bot.py 9.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327
  1. #!/usr/bin/env python3
  2. import random
  3. import threading
  4. import socket
  5. import subprocess
  6. import sys
  7. import time
  8. from unzipper import unar
  9. class IRCClient():
  10. HOST = "irc.irchighway.net"
  11. PORT = 6667
  12. NICK = ""
  13. IDENT = ""
  14. REALNAME = ""
  15. IGNORE = [
  16. "NOTICE",
  17. "PART",
  18. "QUIT",
  19. "332",
  20. "333",
  21. "372",
  22. "353",
  23. "366",
  24. "251",
  25. "252",
  26. "254",
  27. "255",
  28. "265",
  29. "266",
  30. "396"]
  31. CHANNEL = "#ebooks"
  32. STATUS = ""
  33. PROGRESS = 0
  34. TYPE = "SEARCH"
  35. OUTPUT = ""
  36. buffer = []
  37. readbuffer = b''
  38. NOT_EXITED = True
  39. EXTRA_OUTPUT = ""
  40. PATH = ""
  41. joined = False
  42. connected = False
  43. def __init__(self, book, extension, t, cb, logging=False, path="/tmp/"):
  44. self.TYPE = t
  45. self.EXTENSION = extension
  46. self.LOOKING_FOR = book
  47. self.LOGGING = logging
  48. self.PATH = path
  49. name = "bookbot" + random_hash()
  50. self.NICK = name
  51. self.IDENT = name
  52. self.REALNAME = name
  53. self.callback = cb
  54. self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  55. self.socket.connect((self.HOST, self.PORT))
  56. self.handle_connect()
  57. nickstr = "NICK %s" % self.NICK
  58. userstr = "USER %s %s bla :%s" % (self.IDENT, self.HOST, self.REALNAME)
  59. self.t = threading.Thread(target=self.handle_read)
  60. self.t.start()
  61. self.send_queue(nickstr)
  62. self.send_queue(userstr)
  63. def handle_connect(self):
  64. self.log("connected")
  65. self.set_status("CONNECTED")
  66. self.connected = True
  67. def handle_close(self):
  68. self.NOT_EXITED = False
  69. self.log("closed")
  70. self.set_status("DISCONNECTED")
  71. self.connected = False
  72. self.socket.close()
  73. def do_timeout(self):
  74. if not self.NOT_EXITED:
  75. return
  76. self.NOT_EXITED = False
  77. self.log("Abort mission")
  78. self.set_status("TIMED OUT")
  79. self.connected = False
  80. self.socket.close()
  81. def handle_read(self):
  82. while self.NOT_EXITED:
  83. try:
  84. data = self.socket.recv(1024)
  85. except socket.error:
  86. break
  87. if len(data) < 2:
  88. continue
  89. if not (data[-1] == 10 and data[-2] == 13):
  90. self.readbuffer += data
  91. continue
  92. data = self.readbuffer + data
  93. print("DATA", data)
  94. # purge mojibake
  95. data = (data.replace(b'\x95', b'')
  96. .replace(b'0xc2', b'')
  97. .decode('utf-8', 'ignore'))
  98. self.readbuffer = b''
  99. lines = data.splitlines()
  100. for line in lines:
  101. words = line.split(' ')
  102. if len(words) < 2:
  103. continue
  104. msg_from = words[0]
  105. comm = words[1]
  106. if comm in self.IGNORE:
  107. continue
  108. if comm == "JOIN":
  109. if self.NICK not in msg_from: # msg "NICK joined the channel" not about me
  110. continue
  111. self.set_status("JOINED")
  112. self.log("Joined channel %s" % self.CHANNEL)
  113. self.log("Waiting 30s")
  114. time.sleep(31)
  115. self.run_query()
  116. self.joined = True
  117. continue
  118. if comm == "PRIVMSG":
  119. # private message not addressed to me
  120. if words[2] != self.NICK:
  121. continue
  122. self.parse_msg(line)
  123. if comm == "PING" or msg_from == "PING": # respond ping to avoid getting kicked
  124. self.pong(line)
  125. continue
  126. if comm == "376": # END MOTD
  127. # MOTD complete, lets join the channel
  128. self.join_channel(self.CHANNEL)
  129. continue
  130. if not self.joined: # waiting
  131. continue
  132. def run_query(self):
  133. self.set_status("WAITING")
  134. if self.TYPE == "SEARCH":
  135. self.search(self.LOOKING_FOR)
  136. return
  137. self.book(self.LOOKING_FOR)
  138. def parse_msg(self, msg):
  139. msg = msg.split(':')[2]
  140. msg = msg.replace("\x01", "")
  141. if not msg.startswith("DCC"):
  142. self.EXTRA_OUTPUT = msg
  143. self.log(msg)
  144. return
  145. args = msg.replace("DCC SEND ", "").split(" ")
  146. size = int(args.pop())
  147. port = int(args.pop())
  148. ip = ip_from_decimal(int(args.pop()))
  149. filename = "_".join(args).replace('"', '')
  150. self.set_status("RECEIVING")
  151. n = self.netcat(ip, port, size, filename)
  152. files = unar(n, self.PATH)
  153. print("Unarchived files", files)
  154. out = []
  155. for f in files:
  156. if "searchbot" in f.lower() or "searchook" in f.lower():
  157. out.append(self.list_books(f))
  158. print("Current output", out)
  159. self.OUTPUT = []
  160. if len(out) > 0:
  161. if isinstance(out[0], list):
  162. print("1")
  163. self.OUTPUT = [item for sublist in out for item in sublist]
  164. else:
  165. print("2")
  166. self.OUTPUT = out
  167. print("3")
  168. self.OUTPUT = list(set(self.OUTPUT))
  169. else:
  170. print("4")
  171. self.OUTPUT = files
  172. print(self.OUTPUT)
  173. if self.TYPE == "BOOK":
  174. self.log("Files: ")
  175. self.log(self.OUTPUT)
  176. for f in self.OUTPUT:
  177. if f.lower().endswith(".epub"):
  178. print("EPUB %s" % f)
  179. new_fname = f.replace("epub", "mobi")
  180. p = subprocess.Popen(["ebook-convert", f, new_fname])
  181. p.wait()
  182. self.OUTPUT.append(new_fname)
  183. break
  184. self.OUTPUT = [i.replace(self.PATH, "") for i in self.OUTPUT]
  185. self.handle_close()
  186. self.set_status("FINISHED")
  187. def list_books(self, f):
  188. f = open(f, "r")
  189. lines = f.readlines()
  190. lines = [l.strip().replace('\r', '') for l in lines if self.EXTENSION in l.lower(
  191. ) and l.startswith('!') and "htm" not in l.lower()]
  192. for l in lines:
  193. self.log(l)
  194. return lines
  195. def netcat(self, ip, port, size, filename):
  196. s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  197. self.log("Receiving file")
  198. s.connect((ip, port))
  199. fname = ('%s/%s' % (self.PATH, filename)).replace("//", "/")
  200. f = open(fname, 'wb')
  201. count = 0
  202. while True:
  203. data = s.recv(4096)
  204. if len(data) == 0:
  205. self.log("empty")
  206. break
  207. count += len(data)
  208. f.write(data)
  209. perc = int(100 * count / size)
  210. self.PROGRESS = perc
  211. self.set_status("DOWNLOADING") # % perc #progress
  212. self.log("\r%d%%" % perc, instant=True)
  213. if count >= size:
  214. break
  215. s.close()
  216. f.close()
  217. self.log("") # newline
  218. return fname
  219. def search(self, book):
  220. self.log("searching for %s" % book)
  221. self.send_queue("PRIVMSG %s :@searchook %s " % (self.CHANNEL, book))
  222. def pong(self, data):
  223. msg = data.replace("PING ", "")
  224. self.send_queue("PONG %s" % msg)
  225. def join_channel(self, channel):
  226. self.send_queue("JOIN %s" % channel)
  227. def send_queue(self, msg):
  228. add = bytes(str(msg), "utf-8") + bytes([13, 10])
  229. self.socket.send(add)
  230. def book(self, book):
  231. self.log("Asking for '%s'" % book)
  232. self.send_queue("PRIVMSG %s :%s " % (self.CHANNEL, book))
  233. def who(self):
  234. self.send_queue("WHO %s" % self.CHANNEL)
  235. def log(self, val, instant=False):
  236. if self.LOGGING:
  237. if not instant:
  238. print(val)
  239. else:
  240. sys.stdout.write(val)
  241. def set_status(self, value):
  242. self.STATUS = value
  243. self.callback(value=="FINISHED")
  244. def ip_from_decimal(dec):
  245. return ".".join([str(int(b, 2)) for b in b_octets(bin(dec)[2:])])
  246. def b_octets(l):
  247. l = l.zfill(32)
  248. return [l[0:8], l[8:16], l[16:24], l[24:32]]
  249. def random_hash():
  250. return ("%032x" % random.getrandbits(128))[:10]
  251. def usage():
  252. print("USAGE:")
  253. print("\t %s SEARCH <BOOK> <FORMAT>" % sys.argv[0])
  254. print("\t %s BOOK <BOT COMMAND>" % sys.argv[0])
  255. sys.exit(1)
  256. if __name__ == "__main__":
  257. if len(sys.argv) == 4:
  258. if(sys.argv[1] != "SEARCH"):
  259. usage()
  260. mode = sys.argv[1]
  261. query = sys.argv[2]
  262. grep = sys.argv[3]
  263. elif len(sys.argv) == 3:
  264. if(sys.argv[1] != "BOOK"):
  265. usage()
  266. mode = sys.argv[1]
  267. query = sys.argv[2]
  268. grep = ""
  269. else:
  270. usage()
  271. if mode == "SEARCH":
  272. print("Looking for '%s' in format '%s'" % (query, grep))
  273. if mode == "BOOK":
  274. print("Downloading %s" % query)
  275. client = IRCClient(query, grep, mode, print, logging=True)