JSON to SQL Table & INSERT Converter
Free online JSON to SQL converter. Infers column types and generates CREATE TABLE plus escaped INSERT statements for PostgreSQL, MySQL and SQLite. No upload.
To convert JSON to SQL, paste an array of objects on the left. Every key is scanned across all rows to infer a column type such as INTEGER, BOOLEAN, TIMESTAMP or VARCHAR, and the right pane returns a CREATE TABLE statement followed by escaped INSERT rows.
Converting JSON to SQL means two separate jobs: inferring a table definition, and writing the rows. JSON2X scans every object in your array so the inferred column types cover all the data rather than just the first record, emits a CREATE TABLE statement, and follows it with properly escaped INSERT rows for PostgreSQL, MySQL, SQLite, or SQL Server.
How are column types inferred?
By scanning every row, not just the first. Values that are always integers become INTEGER, values with a fractional part become a numeric or double type, booleans become BOOLEAN (or TINYINT(1) in MySQL), and strings matching an ISO 8601 pattern become TIMESTAMP. Everything else becomes TEXT or VARCHAR. A key that holds a number in some rows and a string in others widens to text, which is deliberate — it is the only choice that will not reject data on insert.
What happens to nested objects and arrays?
They have no scalar column type, so they are serialised as JSON text in a single column. On PostgreSQL that column is a good candidate to change to jsonb, which gives you indexing and operators over the nested data; MySQL 5.7+ and SQLite have comparable JSON support. If the nested array is really a relationship, the right answer is a second table with a foreign key — in which case Prisma or Drizzle output will model it better than flat DDL.
Is the generated SQL safe to run as-is?
String literals are escaped, so the script itself is not an injection vector for the data you pasted. But treat the DDL as a draft. Inferred types carry no length limits, no NOT NULL constraints, no indexes, and no primary key beyond an obvious id; a column inferred as INTEGER from a thousand-row sample may need BIGINT in production. Review it, then run it against a scratch database before it goes anywhere that matters.
How to use the JSON to SQL
Takes under a minute. Nothing is uploaded — every step runs in this browser tab.
- Paste a JSON array
Put an array of uniform objects into the left-hand input pane — one object per intended row.
- Pick a dialect and table name
Choose PostgreSQL, MySQL, SQLite, or SQL Server so identifier quoting and type names match, and name the target table.
- Review the DDL and inserts
The right-hand pane shows the CREATE TABLE statement with inferred column types, followed by INSERT statements with string literals escaped.
- Run it against a scratch database
Copy the script and run it somewhere disposable first — inferred types are a starting point, not a schema review.
When to use it
- Loading an API export into Postgres for ad-hoc analysis
- Seeding a development or test database from a JSON fixture
- Drafting a first-pass table definition from a sample payload
- Migrating documents out of a NoSQL store into a relational one
- Producing a reproducible SQL script from data a colleague sent as JSON
Step-by-step guides for this tool
Task-specific walkthroughs covering the most common ways this tool gets used.
Searches this page answers
- json to sql online free
- free json to sql converter
- json to sql converter online
- json to sql insert statements
- json to create table
- json to postgresql converter
- json to mysql insert
- json to sqlite converter
- convert json array to sql
Frequently Asked Questions
How does JSON to SQL conversion work?
The engine inspects all objects in your JSON array to infer column data types (VARCHAR, INTEGER, BOOLEAN, TIMESTAMP) and generates both CREATE TABLE DDL and INSERT statements.
Which SQL dialects are supported?
PostgreSQL, MySQL and SQLite. Pick the dialect in the toolbar and the generated types and quoting follow its conventions; the output is close enough to standard SQL to adapt for other engines by hand.
Can I set the table name for the generated SQL?
Yes. Type it in the table name field and both the CREATE TABLE statement and every INSERT use it, so you can paste the script into a migration without editing.
Are string values escaped safely in the INSERT statements?
Single quotes in string values are escaped so the generated INSERTs parse. Treat the output as a migration or seed script you review, not as a substitute for parameterised queries in application code.
How are nested objects and arrays handled?
SQL columns are flat, so a nested object or array has no direct column type. Flatten or extract those fields first — the JSONPath Tester is useful for pulling out the slice you want — then convert the flat array of objects.
Related Developer Tools
Last reviewed by the JSON2X Engineering Team.