How to open a JSON file and read it as a table

Every app below opens a .json file. The difference is whether you read raw text, a collapsible tree or a table of records.

1. Go to rows.page and drop your .json, .jsonl, .ndjson or .ldjson file on the page, or choose it from your disk.
2. To open JSON from your clipboard instead, click "Paste data" and paste the text.
3. Scroll the grid (rows load in pages of 100, with no row limit), filter columns, or click a row to see its original JSON in a side drawer.

DuckDB-WASM reads the file inside your browser tab, so it stays on your machine unless you press Share. A .json file over 1.5 GB is refused with "Too big for this browser", while a bigger JSONL file loads only its first rows. You also cannot edit cells, join two files or see the whole file as one collapsible tree.

JSON is plain text, so Notepad and TextEdit open any .json file. If the file was saved on a single line to save space, you get one long wall of text that is hard to read.

VS Code is free (MIT-licensed source) and does better: run Format Document to add indentation and line breaks, and it validates the file against JSON schemas. It still has no table view, so comparing values across hundreds of records means scrolling through braces and repeated keys.

Firefox has a built-in JSON viewer that opens when you load a .json file from disk or a JSON URL. It shows a collapsible tree with a filter box, and a Raw Data view with a Pretty Print button.

Chrome has no built-in JSON viewer. For a collapsible tree, install an extension such as JSON Formatter (open source, BSD-3-Clause). Tree viewers work well for config files, but with a list of records you still open one object at a time.

jq 1.8.2 (MIT) indents and colors a single-line JSON file in your shell. Pass the . filter and the file name:

jq . file.json

To go from the terminal to a browser table, push the file to rows.page with curl. Without an account the upload can be up to 100 MB, and the link it prints expires after 7 days.

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

• A single top-level wrapper such as {"data": [...]} is unwrapped to its biggest list (up to 4 levels deep), with the note: Showing the N items in "data"; the other top-level fields were skipped.
• Nested objects flatten into dot-path columns such as user.address.city, up to 8 levels deep and 200 child fields per object.
• Arrays stay as list columns, and the grid shows the first 6 items followed by +N.
• A column that holds JSON as text is expanded when 90% or more of its values in the first 1,000 rows are JSON objects.
• Any row opens in a side drawer with a JSON/Fields toggle and a "Copy as JSON" button.

Filter the table with expressions like country:NL, price:>20, name:* (has a value) or -name:* (is empty), or click a histogram bar in the Columns panel. That panel shows each column's type, distinct count, share of empty values, min and max for numbers and dates, and top 10 values.

For more, type a DuckDB SQL statement that starts with SELECT, WITH or FROM into the same box; the table is called data. Put double quotes around column names that contain dots, and use FROM (SUMMARIZE data) for averages and quartiles. Results show up to 1,000,000 rows.

SELECT "user.address.city", count(*), avg(price)
FROM data
WHERE price > 20
GROUP BY 1

The Export menu saves the current view, including your filter, sort, hidden columns or SQL result:

• CSV: comma-delimited, UTF-8 without a BOM, with a header row.
• Parquet: compressed with zstd.
• JSON: one array of objects; flattened fields use dot-path keys and lists stay real arrays.
• JSONL: one object per line.
• Markdown: a table of the first 1,000 rows.

Filtered views export every matching row, while SQL results stop at 1,000,000 rows. Share ("Upload and get a link") uploads the file and gives you a https://rows.page/<10-character id> link that expires after 7 days without an account. Sign in with Google and choose "Keep on my account" to keep it: Free ($0) holds 10 datasets, Pro ($12 per month) 100 and Max ($39 per month) 1,000.