Add new base validator and base constraint classes

[drmull]

Add new base classes to validator and constraint that take additional arguments, such as schema, path etc. This allows constraints to use non-primitive validators and resolves issue #133

My initial intention was to pull out the logic for the non-primitive validators from the schema file and put them in each respective validator class but then I figured it would be a bit too drastic..

[lgtm-com]

This pull request fixes 6 alerts when merging e179cda into d2d65eb - view on LGTM.com

fixed alerts:

• 6 for __eq__ not overridden when adding attributes

[drmull]

I forgot to change the name of the _validate function, I will fix it

[lgtm-com]

This pull request fixes 6 alerts when merging 5965dc6 into d2d65eb - view on LGTM.com

fixed alerts:

• 6 for __eq__ not overridden when adding attributes

[mildebrandt]

Welcome back @drmull :)

FYI: This may take me a while to get to, it's a busy time over here. Sorry.

[drmull]

Welcome back @drmull :)

Thanks!

FYI: This may take me a while to get to, it's a busy time over here. Sorry.

No problem, I am in no rush :)

[mildebrandt]

Sorry for the long delay.

I grabbed the example from #133, but I couldn't get your code to work with it.

data:

maptest: 
  1: xpto 
  2: qwerty 
  error: wrong

schema:

maptest: map(include('value'), key=include('key')) 
--- 
value: str()
key: int()

This combination should fail validation.

[drmull]

Hm, strange. I updated the test case to match the example but I get:

maptest: Key error - 'error' is not a int.

Do you use the PyYAML parser? Maybe the order of the elements can differ with different parsers and trigger the problem.

I will take a closer look!

[mildebrandt]

Yeah, I'm using PyYaml 5.3.1.

However, I'm beginning to think this isn't something we should support. Allowing for include in the key constraint only makes sense when processing complex keys, for example:

? {one: 1, two: 2}
: [ 2001-01-01, 2002-02-02 ]

Both PyYAML and ruamel fail to load such a document, but it's valid yaml. Instead, I think the correct course of action is to not allow the include validator for the key constraint and raise an error saying something like:

The include validator is not valid for the key constraint.

Other validators can be used to constrain the key and still support all yaml documents that PyYAML and ruamel can load.

[drmull]

Yeah, I can agree with that. I will close this PR and we can reopen it if ruamel implements complex keys in the future 😁

[mildebrandt]

Cool, and thanks for your willingness to always jump in!