Convert Apache Arrow and Feather files to readable JSON locally - each row as an object, nested fields preserved, no upload.
Running locally on your device ...
0%
Your files never left your device
No. Everything runs in your browser - your file never leaves your device. How this is verifiable
Apache Arrow is the widely used columnar in-memory standard for analytical data; the Arrow IPC file format (also Feather v2) keeps such tables on disk and is used by pandas, Polars, R and many data tools. The file is compact, but not readable in an editor. This tool decodes the content and outputs a clear JSON array: one object per data row, with the column names as keys. Unlike CSV, nested fields stay as a real JSON structure.
Everything runs entirely locally in your browser (flechette, pure JavaScript, no upload, even offline). The tool reads the Arrow schema and data blocks and builds an object for each row. For clean values we emit 64-bit integers beyond the safe JavaScript range as a string, timestamps as ISO 8601 text and raw binary values as base64, so no precision is lost. Lists and structs stay as nested JSON.
You need this when pandas, Polars, R or a data tool hands you a .arrow or .feather file and you want to view it, feed it into a tool that speaks JSON (an API, a script, a test), or keep the nested structure. Note: the Arrow IPC file format (Feather v2) is supported; the older Feather v1 is not read. The conversion is one-way (Arrow to JSON).
| Input formats | ARROW, FEATHER, IPC |
|---|---|
| Output format | JSON |
| Batch processing | Yes |
| Processing | Locally in your browser (WebAssembly) |
| File upload | None |
Limitations: The Apache Arrow IPC file format (also Feather v2) is supported; the older Feather v1 is not read. Large 64-bit integers appear as a string (precision), timestamps as ISO 8601, binary values as base64. The conversion is one-way (no JSON back to Arrow). Very large files need memory accordingly.
The Arrow IPC file format (Feather v2) stores a columnar table on disk; it is used by pandas, Polars and R.
No. The conversion happens entirely locally in the browser - even offline.
JSON keeps nested fields as a structure and suits code and APIs; CSV is a flat table for spreadsheets.
64-bit integers beyond the safe JS range are output as a string, timestamps as ISO 8601, binary values as base64.
Not here - this tool is one-way (Arrow to JSON) for inspecting and reusing the data.