heej-ng / gluesql

GlueSQL is quite sticky, it attaches to anywhere.

GlueSQL

SQL Database Engine as a Library

GlueSQL is a SQL database library written in Rust. It provides a parser (sqlparser-rs), execution layer, and optional storage (sled) packaged into a single library. Developers can choose to use GlueSQL to build their own SQL database, or as an embedded SQL database using the default storage engine.

Standalone Mode

You can use GlueSQL as an embedded SQL database. GlueSQL provides sled as a default storage engine.

Installation

In your Cargo.toml:

[dependencies]
gluesql = "0.8"

Usage

use gluesql::*;
fn main() {
    let storage = SledStorage::new("data/doc-db").unwrap();
    let mut glue = Glue::new(storage);
    let sqls = vec![
        "DROP TABLE IF EXISTS Glue;",
        "CREATE TABLE Glue (id INTEGER);",
        "INSERT INTO Glue VALUES (100);",
        "INSERT INTO Glue VALUES (200);",
        "SELECT * FROM Glue WHERE id > 100;",
    ];

    for sql in sqls {
        let output = glue.execute(sql).unwrap();
        println!("{:?}", output)
    }
}

SQL Library Mode (For Custom Storage)

Installation

sled-storage is optional, so it is not required for custom storage makers.

[dependencies.gluesql]
version = "0.8"
default-features = false
features = ["sorter", "alter-table", "index", "transaction"]

Four features below are also optional.

• sorter - ORDER BY support for non-indexed expressions.
• alter-table - ALTER TABLE query support
• index - CREATE INDEX & DROP INDEX, index support
• transaction - BEGIN, ROLLBACK and COMMIT, transaction support

Usage

Two mandatory store traits to implement

• Store & StoreMut

pub trait Store<T: Debug> {
    async fn fetch_schema(..) -> ..;
    async fn scan_data(..) -> ..;
}

pub trait StoreMut<T: Debug> where Self: Sized {
    async fn insert_schema(..) -> ..;
    async fn delete_schema(..) -> ..;
    async fn insert_data(..) -> ..;
    async fn update_data(..) -> ..;
    async fn delete_data(..) -> ..;
}

Optional store traits

• AlterTable, Index & IndexMut and Transaction

pub trait AlterTable where Self: Sized {
    async fn rename_schema(..) -> ..;
    async fn rename_column(..) -> ..;
    async fn add_column(..) -> ..;
    async fn drop_column(..) -> ..;
}

pub trait Index<T: Debug> {
    async fn scan_indexed_data(..) -> ..;
}

pub trait IndexMut<T: Debug> where Self: Sized {
    async fn create_index(..) -> ..;
    async fn drop_index(..) -> ..;
}

pub trait Transaction where Self: Sized {
    async fn begin(..) -> ..;
    async fn rollback(..) -> ..;
    async fn commit(..) -> ..;
}

Use Cases

GlueSQL-js

https://github.com/gluesql/gluesql-js
Use SQL in web browsers! GlueSQL-js provides 3 storage options,

• in-memory
• localStorage
• sessionStorage

GlueSQL Sheets

https://sheets.gluesql.com
Turn Google Sheets into a SQL database!
It uses Google Sheets as a storage.
Data is stored and updated from Google Sheets.

Other expected use cases

• Add SQL layer to NoSQL databases: Redis, CouchDB...
• Build new SQL database management system

SQL Features

GlueSQL currently supports a limited subset of queries. It's being actively developed.

• CREATE TABLE with 8 types: INTEGER, FLOAT, BOOLEAN, TEXT, DATE, TIMESTAMP, TIME and INTERVAL.
• ALTER TABLE with 4 operations: ADD COLUMN, DROP COLUMN, RENAME COLUMN and RENAME TO.
• CREATE INDEX, DROP INDEX
• INSERT, UPDATE, DELETE, SELECT, DROP TABLE
• GROUP BY, HAVING
• ORDER BY
• Transaction queries: BEGIN, ROLLBACK and COMMIT
• Nested select, join, aggregations ...

You can see tests for the currently supported queries in src/tests/*.

Contribution

There are a few simple rules to follow.

• No mut keywords in src/executor and src/data.
• Iterator should not be evaluated in the middle of execution layer.
• Every error must have corresponding integration test cases to generate.
  (except for Unreachable- and Conflict- error types)