3
5

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 5 years have passed since last update.

Rust で並列HTTP Request

Last updated at Posted at 2017-05-28

Rust の練習に、昔 HaskellClojure で書いた、並列にHTTP Request を送るプログラムを書いてみました。

但し、 HTTP ライブラリの hyper の非同期処理は絶賛開発中?のようだったので、今回は非同期IOによる並列処理ではなくて、スレッドによるものです。

extern crate hyper;

use std::io::prelude::*;
use std::io::stderr;
use std::sync::Arc;
use std::thread;
use std::time::{ Duration, SystemTime };
use hyper::Client;

fn to_seconds(duration: &Duration) -> f64 {
    let nanos = duration.subsec_nanos() as u64;
    let secs  = duration.as_secs();
    ((1000_000_000 * secs + nanos) as f64) / 1000_000_000.
}

fn do_http_request(client: Arc<Client>, index: usize, url: &str) {
    let time     = SystemTime::now();
    let result   = client.get(url).send();
    let duration = to_seconds(&time.elapsed().unwrap());

    match result {
        Ok(response) => {
            println!("{:2}: {:5.3}: {} => {}",
                     index, duration, url, response.status);
        }
        Err(e) => {
            let _ = writeln!(&mut stderr(), "{:2}: Error: {}", index, e);
        }
    };
}

fn main() {
    let gen_url = |q: &str| -> String { format!("http://www.google.com/search?q={}", q) };
    let urls: Vec<String> =
        vec!["clojure", "haskell", "ocaml", "rust", "scala"]
            .iter().map(|q| gen_url(q))
            .chain(vec!["http://127.0.0.1:5000/".to_owned()].into_iter())
            .chain(vec!["javascript", "python", "perl", "php", "ruby"]
                        .iter().map(|q| gen_url(q)))
            .collect();
    let client = Arc::new(Client::new());
    let mut children = vec![];

    for (i, url) in urls.into_iter().enumerate() {
        let client = client.clone();
        children.push(thread::spawn(move || {
            do_http_request(client, i+1, &url)
        }));
    }

    for child in children {
        let _ = child.join();
    }
}

//  To check parallel http requests, prepare a long time request:
//  $ rackup -s webrick --port 5000 -b 'run(proc { |env| sleep 5; ["200", {"Content-Type" => "text/html"}, ["Hello World"]] })'

//  $ cargo run
//  ...
//  1: 0.221: http://www.google.com/search?q=clojure => 200 OK
//  9: 0.223: http://www.google.com/search?q=perl => 200 OK
//  8: 0.223: http://www.google.com/search?q=python => 200 OK
// 10: 0.229: http://www.google.com/search?q=php => 200 OK
//  7: 0.235: http://www.google.com/search?q=javascript => 200 OK
//  5: 0.287: http://www.google.com/search?q=scala => 200 OK
// 11: 0.294: http://www.google.com/search?q=ruby => 200 OK
//  3: 0.303: http://www.google.com/search?q=ocaml => 200 OK
//  2: 0.429: http://www.google.com/search?q=haskell => 200 OK
//  4: 0.550: http://www.google.com/search?q=rust => 200 OK
//  6: 5.005: http://127.0.0.1:5000/ => 200 OK
Cargo.toml
[package]
name = "http_request1"
version = "0.1.0"

[dependencies]
hyper = "*"

はまったところ

最初、スレッド一覧を生成するのに、イテレーターを使って以下のようにしていたのですが、

let children = urls.into_iter().enumerate().map(|(i, url)| {
    let client = client.clone();
    thread::spawn(move || do_http_request(client, i+1, &url))
});

これだと、以下のように、常に HTTP Request が逐次実行されてしまい、並列に実行されずに悩んだのですが、

$ cargo run
...
 1: 0.708: http://www.google.com/search?q=clojure => 200 OK
 2: 0.186: http://www.google.com/search?q=haskell => 200 OK
 3: 0.179: http://www.google.com/search?q=ocaml => 200 OK
 4: 0.508: http://www.google.com/search?q=rust => 200 OK
 5: 0.179: http://www.google.com/search?q=scala => 200 OK
 6: 5.045: http://127.0.0.1:5000/ => 200 OK
 7: 0.173: http://www.google.com/search?q=javascript => 200 OK
 8: 0.181: http://www.google.com/search?q=python => 200 OK
 9: 0.177: http://www.google.com/search?q=perl => 200 OK
10: 0.192: http://www.google.com/search?q=php => 200 OK
11: 0.202: http://www.google.com/search?q=ruby => 200 OK

どうやら イテレーターは遅延評価されるらしく 、そのため今回、 map に渡したクロージャが即時評価されず、 child.join() で初めて値が評価されて実行が開始された、という感じのようです。

追記 - 2017-05-29

はまったところ でイテレーターの遅延評価について書きましたが、コメント にて、イテレーターとコンシューマの関係、そして for 以外にも collect メソッドがコンシューマであることを指摘頂いたので、 for の代わりに collect メソッドを呼ぶように修正しました。

extern crate hyper;

use std::io::prelude::*;
use std::io::stderr;
use std::sync::Arc;
use std::thread;
use std::time::{ Duration, SystemTime };
use hyper::Client;

fn to_seconds(duration: &Duration) -> f64 {
    let nanos = duration.subsec_nanos() as u64;
    let secs  = duration.as_secs();
    ((1000_000_000 * secs + nanos) as f64) / 1000_000_000.
}

fn do_http_request(client: Arc<Client>, index: usize, url: &str) {
    let time     = SystemTime::now();
    let result   = client.get(url).send();
    let duration = to_seconds(&time.elapsed().unwrap());

    match result {
        Ok(response) => {
            println!("{:2}: {:5.3}: {} => {}",
                     index, duration, url, response.status);
        }
        Err(e) => {
            let _ = writeln!(&mut stderr(), "{:2}: Error: {}", index, e);
        }
    };
}

fn main() {
    let gen_url = |q: &str| -> String { format!("http://www.google.com/search?q={}", q) };
    let urls: Vec<String> =
        vec![].into_iter()
            .chain(vec!["clojure", "haskell", "ocaml", "rust", "scala"]
                        .iter().map(|q| gen_url(q)))
            .chain(vec!["http://127.0.0.1:5000/".to_owned()].into_iter())
            .chain(vec!["javascript", "python", "perl", "php", "ruby"]
                        .iter().map(|q| gen_url(q)))
            .collect();
    let client = Arc::new(Client::new());

    let children = urls.into_iter().enumerate().map(|(i, url)| {
        let client = client.clone();
        thread::spawn(move || do_http_request(client, i+1, &url))
    }).collect::<Vec<_>>();

    for child in children {
        let _ = child.join();
    }
}
3
5
2

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
3
5

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?