lib: throws when invalid hex string in http request

[juanarbol]

Signed-off-by: Juan José Arboleda soyjuanarbol@gmail.com

May fix: #45150

I don't know why a string that does not follow this condition is considered an invalid hex string, but It is better to throw an exception instead of crashing the whole thing (IMO).

The condition: CHECK(str->Length() % 2 == 0 && "invalid hex string length");

[nodejs-github-bot]

Review requested:

• @nodejs/http
• @nodejs/net

[juanarbol]

cc @Flarna

What do you thing about this?

[mscdex]

What about just throwing a JS error from C instead of aborting?

[mcollina]

lgtm

[juanarbol]

What about just throwing a JS error from C instead of aborting?

I think that makes sense; any other reviewer w/ the same idea? I could throw from C instead.

[anonrig]

I feel like this should can be a function in validators.js like validateHexString()

[Flarna]

I tried to reproduce the same using net instead http but failed.

const net = require("net");
const s = net.createConnection(8000);
s.on("connect", () => {
    s.write("1", "hex");
});

net just discards the incomplete hex char.

Should we aim to be consistent?

Update:
I'm able to reproduce also with net now:

const net = require("net");
const s = net.createConnection(8000);
s.on("connect", () => {
    s.cork()
    s.write("a")
    s.write("1", "hex");
    s.uncork()
});

Seems writev aborts wheres write ignores chars.

As this is not specific to HTTP I would vote for a more generic solution like throwing a JS exception in C++ or ignore invalid chars (likely also in C++).

[mcollina]

Agreed, this solution should be generic and not limited to HTTP.
I think it might be better implemented in streams themselves.

[juanarbol]

Nice! Thanks for all the input; I will transform this into a "throw from C layer" approach instead. Thanks again.

[Flarna]

Thinking once more about this. An exception at this place may be hard to catch. To my understanding the data written might be buffered and passed down to native in a later tick without any usercode to place a try/catch around.

This and the fact that a single, non buffered write results in ignoring invalid data seems like it's maybe better to ignore it also in the writev case.