Convert your Parquet files to SQL INSERT statements for PostgreSQL, MySQL, SQLite, and other databases. Files are processed locally in your browser.
How the Parquet to SQL conversion works
Drop a .parquet file, pick a dialect — PostgreSQL, MySQL, SQLite, or generic ANSI — and download a ready-to-run .sql script: a typed CREATE TABLE (toggleable) followed by INSERTs batched 1,000 rows per statement. Identifiers are quoted per dialect (double quotes vs backticks), strings escape single quotes by doubling, and MySQL additionally escapes backslashes. Everything is generated locally in your browser.
Column types per dialect
BOOLEAN → BOOLEAN / TINYINT(1) / INTEGER (sqlite)
INT32 → INTEGER
INT64 → BIGINT (INTEGER on sqlite)
DOUBLE → DOUBLE PRECISION / DOUBLE / REAL
TIMESTAMP → TIMESTAMP / DATETIME(3) / TEXT
STRING → TEXT
Timestamp literals are ISO 8601; MySQL gets the DATETIME-friendly form automatically. Booleans become TRUE/FALSE or 1/0 as the dialect requires.
Large files and alternatives
Reads cap at the first 100,000 rows. At bigger scale, skip INSERT scripts: DuckDB reads Parquet natively (CREATE TABLE t AS SELECT * FROM 'file.parquet') and pandas to_sql() loads any SQLAlchemy target.