How to share a JSON file as a link, with a table view

rows.page opens .json and .jsonl files in your browser with DuckDB-WASM, so nothing leaves your machine until you press Share.

1. Drop your .json file onto rows.page, pick it from your computer, or click Paste data and paste the JSON.
2. Filter, sort or hide columns if you want the link to open on a specific view.
3. Click Share (Upload and get a link). rows.page uploads the whole file and opens its https://rows.page/<10-character id> link, with your current view kept in the URL.

A link shared from the browser expires after 7 days, even when you are signed in, unless you click Keep it on the page that opens. A JSON file over 16 MB may be stored as Parquet when that makes it at least 20% smaller, and its raw URL then ends in .parquet.

From a terminal or a script, push the file with one command:

curl -T data.json https://rows.page

curl prints the public link, an owner link with a #t= token, a summary line with rows, columns, format, size and expiry date, and the command for the next version. Add -H "Accept: application/json" to get the same reply as JSON.

Recipients get a table and need no account. Nested objects become dot-path columns such as user.address.city, up to 8 levels deep and 200 child fields per object, and arrays stay list columns. A single top-level object such as {"data": [...]} is unwrapped to its biggest list, and clicking a row opens it as JSON.

• Filter with country:NL, price:>20 or -name:* (empty values).
• Query with DuckDB SQL on the data table, quoting dot-path columns like "user.city".
• Export the view as CSV, Parquet, JSON, JSONL or Markdown (first 1,000 rows).
• Click Copy link to share their exact filter, sort and visible columns.

There is no whole-file tree view and no cell editing.

Every public dataset also serves the file itself, for apps, notebooks and scripts:

• https://rows.page/<id>/raw.json or https://rows.page/<id>/raw for the latest version, served with no-cache.
• https://rows.page/<id>/v<n>/raw.json for a pinned version, cacheable for a year.

Responses send access-control-allow-origin: *, so browser code on any site can fetch() the JSON. The extension must match the dataset's format, and a wrong one returns a 404 that names the right URL. Raw URLs of a private dataset open only for its owner.

Without an account, a push can be up to 100 MB, you get 50 pushes and 1 GB per IP address per UTC day, and each link holds up to 20 versions and expires after 7 days. To keep it, open the owner link, sign in with Google and click Keep on my account, and the dataset stops expiring.

• Free ($0): 10 hosted datasets, files up to 100 MB, the last 3 versions visible.
• Pro ($12 per month): 100 datasets, files up to 500 MB, 90 days of versions, private datasets.
• Max ($39 per month): 1,000 datasets, files up to 2 GB, 1 year of versions, private datasets.

A single request takes up to 100 MB on every plan, so Pro and Max reach their bigger limits through multipart uploads (POST /api/v1/uploads). You make a dataset private with the visibility=private push parameter or the REST API.

Push to the same <id> with the rpt_ owner token from your first push. The link keeps working while versions count up as v1, v2, v3:

curl -T data.json -H "Authorization: Bearer rpt_..." https://rows.page/<id>

For scripts and CI, any signed-in account, Free included, can create up to 20 rpk_ API keys. Push with -H "Authorization: Bearer rpk_..." to https://rows.page/?key=<name> and every run updates the same dataset.

When the previous version is 64 MB or less, a banner shows "+N new, N gone, N changed vs vN". Rows are matched by a key column set with ?id_column=<column>, the X-Rows-Id-Column header or in the banner. Without a unique key, changes show as new and gone.

GitHub Gist

A gist is public or secret, and a secret gist is unlisted, not private: anyone with the URL can open it. Raw files are served from gist.githubusercontent.com with access-control-allow-origin: *. It suits small JSON files that people read as code.

rows.page

rows.page suits data people need to filter, sort or query as a table, with row-level diffs between versions and private links on Pro and Max. If your JSON already sits in a public gist, put its raw URL after https://rows.page/ to view it as a table, and nothing is stored.