JSON to MySQL Schema Transformation — Numio

When to use it

• Sketching a normalized table layout for an API payload before writing a real migration.
• Seeing which nested objects and arrays in a document would need their own tables when moving data from a document store to MySQL.

How to use

1. Paste a JSON object, or an array of sample records that are merged into one table, into the left editor, or use the load-file button to open a UTF-8 file.
2. Read the generated tables: the root is generic, children are generic_<key>, linked by generic_id foreign keys.
3. Copy the SQL, then rename tables and adjust column types (for example DECIMAL for prices or VARCHAR lengths) before running it.

{
  "id": 1024,
  "total": 19.5,
  "coupon": null,
  "created_at": "2026-01-05T10:00:00Z",
  "scores": [4, 4.5],
  "items": [{ "sku": "A1", "qty": 2 }]
}

CREATE TABLE generic (
  id INT,
  total DOUBLE,
  coupon TEXT,
  created_at TIMESTAMP,
  PRIMARY KEY (id)
);

CREATE TABLE generic_scores (
  generic_id INT,
  id INT,
  value DOUBLE,
  PRIMARY KEY (id),
  FOREIGN KEY (generic_id) REFERENCES generic(id)
);

CREATE TABLE generic_items (
  generic_id INT,
  id INT,
  sku TEXT,
  qty INT,
  PRIMARY KEY (id),
  FOREIGN KEY (generic_id) REFERENCES generic(id)
);

Details and limitations

Integers become INT (BIGINT outside the 32-bit range), other numbers DOUBLE, strings TEXT and booleans BOOLEAN. No VARCHAR lengths, NOT NULL, AUTO_INCREMENT or indexes are generated.

The primary key is a key named id or containing _id; if none exists an id column is added. Strings longer than 18 characters that parse as dates become TIMESTAMP.

null values become TEXT columns. Arrays of scalars get a child table with a value column typed from all elements (scores above is DOUBLE).

Runs in your browser; the output is not validated against a MySQL server.

FAQ

Related tools

Specifications and references

• MySQL CREATE TABLE statement