CPPSDK: changes to handle enum and x-schema enum without title

[HaseenaSainul]

No description provided.

[jlacivita]

I'm curious if you could write a summary of what this PR is doing?

We have an SDK.config.json feature to add titles to all untitled schemas. Why wouldn't we just use that, instead of this code?

[jlacivita]

Hmm, we currently don't have any enums w/ false or null in the values. we might want to map them to strings, though, e.g.: false -> "false" and null -> "unknown".

"unknown" is our standard value for enums with an unknown possibility.

[HaseenaSainul]

FontEdge and FontFamily enums ending with null
"FontEdge": {
"type": [
"string",
"null"
],
"enum": [
"none",
"raised",
"depressed",
"uniform",
"drop_shadow_left",
"drop_shadow_right",
null
]
}

Here null is generating as empty line, if we are not filtering then it is converting empty line with ','

[jlacivita]

So these two fields need to actually support an unset value. we're using null for that in the JSON-Schema.
What makes sense in Cpp for an unset value?

We could return an error or create a standard enum value, like NONE.

If we filter it out, then the method won't support unset values.

[HaseenaSainul]

Will check about this

[HaseenaSainul]

for cpp there is no Unset values, so just added to set value if there is unset template available. So for javascript this can be used to set unset values. So in this case, for CPP, there is chance to create enum with delimiter for last parameter as well (if last one is null), hence the delimiter handling logic is added after completing the value parsing

[jlacivita]

Same comment here about false & null.

[jlacivita]

Would be better to do this with a feature flag in sdk.config.json

[HaseenaSainul]

ok

[HaseenaSainul]

updated config for this

[HaseenaSainul]

removed enumSufix from the config and used delimiter macro to handle this

[HaseenaSainul]

SDK.config.json

which one you meant here ? config.extractSubSchemas ?
In latest accessibility.json schema we can see properties of Accessibility Schema such as FontFamily, FontEdge etc are changed to enums, but does not have title. these are x-schemas. So while searching for the x-schema property info titles are missing here and it is not generating types/shapes properly, even issue with namespace mapping.

Hence this missing title is fixing from module x-schema generation time itself

[jlacivita]

I think we can get rid of enumSuffix everywhere and just follow the same ${delimiter},${end.delimiter} pattern. See my comment.

[jlacivita]

I was thinking more of a boolean, like removeExtraEnumDelimiter but now that I think about it, we'd want to do this in every language.

The way we've done it in other templates, like Objects and Arrays is like this:

template = template.replace(/\$\{delimiter\}(.*?)\$\{end.delimiter\}/g, i === schema.anyOf.length - 1 ? '' : '$1')

[jlacivita]

Then we can wrap the , in the template w/t that delimiter macro, and it won't get put on the last line.

JavaScript doesn't need the final delimiter, its happy with or without it.

[HaseenaSainul]

sure will check

[HaseenaSainul]

removed enumSufix from the config and used delimiter macro to handle this

[jlacivita]

this looks good, thanks!