CLI tool for chaining JSON API requests.
jaq simplifies running multiple, related API commands such as getting IDs or other fields from one endpoint for use in another query.
- Uses a config file to avoid repetition on the command-line
- Set URL paths, query strings, or headers
- Pipe output from one endpoint to another
- Ability to "explode" a JSON list into separate elements for separate processing
- Clone the repo & run
go install - Setup a small config file at ~/.jaq.json:
- Run commands!
go get github.com/Ericsson/jaq && go install
echo '{"domain":"jsonplaceholder.typicode.com"}' > ~/.jaq.json
jaq get /postsAlthough not a requirement, jaq works really well if you are using jq as well for JSON querying/filtering.
These examples assume you are have the jq tool available since we do not try to duplicate its capabilities.
NOTE jaq consumes the
$as a special character so the commands below escape it so it is not consumed by your shell.
# Examples assume config file
echo '{"domain":"jsonplaceholder.typicode.com"}' > ~/.jaq.json
# Get objects from an endpoint
jaq get /posts
# Delete all the objects by id which have userId == 3
jaq get /posts | jq -c '.[] | select(.userId == 3)' | jaq delete /posts/\${1.id}
# Get comments from the 3 most recent posts
jaq get /posts | jq -c .[-3:] | jaq get /comments -q postId=\${1.id}
jaq uses a configuration file (defaults to ~/.jaq.json but configurable via a flag) to make your commands more succinct.
You can also override most settings in the configuration file by using environment variables with similar keys. Use env vars of the form: JAQ_<KEY_NAME> to set the value of "key-name". (Env vars must be upper cased, prefixed with JAQ_ and hyphens replaced with underscores.
The auth setting can switch which authorization scheme to use by default. Currently the supported types are:
basic- Will use the "user" and "pass" fields to set the Authorization header.token- Will set the header "Authorization: Bearer <token>"
The dry-run setting allows you to try out potentially destructive commands and ensure all the input transformations result in the expected commands.
When running in dry-run mode, all of the input transformations occur but the resulting commands are just printed to stdout rather than actually being run.
Since jaq operates by piping JSON over stdin, if you want to use a header field you must include it in the JSON from the response. To facilitate this, when --print-headers is set, all of the headers are added as new JSON fields on all the JSON objects of the response with the prefix jaq-.
You may want different behavior when encountering an error (HTTP response >= 400). Options are:
silence- Completely ignore those responses and do not print them or error.fatal- Print them to stderr and return an error.continue- Print the responses to stdout as if they were not errors and continue.report- Print the responses to stderr so they do not pollute stdout for other piped commands.
When executing commands you may want an entire dump of the HTTP request/response. By specifying --trace the request/response will be dumped to stderr (so that it doesn't interfere with the JSON on stdout). By default, the body of the requests are NOT dumped. You can set --DEBUG to also add the body of the request.