17monipdb 格式解码

17monipdb.py

#coding: utf8
"""17monipdb 解码"""


import socket
import struct

_unpack_ulong_little = lambda x: struct.unpack("<L", x)
_unpack_ulong_big = lambda x: struct.unpack(">L", x)


class IPDB(object):
    def __init__(self, datafn=None):
        if not datafn:
            datafn = "17monipdb.dat"
        self._datafn = datafn
        self._datafile = None
        self._index_end = None
        self._index = None
        self._cache = dict()

    def _do_init(self):
        if self._datafile is not None:
            return
        self._datafile = open(self._datafn, 'rb')
        # 头4字节保存着索引数据的结束位置
        self._index_end = _unpack_ulong_big(self._datafile.read(4))[0]
        index_len = self._index_end - self._datafile.tell()     # 索引数值的实际长度
        self._index = self._datafile.read(index_len)

    def find(self, host):
        ip_str = socket.gethostbyname(host)
        ip_int = socket.inet_aton(ip_str)

        if self._datafile is None:
            self._do_init()

        if ip_int in self._cache:
            return self._cache[ip_int]

        # 通过IP地址第一段找到一级索引的位置。一个索引数据占4字节，所以要乘以4
        this_index1_pos = int(ip_str.split(".")[0]) * 4

        index2_offset = 1024   # index2 在 index 里的偏移
        index2_len = len(self._index) - index2_offset
        # 通过一级索引找到2级索引的位置
        this_index2_num = _unpack_ulong_little(self._index[this_index1_pos:this_index1_pos+4])[0]
        # 二级索引一个索引数值占8字节
        this_index2_start = this_index2_num * 8

        curr_pos = this_index2_start
        this_index2_pos = None
        while curr_pos < index2_len:
            abs_index_pos = curr_pos + index2_offset
            if self._index[abs_index_pos:abs_index_pos+4] >= ip_int:
                this_index2_pos = abs_index_pos+4
                break
            curr_pos += 8
        else:
            return None

        data_pos = _unpack_ulong_little(self._index[this_index2_pos:this_index2_pos+3]+"\x00")[0]
        data_len = ord(self._index[this_index2_pos+3])

        self._datafile.seek(self._index_end+data_pos-1024)
        self._cache[ip_int] = self._datafile.read(data_len).decode("utf8")
        return self._cache[ip_int]

ipdb = IPDB()
find = ipdb.find

if __name__ == "__main__":
    print find("tool.17mon.cn")