg = msg.replace("\n", " ")
49         if len(msg.encode("utf-8")) > 510:
50             print("NOT SENT LINE TO SERVER (too long): " + msg)
51         print("LINE TO SERVER: "
52               + str(datetime.datetime.now()) + ": " + msg)
53         msg = msg + "\r\n"
54         msg_len = len(msg)
55         total_sent_len = 0
56         while total_sent_len < msg_len:
57             sent_len = self.socket.send(bytes(msg[total_sent_len:], "UTF-8"))
58             if sent_len == 0:
59                 print("SOCKET CONNECTION BROKEN")
60                 raise ExceptionForRestart
61             total_sent_len += sent_len
62
63     def _recv_line_wrapped(self, send_ping=True):
64         if len(self.line_buffer) > 0:
65             return self.line_buffer.pop(0)
66         while True:
67             ready = select.select([self.socket], [], [], int(self.timeout / 2))
68             if not ready[0]:
69                 self._pingtest(send_ping)
70                 return None
71             self.last_pong = time.time()
72             received_bytes = self.socket.recv(1024)
73             try:
74                 received_runes = received_bytes.decode("UTF-8")
75             except UnicodeDecodeError:
76                 received_runes = received_bytes.decode("latin1")
77             if len(received_runes) == 0:
78                 print("SOCKET CONNECTION BROKEN")
79                 raise ExceptionForRestart
80             self.rune_buffer += received_runes
81             lines_split = str.split(self.rune_buffer, "\r\n")
82             self.line_buffer += lines_split[:-1]
83             self.rune_buffer = lines_split[-1]
84             if len(self.line_buffer) > 0:
85                 return self.line_buffer.pop(0)
86
87     def recv_line(self, send_ping=True):
88         line = self._recv_line_wrapped(send_ping)
89         if line:
90             print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " +
91                   line)
92         return line
93
94
95 def init_session(server, port, timeout, nickname, username, channel):
96     print("CONNECTING TO " + server)
97     io = IO(server, port, timeout)
98     io.send_line("NICK " + nickname)
99     io.send_line("USER " + username + " 0 * : ")
100     io.send_line("JOIN " + channel)
101     return io
102
103
104 def lineparser_loop(io, nickname):
105
106     def act_on_privmsg(tokens):
107
108         def notice(msg):
109             io.send_line("NOTICE " + target + " :" + msg)
110
111         def url_check(msg):
112
113             def handle_url(url):
114
115                 def mobile_twitter_hack(url):
116                     re1 = 'https?://(mobile.twitter.com/)[^/]+(/status/)'
117                     re2 = 'https?://mobile.twitter.com/([^/]+)/status/([^\?]+)'
118                     m = re.search(re1, url)
119                     if m and m.group(1) == 'mobile.twitter.com/' \
120                             and m.group(2) == '/status/':
121                         m = re.search(re2, url)
122                         url = 'https://twitter.com/' + m.group(1) + '/status/' \
123                                 + m.group(2)
124                         handle_url(url)
125                         return True
126
127                 try:
128                     r = requests.get(url, timeout=15)
129                 except (requests.exceptions.TooManyRedirects,
130                         requests.exceptions.ConnectionError,
131                         requests.exceptions.InvalidURL,
132                         requests.exceptions.InvalidSchema) as error:
133                     notice("TROUBLE FOLLOWING URL: " + str(error))
134                     return
135                 if mobile_twitter_hack(url):
136                     return
137                 title = bs4.BeautifulSoup(r.text).title
138                 if title:
139                     notice("PAGE TITLE: " + title.string.strip())
140                 else:
141                     notice("PAGE HAS NO TITLE TAG")
142
143             matches = re.findall("(https?://[^\s>]+)", msg)
144             for i in range(len(matches)):
145                 handle_url(matches[i])
146
147         def command_check(msg):
148             if msg[0] != "!":
149                 return
150             tokens = msg[1:].split()
151             hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
152             quotesfile_name = "quotes_" + hash_string
153             if tokens[0] == "addquote":
154                 if not os.access(quotesfile_name, os.F_OK):
155                     quotesfile = open(quotesfile_name, "w")
156                     quotesfile.write("QUOTES FOR " + target + ":\n")
157                     quotesfile.close()
158                 quotesfile = open(quotesfile_name, "a")
159                 quotesfile.write(str.join(" ", tokens[1:]) + "\n")
160                 quotesfile.close()
161                 quotesfile = open(quotesfile_name, "r")
162                 lines = quotesfile.readlines()
163                 quotesfile.close()
164                 notice("ADDED QUOTE #" + str(len(lines) - 1))
165             elif tokens[0] == "quote":
166                 if len(tokens) > 2 or \
167                     (len(tokens) == 2 and not tokens[1].isdigit()):
168                     notice("SYNTAX: !quote [int]")
169                     return
170                 if not os.access(quotesfile_name, os.F_OK):
171                     notice("NO QUOTES AVAILABLE")
172                     return
173                 quotesfile = open(quotesfile_name, "r")
174                 lines = quotesfile.readlines()
175                 quotesfile.close()
176                 lines = lines[1:]
177                 if len(tokens) == 2:
178                     i = int(tokens[1])
179                     if i == 0 or i > len(lines):
180                         notice("THERE'S NO QUOTE OF THAT INDEX")
181                         return
182                     i = i - 1
183                 else:
184                     i = random.randrange(len(lines))
185                 notice("QUOTE #" + str(i + 1) + ": " + lines[i])
186
187         sender = ""
188         for rune in tokens[0]:
189             if rune == "!":
190                 break
191             if rune != ":":
192                 sender += rune
193         receiver = ""
194         for rune in tokens[2]:
195             if rune == "!":
196                 break
197             if rune != ":":
198                 receiver += rune
199         target = sender
200         if receiver != nickname:
201             target = receiver
202         msg = str.join(" ", tokens[3:])[1:]
203         command_check(msg)
204         url_check(msg)
205
206     while True:
207         line = io.recv_line()
208         if not line:
209             continue
210         tokens = line.split(" ")
211         if len(tokens) > 1:
212             if tokens[1] == "PRIVMSG":
213                 act_on_privmsg(tokens)
214             if tokens[0] == "PING":
215                 io.send_line("PONG " + tokens[1])
216
217
218 def parse_command_line_arguments():
219     parser = argparse.ArgumentParser()
220     parser.add_argument("-s, --server", action="store", dest="server",
221                         default=SERVER,
222                         help="server or server net to connect to (default: "
223                         + SERVER + ")")
224     parser.add_argument("-p, --port", action="store", dest="port", type=int,
225                         default=PORT, help="port to connect to (default : "
226                         + str(PORT) + ")")
227     parser.add_argument("-t, --timeout", action="store", dest="timeout",
228                         type=int, default=TIMEOUT,
229                         help="timeout in seconds after which to attempt " +
230                         "reconnect (default: " + str(TIMEOUT) + ")")
231     parser.add_argument("-u, --username", action="store", dest="username",
232                         default=USERNAME, help="username to use (default: "
233                         + USERNAME + ")")
234     parser.add_argument("-n, --nickname", action="store", dest="nickname",
235                         default=NICKNAME, help="nickname to use (default: "
236                         + NICKNAME + ")")
237     parser.add_argument("CHANNEL", action="store", help="channel to join")
238     opts, unknown = parser.parse_known_args()
239     return opts
240
241 opts = parse_command_line_arguments()
242 while True:
243     try:
244         io = init_session(opts.server, opts.port, opts.timeout, opts.nickname,
245                           opts.username, opts.CHANNEL)
246         lineparser_loop(io, opts.nickname)
247     except ExceptionForRestart:
248         io.socket.close()
249         continue