Format SQL for Postgres, MySQL, or BigQuery.

A one-liner SELECT with 5 joins and a CTE is basically unreadable. Dialect-aware formatting knows Postgres's ::type casts, BigQuery's backtick identifiers, and MySQL's backtick-vs-quote rules — a generic formatter mangles them.

Step-by-step

1. Open /sqlfmt/. Paste your SQL into the input.
2. Pick a dialect. Standard SQL, Postgres, MySQL, MariaDB, SQLite, BigQuery, Snowflake, Redshift, Spark, DB2, T-SQL, PL/SQL, N1QL, Hive, Trino — 15 total. This isn't cosmetic; it's about which reserved words are recognized and how identifiers are quoted.
3. Set keyword case. UPPER (most conventions), lower, or Preserve if your team style is already set.
4. Set indent. 2 or 4 spaces, or tab. Match your team's SQL style guide.
5. Click Format. Copy the output.

Why dialect matters

Generic SQL formatters treat every input as ANSI SQL. Real queries aren't ANSI:

• Postgres: ::type casts, DISTINCT ON, RETURNING, array types, WITH RECURSIVE.
• BigQuery: backtick-quoted identifiers, ARRAY_AGG, STRUCT, EXCEPT for column exclusion.
• Snowflake: QUALIFY, table sampling, MERGE semantics.
• T-SQL: bracketed identifiers [dbo].[table], TOP N instead of LIMIT.

Formatting the wrong dialect can silently break the query — a Postgres ::text cast wrapped across a line-break in the wrong place becomes invalid. Pick the dialect.

Example

Before

select id,name,coalesce(email,'-') as email from users u left join orders o on u.id=o.user_id where u.created_at > now() - interval '7 days' group by 1,2,3 order by u.id desc limit 100;

After (Postgres, UPPER, 4-space)

SELECT
    id,
    name,
    COALESCE(email, '-') AS email
FROM
    users u
    LEFT JOIN orders o ON u.id = o.user_id
WHERE
    u.created_at > NOW() - INTERVAL '7 days'
GROUP BY
    1, 2, 3
ORDER BY
    u.id DESC
LIMIT
    100;

Nothing uploaded

The formatter engine (sql-formatter, MIT) runs entirely in your browser. Your queries — including any embedded secrets, table names, or business logic — never leave the tab.