The General Transit Feed Specification (GTFS) is a commonly used model to represent public transit data.
This crates brings serde structures of this model and helpers to read GTFS archives.
This crates has 2 main entry-points.
The most common one is to create a gtfs_structures::Gtfs:
```rust // Gtfs::new will try to guess if you provide a path, a local zip file or a remote zip file. // You can also use Gtfs::frompath or Gtfs::fromurl let gtfs = gtfsstructures::Gtfs::new("pathofazipordirectoryorurl")?; println!("there are {} stops in the gtfs", gtfs.stops.len());
// This structure is the easiest to use as the collections are HashMap,
// thus you can access an object by its id.
let route1 = gtfs.routes.get("1").expect("no route 1");
println!("{}: {:?}", route1.shortname, route1);
```
If you want a lower level model, you can use gtfs_structures::RawGtfs:
rust
let raw_gtfs = RawGtfs::new("fixtures/basic").expect("impossible to read gtfs");
for stop in raw_gtfs.stops.expect("impossible to read stops.txt") {
println!("stop: {}", stop.name);
}
Instead of easy to use HashMap, each collection is a Result with an error if something went wrong during the reading.
This makes it possible for example for a GTFS validator to display better error messages.
By default the feature 'read-url' is activated. It makes it possible to read a Gtfs from an url.
rust
let gtfs = gtfs_structures::Gtfs::new("http://www.metromobilite.fr/data/Horaires/SEM-GTFS.zip")?;
Or you can use the explicit constructor:
rust
let gtfs = gtfs_structures::Gtfs::from_url("http://www.metromobilite.fr/data/Horaires/SEM-GTFS.zip")?;
If you don't want the dependency to reqwest, you can remove this feature.
You need an up to date rust tool-chain (commonly installed with rustup).
Building is done with:
cargo build
You can also run the unit tests:
cargo test
And run the examples by giving their names:
cargo run --example gtfs_reading
If you are interested in transit data, you can also use the really nice crate transit_model that can also handle GTFS data.
The price to pay is a steeper learning curve (and a documentation that could be improved :roll_eyes:), but this crate provides very nice ergonomics to handle transit data and lots of utilities like data format conversions, datasets merge, ...