Additional proxy providers

[Pierce Freeman]

This benchmarking is geared to find a proxy application for use in web crawl caching and test harnesses. As such, we need the following functionality:

• HTTP + HTTP(s) support. HTTP/2 is optional
• Client overrides of defined webpages, so we are able to return a mocked version of the page at test time
• Capture client requests for later use in tests. I sometimes refer to this as "replaying the tape," assuming that a cassette started rolling at request time.

Based on these requirements, especially the https support, we can't leverage normal forward proxies. Standard proxies rely on issuing a TCP handshake between the client and upstream server via http connect tunneling, so the proxy server doesn't have access to the request or response content. This has the benefit of ensuring the security of client connections. It also has the positive upshot of providing "transparency" to the proxy, so if destination servers rely on fingerprinting the TCP handshake of the client, it will appear as if the requests are coming from a valid origin.

These requirements mean that we're looking for a mittm proxy server.

Candidate proxies:

• https://mitmproxy.org/
• https://github.com/bjowes/node-http-mitm-proxy (fork of https://github.com/joeferner/node-http-mitm-proxy) - better certificate support
• https://github.com/AdguardTeam/gomitmproxy
• https://github.com/elazarl/goproxy
• https://github.com/google/martian

Ruled out:

• https://medium.com/@mlowicki/http-s-proxy-in-golang-in-less-than-100-lines-of-code-6a51c2f2c38c
• https://github.com/flomesh-io/pipy
• https://github.com/caddyserver/forwardproxy