JSON Schema SQL Builder

This is a simple tool to generate SQL DDL statements from a JSON Schema.

Install

$ composer require aaron-lin/json-schema-sql-builder

Usage

Build SELECT Statement from JSON Schema

For example, the JSON schema is as following:

{
  "$schema": "http://json-schema.org/draft-06/schema#",
  "@table": "products",
  "@id": "id",
  "type": "object",
  "properties": {
    "id": {
      "type": "string"
    },
    "name": {
      "type": "string"
    },
    "_type_name": {
      "type": "string"
    },
    "_weight": {
      "type": "object",
      "properties": {
        "weight": {
          "type": "number"
        },
        "weight_unit": {
          "type": "string",
          "enum": [
            "g",
            "kg"
          ]
        }
      }
    },
    "_bids": {
      "type": "array",
      "@table": "bids",
      "@joinId": "product_id",
      "@id": "id",
      "@orderBy": "time DESC",
      "items": {
        "type": "object",
        "properties": {
          "id": {
            "type": "string"
          },
          "price": {
            "type": "number"
          },
          "time": {
            "type": "string",
            "format": "date-time"
          }
        }
      }
    }
  }
}

Use the following code to run the SQL SELECT query:

use Lin\JsonSchemaSqlBuilder\Storage;
use Lin\JsonSchemaSqlBuilder\SelectSQLBuilder;

$SchemaURI = 'path/to/schema.json#';
$DSN = 'mysql:host=db;dbname=test;charset=utf8mb4';
$DB = new \PDO($DSN, 'test', 'test');

try {
  Storage::SetSchemaFromURI($SchemaURI);
} catch (\Exception $e) {
  echo $e->getMessage();
  exit;
}
Storage::AddSelectExpression($SchemaURI . '#/properties/_type_name', '(SELECT name FROM product_types WHERE product_types.id = products.type_id LIMIT 1)');
Storage::AddSelectExpression($SchemaURI . '#/properties/_weight/properties/weight', 'products.weight');
Storage::AddSelectExpression($SchemaURI . '#/properties/_weight/properties/weight_unit', 'products.weight_unit');
$Builder = new SelectSQLBuilder($SchemaURI, $DB);
$Builder->SetSelectExpressions()
  ->AddWhere("products.keywords like :keywords", ['keywords' => '%apple%'])
  ->AddOrderBy('products.price', 'DESC')
  ->SetLimit(10)
  ->SetOffset(0);
$Result = $Builder->Execute();
echo json_encode($Result, JSON_PRETTY_PRINT);
// [
//   {
//     "id": "1",
//     "name": "Apple",
//     "_type_name": "Fruit",
//     "_weight": {
//       "weight": "100.00",
//       "weight_unit": "g"
//     },
//     "_bids": [
//       {
//         "id": "7",
//         "price": "400",
//         "time": "2018-01-04 00:00:00"
//       },
//       {
//         "id": "5",
//         "price": "300",
//         "time": "2018-01-03 00:00:00"
//       },
//       {
//         "id": "3",
//         "price": "200",
//         "time": "2018-01-02 00:00:00"
//       },
//       {
//         "id": "1",
//         "price": "100",
//         "time": "2018-01-01 00:00:00"
//       }
//     ]
//   }
// ]

As you can see, after initializing JSON schema storage, you can add select expressions to indirect corresponding properties (direct properties are automatically added to select expressions as table_name.property_key). Then you can build the SQL SELECT statement with SelectSQLBuilder::Build method. For more information, please refer to the test script, you can also find database schema and data in the test directory.

Build INSERT/UPDATE Statement from JSON Schema

For example, the JSON schema is as following:

{
  "$schema": "http://json-schema.org/draft-06/schema#",
  "@table": "products",
  "@id": "id",
  "type": "object",
  "properties": {
    "id": {
      "type": "string"
    },
    "name": {
      "type": "string"
    },
    "_type_name": {
      "type": "string",
      "readonly": true
    },
    "_weight": {
      "type": "object",
      "properties": {
        "weight": {
          "type": "number"
        },
        "weight_unit": {
          "type": "string",
          "enum": [
            "g",
            "kg"
          ]
        }
      }
    },
    "_bids": {
      "type": "array",
      "@table": "bids",
      "@joinId": "product_id",
      "@id": "id",
      "@orderBy": "time DESC",
      "items": {
        "type": "object",
        "properties": {
          "id": {
            "type": "string"
          },
          "price": {
            "type": "number"
          },
          "time": {
            "type": "string",
            "format": "date-time"
          }
        }
      }
    }
  }
}

And the data is as following:

{
  "id": "1",
  "name": "Apple",
  "_type_name": "Fruit",
  "_weight": {
    "weight": "100.00",
    "weight_unit": "g"
  },
  "_bids": [
    {
      "id": "7",
      "price": "400",
      "time": "2018-01-04 00:00:00"
    },
    {
      "id": "5",
      "price": "300",
      "time": "2018-01-03 00:00:00"
    },
    {
      "id": "3",
      "price": "200",
      "time": "2018-01-02 00:00:00"
    },
    {
      "id": "1",
      "price": "100",
      "time": "2018-01-01 00:00:00"
    }
  ]
}

Use the following code to run the SQL INSERT/UPDATE query:

use Lin\JsonSchemaSqlBuilder\Storage;
use Lin\JsonSchemaSqlBuilder\UpsertSQLBuilder;

$SchemaURI = __DIR__ . '/schema.json#';
$DSN = 'mysql:host=db;dbname=test;charset=utf8mb4';
$DB = new \PDO($DSN, 'test', 'test');

try {
  Storage::SetSchemaFromURI($SchemaURI);
} catch (\Exception $e) {
    echo $e->getMessage();
  exit;
}

Storage::AddSelectExpression($SchemaURI . '#/properties/_weight/properties/weight', 'products.weight');
Storage::AddSelectExpression($SchemaURI . '#/properties/_weight/properties/weight_unit', 'products.weight_unit');
$Builder = new UpsertSQLBuilder($SchemaURI, $DB, $Data);
$Builder->SetAssignmentList();
$ResultCount = $Builder->Execute();
echo $ResultCount;
// for every new rows added, the return value is 1,
// for every existing rows updated, the return value is 2.
// data contains 1 row in products, 4 rows in bids,
// therefore, if data are all new rows, the return value is 5,
// or data are all existing rows, the return value is 10

For every new rows added, the return value of UpsertSQLBuilder::Execute is 1, for every existing rows updated, the return value is 2. For more information, please refer to the test script, you can also find database schema and data in the test directory.