使用Python编写基于DHT协议的BT资源爬虫

#!/usr/bin/env python
# encoding: utf-8
import socket
from hashlib import sha1
from random import randint
from struct import unpack
from socket import inet_ntoa
from threading import Timer, Thread
from time import sleep
from collections import deque
from bencode import bencode, bdecode
BOOTSTRAP_NODES = (
  ("router.bittorrent.com", 6881),
  ("dht.transmissionbt.com", 6881),
  ("router.utorrent.com", 6881)
)
TID_LENGTH = 2
RE_JOIN_DHT_INTERVAL = 3
TOKEN_LENGTH = 2
def entropy(length):
  return "".join(chr(randint(0, 255)) for _ in xrange(length))
def random_id():
  h = sha1()
  h.update(entropy(20))
  return h.digest()
def decode_nodes(nodes):
  n = []
  length = len(nodes)
  if (length % 26) != 0:
    return n
  for i in range(0, length, 26):
    nid = nodes[i:i+20]
    ip = inet_ntoa(nodes[i+20:i+24])
    port = unpack("!H", nodes[i+24:i+26])[0]
    n.append((nid, ip, port))
  return n
def timer(t, f):
  Timer(t, f).start()
def get_neighbor(target, nid, end=10):
  return target[:end]+nid[end:]
class KNode(object):
  def __init__(self, nid, ip, port):
    self.nid = nid
    self.ip = ip
    self.port = port
class DHTClient(Thread):
  def __init__(self, max_node_qsize):
    Thread.__init__(self)
    self.setDaemon(True)
    self.max_node_qsize = max_node_qsize
    self.nid = random_id()
    self.nodes = deque(maxlen=max_node_qsize)
  def send_krpc(self, msg, address):
    try:
      self.ufd.sendto(bencode(msg), address)
    except Exception:
      pass
  def send_find_node(self, address, nid=None):
    nid = get_neighbor(nid, self.nid) if nid else self.nid
    tid = entropy(TID_LENGTH)
    msg = {
      "t": tid,
      "y": "q",
      "q": "find_node",
      "a": {
        "id": nid,
        "target": random_id()
      }
    }
    self.send_krpc(msg, address)
  def join_DHT(self):
    for address in BOOTSTRAP_NODES:
      self.send_find_node(address)
  def re_join_DHT(self):
    if len(self.nodes) == 0:
      self.join_DHT()
    timer(RE_JOIN_DHT_INTERVAL, self.re_join_DHT)
  def auto_send_find_node(self):
    wait = 1.0 / self.max_node_qsize
    while True:
      try:
        node = self.nodes.popleft()
        self.send_find_node((node.ip, node.port), node.nid)
      except IndexError:
        pass
      sleep(wait)
  def process_find_node_response(self, msg, address):
    nodes = decode_nodes(msg["r"]["nodes"])
    for node in nodes:
      (nid, ip, port) = node
      if len(nid) != 20: continue
      if ip == self.bind_ip: continue
      if port < 1 or port > 65535: continue
      n = KNode(nid, ip, port)
      self.nodes.append(n)
class DHTServer(DHTClient):
  def __init__(self, master, bind_ip, bind_port, max_node_qsize):
    DHTClient.__init__(self, max_node_qsize)
    self.master = master
    self.bind_ip = bind_ip
    self.bind_port = bind_port
    self.process_request_actions = {
      "get_peers": self.on_get_peers_request,
      "announce_peer": self.on_announce_peer_request,
    }
    self.ufd = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, socket.IPPROTO_UDP)
    self.ufd.bind((self.bind_ip, self.bind_port))
    timer(RE_JOIN_DHT_INTERVAL, self.re_join_DHT)
  def run(self):
    self.re_join_DHT()
    while True:
      try:
        (data, address) = self.ufd.recvfrom(65536)
        msg = bdecode(data)
        self.on_message(msg, address)
      except Exception:
        pass
  def on_message(self, msg, address):
    try:
      if msg["y"] == "r":
        if msg["r"].has_key("nodes"):
          self.process_find_node_response(msg, address)
      elif msg["y"] == "q":
        try:
          self.process_request_actions[msg["q"]](msg, address)
        except KeyError:
          self.play_dead(msg, address)
    except KeyError:
      pass
  def on_get_peers_request(self, msg, address):
    try:
      infohash = msg["a"]["info_hash"]
      tid = msg["t"]
      nid = msg["a"]["id"]
      token = infohash[:TOKEN_LENGTH]
      msg = {
        "t": tid,
        "y": "r",
        "r": {
          "id": get_neighbor(infohash, self.nid),
          "nodes": "",
          "token": token
        }
      }
      self.send_krpc(msg, address)
    except KeyError:
      pass
  def on_announce_peer_request(self, msg, address):
    try:
      infohash = msg["a"]["info_hash"]
      #print msg["a"]
      tname = msg["a"]["name"]
      token = msg["a"]["token"]
      nid = msg["a"]["id"]
      tid = msg["t"]
      if infohash[:TOKEN_LENGTH] == token:
        if msg["a"].has_key("implied_port") and msg["a"]["implied_port"] != 0:
          port = address[1]
        else:
          port = msg["a"]["port"]
          if port < 1 or port > 65535: return
        self.master.log(infohash, (address[0], port),tname)
    except Exception:
      pass
    finally:
      self.ok(msg, address)
  def play_dead(self, msg, address):
    try:
      tid = msg["t"]
      msg = {
        "t": tid,
        "y": "e",
        "e": [202, "Server Error"]
      }
      self.send_krpc(msg, address)
    except KeyError:
      pass
  def ok(self, msg, address):
    try:
      tid = msg["t"]
      nid = msg["a"]["id"]
      msg = {
        "t": tid,
        "y": "r",
        "r": {
          "id": get_neighbor(nid, self.nid)
        }
      }
      self.send_krpc(msg, address)
    except KeyError:
      pass
class Master(object):
  def log(self, infohash,address=None,tname=None):
    hexinfohash = infohash.encode("hex")
    print "info_hash is: %s,name is: %s from %s:%s" % (
      hexinfohash,tname, address[0], address[1]
  )
    print "magnet:?xt=urn:btih:%s&dn=%s" % (hexinfohash, tname)
# using example
if __name__ == "__main__":
  # max_node_qsize bigger, bandwith bigger, speed higher
  dht = DHTServer(Master(), "0.0.0.0", 6882, max_node_qsize=200)
  dht.start()
  dht.auto_send_find_node()

PS： DHT协议中有几个重点的需要澄清的地方：

1. node与infohash同样使用160bit的表示方式，160bit意味着整个节点空间有2^160 = 730750818665451459101842416358141509827966271488，是48位10进制，也就是说有百亿亿亿亿亿个节点空间，这么大的节点空间，是足够存放你的主机节点以及任意的资源信息的。

2. 每个节点有张路由表。每张路由表由一堆K桶组成，所谓K桶，就是桶中最多只能放K个节点，默认是8个。而桶的保存则是类似一颗前缀树的方式。相当于一张8桶的路由表中最多有160-4个K桶。

3. 根据DHT协议的规定，每个infohash都是有位置的，因此，两个infohash之间就有距离一说，而两个infohash的距离就可以用异或来表示，即infohash1 xor infohash2，也就是说，高位一样的话，他们的距离就近，反之则远，这样可以快速的计算两个节点的距离。计算这个距离有什么用呢，在DHT网络中，如果一个资源的infohash与一个节点的infohash越近则该节点越有可能拥有该资源的信息，为什么呢？可以想象，因为人人都用同样的距离算法去递归的询问离资源接近的节点，并且只要该节点做出了回应，那么就会得到一个announce信息，也就是说跟资源infohash接近的节点就有更大的概率拿到该资源的infohash

4. 根据上述算法，DHT中的查询是跳跃式查询，可以迅速的跨越的的节点桶而接近目标节点桶。之所以在远处能够大幅度跳跃，而在近处只能小幅度跳跃，原因是每个节点的路由表中离自身越接近的节点保存得越多，如下图

5. 在一个DHT网络中当爬虫并不容易，不像普通爬虫一样，看到资源就可以主动爬下来，相反，因为得到资源的方式(get_peers, announce_peer)都是被动的，所以爬虫的方式就有些变化了，爬虫所要做的事就是像个正常节点一样去响应其他节点的查询，并且得到其他节点的回应，把其中的数据收集下来就算是完成工作了。而爬虫唯一能做的，是尽可能的去多认识其他节点，这样，才能有更多其他节点来向你询问。

6. 有人说，那么我把DHT爬虫的K桶中的容量K增大是不是就能增加得到资源的机会，其实不然，之前也分析过了，DHT爬虫最重要的信息来源全是被动的，因为你不能增大别人的K，所以距离远的节点保存你自身的概率就越小，当然距离远的节点去请求你的概率相对也比较小。