bot.py 8.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300
  1. #!/usr/bin/env python3
  2. import random
  3. import threading
  4. import socket
  5. import sys
  6. from unzipper import unar
  7. class IRCClient():
  8. HOST = "irc.irchighway.net"
  9. PORT = 6667
  10. NICK = ""
  11. IDENT = ""
  12. REALNAME = ""
  13. IGNORE = [
  14. "NOTICE",
  15. "PART",
  16. "QUIT",
  17. "332",
  18. "333",
  19. "372",
  20. "353",
  21. "366",
  22. "251",
  23. "252",
  24. "254",
  25. "255",
  26. "265",
  27. "266",
  28. "396"]
  29. CHANNEL = "#ebooks"
  30. STATUS = ""
  31. PROGRESS = 0
  32. TYPE = "SEARCH"
  33. OUTPUT = ""
  34. buffer = []
  35. readbuffer = b''
  36. NOT_EXITED = True
  37. EXTRA_OUTPUT = ""
  38. PATH = ""
  39. joined = False
  40. connected = False
  41. def __init__(self, book, extension, t, logging=False, path="/tmp/"):
  42. self.TYPE = t
  43. self.EXTENSION = extension
  44. self.LOOKING_FOR = book
  45. self.LOGGING = logging
  46. self.PATH = path
  47. name = "bookbot" + random_hash()
  48. self.NICK = name
  49. self.IDENT = name
  50. self.REALNAME = name
  51. self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  52. self.socket.connect((self.HOST, self.PORT))
  53. self.handle_connect()
  54. nickstr = "NICK %s" % self.NICK
  55. userstr = "USER %s %s bla :%s" % (self.IDENT, self.HOST, self.REALNAME)
  56. self.t = threading.Thread(target=self.handle_read)
  57. self.t.start()
  58. self.send_queue(nickstr)
  59. self.send_queue(userstr)
  60. def handle_connect(self):
  61. self.log("connected")
  62. self.STATUS = "CONNECTED"
  63. self.connected = True
  64. def handle_close(self):
  65. self.NOT_EXITED = False
  66. self.log("closed")
  67. self.STATUS = "DISCONNECTED"
  68. self.connected = False
  69. self.socket.close()
  70. def do_timeout(self):
  71. if not self.NOT_EXITED:
  72. return
  73. self.NOT_EXITED = False
  74. self.log("Abort mission")
  75. self.STATUS = "TIMED OUT"
  76. self.connected = False
  77. self.socket.close()
  78. def handle_read(self):
  79. while self.NOT_EXITED:
  80. try:
  81. data = self.socket.recv(1024)
  82. except socket.error:
  83. break
  84. if len(data) < 2:
  85. continue
  86. if not (data[-1] == 10 and data[-2] == 13):
  87. self.readbuffer += data
  88. continue
  89. data = self.readbuffer + data
  90. # purge mojibake
  91. data = (data.replace(b'\x95', b'')
  92. .replace(b'0xc2', b'')
  93. .decode('utf-8', 'ignore'))
  94. self.readbuffer = b''
  95. lines = data.splitlines()
  96. for line in lines:
  97. words = line.split(' ')
  98. if len(words) < 2:
  99. continue
  100. msg_from = words[0]
  101. comm = words[1]
  102. if comm in self.IGNORE:
  103. continue
  104. if comm == "JOIN":
  105. if self.NICK not in msg_from: # msg "NICK joined the channel" not about me
  106. continue
  107. self.STATUS = "JOINED"
  108. self.log("Joined channel %s" % self.CHANNEL)
  109. self.run_query()
  110. self.joined = True
  111. continue
  112. if comm == "PRIVMSG":
  113. # private message not addressed to me
  114. if words[2] != self.NICK:
  115. continue
  116. self.parse_msg(line)
  117. if comm == "PING" or msg_from == "PING": # respond ping to avoid getting kicked
  118. self.pong(line)
  119. continue
  120. if comm == "376": # END MOTD
  121. # MOTD complete, lets join the channel
  122. self.join_channel(self.CHANNEL)
  123. continue
  124. if not self.joined: # waiting
  125. continue
  126. def run_query(self):
  127. self.STATUS = "WAITING"
  128. if self.TYPE == "SEARCH":
  129. self.search(self.LOOKING_FOR)
  130. return
  131. self.book(self.LOOKING_FOR)
  132. def parse_msg(self, msg):
  133. msg = msg.split(':')[2]
  134. msg = msg.replace("\x01", "")
  135. if not msg.startswith("DCC"):
  136. self.EXTRA_OUTPUT = msg
  137. self.log(msg)
  138. return
  139. args = msg.replace("DCC SEND ", "").split(" ")
  140. size = int(args.pop())
  141. port = int(args.pop())
  142. ip = ip_from_decimal(int(args.pop()))
  143. filename = "_".join(args).replace('"', '')
  144. self.STATUS = "RECEIVING"
  145. n = self.netcat(ip, port, size, filename)
  146. files = unar(n, self.PATH)
  147. out = []
  148. for f in files:
  149. if "searchbot" in f.lower() or "searchook" in f.lower():
  150. out.append(self.list_books(f))
  151. self.OUTPUT = []
  152. if len(out) > 0:
  153. if isinstance(out[0], list):
  154. self.OUTPUT = [item for sublist in out for item in sublist]
  155. else:
  156. self.OUTPUT = out
  157. self.OUTPUT = list(set(self.OUTPUT))
  158. else:
  159. self.OUTPUT = files
  160. if self.TYPE == "BOOK":
  161. self.log("Files: ")
  162. self.log(self.OUTPUT)
  163. self.OUTPUT = [i.replace(self.PATH, "") for i in self.OUTPUT]
  164. self.handle_close()
  165. def list_books(self, f):
  166. f = open(f, "r")
  167. lines = f.readlines()
  168. lines = [l.strip().replace('\r', '') for l in lines if self.EXTENSION in l.lower(
  169. ) and l.startswith('!') and "htm" not in l.lower()]
  170. for l in lines:
  171. self.log(l)
  172. return lines
  173. def netcat(self, ip, port, size, filename):
  174. s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  175. self.log("Receiving file")
  176. s.connect((ip, port))
  177. fname = ('%s/%s' % (self.PATH, filename)).replace("//", "/")
  178. f = open(fname, 'wb')
  179. count = 0
  180. while True:
  181. data = s.recv(4096)
  182. if len(data) == 0:
  183. self.log("empty")
  184. break
  185. count += len(data)
  186. f.write(data)
  187. perc = int(100 * count / size)
  188. self.STATUS = "DOWNLOADING" # % perc #progress
  189. self.PROGRESS = perc
  190. self.log("\r%d%%" % perc, instant=True)
  191. if count >= size:
  192. break
  193. s.close()
  194. f.close()
  195. self.log("") # newline
  196. return fname
  197. def search(self, book):
  198. self.log("searching for %s" % book)
  199. self.send_queue("PRIVMSG %s :@search %s " % (self.CHANNEL, book))
  200. def pong(self, data):
  201. msg = data.replace("PING ", "")
  202. self.send_queue("PONG %s" % msg)
  203. def join_channel(self, channel):
  204. self.send_queue("JOIN %s" % channel)
  205. def send_queue(self, msg):
  206. add = bytes(str(msg), "utf-8") + bytes([13, 10])
  207. self.socket.send(add)
  208. def book(self, book):
  209. self.log("Asking for '%s'" % book)
  210. self.send_queue("PRIVMSG %s :%s " % (self.CHANNEL, book))
  211. def who(self):
  212. self.send_queue("WHO %s" % self.CHANNEL)
  213. def log(self, val, instant=False):
  214. if self.LOGGING:
  215. if not instant:
  216. print(val)
  217. else:
  218. sys.stdout.write(val)
  219. def ip_from_decimal(dec):
  220. return ".".join([str(int(b, 2)) for b in b_octets(bin(dec)[2:])])
  221. def b_octets(l):
  222. l = l.zfill(32)
  223. return [l[0:8], l[8:16], l[16:24], l[24:32]]
  224. def random_hash():
  225. return ("%032x" % random.getrandbits(128))[:10]
  226. def usage():
  227. print("USAGE:")
  228. print("\t %s SEARCH <BOOK> <FORMAT>" % sys.argv[0])
  229. print("\t %s BOOK <BOT COMMAND>" % sys.argv[0])
  230. sys.exit(1)
  231. if __name__ == "__main__":
  232. if len(sys.argv) == 4:
  233. if(sys.argv[1] != "SEARCH"):
  234. usage()
  235. mode = sys.argv[1]
  236. query = sys.argv[2]
  237. grep = sys.argv[3]
  238. elif len(sys.argv) == 3:
  239. if(sys.argv[1] != "BOOK"):
  240. usage()
  241. mode = sys.argv[1]
  242. query = sys.argv[2]
  243. grep = ""
  244. else:
  245. usage()
  246. if mode == "SEARCH":
  247. print("Looking for '%s' in format '%s'" % (query, grep))
  248. if mode == "BOOK":
  249. print("Downloading %s" % query)
  250. client = IRCClient(query, grep, mode, logging=True)