home · contact · privacy
8813498594f0a3ca4d453a366eac6ea246154bc7
[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 import plomsearch
15
16 # Defaults, may be overwritten by command line arguments.
17 SERVER = "irc.freenode.net"
18 PORT = 6667
19 TIMEOUT = 240
20 USERNAME = "plomlombot"
21 NICKNAME = USERNAME
22
23
24 class ExceptionForRestart(Exception):
25     pass
26
27
28 class IO:
29
30     def __init__(self, server, port, timeout):
31         self.timeout = timeout
32         self.socket = socket.socket()
33         self.socket.connect((server, port))
34         self.socket.setblocking(0)
35         self.line_buffer = []
36         self.rune_buffer = ""
37         self.last_pong = time.time()
38         self.servername = self.recv_line(send_ping=False).split(" ")[0][1:]
39
40     def _pingtest(self, send_ping=True):
41         if self.last_pong + self.timeout < time.time():
42             print("SERVER NOT ANSWERING")
43             raise ExceptionForRestart
44         if send_ping:
45             self.send_line("PING " + self.servername)
46
47     def send_line(self, msg):
48         msg = msg.replace("\r", " ")
49         msg = msg.replace("\n", " ")
50         if len(msg.encode("utf-8")) > 510:
51             print("NOT SENT LINE TO SERVER (too long): " + msg)
52         print("LINE TO SERVER: "
53               + str(datetime.datetime.now()) + ": " + msg)
54         msg = msg + "\r\n"
55         msg_len = len(msg)
56         total_sent_len = 0
57         while total_sent_len < msg_len:
58             sent_len = self.socket.send(bytes(msg[total_sent_len:], "UTF-8"))
59             if sent_len == 0:
60                 print("SOCKET CONNECTION BROKEN")
61                 raise ExceptionForRestart
62             total_sent_len += sent_len
63
64     def _recv_line_wrapped(self, send_ping=True):
65         if len(self.line_buffer) > 0:
66             return self.line_buffer.pop(0)
67         while True:
68             ready = select.select([self.socket], [], [], int(self.timeout / 2))
69             if not ready[0]:
70                 self._pingtest(send_ping)
71                 return None
72             self.last_pong = time.time()
73             received_bytes = self.socket.recv(1024)
74             try:
75                 received_runes = received_bytes.decode("UTF-8")
76             except UnicodeDecodeError:
77                 received_runes = received_bytes.decode("latin1")
78             if len(received_runes) == 0:
79                 print("SOCKET CONNECTION BROKEN")
80                 raise ExceptionForRestart
81             self.rune_buffer += received_runes
82             lines_split = str.split(self.rune_buffer, "\r\n")
83             self.line_buffer += lines_split[:-1]
84             self.rune_buffer = lines_split[-1]
85             if len(self.line_buffer) > 0:
86                 return self.line_buffer.pop(0)
87
88     def recv_line(self, send_ping=True):
89         line = self._recv_line_wrapped(send_ping)
90         if line:
91             print("LINE FROM SERVER " + str(datetime.datetime.now()) + ": " +
92                   line)
93         return line
94
95
96 def handle_command(command, argument, notice, target):
97     hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
98     quotesfile_name = "quotes_" + hash_string
99
100     def addquote():
101         if not os.access(quotesfile_name, os.F_OK):
102             quotesfile = open(quotesfile_name, "w")
103             quotesfile.write("QUOTES FOR " + target + ":\n")
104             quotesfile.close()
105         quotesfile = open(quotesfile_name, "a")
106         quotesfile.write(argument + "\n")
107         quotesfile.close()
108         quotesfile = open(quotesfile_name, "r")
109         lines = quotesfile.readlines()
110         quotesfile.close()
111         notice("ADDED QUOTE #" + str(len(lines) - 1))
112
113     def quote():
114
115         def help():
116             notice("SYNTAX: !quote [int] OR !quote search QUERY")
117             notice("QUERY may be a boolean grouping of quoted or unquoted " +
118                    "search terms, examples:")
119             notice("!quote search foo")
120             notice("!quote search foo AND (bar OR NOT baz)")
121             notice("!quote search \"foo\\\"bar\" AND ('NOT\"' AND \"'foo'\"" +
122                    " OR 'bar\\'baz')")
123
124         if "" == argument:
125             tokens = []
126         else:
127             tokens = argument.split(" ")
128         if (len(tokens) > 1 and tokens[0] != "search") or \
129             (len(tokens) == 1 and
130                 (tokens[0] == "search" or not tokens[0].isdigit())):
131             help()
132             return
133         if not os.access(quotesfile_name, os.F_OK):
134             notice("NO QUOTES AVAILABLE")
135             return
136         quotesfile = open(quotesfile_name, "r")
137         lines = quotesfile.readlines()
138         quotesfile.close()
139         lines = lines[1:]
140         if len(tokens) == 1:
141             i = int(tokens[0])
142             if i == 0 or i > len(lines):
143                 notice("THERE'S NO QUOTE OF THAT INDEX")
144                 return
145             i = i - 1
146         elif len(tokens) > 1:
147             query = str.join(" ", tokens[1:])
148             try:
149                 results = plomsearch.search(query, lines)
150             except plomsearch.LogicParserError as err:
151                 notice("FAILED QUERY PARSING: " + str(err))
152                 return
153             if len(results) == 0:
154                 notice("NO QUOTES MATCHING QUERY")
155             else:
156                 for result in results:
157                     notice("QUOTE #" + str(result[0] + 1) + " : " + result[1])
158             return
159         else:
160             i = random.randrange(len(lines))
161         notice("QUOTE #" + str(i + 1) + ": " + lines[i])
162
163     def markov():
164         from random import shuffle
165         select_length = 2
166         selections = []
167
168         def markov(snippet):
169             usable_selections = []
170             for i in range(select_length, 0, -1):
171                 for selection in selections:
172                     add = True
173                     for j in range(i):
174                         if snippet[j] != selection[j]:
175                             add = False
176                             break
177                     if add:
178                         usable_selections += [selection]
179                 if [] != usable_selections:
180                     break
181             if [] == usable_selections:
182                 usable_selections = selections
183             shuffle(usable_selections)
184             return usable_selections[0][select_length]
185
186         hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
187         markovfeed_name = "markovfeed_" + hash_string
188         if not os.access(markovfeed_name, os.F_OK):
189             notice("NOT ENOUGH TEXT TO MARKOV.")
190             return
191         file = open(markovfeed_name, "r")
192         lines = file.readlines()
193         file.close()
194         tokens = []
195         for line in lines:
196             line = line.replace("\n", "")
197             tokens += line.split()
198         if len(tokens) <= select_length:
199             notice("NOT ENOUGH TEXT TO MARKOV.")
200             return
201         for i in range(len(tokens) - select_length):
202             token_list = []
203             for j in range(select_length + 1):
204                 token_list += [tokens[i + j]]
205             selections += [token_list]
206         snippet = []
207         for i in range(select_length):
208             snippet += [""]
209         msg = ""
210         while 1:
211             new_end = markov(snippet)
212             if len(msg) + len(new_end) > 400:
213                 break
214             msg += new_end + " "
215             for i in range(select_length - 1):
216                 snippet[i] = snippet[i + 1]
217             snippet[select_length - 1] = new_end
218         notice(msg.lower() + "malkovich.")
219
220     if "addquote" == command:
221         addquote()
222     elif "quote" == command:
223         quote()
224     elif "markov" == command:
225         markov()
226
227
228 def handle_url(url, notice, show_url=False):
229
230     def mobile_twitter_hack(url):
231         re1 = 'https?://(mobile.twitter.com/)[^/]+(/status/)'
232         re2 = 'https?://mobile.twitter.com/([^/]+)/status/([^\?/]+)'
233         m = re.search(re1, url)
234         if m and m.group(1) == 'mobile.twitter.com/' \
235                 and m.group(2) == '/status/':
236             m = re.search(re2, url)
237             url = 'https://twitter.com/' + m.group(1) + '/status/' + m.group(2)
238             handle_url(url, notice, True)
239             return True
240
241     try:
242         r = requests.get(url, timeout=15)
243     except (requests.exceptions.TooManyRedirects,
244             requests.exceptions.ConnectionError,
245             requests.exceptions.InvalidURL,
246             requests.exceptions.InvalidSchema) as error:
247         notice("TROUBLE FOLLOWING URL: " + str(error))
248         return
249     if mobile_twitter_hack(url):
250         return
251     title = bs4.BeautifulSoup(r.text).title
252     if title:
253         prefix = "PAGE TITLE: "
254         if show_url:
255             prefix = "PAGE TITLE FOR <" + url + ">: "
256         notice(prefix + title.string.strip())
257     else:
258         notice("PAGE HAS NO TITLE TAG")
259
260
261 class Session:
262
263     def __init__(self, io, username, nickname, channel):
264         self.io = io
265         self.nickname = nickname
266         self.io.send_line("NICK " + self.nickname)
267         self.io.send_line("USER " + username + " 0 * : ")
268         self.io.send_line("JOIN " + channel)
269
270     def loop(self):
271
272         def handle_privmsg(tokens):
273
274             def handle_input(msg, target):
275
276                 def notice(msg):
277                     self.io.send_line("NOTICE " + target + " :" + msg)
278
279                 matches = re.findall("(https?://[^\s>]+)", msg)
280                 for i in range(len(matches)):
281                     handle_url(matches[i], notice)
282                 if "!" == msg[0]:
283                     tokens = msg[1:].split()
284                     argument = str.join(" ", tokens[1:])
285                     handle_command(tokens[0], argument, notice, target)
286                     return
287                 hash_string = hashlib.md5(target.encode("utf-8")).hexdigest()
288                 markovfeed_name = "markovfeed_" + hash_string
289                 file = open(markovfeed_name, "a")
290                 file.write(msg + "\n")
291                 file.close()
292
293             sender = ""
294             for rune in tokens[0]:
295                 if rune == "!":
296                     break
297                 if rune != ":":
298                     sender += rune
299             receiver = ""
300             for rune in tokens[2]:
301                 if rune == "!":
302                     break
303                 if rune != ":":
304                     receiver += rune
305             target = sender
306             if receiver != self.nickname:
307                 target = receiver
308             msg = str.join(" ", tokens[3:])[1:]
309             handle_input(msg, target)
310
311         while True:
312             line = self.io.recv_line()
313             if not line:
314                 continue
315             tokens = line.split(" ")
316             if len(tokens) > 1:
317                 if tokens[0] == "PING":
318                     self.io.send_line("PONG " + tokens[1])
319                 elif tokens[1] == "PRIVMSG":
320                     handle_privmsg(tokens)
321
322
323 def parse_command_line_arguments():
324     parser = argparse.ArgumentParser()
325     parser.add_argument("-s, --server", action="store", dest="server",
326                         default=SERVER,
327                         help="server or server net to connect to (default: "
328                         + SERVER + ")")
329     parser.add_argument("-p, --port", action="store", dest="port", type=int,
330                         default=PORT, help="port to connect to (default : "
331                         + str(PORT) + ")")
332     parser.add_argument("-t, --timeout", action="store", dest="timeout",
333                         type=int, default=TIMEOUT,
334                         help="timeout in seconds after which to attempt " +
335                         "reconnect (default: " + str(TIMEOUT) + ")")
336     parser.add_argument("-u, --username", action="store", dest="username",
337                         default=USERNAME, help="username to use (default: "
338                         + USERNAME + ")")
339     parser.add_argument("-n, --nickname", action="store", dest="nickname",
340                         default=NICKNAME, help="nickname to use (default: "
341                         + NICKNAME + ")")
342     parser.add_argument("CHANNEL", action="store", help="channel to join")
343     opts, unknown = parser.parse_known_args()
344     return opts
345
346
347 opts = parse_command_line_arguments()
348 while True:
349     try:
350         io = IO(opts.server, opts.port, opts.timeout)
351         session = Session(io, opts.username, opts.nickname, opts.CHANNEL)
352         session.loop()
353     except ExceptionForRestart:
354         io.socket.close()
355         continue