bugfix: acl rule pattern match error for tree rule

[chuxi]

the generated acl file by running genacl_proxy_gfw_bypass_china_ip.py does not correctly match the pattern in src/acl/mod.rs.

the old version will use heavy regex set rules and it will compile all patterns when the first time used. Instead of using tree subdomain methods.

Actually, most rules are generated for subdomain tree. By removing the last duplicated \$, we can make it right.

If it is not right or has some problem, please comment here.

[zonyitoo]

You shouldn't just remove the last \$ instead of making it optional: \$?

[zonyitoo]

Hmm, I think it is just a special style only in https://raw.githubusercontent.com/NateScarlet/gfwlist.acl/master/gfwlist.acl.json . He omitted the tailing $ for simple domain rules, but kept the $ for complex rules:

\\.google\\.ws
^([^/]+\\.)*google\\.(ac|ad|ae|af|ai|al|am|as|at|az)$

It should be a special case, because if he wrote the first rule as ^([^/]+\\.)*\\.google\\.ws$, then it could also be handled as a tree rule.

[chuxi]

I had tried the two solution: removing \$ or updating \$?.

And the source file https://github.com/shadowsocks/shadowsocks-libev/blob/master/acl/gfwlist.acl it has the ending $. shadowsocks-rust does not have the ending $ for acl rules.

So to keep compatibility, I update it again to use \$?.

[chuxi]

Hmm, I think it is just a special style only in https://raw.githubusercontent.com/NateScarlet/gfwlist.acl/master/gfwlist.acl.json . He omitted the tailing $ for simple domain rules, but kept the $ for complex rules:

\\.google\\.ws
^([^/]+\\.)*google\\.(ac|ad|ae|af|ai|al|am|as|at|az)$

It should be a special case, because if he written the first rule as ^([^/+\\.)*\\.google\\.ws$, then it could also be handled as a tree rule.

Yes. I see. So I updated again.

[zonyitoo]

Let me think twice..

[chuxi]

Let me think twice..

If you decided, just tell me here...

I can accept both solutions...