divisibleBy with floating point numbers

[apple314159]

import jsonschema

s = {
                "type": "object",
                "properties": {
                    "price": {
                        "type": "number",
                        "divisibleBy": 0.01
                    }
                }
}

jsonschema.validate({"price":1.09}, s) # fail
jsonschema.validate({"price":1.89}, s) # OK

[apple314159]

One possible solution using the decimal package:

from decimal import Decimal
...
    def validate_divisibleBy(self, dB, instance, schema):
        x = Decimal(str(instance))
        y = Decimal(str(dB))
        failed = x % y != Decimal(0)

        if failed:
            self.error(u"%r is not divisible by %r" % (instance, dB))

[Julian]

I upped the acceptable range for floating point error in 35aeb62. I'm uncomfortable making this always use decimal.Decimal, but doing so yourself is certainly possible:

import decimal
import jsonschema

s = {
                "type": "object",
                "properties": {
                    "price": {
                        "type": "number",
                        "divisibleBy": decimal.Decimal("0.01")
                    }
                }
}

jsonschema.validate({"price": decimal.Decimal("1.09")}, s, number_types=(int, float, decimal.Decimal)) # OK
jsonschema.validate({"price": decimal.Decimal("1.89")}, s, number_types=(int, float, decimal.Decimal)) # OK

With that commit, you should see better behavior for floats anyhow. Let me know how it goes.

[apple314159]

Agreed that using decimal.Decimal is unnecessarily heavy. Your solution does the job.