Convert JSON to Parquet Online — Flatten and Type in the Browser

JSON is easy to produce and painful to analyze. Parquet is the opposite. PondPilot converts JSON to Parquet in the browser — flattening nested structures and locking in types along the way.

How to Convert

  1. Open app.pondpilot.io
  2. Drop in your .json or .ndjson file
  3. Flatten and type-check with SQL
  4. Export as Parquet

The conversion runs entirely in your browser via DuckDB WebAssembly. Your JSON file stays local.

Schema Inference

DuckDB scans your JSON and infers a schema — column names, types, and nested structure. For large files it samples; for small ones it reads everything. You can inspect the inferred schema before exporting:

DESCRIBE SELECT * FROM read_json_auto('events.json');

If the inference got something wrong (numbers parsed as strings, dates as text), you can fix it before export.

Flatten Nested JSON

API responses and log exports are usually deeply nested. Parquet prefers flat, typed columns. SQL bridges the gap:

SELECT
  id,
  created_at::TIMESTAMP AS created_at,
  user->>'id' AS user_id,
  user->>'country' AS country,
  UNNEST(events) AS event
FROM read_json_auto('payload.ndjson')

Extract the fields you care about, unnest arrays, and write out a flat Parquet file ready for analytics.

Why Parquet?

  • Typed — integers stay integers, timestamps stay timestamps
  • Small — columnar compression typically beats gzipped JSON by a wide margin
  • Fast — analytical engines read columns, not rows

NDJSON Works Too

Newline-delimited JSON is common in logs and streaming exports. PondPilot handles both regular JSON arrays and NDJSON transparently.

Privacy

Log exports and API dumps are often sensitive. PondPilot runs in your browser — the file never leaves your machine.

Get Started

Visit app.pondpilot.io and convert your JSON to Parquet.

Back to home