Any known python bindings?

[ghost]

Hi! I was wondering if there are any known python bindings or api's for the same validator? I would be a great effort if there is any.

Thanks

[DavidUnderdown]

I've got a very simplistic script (not currently public) that launches the Scala validator as a subprocess from Python 3, checks the return code and writes out the validator output. Inputs and outputs mostly hard-coded at the moment so it would need a little work to turn into a proper module which would be a nicer approach. I'm certainly not aware of any pure Python implementation or more sophisticated API being available though.

The calling module was something I vaguely had in mind as a task for a kind of Python code club we've got running within the organisation at the moment, once we're a bit further on.

[ghost]

yeah i have similar usage of it too as a subprocess. But you know using subprocess is kind of hard approach to me.

[DavidUnderdown]

Probably the most important thing is understanding the return codes. I have

try :
	csvvalidator.check_returncode();
except subprocess.CalledProcessError as err:
	if csvvalidator.returncode == 3 :
		#this is not unexpected (a data validation issue), flag to user and continue
		print(batch,"has CSV validation errors reported see",reportFile,"for details");
	elif csvvalidator.returncode == 2 :
		#something wrong with the schema file, all validation uses the same schema, so stop run, raising the error
		print("CSV schema parsing error:",csvvalidator.stdout)
		raise err;
except subprocess.SubprocessError as suberr:
	## some completely unexpected error has occurred with the subprocess
	raise suberr;
## write out the report file and if no error confirm pass/pass with warnings for the batch;
if csvvalidator.returncode == 0 :
	if csvvalidator.stdout == "PASS\n" :
		print(batch,"passed CSV validation");
	else :
		print(batch,"passed CSV validation with warnings see",reportFile,"for details");

[afranke]

Thanks @DavidUnderdown, I’m on the subprocess train as well but my code was not as sophisticated as yours. A native implementation would indeed be nice, not only because of the more idiomatic use or for performance reasons, but also because having more than one implementation around is always a good thing for something that tries to establish itself as a standard.