pg-worm

PostgreSQL's Worst ORM

pg-worm is a straightforward, fully typed, async ORM and Query Builder for PostgreSQL. Well, at least that's the goal.

Motivation

Existing ORMs are not async, require you to write migrations or use a cli. pg_worm's explicit goal is to be easy and to require no setup beyond defining your types.

pg_worm also has no DSL which you need to learn. Everything you can do maps intuitively and directly to SQL expressions.

And, last but not least, pg_worm does not get in your way. Instead of relying on weird DSL tricks to achieve complex queries, you can simply write your own SQL and still profit off automatic parsing, etc.

But that's enough ranting, go ahead and try it!

Usage

This library is based on tokio_postgres and is intended to be used with tokio.

Fortunately, using pg_worm is very easy.

Simply derive the Model trait for your type, connect to your database and you are ready to go!

Here's a quick example:

use pg_worm::prelude::*;

#[derive(Model)]
struct Book {
    // An auto-generated primary key
    #[column(primary_key, auto)]
    id: i64,
    title: String
    author_id: i64
}

#[derive(Model)]
struct Author {
    #[column(primary_key, auto)]
    id: i64,
    name: String,
    age: i64
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // First create a connection. This can be only done once.
    connect!("postgres://postgres:postgres@localhost:5432", NoTls).await?;

    // Then, create tables for your models. 
    // Use `register!` if you want to fail if a
    // table with the same name already exists.
    //
    // `force_register` drops the old table,
    // which is useful for development.
    //
    // If your tables already exist, skip this part.
    force_register!(Author, Book)?;

    // Next, insert some data.
    // This works by passing values for all
    // fields which aren't autogenerated.
    Author::insert("Stephen King").await?;
    Author::insert("Martin Luther King").await?;
    Author::insert("Karl Marx").await?;
    Book::insert("Foo - Part I", 1).await?;
    Book::insert("Foo - Part II", 2).await?;
    Book::insert("Foo - Part III", 3).await?;

    // Let's start with a simple query for all books:
    let books = Book::select().await?; // Vec<Book>
    assert_eq!(books.len(), 3);

    // You can also search for a specific book.
    // Adding a `WHERE` clause is as simple as
    // calling a method on the respective field:
    let book = Book::select_one()
        .where_(Book::title.eq(&"Foo - Part I".to_string()))
        .await?; // Option<Book>
    assert!(book.is_some());

    // Or update exsisting records:
    let books_updated = Book::update()
        .set(Book::title, &"Foo - Part III".to_string())
        .where_(Book::title.eq(&"Foo - Part II".to_string()))
        .await?; // u64
    assert_eq!(books_updated, 1);

    // Or delete a book, you don't like:
    let books_deleted = Book::delete()
        .where_(Book::title.eq(&"Foo - Part III".to_string()))
        .await?; // u64
    assert_eq!(books_deleted, 2);

    Ok(())
}

If you want to see more code examples, have a look at the tests directory.

Query Builders

As you can see in the above example, pg_worm allows you to build queries by chaining methods on so called 'builders'. For each query type pg_worm provides a respective builder (except for INSERT which is handled differently).

These builders expose a set of methods for building queries. Here's a list of them:

Filtering using WHERE

where_() can be used to easily include WHERE clauses in your queries.

This is done by passing a Where object which can be constructed by calling methods on the respective column. pg_worm automatically constructs a constant for each field of your Model.

A practical example would look like this:

let where_: Where<'_> = MyModel::my_field.eq(&5);

Available methods

Currently, the following methods are implemented:

Boolean logic

You can also chain/modify these filters with standard boolean logic:

Executing a query

After having finished building your query, you can simply call .await. This will turn the builder into a Query object which is then executed asynchronously.

Executing a query will always result in a Result.

Raw queries

Though these features are nice, they are not sufficient for most applications. This is why you can easily execute custom queries and still take advantage of automatic parsing, etc:

// NOTE: You have to pass the exact type that Postgres is 
// expecting. Doing otherwise will result in a runtime error.
let king_books = Book::query(r#"
        SELECT * FROM book 
        JOIN author ON author.id = book.author_id
        WHERE POSITION(? in author.name) > 0 
    "#, 
    vec![&"King".to_string()]
).await?;
assert_eq!(king_books.len(), 2);

License

This project is dual-licensed under the MIT and Apache 2.0 licenses.