Interactive table viewer for parquet/csv/jsonl etc

1. Start in terminal#

sense table /path/to/file.csv          # Open table viewer with given file

or open file on S3:

sense table s3://smoosense-demo/datasets/COCO2017/images-emb-2d.parquet  # Open S3 file

2. File information for columnar formats#

2.1 Parquet row group and compression#

When you care about I/O efficiency, query performance or want to optimize data processing, you need to pay attention to parquet row groups. Each group stores a chunk of rows that share the same schema, and within it, each column is stored as a contiguous block.

With one click on Info icon in the top right corner, SmooSense can display parquet row group metadata, compression and writer version etc.

Parquet info with mostly numeric data

Parquet info with mostly string data

2.2 Lance indices and versions#

Lancedb supports data mutation, schema evolution, indexing and version controls. You can conveniently see these by clicking the Info icon in the top right corner.

Display indcies in lance table

Display versions along with differences in lance table

3. Gallery#

Gallery provides a convenient way to browse visual data, including images, videos, and domain-specific visualization.

4. Freeform query#

When you need flexibility, SmooSense allows you to write your own SQL query using duckdb and renders the result intelligently.

5. Editing and quick labeling#

Coming