JSONL Query (jq-style)

updated 11 May 2026 · live query playground

100% client-side. No upload.

Query

Query:

Supported: .foo, .foo.bar, .[], .[0], .[0:5], | pipes, select(expr), map(expr), length, keys, values, type, has("k"), not, {a, b: .x} object construction, [expr] array construction, arithmetic (+ - * /), comparisons (== != < <= > >=), boolean (and or), string functions tostring, ascii_downcase, ascii_upcase, contains, startswith, endswith, plus sort_by(.k), group_by(.k), unique, min, max, add.

Drop a .jsonl file here, or

Input

Output

JSONL Query Playground

An interactive jq-style query environment for JSONL streams, running entirely in your browser — no install, no upload. Supports the subset of jq that covers the day-to-day use cases: path access, pipes, select, map, object/array construction, sort_by, group_by, and string functions. For the long tail of jq features, fall back to the jq CLI; for everything else, this is faster than spinning up a terminal. 100% in-browser.

— S., [email protected]

Five queries that cover 90% of jq usage

Pluck fields

.[] | {id, name}

For each record in the stream, emit an object with just id and name.

Filter

.[] | select(.age >= 30 and .role == "admin")

Keep records matching a condition.

Top-N by a field

[.] | sort_by(.age) | reverse | .[0:5]

Slurp into an array, sort descending by age, take the first five. (Or hit the Slurp button so you can start the query with sort_by(...) directly.)

Group and count

group_by(.category) | map({key: .[0].category, count: length})

One row per category with the count.

Project + rename

.[] | {user_id: .id, email: .user.email}

Build a flat output schema from a nested input.

Slurp vs stream

Stream mode (default): treats each input line as a separate JSON value and applies the query to each. Match this with .[] in the query when you want the "outer loop" semantics.

Slurp mode: reads the whole input into a single JSON array first, then applies the query once. Required for aggregations across the whole file — sort_by, group_by, add, etc.

Tips & common pitfalls

This isn't full jq. The supported subset is documented above. Recursion, def, variables (as $x), regex functions, foreach, and module imports are not supported — you'll get a clear "unsupported feature" error rather than silent wrong answers.
Error messages are precise. Parse errors point to the column of the offending token.
Performance. Compiled in JS, so streaming through 100k–1M small records is fine. For multi-GB files, use real jq on the CLI.