Ultimate transformation library that supports validation, contexts and aiohttp.

Upgraded to 0.11 and seeing a SyntaxError:

File "/home/foo/.virtualenvs/bar/lib/python3.5/site-packages/trafaret_config/__init__.py", line 3, in <module>
    from .simple import read_and_validate, parse_and_validate
  File "/home/foo/.virtualenvs/bar/lib/python3.5/site-packages/trafaret_config/simple.py", line 4, in <module>
    import trafaret as _trafaret
  File "/home/foo/.virtualenvs/bar/lib/python3.5/site-packages/trafaret/__init__.py", line 1, in <module>
    from .base import (
  File "/home/foo/.virtualenvs/bar/lib/python3.5/site-packages/trafaret/base.py", line 20, in <module>
    from .async import (
  File "/home/foo/.virtualenvs/bar/lib/python3.5/site-packages/trafaret/async.py", line 176
    yield (
    ^
SyntaxError: 'yield' inside async function

In [19]: t.Bytes().check(b'test') 
Out[19]: b'test'

In [18]: t.Bytes().check('test')
... 
DataError: value is not a string

In my opinion we need to change string to bytes string.

Hello and thank you for your work fiirst of all.

I have a small suggestion to use trafaret in more uniform way. I have some data with assigned trafaret and some without constraints. So I have to do something like this: if data.trafaret: val = data.trafaret.check_and_return(rawval) else: val = rawval

If you'll kindly add an object something like "NoCheck" which will pass arguments out without any check it will be possible to always do: val = data.trafaret.check_and_return(rawval)

while data without constraints will be initialized as: data.trafaret = trafaret.NoCheck()

Basic trafarets should be called directly, instead of calling check method.

Reason: method __call__ is an entrypoint to trafaret

If we need to redefine method __call__ in custom trafaret, and use this trafaret in t.Or or any other containers, then its method __call__ will never be called.

Example:

import trafaret as t

class CustomString(t.String):
    def __call__(self, val, context=None):
        print('hello')
        return super().__call__(val, context)


Example = t.Or(CustomString(), t.Int())
Example('foo')

hello wil never be printed

Steps to reproduce: install latest version from pypi

There is no extras.py in source code anymore but it is still present after installation. This creates certain confusion as two versions of KeysSubset are present (in keys.py and extras). Moreover, documentation is still pointing to the extras, which is even more confusing

I'd like to rely on this software for important world changing projects, but it's still in 0.X version which means minor version bumps include breaking changes which is not good.

Are there any plans to bump to 1.X so I can pin to trafaret>=1,<2? I'd like some sort of agreement of stability before I include start using this to change the world.

I'm not sure if it's the intended behavior, but in 0.12 it worked without wrapping into t.Dict:

>>> rule, raw = {"foo": int}, {"foo": "1"}
>>> construct(rule)(raw)
{'foo': '1'}
>>> construct(t.Dict(rule))(raw)
{'foo': 1}

In version2 (branch) u need to add to introduction part documentation about Iterable.

For that u can use tests which connected with current part. As example u can use this PR.

Please don't forget:

• your MR u need to do to the version2 (branch)
• add screenshot with your part of documentation
• write below that u want to take it to work

In python 3.7 I'm getting the following deprecation warning:

DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working

collections.abc is available starting from python 3.3.

Are there any plans on fixing it? (I guess you want to retain python2 support yet)

Also I don't mind to create PR to fix this. If you agree, do you have any preferred way this to be done?

Sorry for disturbing again.

I wonder is it possible to convet individually supplied value to a list.

I want to do something like: t = trafaret.List(trafaret.String()) t(['a','b'])

['a','b'] t('a') ['a']

It will be amazingly cosily to pass FieldStorage.getvalue() directly to trafaret.

In version2 (branch) u need to add to introduction part documentation about Enum.

For that u can use old documentation and also tests which connected with current part. As example u can use this PR.

In this task i recommend to add some example of usage and some reach description.

Please don't forget:

• your MR u need to do to the version2 (branch)
• add screenshot with your part of documentation
• write below that u want to take it to work

When importing trafaret from a type-checked codebase, it often generates type check failures (mypy) due to missing explicit __rshift__() method declaration and -> NoReturn for the Trafaret._failure() method.

I'm currently workarounding the issue by adding a custom stubs: https://github.com/lablup/backend.ai-common/tree/main/stubs/trafaret

It would be nice if we could embed the type stubs or provide a separate type stubs package. You may start from my privately written type stubs.

For details about writing/distributing type stubs, refer PEP-561.

import trafaret as trf


@trf.guard(a=trf.Int)
def foo(*, a=1):
    pass


foo()

# GuardError: {'a': DataError('is required')}

problem related to defaults property from FullArgSpec class

def foo(*, a=1): pass
def bar(a=1): pass

getargspec(foo)
# FullArgSpec(args=[], varargs=None, varkw=None, defaults=None, kwonlyargs=['a'], kwonlydefaults={'a': 1}, annotations={})

getargspec(bar)
# FullArgSpec(args=['a'], varargs=None, varkw=None, defaults=(1,), kwonlyargs=[], kwonlydefaults=None, annotations={})

https://github.com/Deepwalker/trafaret/blob/master/trafaret/base.py#L1493-L1496

here we need try to use the kwonlydefaults mapping

Hi, thanks for the useful package!

Is there any technical reason behind Set support absence?

Or I'm a very first person who actually needs it?

If it's missed because no one needs it yet, and maintainers are not against it, I would like to provide pull request with such implementation. Documentation and tests included.

Hopefully, new minor release would be issued after this contribution.

Thoughts?

Regards, Artem.

Hi,

Would it be possible to generate a JSON schema (https://json-schema.org/) from the trafaret definition? This would be really useful because the schema can be used in the IDE (vscode in my case) to autocomplete the yaml file as you are typing (https://github.com/redhat-developer/yaml-language-server). Any thoughts on this?

When using a default value such as: t.Key("integration", default=None, trafaret=t.Regexp(r"\w+"))

This will result in an error, as it checks the default argument against the trafaret object. It seems sensible to skip any checking of the default argument for efficiency and to allow code as shown above.