S3 & URL
Checking one Parquet file in a bucket should not require aws s3 cp, a notebook, or an Athena query. viewparquet points DuckDB-WASM straight at the object: paste a public HTTPS or presigned URL, an s3:// path, or connect a private bucket with access keys.
DuckDB reads only the byte ranges a query needs — footer first, then just the relevant columns and row groups — so a quick look at a multi-gigabyte remote file transfers megabytes, not gigabytes. In the direct S3/URL connection mode, requests go from your browser to the bucket and credentials stay in browser storage. AWS S3, Cloudflare R2, Google Cloud Storage, and MinIO all work.
On the home page or inside the viewer. Public HTTPS links and presigned URLs need no credentials at all.
Paste the URL or s3:// path, or connect a private bucket with access keys or temporary STS tokens — they stay in your browser.
Run DuckDB SQL against the remote file. Only the bytes your query touches are streamed. Export results to Parquet, CSV, or JSON locally.
Yes. viewparquet uses DuckDB’s HTTP range reads: it fetches the Parquet footer first, then only the columns and row groups your query needs. Previewing or filtering a large remote file typically transfers a small fraction of its size.
In the direct S3 connection mode, access keys or temporary STS tokens stay in browser storage and requests go from the browser to S3. Presigned URLs are an alternative that requires no keys in the browser. This is separate from the optional ViewParquet Cloud-copy feature.
Any S3-compatible store: Cloudflare R2, Google Cloud Storage (via HMAC keys), and MinIO, plus plain HTTPS URLs and presigned URLs. Point the viewer at the endpoint and the same streaming reads apply.
The most common cause is CORS: browsers require the bucket to allow cross-origin GET and HEAD requests with Range headers. Add a CORS rule to the bucket (or use a presigned URL from a bucket that has one) and the file will stream. Expired presigned URLs and missing read permissions are the other usual suspects.
With the direct S3/URL connector, the data path is bucket to browser and SQL runs locally in DuckDB-WASM. This is separate from explicitly adding a private ViewParquet Cloud copy or choosing Cloud query execution.
Drop a Parquet, GeoParquet, CSV, or JSON file into viewparquet: browse rows, run DuckDB SQL, and export results locally in your browser. A file is uploaded only if you explicitly add a private Cloud copy.