lta-rs

🚍 Singapore LTA Datamall Rust Client written in pure rust with support for asynchronous requests. lta-rs is used to interact with the lta-datamall

lta-rs in action

Cargo.toml setup

There are various versions available. If you omit branch = "version_no", you are taking it from master branch The library is also available on crates.io toml [dependencies] lta = "0.3.0-async-preview-3"

API key setup

You can get your API key from here

```rust extern crate lta;

use lta::lta_client::*;

fn main() { let apikey = "MYAPIKEY"; let client = LTAClient::withapikey(apikey); } ```

Examples

Getting bus timings ```rust use lta::prelude::*; use lta::ltaclient::LTAClient; use lta::bus::getarrival; use lta::Result;

fn getbusarrival() -> Result<()> { let apikey = std::env::var("APIKEY").unwrap(); let client = LTAClient::withapikey(apikey); let arrivals: BusArrivalResp = getarrival(&client, 83139, "15")?; println!("{:?}", arrivals); Ok(()) } ```

Getting anything else ``rust // All the APIs in this library are designed to be used like this //module::getsomething` // All of them return lta::utils::Result> // The example below is bus::getbusservices() // and traffic::geterprates() // Do note that the API calling convention is similar across all the APIs except for // bus::getarrival // prefer lta::prelude::* over glob imports use lta::prelude::*; use lta::ltaclient::LTAClient; use lta::traffic::geterprates; use lta::bus::getbus_services; use lta::Result;

fn busservices() -> Result<()> { let apikey = std::env::var("APIKEY").unwrap(); let client = LTAClient::withapikey(apikey); let busservices: Vec = getbusservices(&client)?; println!("{:?}", busservices); Ok(()) }

fn geterp() -> Result<()> { let apikey = std::env::var("APIKEY").unwrap(); let client = LTAClient::withapikey(apikey); let erprates: Vec = geterprates(&client)?; println!("{:?}", erprates); Ok(()) } ```

Async Example

```rust use lta::r#async::{ prelude::*, ltaclient::LTAClient, bus::getarrival, traffic::geterprates }; use std::env::var; use tokio::run;

fn asyncexample(client: &LTAClient) -> impl Future, BusArrivalResp); let fut = geterprates(client); let fut2 = getarrival(client, 83139, "15"); fut.join(fut2) .map(|(a,b): Req| { println!("{:?}", a); println!("{:?}", b); }) .map_err(|e| println!("Request failed ${:?}", e)) }

fn multipleasyncrequests() { let apikey = var("APIKEY").unwrap(); let client = &LTAClient::withapikey(apikey); let fut = asyncexample(client); run(fut); } ```

Custom Client

There are some instance where you might need to customise the client more due to certain limitations. ```rust use std::time::Duration; use lta::reqwest::ClientBuilder; use lta::lta_client::LTAClient; use lta::utils::commons::Client;

fn mycustomclient() -> LTAClient { let client = ClientBuilder::new() .gzip(true) .connect_timeout(Some(Duration::new(420,0))) .build() .unwrap();

LTAClient::new(Some("api_key".to_string()), client)

} ```

Concurrent requests without `Futures`

```rust use std::sync::Arc; use std::thread::spawn; use lta::lta_client::LTAClient; use lta::utils::commons::Client;

fn concurrent() { let apikey = env::var("APIKEY").unwrap(); let c1 = Arc::new(LTAClient::withapikey(api_key)); let c2 = c1.clone();

let child = spawn(move || {
    let res = get_carpark_avail(&c1).unwrap();
    println!("{:?}", res)
});

let vms = traffic::get_vms_emas(&c2).unwrap();
println!("{:?}", vms);

child.join();

} ```

General advice

Reuse LTAClient as it holds a connection pool internally
Reduce the number of times you call the API, take a look at the Update Freq in the documentation andprevent yourself from getting blacklisted. Use a caching mechanism.
Prefer async APIs over writing your own implementation for concurrent requests.

Getting help

You can get help via github issues. I will try my best to respond to your queries :smile:

Design decisions

Made sure that Rust structs are as close to the original response as possible to make sure that people can reference the original docs if there are any issues
Simple and no additional baggage. Only the client is included. E.g If anyone wants to add concurrency, they have to do it on their own
Predictable API usage

Changelog

Version 0.1 - All endpoints that are available from lta datamall website - Configuration using API

Version 0.2 [ Breaking Changes ] - Changed all API to take in &LTAClient rather than using a global LTAClient

Version 0.2.1 - Updated dependencies to latest version as of 21 July 2019

Version 0.2.2 [ Broken getbusstops, yanked from crates.io ] - Updated LTAClient::with_api_key to create a LTAClient

Version 0.2.3 - Hotfix for broken lta::bus::get_bus_stops which will panic due to typo in serde rename

Version 0.3.0-async-preview-1 [ Breaking Changes ] - Client trait, now has 2 clients, one with async capabilities - Currently using futures-preview = "0.3.0-alpha.17" and tokio = "0.1.22"

Version 0.3.0-async-preview-2 [ Breaking Changes ] - Re-exports to ensure computability - Removed futures-preview = "0.3.0-alpha.17" - Examples for all API, with the exception of async

Version 0.3.0-async-preview-3 [ Breaking Changes ] - Removed some re-exports to avoid confusion - Removed futures-preview = "0.3.0-alpha.17" - Removed tokio as dependency and make it dev-dependency - Added futures = "0.1.28"

Todo (excluding bugs from issues)

[x] Proper date types using chrono library
[x] Utils cleanup
[x] Host on crates.io
[ ] Static website to showcase project
[x] Documentation
[x] More idiomatic Rust code
[x] Asynchronous requests
[x] Travis CI
[x] Documentation for async
[ ] std::future
[x] Customisable Client
[ ] Better testing, reduce API spam and cache data for testing
[ ] Deserialization benchmark

License

lta-rs is licensed under MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

Frequently Asked Questions

Is this library being actively developed?

Yes. However, development will slow down from mid August 2019 onwards due to my NS commitments.

What are the APIs available?

Take a look at the offical LTA docs.

Where do I get the official docs from lta?

You can get them here

Why are some of the data types different from the lta documentation?

Some of the data types returned are not ideal such as returning lat and lang as string rather than number. Some of the types are also converted to enums to reduce the number of stringly typed stuff

My application panicked.

Check if your API key is valid, if it is and your application still panics because of this library, create a github issue

Why is the most fully featured LTA client library implemented in a language not many people use?

Friendship ended with Kotlin. Now Rust is my best friend ❤️.

Is this project affiliated to LTA or any government body?

No.

What is the plan to move to std::future?

Currently waiting for dependencies to move to std::future. However, this might take some time and different libraries might update at different times, so I am currently experimenting on making the APIs exposed to users use std::future while the internal implementation depends on the compat layer provided by futures-preview.

All the async stuff is currently on preview and will be released for 0.3.0. I do not want to rush the implementation of async APIs to ensure that the ergonomics of them are user friendly. Considering that a lot of libraries are currently moving to std::future, this can be very confusing to beginners that want to take a look into futures.

Development of this happen on master branch.