Support for other file formats in the CLI

[nightwatchcyber]

Would you consider adding support for validating other file formats (YAML, TOML, INI, etc) via JSON Schema? This would be similar to the functionality provided by this NodeJS project:
https://github.com/json-schema-everywhere/pajv

It is possible to add this functionality via the "anymarkup" module:
https://github.com/bkabrda/anymarkup

I can provide a pull request with these changes

[Julian]

Hi! Thanks for the idea and offer --

So, I do this personally with remarshal.

E.g. by running jsonschema -i <(yml2json schema.yml) etc

I am... not completely against having that more native (though yeah for me what I do already works quite OK), but I have reservations about "magic". It's easy (as anymarkup says itself) to have ambiguous say INI vs TOML, so it's introducing room for complexity and incorrectness there.

Happy to hear your thoughts on that if you have any.

[thebjorn]

(not OP) I would be happy to have this for only yaml to start with since it doesn't introduce ambiguity, schemas are much nicer to write in yaml (and our own config files are in yaml and needs validation). Since yaml is a subset of json there shouldn't be much, if any, magic need :-)

remarshal is a great tool, but your example is not quite so convenient on windows (or when both schema and instance are yaml).

[Julian]

That all makes sense to me! Definitely OK with that much.

[ssbarnea]

@julesjulian I was looking to create a YAML linter that would use data from http://schemastore.org/json/ (which applies to yaml too) but without having native support for yaml I am kinda stuck.

Once this is covered, I can start implementing the tool which should be able to detect schema to use, use internal cache of schemas, lint all files from a repo. If you think that this could be desirable for your tool, I don't mind extending it, otherwise I would make another tool that only calls jsonschema for validating the these files.

[Julian]

@ssbarnea that's.. interesting, I don't think I'd seen that before.

I think I'm not opposed to adding support for that if it's reasonably small to implement.

Can we tackle that separately though? I.e., one ticket (this one) for just switching to parsing via YAML, and a second one for adding support for schema store to the CLI?

[gsemet]

I would be interested in having a transparent yaml loader with $ref resolution. It would require to restrict to the YAML syntax that is 1:1 transposable to json.

[ssbarnea]

I think that for start trying using the pyyaml loader when file extension is either .yml or .yaml should count as easy and safe to implement.

We could even make the dependency on pyyaml soft and enable the feature only when it is installed.

[ioggstream]

While I'm not sure of the "anymarkup", given that YAML is widely used when developing json-schema and OpenAPI3 files, even a simple YAML loader (eg. yaml.safe_load) would be great.

I think that a conservative solution could even rely on a specify CLI parameter (eg. --yaml) that will do something like

jsonschema/cli.py

class Outputter():
   def from_arguments(...):
      if ... yaml in arguments ... :
            ... set loader ...
      
...
def run(...):

   schema_loader = json.load
   if ... yaml in arguments ... :
       schema_loader = yaml.safe_load

Probably a way to specify the loader in a single place could work too (eg. outputter.schema_loader )

If you are interested, I'm happy to PR :)

[sirosen]

Providing this feature is one of the main goals for check-jsonschema.

It provides json and yaml support by default, and if pyjson5 or json5 is installed, it will support JSON5 format files as well. If TOML is really wanted, I'd probably add tomli support following the same path laid out by JSON5. I'm less enthusiastic about INI support (though configparser makes it easy), but could be convinced if users really want it.

YAML for schema files is only supported for local files. Because a RefResolver won't understand YAML, it didn't seem very safe to try to load remote schemas as YAML. I'm happy to revisit or discuss this behavior.

I'd love to hear from any of the users in this thread if check-jsonschema does (or does not!) meet their needs.

I am the author of check-jsonschema and we are currently discussing making it the official CLI for jsonschema. For details, you can follow #910. If check-jsonschema has a behavior which you believe should be changed to handle your use-case, please open an issue!

[sirosen]

We're going to move forward and begin calling check-jsonschema the official CLI. I'm therefore going to close this as a wontfix.