CANYON-SQL

A full written in Rust ORM for multiple databases.

Canyon-SQL is a high level abstraction for working with multiple databases concurrently. Is build on top of the async language features to provide a high speed, high performant library to handling data access for consumers.

Early stage disclaimer

The library it's still on a early stage state. Any contrib via fork + PR it's really appreciated. Currently we are involved in a really active development on the project.

Full documentation resources

There is a work-in-progress web page, build with mdBook containing the official documentation. Here is where you will find all the technical documentation for Canyon-SQL. You can read it by clicking this link

Most important features

• Async by default. Almost every functionality provided is ready to be consumed concurrently.
• Use of multiple datasources. You can query multiple databases at the same time, even different ones!. This means that you will be able to query concurrently a PostgreSQL database and an SqlServer one in the same project.
• Is macro based. With a few annotations and a configuration file, you are ready to write your data access.
• Allows migrations. Canyon-SQL comes with a god-mode that will manage every table on your database for you. You can modify in Canyon code your tables internally, altering columns, setting up constraints... Also, in the future, we have plans to allow you to manipulate the whole server, like creating databases, altering configurations... everything, but in a programmatically approach with Canyon!

Supported databases

Canyon-SQL currently has support for work with the following databases:

• PostgreSQL (via tokio-postgres crate)
• SqlServer (via tiberius crate)

Every crate listed above is an async based crate, in line with the guidelines of the Canyon-SQL design.

There are plans for include more databases engines.

Better by example

Let's take a look to see how the Canyon code looks like!

The classical SELECT * FROM {table_name}

let find_all_result: Result<Vec<League>, Box<dyn Error + Send + Sync>> =  League::find_all().await;

// Connection doesn't return an error
assert!(find_all_result.is_ok());
// We retrieved elements from the League table
assert!(!find_all_result.unwrap().is_empty());

Performing a search over the primary key column

let find_by_pk_result: Result<Option<League>, Box<dyn Error + Send + Sync>> = League::find_by_pk(&1).await;

assert!(find_by_pk_result.as_ref().unwrap().is_some());

let some_league = find_by_pk_result.unwrap().unwrap();
assert_eq!(some_league.id, 1);
assert_eq!(some_league.ext_id, 100695891328981122_i64);
assert_eq!(some_league.slug, "european-masters");
assert_eq!(some_league.name, "European Masters");
assert_eq!(some_league.region, "EUROPE");
assert_eq!(
    some_league.image_url,
    "http://static.lolesports.com/leagues/EM_Bug_Outline1.png"
);

Note the leading reference on the find_by_pk(...) parameter. This associated function receives an &dyn QueryParameter<'_> as argument, not a value.

Building more complex queries

For exemplify the capabilities of Canyon, we will use SelectQueryBuilder<T>, which implements the QueryBuilder<T> trait for build a more complex where, filteing data and joining tables.

let mut select_with_joins = LeagueTournament::select_query();
    select_with_joins
        .inner_join("tournament", "league.id", "tournament.league_id")
        .left_join("team", "tournament.id", "player.tournament_id")
        .r#where(LeagueFieldValue::id(&7), Comp::Gt)
        .and(LeagueFieldValue::name(&"KOREA"), Comp::Eq)
        .and_values_in(LeagueField::name, &["LCK", "STRANGER THINGS"]);
    // NOTE: We don't have in the docker the generated relationships
    // with the joins, so for now, we are just going to check that the
    // generated SQL by the SelectQueryBuilder<T> is the spected
    assert_eq!(
        select_with_joins.read_sql(),
        "SELECT * FROM league INNER JOIN tournament ON league.id = tournament.league_id LEFT JOIN team ON tournament.id = player.tournament_id WHERE id > $1 AND name = $2  AND name IN ($2, $3) "
    )

Note: For now, when you use joins, you will need to create a new model with the columns in both tables (in case that you desire the data in such columns), but just follows the habitual process with the CanyonMapper. It will try to retrieve the data for every field declared. If you don't declare a field that is in the open clause, in this case (*), that field won't be retrieved. No problem. But if you have fields that aren't map able with some column in the database, the program will panic.

More examples

If you want to see more examples, you can take a look into the tests folder, at the root of this repository. Every available database operation is tested there, so you can use it to find the usage of the described operations in the documentation mentioned above

Contributing to CANYON-SQL

First of all, thanks for take in consideration help us with the project. You can take a look to our templated guide.

But, to summarize:

• Take a look at the already opened issues, to see if already exists of it's someone already taking care about solving it. Even tho, you can enter to participate and explain your point of view, or even help to accomplish the task
• Make a fork of Canyon-SQL
• If you opened an issue, create a branch from the base branch of the repo (that's the default), and point it to your fork
• After complete your changes, open a PR to the default branch. Fill the template provided in the best way you're able to do it
• Wait for the approval. In most of cases, a test over the feature will be required before approve your changes

What about the tests?

Typically in Canyon, isolated unit tests are written as doc-tests, and the integration ones are under the folder ./tests

If you want to run the tests (because this is the first thing that you want to do after fork the repo), a couple of things have to be considered before.

• You will need Docker installed in the target machine
• If you have Docker, and Canyon-SQL cloned of forked, you can run our docker-compose file (docker/docker-compose.yml), which will initialize a PostgreSQL database and will put content on it to make the tests able to work.
• Finally, some tests runs against MSSQL. We didn't found a nice way of inserting data directly when the Docker wakes up, but instead, we run a very special test located at tests/crud/mod.rs, that is named initialize_sql_server_docker_instance. When you run this one, initial data will be inserted into the tables that are created when this test run. (If you know a better way of doing this, please, open a issue to let us know it, and improve this process!)