Parquet to XLSX creates a workbook for review, but it necessarily loses Parquet's columnar storage metadata.
From real runs
| Element | Result | What happens |
|---|---|---|
| Worksheet | Changed | The generated workbook contained one sheet named parquet, taken from the parsed table name. |
| Headers | Kept | The first row contained the Parquet field names id, created_at, active, amount, note, and optional. |
| Numbers | Kept | Numeric Parquet values appeared as workbook cell values, with 101 and 1234.5 visible in the first data row. |
| Booleans | Changed | Boolean values were visible as TRUE and FALSE when the output workbook was inspected through our table engine. |
| Nulls | Changed | Nullable Parquet values became blank worksheet cells, so the first row's optional cell was empty. |
| Timestamps | Changed | Timestamp values appeared as ISO strings in cells rather than Excel date serial formatting. |
| Parquet internals | Lost | Compression, physical schema, row groups, and column encodings were not transferred into the XLSX package. |
The generated workbook contained one sheet named parquet, taken from the parsed table name.
The first row contained the Parquet field names id, created_at, active, amount, note, and optional.
Numeric Parquet values appeared as workbook cell values, with 101 and 1234.5 visible in the first data row.
Boolean values were visible as TRUE and FALSE when the output workbook was inspected through our table engine.
Nullable Parquet values became blank worksheet cells, so the first row's optional cell was empty.
Timestamp values appeared as ISO strings in cells rather than Excel date serial formatting.
Compression, physical schema, row groups, and column encodings were not transferred into the XLSX package.
parquet and rows like ["101","2026-09-10T18:30:00.000Z","TRUE","1234.5","alpha, quoted"] when inspected.nullable, matching the source null semantics.We converted our Apache Parquet sample (816 B) with Table engine on 2026-10-20. It took 2 ms and produced a 17 KB Excel XLSX file, 21× larger than the input. The engine raised no warnings.
Input · sample.parquet · 816 B
Parquet file (PAR1 magic present)
Output · sample.xlsx · 17 KB
Workbook with 1 sheet (parquet); first sheet 5 rows, 25 cells, 0 formulas, header filter, fitted column widths
Python pandas
python - <<'PY'
import pandas as pd
pd.read_parquet("in.parquet").to_excel("out.xlsx", index=False, sheet_name="parquet")
PYInstall pyarrow or fastparquet, and watch Excel row limits for large Parquet datasets.
The HTTP API and the MCP server run the exact engine this page uses (Table engine); the response carries the same warnings the web converter shows. Check support first with GET /api/v1/pairs?from=parquet&to=xlsx.
curl -X POST https://filetypeconverters.com/api/v1/convert \
-H "Authorization: Bearer $FTC_API_KEY" \
-F "file=@input.parquet" -F "from=parquet" -F "to=xlsx" \
-F 'options={"encoding":"utf-8"}' \
--output output.xlsx| Key | Applies to | Type | Default |
|---|---|---|---|
| encoding | Reading Apache Parquet | utf-8 · latin1 | utf-8 |
| sheetName | Writing Excel XLSX | string |
/convert/parquet-to-csv when the receiving system imports delimited text more reliably than XLSX.No. The converter maps values into worksheet cells. Complex Parquet metadata and some specialized logical types are outside XLSX's table model.
Nulls become blank worksheet cells. In the verified workbook, optional was blank on the first row and amount was blank on the second row.
No. The output is one worksheet for the parsed table. Row group boundaries are storage details and are not shown as sheet divisions.