Conversion Tools — yq, jq, dasel, gron

The CLI toolchain that bridges the four formats

jq — JSON-only query language

The pioneer. jq parses JSON, applies a query, emits JSON. Its filter syntax (.foo[], map, select) became the lingua franca for the others.

yq — same syntax, all four formats

Mike Farah's yq applies jq's filter language to YAML, JSON, TOML, XML, and CSV. yq -p toml -o yaml file.toml reads TOML and emits YAML. Best general-purpose conversion tool.

dasel — selector across many formats

dasel uses a unified selector syntax (.config.server.port) and supports JSON/YAML/TOML/XML/CSV. Lighter than yq for simple lookups; comparable for conversions.

gron — flatten JSON for grep

gron turns nested JSON into one assignment per line: json.users[0].name = "Pippa";. Then grep works the way you expect on JSON. gron --ungron rebuilds the original. Surgical for finding-then-extracting fields in deeply nested API responses.

Principle: learn jq's filter language once; reuse it via yq for YAML/TOML. The investment pays off across config audit, log spelunking, and CI pipeline glue. The minute you reach for Python to massage a YAML file, you've probably forgotten yq exists.

Code

yq — every conversion you'll need·bash

# YAML → JSON
yq -o json file.yaml

# JSON → YAML
yq -p json -o yaml file.json

# TOML → YAML
yq -p toml -o yaml pyproject.toml

# YAML → TOML
yq -p yaml -o toml file.yaml

# JSON → TOML
yq -p json -o toml file.json

# Round-trip — pretty-print in place
yq -i . config.yaml

jq + yq combos·bash

# Find all replicas across multi-document YAML
yq '.. | select(has("replicas")) | .replicas' deploy/*.yaml

# Filter PR JSON for open ones with a label
gh pr list --json number,title,labels | \
  jq '[.[] | select(.labels[]?.name == "ready-for-review")]'

# Convert pyproject.toml [project] block to JSON
yq -p toml -o json '.project' pyproject.toml

dasel — quick lookups·bash

# Get a value
dasel -f config.yaml '.server.port'

# Set a value (in place)
dasel put -f config.yaml -v 9000 '.server.port'

# Convert via -r/-w flags
dasel -r yaml -w json < config.yaml > config.json

gron — grep deeply nested JSON·bash

# Flatten
gh api repos/cwk/foo | gron | head
# json.full_name = "cwk/foo";
# json.private = false;
# json.owner.login = "cwk";

# Find every URL field in the response:
gh api repos/cwk/foo | gron | grep '_url'

# Edit the gron output and ungron back to JSON
gh api repos/cwk/foo | gron > flat.txt
# (edit flat.txt)
gron --ungron < flat.txt > modified.json

Exercise

Take a real config (a pyproject.toml, an Ansible inventory, a Kubernetes manifest). Convert it to each of the other three formats with yq. Diff the results. Notice what gets lost (comments, key order in some round-trips). Save the conversion incantation you used most often as a shell function in your dotfiles.

Conversion Tools — yq, jq, dasel, gron

The CLI toolchain that bridges the four formats

jq — JSON-only query language

yq — same syntax, all four formats

dasel — selector across many formats

gron — flatten JSON for grep

Code

External links

Exercise

Progress

Comments 0