To do so it defines a simple input/output protocol (specified in this JSON Schema which validator implementations can implement, and it provides a CLI which can execute supported implementations.
It’s called Bowtie because it fans in lots of JSON then fans out lots of results:
Looks like a bowtie, no?
Also because it’s elegant – we hope.
Via Homebrew (macOS or Linuxbrew)#
brew install bowtie-json-schema/tap/bowtie
Manual Installation via PyPI#
Bowtie is written in Python, and uses a container runtime to execute JSON Schema implementations, so you’ll need both Python and a suitable container runtime installed.
If you have no previous container runtime installed (e.g. Docker), follow the installation instructions for podman specific to your operating system. Ensure you’ve started a Podman VM if you are on macOS.
Then follow the pipx installation process to install
pipx, and finally use it to install Bowtie via:
pipx install bowtie-json-schema
which should give you a working Bowtie installation, which you can check via:
Further usage details of the command-line interface can be found here.
In general, executing Bowtie consists of providing 2 pieces of input:
The names of one or more supported implementations to execute
One or more test cases to run against these implementations (schemas, instances and optionally, expected validation results)
Given these, Bowtie will report on the result of executing each implementation against the input schema/instance pairs. If expected results are provided, it will compare the results produced against the expected ones, reporting on any implementations which differ from the expected output.
A key use of Bowtie is in executing as input the official test suite and comparing the results produced by implementations to the expected ones from the suite.
Bowtie however isn’t limited to just the test cases in the test suite. It can be used to compare the validation results of any JSON Schema input across its supported implementations.