Inspect a Parquet schema
Read a Parquet file's footer metadata: every top-level column's physical type, logical type, and compression codec, plus row count, row-group count, and the writer named in created-by. Runs entirely on your device.
Inspect a Parquet schema opens a file's footer rather than any row data. It lists each column's physical type, a logical type like DECIMAL or TIMESTAMP when present, and its compression codec. Total rows and the row group tally round out the summary, alongside the writer name from created-by. The first row group's chunks alone decide the codec shown per column.
Questions
What does it show per column?
Its physical type, its logical type such as DECIMAL, TIME or TIMESTAMP when present, and the compression codec used in the file's first row group.
Where does the codec come from?
The first row group's column chunks only; a later group using a different codec is not reflected.
What else does the summary line report?
The total row count, the row group count, and the writer named in the file's created-by field.
What happens with an unreadable file?
It is refused with the underlying parquetMetadata parse error attached.
Does it read the actual row data?
No, only the footer metadata and schema; no row values are decoded.
Compare
| Tool | Runs on the CLI | Live as you type |
|---|---|---|
| Inspect a Parquet schema | Yes | No |
| CSV to Parquet | Yes | No |
| Parquet to CSV | Yes | No |