jq, yq, dasel — Structured Data

Three tools, one shape

Each lets you query and transform structured data with a small expression language:

• jq — JSON. The original, most widely deployed.
• yq — YAML, with a jq-compatible syntax. By Mike Farah's go-yq (different from the older python-yq).
• dasel — JSON, YAML, TOML, XML, CSV in one tool with a single jq-like syntax. Newest, broadest format support.

jq — already covered

See textproc.lesson7. The jq filter language is the lingua franca; every other tool below either copies it or improves on it.

yq — YAML the same way

brew install yq
yq '.metadata.name' deployment.yaml
yq '.spec.replicas = 5' -i deployment.yaml   # in-place edit
yq -o=json '.' config.yaml | jq    # convert to JSON

dasel — universal

brew install dasel
dasel -f config.yaml '.database.host'
dasel -f Cargo.toml '.package.version'
dasel -f data.csv '.[0].name'
dasel -f doc.xml '.root.item.name'

One tool to query any structured file format you'll meet daily.

When to reach for which

• JSON only → jq (most ubiquitous, most docs).
• YAML primarily → yq (matches jq syntax exactly, in-place edit).
• Mixed formats in one script → dasel.

Rewriting the same YAML can change its shape

An in-place edit can reserialize comments, anchors, key order, and quoting even when one value changes. Generate to stdout, inspect a diff, and know which document features the tool cannot preserve.

Identify the implementation, not only the command name

Several projects can install an executable named yq. The path and --version identify the syntax contract. Check the installed help before pasting flags or expressions from the web.