SQL over HTTP.

sqld supports MySQL (-type mysql), Postgres (-type postgres), and SQLite (-type sqlite3) databases.

go get github.com/mmaelzer/sqld

Usage of 'sqld':
	sqld -u root -db database_name -h localhost:3306 -type mysql

Flags:
  -db string
    	database name
  -dsn string
    	database source name
  -h string
    	database host
  -p string
    	database password
  -port int
    	http port (default 8080)
  -raw
    	allow raw sql queries
  -type string
    	database type (default "mysql")
  -u string
    	database username (default "root")
  -url string
    	url prefix (default "/")

The name of the database. Just like use my_database.

The dsn is the data source name for the database, used when making the initial connection to the database. If specified, any host (h), user (u), or password (p) values will be ignored in favor of the dsn.

• MySQL : For MySQL the format looks like {user}/{password}@({host})/{database_name}?parseTime=true. More info on MySQL dsn values: https://github.com/go-sql-driver/mysql#dsn-data-source-name

• Postgres : For Postgres the format looks like postgres://{user}:{password}@{host}/{database_name}?sslmode=disable. More info on Postgres dsn values: https://godoc.org/github.com/lib/pq#hdr-Connection_String_Parameters

• SQLite : For SQLite the format can be a file name test.db or file:test.db?cache=shared&mode=memory or an in-memory store with :memory:. More info on SQLite dsn values: https://godoc.org/github.com/mattn/go-sqlite3#SQLiteDriver.Open

The database hostname. For example, running locally, MySQL will generally be localhost:3306 and for Postgres localhost:5432.

The database password.

The HTTP port to serve requests from.

The database type. Currently supported types are mysql, postgres, and sqlite3.

The database username.

The url prefix to use. For example -url api will serve requests from http://hostname:port/api/table or -url foo/bar will serve requests from http://hostname:port/foo/bar/table.

Interact with the database via URLs.

http://localhost:8080/table_name

The following equivalent to a request with table_name?id=10

http://localhost:8080/table_name/10

http://localhost:8080/table_name?id=10
http://localhost:8080/table_name?name=fred&age=67

http://localhost:8080/table_name?__limit__=20&name=bob

http://localhost:8080/table_name?__limit__=20&__offset__=100

http://localhost:8080/table_name?__order_by__=id+DESC

Create rows in the database via POST requests.

POST http://localhost:8080/table_name

{
  "name": "jim",
  "age": 54
}

{
  "id": 10,
  "name": "jim",
  "age": 54
}

Update a row in the database with PUT requests.

PUT http://localhost:8080/table_name/:id?where=clause

{
  "name": "jill"
}

Empty

Delete a row in the database with DELETE requests.

DELETE http://localhost:8080/table_name/:id?where=clause

Empty

If you use the -raw flag when launching sqld, you can POST raw SQL queries that will be evaluated and returned. Queries are provided inside of the JSON request body with either read or write keys and string values that contain the SQL to execute.

For example, if we run sqld -name=my_db -raw we can perform queries like:

POST http://localhost:8080

{
  "read": "SELECT * FROM user WHERE name LIKE %ji%"
}

[
  {
    "id": 66,
    "name": "jill"
  },
  {
    "id": 71,
    "name": "jim"
  }
]

{
  "write": "CREATE TABLE number (id INT NOT NULL AUTO_INCREMENT, num INT NOT NULL, PRIMARY KEY ( id ) )"
}

{
    "last_insert_id": 0,
    "rows_affected": 0
}

For a completely unscientific benchmark, on my Core i5 laptop (2 cores), I ran wrk against a local sqld / mysql instance on a 2-column table with 10 rows. The corresponding SQL query SELECT * FROM user takes ~250μs when run in the mysql console.

❯ wrk -t10 -d10 http://localhost:8080/user
Running 10s test @ http://localhost:8080/user
  10 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     3.82ms   10.28ms 114.96ms   97.53%
    Req/Sec   484.34    124.50     0.86k    73.02%
  48213 requests in 10.02s, 16.60MB read
Requests/sec:   4811.80
Transfer/sec:      1.66MB

MIT