TRACER RECARTParquet Inspector
LOCAL PARQUET TOOL

View large Parquet files without uploading them.

Parquet Inspector is built for the common case where a Parquet file is too awkward for a spreadsheet, too sensitive for an upload service, or too large to render as one giant HTML table. Inspect metadata first, browse rows in a virtualized grid, and run DuckDB SQL locally in your browser.

Open Parquet Inspector

Why large Parquet files are awkward to inspect

Privacy
Company and customer datasets often should not be sent to a third-party converter or viewer.
Memory
Rendering millions of cells at once can freeze a tab even when the underlying file is efficient.
Structure
Row groups, codecs, statistics and nested types matter when diagnosing scan performance.
Workflow
Writing a one-off Python script just to inspect or filter a file is slow when you only need a quick answer.

How Parquet Inspector handles the job

Inspect metadata before reading rows

File-level metadata, schema, row groups, column chunks, compression codecs and statistics can be inspected without turning the entire dataset into a browser table.

For a focused explanation of layout and statistics, see the Parquet row group viewer guide.

Browse with virtualized rows

The workbench renders only the visible region and pages through data instead of mounting every row into the DOM.

Query locally with DuckDB-WASM

Use read-only SQL against the selected local Parquet file. This is useful for filtering, aggregation, spot checks and schema exploration without installing a desktop database tool.

Edit or export when needed

Supported values can be edited locally, then exported as CSV, JSON or Parquet. The original file is not uploaded to TRACER RECART.

Practical limits to know about

“Large file” does not mean unlimited. Browsers and WebAssembly still have memory and address-space constraints, and operations such as full exports, wide sorts, complex nested data or broad searches may require much more memory than inspecting metadata or paging through rows.

For very large multi-gigabyte datasets, remote object-store scans or native DuckDB can be a better fit. Parquet Inspector is most useful when you want a zero-install, privacy-first first look at a local file.

Common searches this page answers

large parquet viewer · open parquet file online without upload · local parquet viewer · parquet SQL browser · inspect parquet row groups · view parquet schema · private parquet viewer

Open the local viewer