stricter str validation

[samuelcolvin]

fix #45

In the end, I thought about this and decided whitelisting things which can be coerced to strings made more sense.

It shouldn't affect performance much as most string-like things will be strings and returned before this logic kicks in.

[codecov]

Codecov Report

Merging #52 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master    #52   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files          10     10           
  Lines         694    704   +10     
  Branches      155    157    +2     
=====================================
+ Hits          694    704   +10

[nielskou]

Thank you for the fix. I appreciate it.

However, I still think that "strict by default" would be better. I would expect pydantic to check the types, instead of cleaning the data: If I pass my data to a model, and it doesn't raise a Validation error, I expect my data to be of the right types. I expected Pydantic to more like mypy: a typechecker. Mypy would not allow an int were a str is required.

But maybe I shouldn't complain: dict(Model(**my_data)) will have the right types.

[samuelcolvin]

This is strict by default. What were you expecting?

…

On 21 Jun 2017 3:53 pm, "Zweedeend" ***@***.***> wrote: Thank you for the fix. I appreciate it. However, I still think that "strict by default" would be better. I would expect pydantic to check the types, instead of cleaning the data: If I pass my data to a model, and it doesn't raise a Validation error, I expect my data to be of the right types. I expected Pydantic to more like mypy: a typechecker. Mypy would not allow an int were a str is required. But maybe I shouldn't complain: dict(Model(**my_data)) will have the right types. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#52 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AD2jGfRfnUw-g92IVevEjPYi2Wld5naPks5sGS6HgaJpZM4OA5XB> .

[samuelcolvin]

oh, I see what you mean now. I'll add StrictStr now.

But I don't want to change the default as I think if as user submits {"telephone": 1234567} it shouldn't throw an error, but rather be processed the same as {"telephone": "1234567"}.

[nielskou]

Nice work 👍