add float truncation

[Eomm]

Fix #277

Note that to implement a rounding input option the most straightforward way is to generate the output function calling a global named function.

  let customAsInteger = $asInteger.toString()
  if (options.rounding) {
    customAsInteger = customAsInteger.replace('parseInt', options.rounding)
  }

In this case the input would be the ugly (note the Math.round as 'string`):

const debugCompiled = fastJson(schema, { debugMode: true, rounding: 'Math.round' })

Or:

function round(in){ return Math.round(in) }
const debugCompiled = fastJson(schema, { debugMode: true, rounding: round })

Where

-  customAsInteger = customAsInteger.replace('parseInt', options.rounding)
+ customAsInteger = customAsInteger.replace('parseInt', options.rounding.name)

Which one do you think would be better? Or do you see nicest solutions?

Checklist

• run npm run test and npm run benchmark
• tests and/or benchmarks are included
• documentation is changed or added
• commit message and code follows the Developer's Certification of Origin
  and the Code of conduct

[mcollina]

I would support only strings: 'round', 'floor', 'ceil'.

[jsumners]

There isn't any concept of "rounding" defined in the spec https://json-schema.org/draft/2019-09/json-schema-validation.html#rfc.section.6.2

[mcollina]

so you think the default behavior should be to throw if it's not an integer?

[jsumners]

I believe that is the consensus we reached the last time we discussed how to handle non-spec compliant values (i.e. undefined).

[mcollina]

I would actually prefer to not do anything in this case if we want to throw.

[jsumners]

Looking over this again, it seems this PR is making use of an undocumented second parameter to the FJS function: const stringifierFunction = FJS(jsonSchema, {...options}). So I am okay with this PR since it isn't adding new syntax to the JSON Schema spec. However, this PR should also update the documentation for this mysterious options object or there should be a fast follow-up PR that does so.

[Eomm]

Yes, this pr doesn't force the user to add non-spec fields to the input schema.

This pr force the type coertion of those properties defined by user as "integer"

The second parameter is documented to add external schemas and debug the stringify function, but I agree on a follow up pr to document better this option that is spread across the readme and I will take care of it 👍🏼

[Eomm]

Are there any concerns for this PR?

[mcollina]

lgtm