Open a CSV from a URL in your browser, without downloading it

1. Copy the direct link to the file, not to a web page that shows it.
2. Type https://rows.page/ in front of it in your browser's address bar. Leave the link as it is, query string included; you don't need to encode it.
3. Press Enter. The table opens, and you can filter, sort or run SQL against data.

https://rows.page/https://example.com/data.csv

Many share links open a web page instead of the file. These patterns point to the file itself:

Your browser first downloads the file straight from the host. That works when the host allows cross-origin reads, as raw.githubusercontent.com does with access-control-allow-origin: *. The whole file downloads into your tab, DuckDB-WASM reads it there, and rows.page keeps no copy.

When the host blocks cross-origin requests, rows.page fetches the file through its proxy instead. The proxy has fixed limits:

• 100 MB per file.
• 30 seconds per request.
• 5 redirects.
• No private, local or rows.page addresses, such as localhost, 192.168.x.x or .internal hosts.

rows.page picks the format from the extension in the URL and otherwise from the bytes, ignoring the Content-Type header. That's why a Google Sheets export?format=csv link opens as CSV without a .csv ending.

Once downloaded, a remote file follows the same rules as one you drop on the page: up to 1.5 GB per tab, with bigger CSV, TSV and JSONL files loading their first rows and a bigger .json file refused with "Too big for this browser". Parquet skips the 1.5 GB check, but a remote Parquet file still downloads in full.

• You see rows of HTML: the link points to a web page, such as a GitHub /blob/ page. Click Raw on GitHub, or use the raw.githubusercontent.com pattern above.
• Google Sheets shows HTML or an error instead of rows: the sheet isn't shared publicly. Set link sharing to Anyone with the link and use the export?format=csv URL.
• "The remote file is over 100 MB." or "The remote server took longer than 30 seconds.": the host blocks cross-origin requests, so the proxy limits apply. Download the file and drop it on rows.page, or add access-control-allow-origin: * on your bucket so your browser can fetch it directly.
• A 403 or 404 answer: the link has expired, needs a login or is wrong. Presigned S3 links answer 403 once they expire.
• "Couldn't open this data" on a CSV: the file is probably saved in a Windows code page, and characters like the euro sign or curly quotes break the read. Re-save it as CSV UTF-8.
• Excel .xlsx, XML and zip files don't open at all. Export the sheet to CSV first, or use the Google Sheets export link.

The grid has no row limit and loads 100 rows at a time as you scroll. Type filters like country:NL, price:>20 or name:* in the search box, or SQL that starts with SELECT, WITH or FROM against the table data:

SELECT country, count(*), avg(price)
FROM data
WHERE price > 20
GROUP BY country
ORDER BY 2 DESC

SQL results stop at 1,000,000 rows, while a filtered view keeps every matching row. Export either one as CSV, Parquet (zstd), JSON, JSONL or Markdown (first 1,000 rows).

Your filter or SQL, sort and visible columns are kept in the part of the address after #. Copy the address bar to share that exact view. The other person's browser downloads the file from the original URL again, so a presigned link stops working once its signature expires.

For a link that doesn't depend on the original server, click Save to rows.page, which keeps a copy with its own link, versions and diffs. From a terminal, push the file instead:

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

Without an account, a push can be up to 100 MB, links expire after 7 days, and each IP address gets 50 pushes and 1 GB per UTC day. A free account keeps up to 10 datasets, and Pro ($12 per month) and Max ($39 per month) add private datasets and files up to 500 MB and 2 GB through multipart uploads. Every hosted dataset also gets a raw file URL like https://rows.page/<id>/raw.csv that other sites and scripts can read.