bot.py 6.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216
  1. #!/usr/bin/env python3
  2. import threading
  3. import re
  4. import socket
  5. import sys
  6. from unzipper import unar
  7. class IRCClient():
  8. HOST="irc.irchighway.net"
  9. PORT=6667
  10. NICK="booksbot3321321"
  11. IDENT="booksbot"
  12. REALNAME="NotABot"
  13. IGNORE=["NOTICE","PART","QUIT", "332","333", "372", "353","366", "251", "252", "254", "255","265","266","396"]
  14. CHANNEL="#ebooks"
  15. STATUS=""
  16. OUT_DIR="/tmp"
  17. TYPE="SEARCH"
  18. buffer = []
  19. readbuffer=b''
  20. NOT_EXITED=True
  21. joined=False
  22. connected=False
  23. def __init__(self,book,extension,t,logging=False,q=None):
  24. self.TYPE=t
  25. self.EXTENSION=extension
  26. self.LOOKING_FOR=book
  27. self.LOGGING=logging
  28. self.OUTPUT=q
  29. self.socket=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  30. self.socket.connect( (self.HOST, self.PORT) )
  31. self.handle_connect()
  32. nickstr="NICK %s" % self.NICK
  33. userstr="USER %s %s bla :%s" % (self.IDENT, self.HOST, self.REALNAME)
  34. self.t=threading.Thread(target=self.handle_read)
  35. self.t.start()
  36. self.send_queue(nickstr)
  37. self.send_queue(userstr)
  38. self.t.join()
  39. def handle_connect(self):
  40. self.log("connected")
  41. self.STATUS="CONNECTED"
  42. self.connected=True
  43. pass
  44. def handle_close(self):
  45. self.NOT_EXITED=False
  46. self.log("closed")
  47. self.STATUS="DISCONNECTED"
  48. self.connected=False
  49. self.socket.close()
  50. def handle_read(self):
  51. while self.NOT_EXITED:
  52. try:
  53. data=self.socket.recv(1024)
  54. except e:
  55. break
  56. if not (data[-1]==10 and data[-2]==13):
  57. self.readbuffer+=data
  58. continue
  59. data=self.readbuffer+data
  60. data=data.replace(b'\x95', b'').replace(b'0xc2',b'').decode('utf-8','ignore') #purge mojibake
  61. self.readbuffer=b''
  62. lines=data.splitlines()
  63. for line in lines:
  64. words=line.split(' ')
  65. if len(words)<2:
  66. continue
  67. msg_from=words[0]
  68. comm=words[1]
  69. if comm in self.IGNORE:
  70. continue
  71. if comm=="JOIN":
  72. if self.NICK not in msg_from:
  73. continue
  74. self.STATUS="JOINED"
  75. self.log("Joined channel %s" % self.CHANNEL)
  76. self.run_query()
  77. self.joined=True
  78. continue
  79. if comm=="PRIVMSG":
  80. if words[2] != self.NICK:
  81. continue
  82. self.parse_msg(line);
  83. if comm=="PING" or msg_from=="PING":
  84. self.pong(line)
  85. continue
  86. if comm=="376": #END MOTD
  87. self.join_channel(self.CHANNEL)
  88. continue
  89. if not self.joined:
  90. continue
  91. def run_query(self):
  92. if self.TYPE=="SEARCH":
  93. self.search(self.LOOKING_FOR)
  94. return
  95. self.book(self.LOOKING_FOR)
  96. def parse_msg(self,msg):
  97. msg=msg.split(':')[2]
  98. msg=msg.replace("\x01","")
  99. if not msg.startswith("DCC"):
  100. self.log(msg)
  101. return
  102. args=msg.replace("DCC SEND ","").split(" ")
  103. size=int(args.pop())
  104. port=int(args.pop())
  105. ip=self.ip_from_decimal(int(args.pop()))
  106. filename="_".join(args)
  107. self.STATUS="RECEIVING"
  108. n=self.netcat(ip,port,size,filename)
  109. files=unar(n,"/tmp/")
  110. out=[]
  111. for f in files:
  112. if "searchbot" in f.lower():
  113. out.append(self.list_books(f))
  114. if self.OUTPUT is not None:
  115. self.OUTPUT.put(out)
  116. self.handle_close()
  117. def list_books(self,f):
  118. f=open(f,"r")
  119. lines=f.readlines()
  120. lines=[l.strip() for l in lines if self.EXTENSION in l.lower() ]
  121. for l in lines:
  122. self.log(l)
  123. return lines
  124. def netcat(self,ip,port,size,filename):
  125. s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  126. s.connect((ip,port))
  127. #s.shutdown(socket.SHUT_WR)
  128. sizelen=len(str(size))
  129. fname='%s/%s' % (self.OUT_DIR,filename)
  130. f=open(fname, 'wb')
  131. count=0
  132. while True:
  133. data = s.recv(1024)
  134. if len(data)==0:
  135. self.log("empty")
  136. break
  137. count+=len(data)
  138. f.write(data)
  139. sys.stdout.write("\r%s/%d" % (str(count).zfill(sizelen),size)) #progress
  140. if count >= size:
  141. break
  142. s.close()
  143. f.close()
  144. return fname
  145. def search(self,book):
  146. self.log("searching for %s" % book)
  147. self.send_queue("PRIVMSG %s :@search %s " % (self.CHANNEL,book))
  148. def pong(self,data):
  149. msg=data.replace("PING ","")
  150. self.send_queue("PONG %s" % msg)
  151. def join_channel(self,channel):
  152. self.send_queue("JOIN %s" % channel)
  153. def send_queue(self,msg):
  154. add=bytes(str(msg),"utf-8")+bytes([13,10])
  155. self.socket.send(add)
  156. # self.buffer.append(add)
  157. def book(self,book):
  158. self.log("DOWNLOADING %s" % book)
  159. self.send_queue("PRIVMSG %s :%s " % (self.CHANNEL,book))
  160. def who(self):
  161. self.send_queue("WHO %s" % self.CHANNEL)
  162. def ip_from_decimal(self,dec):
  163. return ".".join([ str(int(b,2)) for b in self.b_octets(bin(dec)[2:]) ])
  164. def b_octets(self,l):
  165. l=l.zfill(32)
  166. return [l[0:8],l[8:16],l[16:24],l[24:32]]
  167. def log(self,val):
  168. if self.LOGGING:
  169. print(val)
  170. if __name__ == "__main__":
  171. if len(sys.argv)!=3:
  172. print("USAGE: %s <BOOK> <FORMAT>" % sys.argv[0])
  173. sys.exit(1)
  174. print("Looking for %s in format %s" % (sys.argv[1],sys.argv[2]))
  175. #client = IRCClient(sys.argv[1],sys.argv[2],"SEARCH",True)
  176. client = IRCClient(sys.argv[1],sys.argv[2],"BOOK",True)