ipipdotnet / ipdb-luajit Goto Github PK

ipipfree.ipdb
local ipiploc = ipdb:find(ipipargs["ip"], "CN");

怎么解决类似下面的问题呢？都是按git里的方法使用的

2020/01/15 10:10:36 [error] 31256#0: *111517685 lua entry thread aborted: memory allocation error: not enough memory
stack traceback:
coroutine 0:
[C]: in function 'read'
/home/work/local/openresty/lualib/resty/ipdb/reader.lua:97: in function 'new'
/home/work/local/openresty/lualib/resty/ipdb/city.lua:14: in function 'new'

您好，我使用您们提供的ipdb-luajit解析ipv4地址001.000.128.000得到结果为：


然后，在使用相同的ipdb文件的情况下，使用ipdb-go解析001.000.128.000的结果为：

经过对比发现ipdb-luajit reader.lua里调用的C函数inet_pton：

无法正确处理001.000.128.000这种有前导0的ip，以下是inet_pton分别对001.000.128.000和1.0.128.0的处理结果：
1)001.000.128.000 不正常

2)1.0.128.0 正常

After reading several ipdb-* implementations, like:
Java: https://github.com/ipipdotnet/ipdb-java
Python: https://github.com/ipipdotnet/ipdb-python
C: https://github.com/ipipdotnet/ipdb-c
LuaJIT: https://github.com/ipipdotnet/ipdb-luajit (This one)

I realize that this implementation is different from the others in some details.

For example,
https://github.com/ipipdotnet/ipdb-luajit/blob/master/resty/ipdb/reader.lua#L148

for i = 0, 95 do
    if i >= 80 then

Unlike other implementations, this one doesn't check if current node is larger than the nodeCount.

https://github.com/ipipdotnet/ipdb-luajit/blob/master/resty/ipdb/reader.lua#L161

for i = 0, 128 do
    local val = band(1, rshift(ip_address[rshift(i, 3)], 7 - (i % 8)))

Other implementations use bitCount (which is 32 for IPv4, 128 for IPv6) as the upper boundary. But this one use a hard-code 128. BTW, it should be i = 0, 127.

https://github.com/ipipdotnet/ipdb-luajit/blob/master/resty/ipdb/reader.lua#L163

node = read_node(self.data, node, val)
if node > self.node_count then
    break
end

While other implementations check node > nodeCount before read_node, this implementation checks it after read_node.