bot.py 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239
  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. TYPE="SEARCH"
  18. OUTPUT=""
  19. buffer = []
  20. readbuffer=b''
  21. NOT_EXITED=True
  22. EXTRA_OUTPUT=""
  23. PATH=""
  24. joined=False
  25. connected=False
  26. def __init__(self,book,extension,t,logging=False,path="/tmp/"):
  27. self.TYPE=t
  28. self.EXTENSION=extension
  29. self.LOOKING_FOR=book
  30. self.LOGGING=logging
  31. self.PATH=path
  32. self.NICK="bookbot"+self.random_hash()
  33. self.IDENT="bookbot"+self.random_hash()
  34. self.REALNAME="bookbot"+self.random_hash()
  35. self.socket=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  36. self.socket.connect( (self.HOST, self.PORT) )
  37. self.handle_connect()
  38. nickstr="NICK %s" % self.NICK
  39. userstr="USER %s %s bla :%s" % (self.IDENT, self.HOST, self.REALNAME)
  40. self.t=threading.Thread(target=self.handle_read)
  41. self.t.start()
  42. self.send_queue(nickstr)
  43. self.send_queue(userstr)
  44. def random_hash(self):
  45. return ("%032x" % random.getrandbits(128))[:10]
  46. def handle_connect(self):
  47. self.log("connected")
  48. self.STATUS="CONNECTED"
  49. self.connected=True
  50. pass
  51. def handle_close(self):
  52. self.NOT_EXITED=False
  53. self.log("closed")
  54. self.STATUS="DISCONNECTED"
  55. self.connected=False
  56. self.socket.close()
  57. def handle_read(self):
  58. while self.NOT_EXITED:
  59. try:
  60. data=self.socket.recv(1024)
  61. except e:
  62. break
  63. if len(data) < 2:
  64. continue
  65. if not (data[-1]==10 and data[-2]==13):
  66. self.readbuffer+=data
  67. continue
  68. data=self.readbuffer+data
  69. data=data.replace(b'\x95', b'').replace(b'0xc2',b'').decode('utf-8','ignore') #purge mojibake
  70. self.readbuffer=b''
  71. lines=data.splitlines()
  72. for line in lines:
  73. words=line.split(' ')
  74. if len(words)<2:
  75. continue
  76. msg_from=words[0]
  77. comm=words[1]
  78. if comm in self.IGNORE:
  79. continue
  80. if comm=="JOIN":
  81. if self.NICK not in msg_from:
  82. continue
  83. self.STATUS="JOINED"
  84. self.log("Joined channel %s" % self.CHANNEL)
  85. self.run_query()
  86. self.joined=True
  87. continue
  88. if comm=="PRIVMSG":
  89. if words[2] != self.NICK:
  90. continue
  91. self.parse_msg(line);
  92. if comm=="PING" or msg_from=="PING":
  93. self.pong(line)
  94. continue
  95. if comm=="376": #END MOTD
  96. self.join_channel(self.CHANNEL)
  97. continue
  98. if not self.joined:
  99. continue
  100. def run_query(self):
  101. self.STATUS="WAITING"
  102. if self.TYPE=="SEARCH":
  103. self.search(self.LOOKING_FOR)
  104. return
  105. self.book(self.LOOKING_FOR)
  106. def parse_msg(self,msg):
  107. msg=msg.split(':')[2]
  108. msg=msg.replace("\x01","")
  109. if not msg.startswith("DCC"):
  110. self.EXTRA_OUTPUT=msg;
  111. self.log(msg)
  112. return
  113. args=msg.replace("DCC SEND ","").split(" ")
  114. size=int(args.pop())
  115. port=int(args.pop())
  116. ip=self.ip_from_decimal(int(args.pop()))
  117. filename="_".join(args).replace('"','')
  118. self.STATUS="RECEIVING"
  119. n=self.netcat(ip,port,size,filename)
  120. files=unar(n,self.PATH)
  121. out=[]
  122. for f in files:
  123. if "searchbot" in f.lower():
  124. out.append(self.list_books(f))
  125. self.OUTPUT=[]
  126. if len(out)>0:
  127. if type(out[0]) is list:
  128. self.OUTPUT=[item for sublist in out for item in sublist]
  129. else:
  130. self.OUTPUT=out
  131. self.OUTPUT=list(set(self.OUTPUT))
  132. else:
  133. self.OUTPUT=files
  134. self.OUTPUT=[i.replace(self.PATH,"") for i in self.OUTPUT ]
  135. self.handle_close()
  136. def list_books(self,f):
  137. f=open(f,"r")
  138. lines=f.readlines()
  139. lines=[l.strip() for l in lines if self.EXTENSION in l.lower() ]
  140. for l in lines:
  141. self.log(l)
  142. return lines
  143. def netcat(self,ip,port,size,filename):
  144. s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  145. s.connect((ip,port))
  146. #s.shutdown(socket.SHUT_WR)
  147. sizelen=len(str(size))
  148. fname=('%s/%s' % (self.PATH,filename)).replace("//","/")
  149. f=open(fname, 'wb')
  150. count=0
  151. while True:
  152. data = s.recv(4096)
  153. if len(data)==0:
  154. self.log("empty")
  155. break
  156. count+=len(data)
  157. f.write(data)
  158. perc=int(100*count/size)
  159. self.STATUS="DOWNLOADING %d%%" % perc #progress
  160. # sys.stdout.write("\r%s/%d" % (str(count).zfill(sizelen),size)) #progress
  161. if count >= size:
  162. break
  163. s.close()
  164. f.close()
  165. return fname
  166. def search(self,book):
  167. self.log("searching for %s" % book)
  168. self.send_queue("PRIVMSG %s :@search %s " % (self.CHANNEL,book))
  169. def pong(self,data):
  170. msg=data.replace("PING ","")
  171. self.send_queue("PONG %s" % msg)
  172. def join_channel(self,channel):
  173. self.send_queue("JOIN %s" % channel)
  174. def send_queue(self,msg):
  175. add=bytes(str(msg),"utf-8")+bytes([13,10])
  176. self.socket.send(add)
  177. # self.buffer.append(add)
  178. def book(self,book):
  179. self.log("DOWNLOADING %s" % book)
  180. self.send_queue("PRIVMSG %s :%s " % (self.CHANNEL,book))
  181. def who(self):
  182. self.send_queue("WHO %s" % self.CHANNEL)
  183. def ip_from_decimal(self,dec):
  184. return ".".join([ str(int(b,2)) for b in self.b_octets(bin(dec)[2:]) ])
  185. def b_octets(self,l):
  186. l=l.zfill(32)
  187. return [l[0:8],l[8:16],l[16:24],l[24:32]]
  188. def log(self,val):
  189. if self.LOGGING:
  190. print(val)
  191. if __name__ == "__main__":
  192. if len(sys.argv)!=3:
  193. print("USAGE: %s <BOOK> <FORMAT>" % sys.argv[0])
  194. sys.exit(1)
  195. print("Looking for %s in format %s" % (sys.argv[1],sys.argv[2]))
  196. client = IRCClient(sys.argv[1],sys.argv[2],"SEARCH",True)
  197. #client = IRCClient(sys.argv[1],sys.argv[2],"BOOK",True)