Skip to content

Latest commit

 

History

History
420 lines (315 loc) · 14.3 KB

README.md

File metadata and controls

420 lines (315 loc) · 14.3 KB

ureq

A simple, safe HTTP client.

Ureq's first priority is being easy for you to use. It's great for anyone who wants a low-overhead HTTP client that just gets the job done. Works very well with HTTP APIs. Its features include cookies, JSON, HTTP proxies, HTTPS, charset decoding, and is based on the API of the http crate.

Ureq is in pure Rust for safety and ease of understanding. It avoids using unsafe directly. It uses blocking I/O instead of async I/O, because that keeps the API simple and keeps dependencies to a minimum. For TLS, ureq uses rustls or native-tls.

See the changelog for details of recent releases.

Usage

In its simplest form, ureq looks like this:

let body: String = ureq::get("http://example.com")
    .header("Example-Header", "header value")
    .call()?
    .body_mut()
    .read_to_string()?;

For more involved tasks, you'll want to create an [Agent]. An Agent holds a connection pool for reuse, and a cookie store if you use the cookies feature. An Agent can be cheaply cloned due to internal Arc and all clones of an Agent share state among each other. Creating an Agent also allows setting options like the TLS configuration.

use ureq::Agent;
use std::time::Duration;

let mut config = Agent::config_builder()
    .timeout_global(Some(Duration::from_secs(5)))
    .build();

let agent: Agent = config.into();

let body: String = agent.get("http://example.com/page")
    .call()?
    .body_mut()
    .read_to_string()?;

// Reuses the connection from previous request.
let response: String = agent.put("http://example.com/upload")
    .header("Authorization", "example-token")
    .send("some body data")?
    .body_mut()
    .read_to_string()?;

JSON

Ureq supports sending and receiving json, if you enable the json feature:

use serde::{Serialize, Deserialize};

#[derive(Serialize)]
struct MySendBody {
   thing: String,
}

#[derive(Deserialize)]
struct MyRecvBody {
   other: String,
}

let send_body = MySendBody { thing: "yo".to_string() };

// Requires the `json` feature enabled.
let recv_body = ureq::post("http://example.com/post/ingest")
    .header("X-My-Header", "Secret")
    .send_json(&send_body)?
    .body_mut()
    .read_json::<MyRecvBody>()?;

Error handling

ureq returns errors via Result<T, ureq::Error>. That includes I/O errors, protocol errors. By default, also HTTP status code errors (when the server responded 4xx or 5xx) results in Error.

This behavior can be turned off via [http_status_as_error()][crate::config::ConfigBuilder::http_status_as_error].

use ureq::Error;

match ureq::get("http://mypage.example.com/").call() {
    Ok(response) => { /* it worked */},
    Err(Error::StatusCode(code)) => {
        /* the server returned an unexpected status
           code (such as 400, 500 etc) */
    }
    Err(_) => { /* some kind of io/transport/etc error */ }
}

Features

To enable a minimal dependency tree, some features are off by default. You can control them when including ureq as a dependency.

ureq = { version = "3", features = ["socks-proxy", "charset"] }

The default enabled features are: rustls, gzip and json.

  • rustls enabled the rustls TLS implementation. This is the default for the the crate level convenience calls (ureq::get etc)
  • native-tls enables the native tls backend for TLS. Due to the risk of diamond dependencies accidentally switching on an unwanted TLS implementation, native-tls is never picked up as a default or used by the crate level convenience calls (ureq::get etc) – it must be configured on the agent
  • platform-verifier enables verifying the server certificates using a method native to the platform ureq is executing on. See rustls-platform-verifier crate
  • socks-proxy enables proxy config using the socks4://, socks4a://, socks5:// and socks:// (equal to socks5://) prefix
  • cookies enables cookies
  • gzip enables requests of gzip-compressed responses and decompresses them
  • brotli enables requests brotli-compressed responses and decompresses them
  • charset enables interpreting the charset part of the Content-Type header (e.g. Content-Type: text/plain; charset=iso-8859-1). Without this, the library defaults to Rust's built in utf-8
  • json enables JSON sending and receiving via serde_json
  • vendored compiles and statically links to a copy of non-Rust vendors (e.g. OpenSSL from native-tls)

TLS (https)

rustls

By default, ureq uses rustls crate with the ring cryptographic provider. As of Sep 2024, the ring provider has a higher chance of compiling successfully. If the user installs another default provider, that choice is respected.

// This uses rustls
ureq::get("https://www.google.com/").call().unwrap();

native-tls

As an alternative, ureq ships with native-tls as a TLS provider. This must be enabled using the native-tls feature. Due to the risk of diamond dependencies accidentally switching on an unwanted TLS implementation, native-tls is never picked up as a default or used by the crate level convenience calls (ureq::get etc) – it must be configured on the agent.

use ureq::config::Config;
use ureq::tls::{TlsConfig, TlsProvider};

let mut config = Config::builder()
    .tls_config(
        TlsConfig::builder()
            // requires the native-tls feature
            .provider(TlsProvider::NativeTls)
            .build()
    )
    .build();

let agent = config.new_agent();

agent.get("https://www.google.com/").call().unwrap();

Root certificates

webpki-roots

By default, ureq uses Mozilla's root certificates via the webpki-roots crate. This is a static bundle of root certificates that do not update automatically. It also circumvents whatever root certificates are installed on the host running ureq, which might be a good or a bad thing depending on your perspective. There is also no mechanism for SCT, CRLs or other revocations. To maintain a "fresh" list of root certs, you need to bump the ureq dependency from time to time.

The main reason for chosing this as the default is to minimize the number of dependencies. More details about this decision can be found at PR 818

If your use case for ureq is talking to a limited number of servers with high trust, the default setting is likely sufficient. If you use ureq with a high number of servers, or servers you don't trust, we recommend using the platform verifier (see below).

platform-verifier

The rustls-platform-verifier crate provides access to natively checking the certificate via your OS. To use this verifier, you need to enable it using feature flag platform-verifier as well as configure an agent to use it.

use ureq::Agent;
use ureq::tls::{TlsConfig, RootCerts};

let agent = Agent::config_builder()
    .tls_config(
        TlsConfig::builder()
            .root_certs(RootCerts::PlatformVerifier)
            .build()
    )
    .build()
    .new_agent();

let response = agent.get("https://httpbin.org/get").call()?;

Setting RootCerts::PlatformVerifier together with TlsProvider::NativeTls means also native-tls will use the OS roots instead of webpki-roots crate. Whether that results in a config that has CRLs and revocations is up to whatever native-tls links to.

JSON

By enabling the json feature, the library supports serde json.

This is enabled by default.

  • [request.send_json()][RequestBuilder::send_json()] send body as json.
  • [body.read_json()][Body::read_json()] transform response to json.

Sending body data

HTTP/1.1 has two ways of transfering body data. Either of a known size with the Content-Length HTTP header, or unknown size with the Transfer-Encoding: chunked header. ureq supports both and will use the appropriate method depending on which body is being sent.

ureq has a [AsSendBody] trait that is implemented for many well known types of data that we might want to send. The request body can thus be anything from a String to a File, see below.

Content-Length

The library will send a Content-Length header on requests with bodies of known size, in other words, if the body to send is one of:

  • &[u8]
  • &[u8; N]
  • &str
  • String
  • &String
  • Vec<u8>
  • &Vec<u8>)
  • [SendBody::from_json()] (implicitly via [RequestBuilder::send_json()])

Transfer-Encoding: chunked

ureq will send a Transfer-Encoding: chunked header on requests where the body is of unknown size. The body is automatically converted to an [std::io::Read] when the type is one of:

  • File
  • &File
  • TcpStream
  • &TcpStream
  • Stdin
  • UnixStream (not on windows)

From readers

The chunked method also applies for bodies constructed via:

  • [SendBody::from_reader()]
  • [SendBody::from_owned_reader()]

Proxying a response body

As a special case, when ureq sends a [Body] from a previous http call, the use of Content-Length or chunked depends on situation. For input such as gzip decoding (gzip feature) or charset transformation (charset feature), the output body might not match the input, which means ureq is forced to use the chunked method.

  • Response<Body>

Sending form data

[RequestBuilder::send_form()] provides a way to send application/x-www-form-urlencoded encoded data. The key/values provided will be URL encoded.

Overriding

If you set your own Content-Length or Transfer-Encoding header before sending the body, ureq will respect that header by not overriding it, and by encoding the body or not, as indicated by the headers you set.

let resp = ureq::put("https://httpbin.org/put")
    .header("Transfer-Encoding", "chunked")
    .send("Hello world")?;

Character encoding

By enabling the charset feature, the library supports receiving other character sets than utf-8.

For [Body::read_to_string()] we read the header like:

Content-Type: text/plain; charset=iso-8859-1

and if it contains a charset specification, we try to decode the body using that encoding. In the absence of, or failing to interpret the charset, we fall back on utf-8.

Currently ureq does not provide a way to encode when sending request bodies.

Lossy utf-8

When reading text bodies (with a Content-Type starting text/ as in text/plain, text/html, etc), ureq can ensure the body is possible to read as a String also if it contains characters that are not valid for utf-8. Invalid characters are replaced with a question mark ? (NOT the utf-8 replacement character).

For [Body::read_to_string()] this is turned on by default, but it can be disabled and conversely for [Body::as_reader()] it is not enabled, but can be.

To precisely configure the behavior use [Body::with_config()].

Proxying

ureq supports two kinds of proxies, HTTP (CONNECT), SOCKS4/SOCKS5, the former is always available while the latter must be enabled using the feature socks-proxy.

Proxies settings are configured on an [Agent]. All request sent through the agent will be proxied.

Example using HTTP

use ureq::{Agent, Proxy};
// Configure an http connect proxy.
let proxy = Proxy::new("http://user:[email protected]:9090")?;
let agent: Agent = Agent::config_builder()
    .proxy(Some(proxy))
    .build()
    .into();

// This is proxied.
let resp = agent.get("http://cool.server").call()?;

Example using SOCKS5

use ureq::{Agent, Proxy};
// Configure a SOCKS proxy.
let proxy = Proxy::new("socks5://user:[email protected]:9090")?;
let agent: Agent = Agent::config_builder()
    .proxy(Some(proxy))
    .build()
    .into();

// This is proxied.
let resp = agent.get("http://cool.server").call()?;

Versioning

Semver and unversioned

ureq follows semver. From ureq 3.x we strive to have a much closer adherence to semver than 2.x. The main mistake in 2.x was to re-export crates that were not yet semver 1.0. In ureq 3.x TLS and cookie configuration is shimmed using our own types.

ureq 3.x is trying out two new traits that had no equivalent in 2.x, [Transport][unversioned::transport::Transport] and [Resolver][unversioned::resolver::Resolver]. These allow the user write their own bespoke transports and (DNS name) resolver. The API:s for these parts are not yet solidified. They live under the [unversioned] module, and do not follow semver. See module doc for more info.

Minimum Supported Rust Version (MSRV)

From time to time we will need to update our minimum supported Rust version (MSRV). This is not something we do lightly; our ambition is to be as conservative with MSRV as possible.

  • For some dependencies, we will opt for pinning the version of the dep instead of bumping our MSRV.
  • For important dependencies, like the TLS libraries, we cannot hold back our MSRV if they change.
  • We do not consider MSRV changes to be breaking for the purposes of semver.
  • We will not make MSRV changes in patch releases.
  • MSRV changes will get their own minor release, and not be co-mingled with other changes.