gosqlapi

Turns any SQL database into a RESTful API. Currently supports MySQL, MariaDB, PostgreSQL, Oracle, Microsoft SQL Server and SQLite.

The intention of gosqlapi is not to replace a full-fledged backend server, but to provide a quick and easy way to expose any SQL databases as a RESTful API without writing any server code, except for only SQL scripts.

Installation

Install gosqlapi with one of the following commands, depending on the databases you want to use:

# Install gosqlapi for all databases.
$ go install github.com/elgs/gosqlapi@all

# Install gosqlapi for MySQL and MariaDB - https://github.com/go-sql-driver/mysql
$ go install github.com/elgs/gosqlapi@mysql

# Install gosqlapi for PostgreSQL - https://github.com/lib/pq
$ go install github.com/elgs/gosqlapi@postgres

# Install gosqlapi for PostgreSQL - https://github.com/jackc/pgx
$ go install github.com/elgs/gosqlapi@pgx

# Install gosqlapi for Oracle - https://github.com/sijms/go-ora
$ go install github.com/elgs/gosqlapi@oracle

# Install gosqlapi for Microsoft SQL Server - https://github.com/microsoft/go-mssqldb
$ go install github.com/elgs/gosqlapi@sqlserver

# Install gosqlapi for SQLite - https://pkg.go.dev/modernc.org/sqlite
$ go install github.com/elgs/gosqlapi@sqlite

# Install gosqlapi for all databases from the latest commit. Things may break.
$ go install github.com/elgs/gosqlapi@latest

To uninstall gosqlapi:

$ go clean -i github.com/elgs/gosqlapi

Download pre-built binaries

If you don't have Go installed, you can download pre-built binaries from the releases page.

Usage

$ gosqlapi

or if you don't have gosqlapi.json in the current directory:

$ gosqlapi -c /path/to/gosqlapi.json

Hello World

On the server side

Prepare gosqlapi.json and init.sql in the current directory, and run gosqlapi:

gosqlapi.json:

{
  "web": {
    "http_addr": "127.0.0.1:8080",
    "cors": true
  },
  "databases": {
    "test_db": {
      "type": "sqlite",
      "url": ":memory:"
    }
  },
  "scripts": {
    "init": {
      "database": "test_db",
      "path": "init.sql",
      "public_exec": true
    }
  },
  "tables": {
    "test_table": {
      "database": "test_db",
      "name": "TEST_TABLE",
      "public_read": true,
      "public_write": true
    }
  }
}

init.sql:

drop TABLE IF EXISTS TEST_TABLE;
create TABLE IF NOT EXISTS TEST_TABLE(
    ID INTEGER NOT NULL PRIMARY KEY,
    NAME VARCHAR(50)
);

insert INTO TEST_TABLE (ID, NAME) VALUES (1, 'Alpha');
insert INTO TEST_TABLE (ID, NAME) VALUES (2, 'Beta');
insert INTO TEST_TABLE (ID, NAME) VALUES (3, 'Gamma');

-- @label: data
SELECT * FROM TEST_TABLE WHERE ID > ?low? AND ID < ?high?;

On the client side

Run a pre-defined SQL query

$ curl -X PATCH 'http://localhost:8080/test_db/init/' \
  --header 'Content-Type: application/json' \
  --data-raw '{
  "low": 0,
  "high": 3
}'

{
  "data": [
    { "id": 1, "name": "Alpha" },
    { "id": 2, "name": "Beta" }
  ]
}

Get a record

$ curl -X GET 'http://localhost:8080/test_db/test_table/1'

{ "id": 1, "name": "Alpha" }

Create a new record

$ curl -X POST 'http://localhost:8080/test_db/test_table' \
  --header 'Content-Type: application/json' \
  --data-raw '{
  "id": 4,
  "name": "Delta"
}'

{ "last_insert_id": 4, "rows_affected": 1 }

Update a record

$ curl -X PUT 'http://localhost:8080/test_db/test_table/4' \
  --header 'Content-Type: application/json' \
  --data-raw '{
  "name": "Omega"
}'

{ "last_insert_id": 4, "rows_affected": 1 }

Delete a record

$ curl -X DELETE 'http://localhost:8080/test_db/test_table/4'

{ "last_insert_id": 4, "rows_affected": 1 }

Primary Key

If the table's primary key is not ID, you can specify the primary key for a table in the gosqlapi.json file:

{
  "tables": {
    "test_table": {
      "database": "test_db",
      "name": "TEST_TABLE",
      "primary_key": "UID"
    }
  }
}

Search for records

$ curl -X GET 'http://localhost:8080/test_db/test_table?name=Beta'

[{ "id": 2, "name": "Beta" }]

Search for records with .page_size, .offset and .order_by

$ curl --request GET \
  --url 'http://localhost:8080/test_db/test_table?.page_size=2&.offset=1&.show_total=1'

{
  "data": [
    {
      "id": 2,
      "name": "Beta"
    },
    {
      "id": 3,
      "name": "Gamma"
    }
  ],
  "offset": 1,
  "page_size": 2,
  "total": 3
}

You can use the following parameters:

.page_size: maximum number of records returned
.offset: offset the number of records returned
.order_by: order the records returned
.show_total: show the total number of records

You can give a table a default page_size, order_by, show_total and exported_columns by setting page_size and order_by in gosqlapi.json:

{
  "tables": {
    "test_table": {
      "database": "test_db",
      "name": "TEST_TABLE",
      "public_read": true,
      "public_write": true,
      "page_size": 10,
      "order_by": "NAME DESC, ID ASC",
      "show_total": true,
      "exported_columns": ["ID", "NAME AS USERNAME"]
    }
  }
}

if exported_columns is not set or is empty, all columns will be exported.

Passing SQL `NULL` from URL parameters

You can pass SQL NULL from URL parameters by setting null_value in gosqlapi.json. For example:

{
  "null_value": "null"
}

Then you can pass null from URL parameters:

$ curl -X GET 'http://localhost:8080/test_db/test_table?name=null'

[{ "id": 1, "name": null }]

Access Control

When a script has public_exec set to true, it can be executed by public users. When a table has public_read set to true, it can be read by public users. When a table has public_write set to true, it can be written by public users.

When a script or table is set to not be accessible by public users, an auth token is required to access the script or table. The client should send the auth token back to the server in the Authorization header. The server will verify the auth token and return an error if the auth token is invalid.

Simple Tokens

Simple tokens are configured in gosqlapi.json:

{
  "tokens": {
    "401d2fe0a18b26b4ce5f16c76cca6d484707f70a3a804d1c2f5e3fa1971d2fc0": [
      {
        "target_database": "test_db",
        "target_objects": ["test_table"],
        "read_private": true,
        "write_private": true,
        "allowed_origins": ["localhost", "*.example.com"]
      },
      {
        "target_database": "test_db",
        "target_objects": ["init"],
        "exec_private": true
      }
    ]
  }
}

In the example above, the auth token is configured to allow users to read and write test_table and execute init script in test_db.

The allowed_origins field is optional. If it is set, the server will only allow requests from the specified origins. It checks the Origin or Referer header to determine the origin of the request. If it is not set, it will reject all requests. If it is set to *, it will allow requests from all origins or referers.

You can use * to match all databases or all target objects:

{
  "tokens": {
    "401d2fe0a18b26b4ce5f16c76cca6d484707f70a3a804d1c2f5e3fa1971d2fc0": [
      {
        "target_database": "*",
        "target_objects": ["*"],
        "read_private": true,
        "write_private": true,
        "exec_private": true
      }
    ]
  }
}

This token will have super power.

Managed Tokens

Token Table

Managed tokens are stored in the database. The table and database that will store managed tokens are configured as managed_tokens in gosqlapi.json.

{
  "managed_tokens": {
    "database": "test_db",
    "table_name": "TOKENS"
  }
}

The table that stores managed tokens should have the following schema:

CREATE TABLE IF NOT EXISTS `TOKENS` (
  `ID` CHAR(36) NOT NULL,
  `USER_ID` CHAR(36) NOT NULL,
  `TOKEN` VARCHAR(255) NOT NULL,              -- required, auth token
  `TARGET_DATABASE` VARCHAR(255) NOT NULL,    -- required, target database
  `TARGET_OBJECTS` TEXT NOT NULL,             -- required, target objects, separated by whitespace
  `READ_PRIVATE` INT NOT NULL DEFAULT 0 ,     -- required, 1: read, 0: no read
  `WRITE_PRIVATE` INT NOT NULL DEFAULT 0 ,    -- required, 1: write, 0: no write
  `EXEC_PRIVATE` INT NOT NULL DEFAULT 0 ,     -- required, 1: exec, 0: no exec
  `ALLOWED_ORIGINS` TEXT NOT NULL,            -- required, allowed origins or referers, separated by whitespace
  CONSTRAINT `PRIMARY` PRIMARY KEY (`ID`)
);
create INDEX TOKEN_INDEX ON TOKENS (TOKEN);

Please feel free to change the ID to a different type, such as INT, or add more columns to the table. The only requirement is that the table should have the required columns listed above. Also consider adding an index to the TOKEN column.

When managed_tokens is configured in gosqlapi.json, the tokens in gosqlapi.json will be ignored.

If you already have a table that stores managed tokens, you can map the fields in that token table as follows:

{
  "managed_tokens": {
    "database": "test_db",
    "table_name": "TOKENS",
    "token": "AUTH_TOKEN",
    "target_database": "TARGET_DATABASE",
    "target_objects": "TARGET_OBJECTS",
    "read_private": "READ_PRIVATE",
    "write_private": "WRITE_PRIVATE",
    "exec_private": "EXEC_PRIVATE",
    "allowed_origins": "ALLOWED_ORIGINS"
  }
}

For example, if your token table has the field AUTH_TOKEN instead of TOKEN, you can use the configuration above to map the field AUTH_TOKEN to TOKEN.

Token Query

Instead of specifying the table_name, you can use a query in the config. The query should return the same columns as the token table.

{
  "managed_tokens": {
    "database": "test_db",
    "query": "SELECT TARGET_DATABASE AS target_database, TARGET_OBJECTS AS target_objects, READ_PRIVATE AS read_private, WRITE_PRIVATE AS write_private, EXEC_PRIVATE AS exec_private, ALLOWED_ORIGINS AS allowed_origins FROM TOKENS WHERE TOKEN=?token?"
  }
}

The placeholder will be replaced with the auth token. If the query is getting too long, you can use a separate file to store the query.

{
  "managed_tokens": {
    "database": "test_db",
    "query_path": "token_query.sql"
  }
}

Cache Managed Tokens

In production, you may want to cache the managed tokens in memory. To enable caching, set cache_tokens to true in gosqlapi.json. This will prevent the server from querying the database for tokens for every request.

{
  "cache_tokens": true
}

When any token is updated, an update to the cache is necessary. To update the token cache, send a POST request to /.clear-tokens with the following header:

Authorization: Bearer <auth token>

Pre-defined SQL Queries

There are a few things to note when defining a pre-defined SQL query in a script:

drop TABLE IF EXISTS TEST_TABLE;
create TABLE IF NOT EXISTS TEST_TABLE(
    ID INTEGER NOT NULL PRIMARY KEY,
    NAME TEXT
);

insert INTO TEST_TABLE (ID, NAME) VALUES (1, 'Alpha');
insert INTO TEST_TABLE (ID, NAME) VALUES (2, 'Beta');
insert INTO TEST_TABLE (ID, NAME) VALUES (3, 'Gamma');

-- @label: data
SELECT * FROM TEST_TABLE WHERE ID > ?low? AND ID < ?high?;

You can define multiple SQL statements in a single script. The statements will be executed in the order they appear in the script. The script will be executed in a transaction. If any statement fails, the transaction will be rolled back, and if all statements succeed, the transaction will be committed. Statements in the script are separated by ;.
The results of the statements that start with an uppercase letter will be returned to the client. The results of the statements that start with a lowercase letter will not be returned to the client.
You can label a statement with -- @label: label_name. The label_name will be the key of the result in the returned JSON object.
You can use ?param_name? to define a parameter. The param_name will be the key of the parameter in the JSON object sent to the server.

Inline scripts

You have the option to define a script inline in the gosqlapi.json file. This is useful when you want to define a script that is short or you don't want to create a separate file for the script. The script can be defined in the gosqlapi.json file as follows:

{
  "scripts": {
    "init": {
      "database": "test_db",
      "sql": "drop TABLE IF EXISTS TEST_TABLE; create TABLE IF NOT EXISTS TEST_TABLE( ID INTEGER NOT NULL PRIMARY KEY, NAME TEXT ); insert INTO TEST_TABLE (ID, NAME) VALUES (1, 'Alpha'); insert INTO TEST_TABLE (ID, NAME) VALUES (2, 'Beta'); insert INTO TEST_TABLE (ID, NAME) VALUES (3, 'Gamma'); -- @label: data \n SELECT * FROM TEST_TABLE WHERE ID > ?low? AND ID < ?high?;"
    }
  }
}

When both sql and path are defined, path will be used, and sql will be ignored.

Edit scripts in dev mode

When the server is running in dev mode, the server will not cache the scripts and will reload the scripts every time a request is made. This is useful when you are editing the scripts so that you don't have to restart the server every time you make a change. To run the server in dev mode, set the env environment variable to dev when starting the server:

$ env=dev gosqlapi

dev mode is only effective for scripts defined in gosqlapi.json by path. For scripts defined in gosqlapi.json by sql, dev mode will not be effective.

Do not use dev mode in production, as it will read the scripts from the disk every time a request is made.

Request Metadata in Pre-defined SQL Queries

You can access the request metadata in pre-defined SQL queries. The request metadata includes the request method, the request path, the request query string, and the request headers. The request metadata can be accessed in the pre-defined SQL queries as follows:

SELECT
!remote_addr! as "REMOTE_ADDRESS",
!host! as "HOST",
!method! as "METHOD",
!path! as "PATH",
!query! as "QUERY",
!user_agent! as "USER_AGENT",
!referer! as "REFERER",
!accept! as "ACCEPT",
!AUThorization! as "AUTHORIZATION";

The request metadata parameters are case-insensitive. The request metadata parameters are surrounded by ! characters.

Database Configuration

SQLite

{
  "databases": {
    "test_db": {
      "type": "sqlite",
      "url": "./test_db.sqlite3"
    }
  }
}

https://pkg.go.dev/modernc.org/sqlite

MySQL and MariaDB

{
  "databases": {
    "test_db": {
      "type": "mysql",
      "url": "user:pass@tcp(localhost:3306)/test_db"
    }
  }
}

https://github.com/go-sql-driver/mysql

PostgreSQL (pq or pgx)

{
  "databases": {
    "test_db": {
      "type": "pq",
      "url": "postgres://user:pass@localhost:5432/test_db?sslmode=disable"
    }
  }
}

https://github.com/lib/pq

{
  "databases": {
    "test_db": {
      "type": "pgx",
      "url": "postgres://user:pass@localhost:5432/test_db"
    }
  }
}

https://github.com/jackc/pgx

Microsoft SQL Server

{
  "databases": {
    "test_db": {
      "type": "sqlserver",
      "url": "sqlserver://user:pass@localhost:1433?database=test_db"
    }
  }
}

https://github.com/microsoft/go-mssqldb

Oracle

{
  "databases": {
    "test_db": {
      "type": "oracle",
      "url": "oracle://user:pass@localhost:1521/test_db"
    }
  }
}

Oracle Cloud TLS

{
  "databases": {
    "test_db": {
      "type": "oracle",
      "url": "oracle://user:pass@:0/?SSL VERIFY=FALSE&connStr=(description=(retry_count=20)(retry_delay=3)(address=(protocol=tcps)(port=1521)(host=host))(connect_data=(service_name=service_name))(security=(ssl_server_dn_match=yes)))"
    }
  }
}

https://github.com/sijms/go-ora

Store passwords in environment variables

If you don't want to expose the database password in the gosqlapi.json file, you can store the password in an environment variable and reference the environment variable in the gosqlapi.json file with env:. For example:

{
  "databases": {
    "test_db": {
      "type": "env:db_type",
      "url": "env:db_url"
    }
  }
}

The environment variables db_type and db_url will be used to configure the database.

$ db_type=sqlite db_url=./test_db.sqlite3 gosqlapi

HTTPS

Here is an example of how to configure HTTPS:

{
  "web": {
    "http_addr": "127.0.0.1:8080",
    "https_addr": "127.0.0.1:8443",
    "cert_file": "/path/to/cert.pem",
    "key_file": "/path/to/key.pem",
    "cors": true
  }
}

Custom HTTP Headers

You can add custom HTTP headers to the response. For example, you can add the following to the gosqlapi.json file:

{
  "web": {
    "http_addr": "127.0.0.1:8080",
    "cors": false,
    "http_headers": {
      "Access-Control-Allow-Origin": "https://example.com"
    }
  }
}

Auto start with systemd

Create service unit file /etc/systemd/system/gosqlapi.service with the following content:

[Unit]
After=network.target

[Service]
WorkingDirectory=/home/user/gosqlapi/
ExecStart=/home/user/go/bin/gosqlapi -c /home/user/gosqlapi/gosqlapi.json

[Install]
WantedBy=default.target

Enable the service:

$ sudo systemctl enable gosqlapi

Remove the service:

$ sudo systemctl disable gosqlapi

Start the service

$ sudo systemctl start gosqlapi

Stop the service

$ sudo systemctl stop gosqlapi

Check service status

$ sudo systemctl status gosqlapi

Why is mattn/go-sqlite3 removed

Because I cannot get cross compile to work for Windows ARM64. If you have a clue, please let me know. Thanks. If you are a macOS or Linux user, you can still use mattn/go-sqlite3 by:

$ git clone https://github.com/elgs/gosqlapi
$ cd gosqlapi
$ git checkout sqlite3
$ go build

If you don't particularly want to use mattn/go-sqlite3, sqlite driver from modernc.org/sqlite is used by default.

License

MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

elgs / gosqlapi

readme