actix / actix/actix-web

actix-web returns 400 bad request for http requests emitted by many user agents

Ouverte
#3,102 3 commentaires 0 réactions 0 personnes assignées Voir sur GitHub
A-http needs-investigation
Langage dominant
Rust
Étoiles
24.8k
Forks
1.9k
Merge moyen
23 h 10 min
PR mergées (30 j)
26

Description

Hello, and first, thank you for this great library !

Recently, I published a blog post titled [*I’m sorry I forked you*](https://sql.ophir.dev/blog.sql?post=I%E2%80%99m%20sorry%20I%20forked%20you). In the title, the second character is a [curly apostrophe](https://practicaltypography.com/straight-and-curly-quotes.html) (`’` U+2019 Right Single Quotation Mark).

I shared it online and started getting hits from a lot of different browsers. I significant portion of hits (I don't know which browsers exactly), did not encode the apostrophe (as `%E2%80%99`), but included the `’` directly in the HTTP query.

There are two layers between the web and my actix service:
- cloudflare, which parsed and understood the HTTP query perfectly well, and forwarded it with the curved apostrophe
- nginx, which also parsed and forwarded the query without issue.

But when it got to actix-web, it failed to parse the query, and returned a 400 back without even invoking my code.
The very confusing error message I got was: `[ERROR actix_http::h1::dispatcher] stream error: Request parse error: Invalid Header provided` (confusing because the problem did not state what the problem was exactly, and said it came from headers instead of the query string).

See: https://en.wikipedia.org/wiki/Internationalized_Resource_Identifier

## Expected Behavior

Since clients in the real world emit http requests with unicode characters, I think actix-web should accept them, and just invoke the user code with the unicode query string.

And when it encounters a real issue with the query string, it should say it comes from the query string, not from the headers, and give more details than just `Request parse error`.

## Current Behavior

logs `[ERROR actix_http::h1::dispatcher] stream error: Request parse error: Invalid Header provided`

and returns an HTTP 400 bad request response to the client.

## Steps to Reproduce (for bugs)

```rust
#[actix_web::main]
async fn main() -> std::io::Result<()> {
actix_web::HttpServer::new(|| actix_web::App::new())
.bind(("127.0.0.1", 8080))?
.run()
.await
}
```

```
❯ curl -v 'localhost:8080/’'
* Trying 127.0.0.1:8080...
* Connected to localhost (127.0.0.1) port 8080 (#0)
> GET /’ HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/7.81.0
> Accept: */*
>
* Mark bundle as not supporting multiuse
< HTTP/1.1 400 Bad Request
< content-length: 0
< connection: close
< date: Sun, 13 Aug 2023 20:01:26 GMT
<
* Closing connection 0
```

Guide de contribution

Ouvrir le guide de contribution

Piste de recherche

Start with the minimal actix_web::HttpServer example in the issue and reproduce the failure using curl with the unencoded curly apostrophe in the request path. Trace the request parsing that produces “Invalid Header provided,” then verify that valid Unicode requests reach user code and that malformed query or request input reports the correct source and details.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Évaluation

Stack technique
rust
Domaine
api, backend, networking
Type d'issue
Bug
Difficulté
4/5
Temps estimé
3-5 jours
Activité
À l'abandon
Clarté
Plutôt claire
Accessibilité débutants
35/100

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.