bot.py 7.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252
  1. #!/usr/bin/env python3
  2. import random
  3. import threading
  4. import re
  5. import socket
  6. import sys
  7. from unzipper import unar
  8. class IRCClient():
  9. HOST="irc.irchighway.net"
  10. PORT=6667
  11. NICK=""
  12. IDENT=""
  13. REALNAME=""
  14. IGNORE=["NOTICE","PART","QUIT", "332","333", "372", "353","366", "251", "252", "254", "255","265","266","396"]
  15. CHANNEL="#ebooks"
  16. STATUS=""
  17. PROGRESS=0
  18. TYPE="SEARCH"
  19. OUTPUT=""
  20. buffer = []
  21. readbuffer=b''
  22. NOT_EXITED=True
  23. EXTRA_OUTPUT=""
  24. PATH=""
  25. joined=False
  26. connected=False
  27. def __init__(self,book,extension,t,logging=False,path="/tmp/"):
  28. self.TYPE=t
  29. self.EXTENSION=extension
  30. self.LOOKING_FOR=book
  31. self.LOGGING=logging
  32. self.PATH=path
  33. self.NICK="bookbot"+self.random_hash()
  34. self.IDENT="bookbot"+self.random_hash()
  35. self.REALNAME="bookbot"+self.random_hash()
  36. self.socket=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  37. self.socket.connect( (self.HOST, self.PORT) )
  38. self.handle_connect()
  39. nickstr="NICK %s" % self.NICK
  40. userstr="USER %s %s bla :%s" % (self.IDENT, self.HOST, self.REALNAME)
  41. self.t=threading.Thread(target=self.handle_read)
  42. self.t.start()
  43. self.send_queue(nickstr)
  44. self.send_queue(userstr)
  45. def random_hash(self):
  46. return ("%032x" % random.getrandbits(128))[:10]
  47. def handle_connect(self):
  48. self.log("connected")
  49. self.STATUS="CONNECTED"
  50. self.connected=True
  51. pass
  52. def handle_close(self):
  53. self.NOT_EXITED=False
  54. self.log("closed")
  55. self.STATUS="DISCONNECTED"
  56. self.connected=False
  57. self.socket.close()
  58. def do_timeout(self):
  59. if not self.NOT_EXITED:
  60. return
  61. self.NOT_EXITED=False
  62. self.log("Abort mission")
  63. self.STATUS="TIMED OUT"
  64. self.connect=False
  65. self.socket.close()
  66. def handle_read(self):
  67. while self.NOT_EXITED:
  68. try:
  69. data=self.socket.recv(1024)
  70. except e:
  71. break
  72. if len(data) < 2:
  73. continue
  74. if not (data[-1]==10 and data[-2]==13):
  75. self.readbuffer+=data
  76. continue
  77. data=self.readbuffer+data
  78. data=data.replace(b'\x95', b'').replace(b'0xc2',b'').decode('utf-8','ignore') #purge mojibake
  79. self.readbuffer=b''
  80. lines=data.splitlines()
  81. for line in lines:
  82. words=line.split(' ')
  83. if len(words)<2:
  84. continue
  85. msg_from=words[0]
  86. comm=words[1]
  87. if comm in self.IGNORE:
  88. continue
  89. if comm=="JOIN":
  90. if self.NICK not in msg_from:
  91. continue
  92. self.STATUS="JOINED"
  93. self.log("Joined channel %s" % self.CHANNEL)
  94. self.run_query()
  95. self.joined=True
  96. continue
  97. if comm=="PRIVMSG":
  98. if words[2] != self.NICK:
  99. continue
  100. self.parse_msg(line);
  101. if comm=="PING" or msg_from=="PING":
  102. self.pong(line)
  103. continue
  104. if comm=="376": #END MOTD
  105. self.join_channel(self.CHANNEL)
  106. continue
  107. if not self.joined:
  108. continue
  109. def run_query(self):
  110. self.STATUS="WAITING"
  111. if self.TYPE=="SEARCH":
  112. self.search(self.LOOKING_FOR)
  113. return
  114. self.book(self.LOOKING_FOR)
  115. def parse_msg(self,msg):
  116. msg=msg.split(':')[2]
  117. msg=msg.replace("\x01","")
  118. if not msg.startswith("DCC"):
  119. self.EXTRA_OUTPUT=msg;
  120. print(msg)
  121. self.log(msg)
  122. return
  123. args=msg.replace("DCC SEND ","").split(" ")
  124. size=int(args.pop())
  125. port=int(args.pop())
  126. ip=self.ip_from_decimal(int(args.pop()))
  127. filename="_".join(args).replace('"','')
  128. self.STATUS="RECEIVING"
  129. n=self.netcat(ip,port,size,filename)
  130. files=unar(n,self.PATH)
  131. out=[]
  132. for f in files:
  133. if "searchbot" in f.lower():
  134. out.append(self.list_books(f))
  135. self.OUTPUT=[]
  136. if len(out)>0:
  137. if type(out[0]) is list:
  138. self.OUTPUT=[item for sublist in out for item in sublist]
  139. else:
  140. self.OUTPUT=out
  141. self.OUTPUT=list(set(self.OUTPUT))
  142. else:
  143. self.OUTPUT=files
  144. self.OUTPUT=[i.replace(self.PATH,"") for i in self.OUTPUT ]
  145. self.handle_close()
  146. def list_books(self,f):
  147. f=open(f,"r")
  148. lines=f.readlines()
  149. lines=[l.strip().replace('\r','') for l in lines if self.EXTENSION in l.lower() and l.startswith('!') and "htm" not in l.lower() ]
  150. for l in lines:
  151. self.log(l)
  152. return lines
  153. def netcat(self,ip,port,size,filename):
  154. s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  155. s.connect((ip,port))
  156. #s.shutdown(socket.SHUT_WR)
  157. sizelen=len(str(size))
  158. fname=('%s/%s' % (self.PATH,filename)).replace("//","/")
  159. f=open(fname, 'wb')
  160. count=0
  161. while True:
  162. data = s.recv(4096)
  163. if len(data)==0:
  164. self.log("empty")
  165. break
  166. count+=len(data)
  167. f.write(data)
  168. perc=int(100*count/size)
  169. self.STATUS="DOWNLOADING" #% perc #progress
  170. self.PROGRESS=perc
  171. # sys.stdout.write("\r%s/%d" % (str(count).zfill(sizelen),size)) #progress
  172. if count >= size:
  173. break
  174. s.close()
  175. f.close()
  176. return fname
  177. def search(self,book):
  178. self.log("searching for %s" % book)
  179. self.send_queue("PRIVMSG %s :@search %s " % (self.CHANNEL,book))
  180. def pong(self,data):
  181. msg=data.replace("PING ","")
  182. self.send_queue("PONG %s" % msg)
  183. def join_channel(self,channel):
  184. self.send_queue("JOIN %s" % channel)
  185. def send_queue(self,msg):
  186. add=bytes(str(msg),"utf-8")+bytes([13,10])
  187. self.socket.send(add)
  188. # self.buffer.append(add)
  189. def book(self,book):
  190. self.log("DOWNLOADING %s" % book)
  191. self.send_queue("PRIVMSG %s :%s " % (self.CHANNEL,book))
  192. def who(self):
  193. self.send_queue("WHO %s" % self.CHANNEL)
  194. def ip_from_decimal(self,dec):
  195. return ".".join([ str(int(b,2)) for b in self.b_octets(bin(dec)[2:]) ])
  196. def b_octets(self,l):
  197. l=l.zfill(32)
  198. return [l[0:8],l[8:16],l[16:24],l[24:32]]
  199. def log(self,val):
  200. if self.LOGGING:
  201. print(val)
  202. if __name__ == "__main__":
  203. if len(sys.argv)!=3:
  204. print("USAGE: %s <BOOK> <FORMAT>" % sys.argv[0])
  205. sys.exit(1)
  206. print("Looking for %s in format %s" % (sys.argv[1],sys.argv[2]))
  207. client = IRCClient(sys.argv[1],sys.argv[2],"SEARCH",True)
  208. #client = IRCClient(sys.argv[1],sys.argv[2],"BOOK",True)