avoid reading in whole blob during NamedBlobFile validation (slows TUS uploads)

[djay]

following on from plone/plone.restapi#1690

Currently NamedBlobFile and NamedImageFile fields read in the entire file contents during validation which can be slow and use a lot of memory for large files.
The validation itself doesn't do a lot. It's checking the data field has 0 or more length.
This fix bypasses this check and instead only checks the object provides the right schema.

[mister-roboto]

@djay thanks for creating this Pull Request and helping to improve Plone!

TL;DR: Finish pushing changes, pass all other checks, then paste a comment:

@jenkins-plone-org please run jobs

To ensure that these changes do not break other parts of Plone, the Plone test suite matrix needs to pass, but it takes 30-60 min. Other CI checks are usually much faster and the Plone Jenkins resources are limited, so when done pushing changes and all other checks pass either start all Jenkins PR jobs yourself, or simply add the comment above in this PR to start all the jobs automatically.

Happy hacking!

[djay]

@davisagli any feedback on this solution?
plone/plone.restapi#1690 is dependent on a fix to this problem. or the test is at least.

[davisagli]

@djay Sorry for the delay looking at this. I think you're on to something but have a suggestion to make sure we don't get rid of other parts of the validation.

[davisagli]

@djay Could you add a changelog entry please?

[davisagli]

@jenkins-plone-org please run jobs

[djay]

@davisagli looks like restapi is using that schema to generate the json for the field.

Failing test is (and same for image)

    def test_file_type(self):
        response = self.api_session.get("/@types/File")
        response = response.json()
        self.assertIn("fieldsets", response)
        self.assertIn("file.data", response["properties"]["file"]["properties"])  # noqa

[davisagli]

@djay We're getting into hack territory now but maybe you could do something like this:

def _validate(self, value):
    self.schema = INamedTyped
    try:
        super()._validate(value)
    finally:
        self.schema = INamedBlobFile

But probably it would be cleaner to copy the code from Object._validate so you can pass in INamedTyped instead of self.schema

[djay]

@jenkins-plone-org please run jobs

[djay]

But probably it would be cleaner to copy the code from Object._validate so you can pass in INamedTyped instead of self.schema

@davisagli I did this. didn't like copying all that code but seemed less hacky. It now passes everything.