bot.py 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226
  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. OUT_DIR="/tmp"
  18. TYPE="SEARCH"
  19. OUTPUT=""
  20. buffer = []
  21. readbuffer=b''
  22. NOT_EXITED=True
  23. joined=False
  24. connected=False
  25. def __init__(self,book,extension,t,logging=False):
  26. self.TYPE=t
  27. self.EXTENSION=extension
  28. self.LOOKING_FOR=book
  29. self.LOGGING=logging
  30. self.NICK="bookbot"+self.random_hash()
  31. self.IDENT="bookbot"+self.random_hash()
  32. self.REALNAME="bookbot"+self.random_hash()
  33. self.socket=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  34. self.socket.connect( (self.HOST, self.PORT) )
  35. self.handle_connect()
  36. nickstr="NICK %s" % self.NICK
  37. userstr="USER %s %s bla :%s" % (self.IDENT, self.HOST, self.REALNAME)
  38. self.t=threading.Thread(target=self.handle_read)
  39. self.t.start()
  40. self.send_queue(nickstr)
  41. self.send_queue(userstr)
  42. def random_hash(self):
  43. return ("%032x" % random.getrandbits(128))[:10]
  44. def handle_connect(self):
  45. self.log("connected")
  46. self.STATUS="CONNECTED"
  47. self.connected=True
  48. pass
  49. def handle_close(self):
  50. self.NOT_EXITED=False
  51. self.log("closed")
  52. self.STATUS="DISCONNECTED"
  53. self.connected=False
  54. self.socket.close()
  55. def handle_read(self):
  56. while self.NOT_EXITED:
  57. try:
  58. data=self.socket.recv(1024)
  59. except e:
  60. break
  61. if len(data) < 2:
  62. continue
  63. if not (data[-1]==10 and data[-2]==13):
  64. self.readbuffer+=data
  65. continue
  66. data=self.readbuffer+data
  67. data=data.replace(b'\x95', b'').replace(b'0xc2',b'').decode('utf-8','ignore') #purge mojibake
  68. self.readbuffer=b''
  69. lines=data.splitlines()
  70. for line in lines:
  71. words=line.split(' ')
  72. if len(words)<2:
  73. continue
  74. msg_from=words[0]
  75. comm=words[1]
  76. if comm in self.IGNORE:
  77. continue
  78. if comm=="JOIN":
  79. if self.NICK not in msg_from:
  80. continue
  81. self.STATUS="JOINED"
  82. self.log("Joined channel %s" % self.CHANNEL)
  83. self.run_query()
  84. self.joined=True
  85. continue
  86. if comm=="PRIVMSG":
  87. if words[2] != self.NICK:
  88. continue
  89. self.parse_msg(line);
  90. if comm=="PING" or msg_from=="PING":
  91. self.pong(line)
  92. continue
  93. if comm=="376": #END MOTD
  94. self.join_channel(self.CHANNEL)
  95. continue
  96. if not self.joined:
  97. continue
  98. def run_query(self):
  99. self.STATUS="WAITING"
  100. if self.TYPE=="SEARCH":
  101. self.search(self.LOOKING_FOR)
  102. return
  103. self.book(self.LOOKING_FOR)
  104. def parse_msg(self,msg):
  105. msg=msg.split(':')[2]
  106. msg=msg.replace("\x01","")
  107. if not msg.startswith("DCC"):
  108. self.log(msg)
  109. return
  110. args=msg.replace("DCC SEND ","").split(" ")
  111. size=int(args.pop())
  112. port=int(args.pop())
  113. ip=self.ip_from_decimal(int(args.pop()))
  114. filename="_".join(args).replace('"','')
  115. self.STATUS="RECEIVING"
  116. n=self.netcat(ip,port,size,filename)
  117. files=unar(n,"/tmp/")
  118. out=[]
  119. for f in files:
  120. if "searchbot" in f.lower():
  121. out.append(self.list_books(f))
  122. self.OUTPUT=out
  123. self.handle_close()
  124. def list_books(self,f):
  125. f=open(f,"r")
  126. lines=f.readlines()
  127. lines=[l.strip() for l in lines if self.EXTENSION in l.lower() ]
  128. for l in lines:
  129. self.log(l)
  130. return lines
  131. def netcat(self,ip,port,size,filename):
  132. s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  133. s.connect((ip,port))
  134. #s.shutdown(socket.SHUT_WR)
  135. sizelen=len(str(size))
  136. fname='%s/%s' % (self.OUT_DIR,filename)
  137. f=open(fname, 'wb')
  138. count=0
  139. while True:
  140. data = s.recv(4096)
  141. if len(data)==0:
  142. self.log("empty")
  143. break
  144. count+=len(data)
  145. f.write(data)
  146. self.STATUS="DOWNLOADING %s/%d" % (str(count).zfill(sizelen),size) #progress
  147. # sys.stdout.write("\r%s/%d" % (str(count).zfill(sizelen),size)) #progress
  148. if count >= size:
  149. break
  150. s.close()
  151. f.close()
  152. return fname
  153. def search(self,book):
  154. self.log("searching for %s" % book)
  155. self.send_queue("PRIVMSG %s :@search %s " % (self.CHANNEL,book))
  156. def pong(self,data):
  157. msg=data.replace("PING ","")
  158. self.send_queue("PONG %s" % msg)
  159. def join_channel(self,channel):
  160. self.send_queue("JOIN %s" % channel)
  161. def send_queue(self,msg):
  162. add=bytes(str(msg),"utf-8")+bytes([13,10])
  163. self.socket.send(add)
  164. # self.buffer.append(add)
  165. def book(self,book):
  166. self.log("DOWNLOADING %s" % book)
  167. self.send_queue("PRIVMSG %s :%s " % (self.CHANNEL,book))
  168. def who(self):
  169. self.send_queue("WHO %s" % self.CHANNEL)
  170. def ip_from_decimal(self,dec):
  171. return ".".join([ str(int(b,2)) for b in self.b_octets(bin(dec)[2:]) ])
  172. def b_octets(self,l):
  173. l=l.zfill(32)
  174. return [l[0:8],l[8:16],l[16:24],l[24:32]]
  175. def log(self,val):
  176. if self.LOGGING:
  177. print(val)
  178. if __name__ == "__main__":
  179. if len(sys.argv)!=3:
  180. print("USAGE: %s <BOOK> <FORMAT>" % sys.argv[0])
  181. sys.exit(1)
  182. print("Looking for %s in format %s" % (sys.argv[1],sys.argv[2]))
  183. client = IRCClient(sys.argv[1],sys.argv[2],"SEARCH",True)
  184. #client = IRCClient(sys.argv[1],sys.argv[2],"BOOK",True)