Why this tool is useful
Check the schema and sample values in an analytics, machine-learning or data-lake export before opening it in a heavier desktop environment.
Local File Tools · runs locally
Private Parquet File Viewer
Inspect a bounded Apache Parquet file locally, including its row count, top-level columns and a 100-row preview, without uploading the dataset.
Ready. Your working data stays in this browser.
Files and capture streams stay on this device. Processing uses browser APIs only; nothing is uploaded to YTSave, the media API or a proxy.
How it works
The browser verifies both Parquet magic markers, reads metadata with pinned Hyparquet 1.29.2, limits the selection to 50 top-level columns and decodes only the first 100 rows. The selected file never leaves the tab.
Practical example
Choose a local .parquet report to confirm its total rows, visible column names and representative values before handing it to an ETL or notebook workflow.
Validation
The calculation is deterministic and explains validation errors instead of silently changing invalid input.
Privacy boundary
Input and output remain in this tab. Copy and download happen through browser APIs without a server upload.
Result passport
Every run reports its tool mode, timestamp and input/output size so transformed data can be audited.