Convert data files — Parquet, CSV, JSON, Excel & more
A fast, private file converter that runs entirely in your browser. Drop a Parquet, CSV, TSV, JSON, JSONL, XML or Excel file, pick a target format, and download the result — no SQL, no sign-up, and nothing uploaded. Everything runs with DuckDB-WASM, so even sensitive data never leaves your machine.
Use it to hand a stakeholder a CSV from a Parquet table, shrink a bulky CSV into columnar Parquet for your warehouse, or pull a JSON fixture out of a dataset. Need to filter or reshape first? The Explorer lets you run SQL, then export the result here.
Drop a file or click to browse
Parquet, Avro, CSV, TSV, JSON, JSONL, XML, Excel — processed locally, never uploaded
Supported formats
Read (input)
- Parquet — a columnar binary format built for analytics — compressed, strongly typed, and splittable
- Avro — a row-oriented binary format with a rich embedded schema — common in Kafka and Hadoop pipelines
- CSV — comma-separated text — the universal, human-readable format every tool can open
- TSV — tab-separated text — like CSV but tab-delimited, safer when values contain commas
- JSON — a single JSON array of records — ubiquitous in APIs and easy to eyeball
- JSONL — JSON Lines (NDJSON) — one JSON object per line, ideal for streaming and append-only logs
- Excel — Excel .xlsx workbooks — the format analysts and business stakeholders live in
- XML — hierarchical markup common in legacy systems, SOAP APIs, RSS and config — flattened by taking its repeating element as rows
Write (output)
- Parquet — a columnar binary format built for analytics — compressed, strongly typed, and splittable
- CSV — comma-separated text — the universal, human-readable format every tool can open
- TSV — tab-separated text — like CSV but tab-delimited, safer when values contain commas
- JSON — a single JSON array of records — ubiquitous in APIs and easy to eyeball
- JSONL — JSON Lines (NDJSON) — one JSON object per line, ideal for streaming and append-only logs
- Excel — Excel .xlsx workbooks — the format analysts and business stakeholders live in
ORC and Arrow/Feather reading is on the roadmap. (Avro reading fetches DuckDB’s avro extension, so it needs a network connection.)
Popular conversions
- Parquet → CSV
- Parquet → TSV
- Parquet → JSON
- Parquet → JSONL
- Parquet → Excel
- Avro → Parquet
- Avro → CSV
- Avro → TSV
- Avro → JSON
- Avro → JSONL
- Avro → Excel
- CSV → Parquet
- CSV → TSV
- CSV → JSON
- CSV → JSONL
- CSV → Excel
- TSV → Parquet
- TSV → CSV
- TSV → JSON
- TSV → JSONL
- TSV → Excel
- JSON → Parquet
- JSON → CSV
- JSON → TSV
- JSON → JSONL
- JSON → Excel
- JSONL → Parquet
- JSONL → CSV
- JSONL → TSV
- JSONL → JSON
- JSONL → Excel
- Excel → Parquet
- Excel → CSV
- Excel → TSV
- Excel → JSON
- Excel → JSONL
- XML → Parquet
- XML → CSV
- XML → TSV
- XML → JSON
- XML → JSONL
- XML → Excel
Frequently asked questions
- Are my files uploaded anywhere?
- No. Conversion runs entirely in your browser via DuckDB-WASM — the file is read, converted, and downloaded locally. Nothing is sent to a server, so it works offline and keeps sensitive data private.
- How large a file can I convert?
- It's bounded by your browser's memory rather than any upload limit — files in the hundreds of MB are routine. DuckDB runs single-threaded here, so very large files just take a little longer.
- What about ORC and Arrow?
- Today you can read Parquet, Avro, CSV, TSV, JSON, JSONL, XML and Excel, and write to Parquet, CSV, TSV, JSON, JSONL and Excel. ORC and Arrow reading are on the roadmap; when they land, conversions from them light up automatically. (ORC has no in-browser reader yet, so it may stay a self-hosted-only format.)
- Do types survive the conversion?
- Parquet carries a full schema, so converting to/from it preserves types. Text formats (CSV/TSV) are untyped, and JSON has no date/decimal types — so expect some coercion in those directions. The schema panel shows exactly what was inferred.