Escape JSON pointer tokens

[davishmcclurg]

When building data_pointer and schema_pointer in error objects.

JSON Pointer RFC:

Because the characters '~' (%x7E) and '/' (%x2F) have special
meanings in JSON Pointer, '~' needs to be encoded as '~0' and '/'
needs to be encoded as '~1' when these characters appear in a
reference token.

Fixes: #121

[eraffel-MDSol]

@davishmcclurg This breaks when the token is not a string, for example a boolean

[davishmcclurg]

@eraffel-MDSol do you mind providing an example that I can test with?

[eraffel-MDSol]

Ok, correction, the issue is when the key is a boolean.
example: { true => 'asdf' }

I guess my question is this, since the library is accepting a hash, should it do the conversion from true to 'true', or should the caller be responsible for that.

Regardless, I would assume you don't want to throw an exception, so there should be a check that the key is a string, right?

[eraffel-MDSol]

@davishmcclurg Any thoughts on the above?

[davishmcclurg]

For reference:

>> schema = { 'type' => 'object' }
=> {"type"=>"object"}
>> data = { true => 'value' }
=> {true=>"value"}
>> JSONSchemer.schema(schema).valid?(data)
/Users/dharsha/repos/json_schemer/lib/json_schemer/schema/base.rb:625:in `escape_json_pointer_token': undefined method `gsub' for true:TrueClass (NoMethodError)

        token.gsub(JSON_POINTER_TOKEN_ESCAPE_REGEXP, JSON_POINTER_TOKEN_ESCAPE_CHARS)
             ^^^^^
        from /Users/dharsha/repos/json_schemer/lib/json_schemer/schema/base.rb:552:in `block in validate_object'

I guess my question is this, since the library is accepting a hash, should it do the conversion from true to 'true', or should the caller be responsible for that.

Hmm, good question. I think ideally the caller should be responsible for that since this library assumes data is valid JSON (ie, string keys). For example, versions prior to this change have a similar problem with patternProperties:

>> schema = { 'patternProperties' => { '.+' => { 'type' => 'string' } } }
=> {"patternProperties"=>{".+"=>{"type"=>"string"}}}
>> data = { true => 'value' }
=> {true=>"value"}
>> JSONSchemer.schema(schema).valid?(data)
/Users/dharsha/repos/json_schemer/lib/json_schemer/schema/base.rb:583:in `match?': no implicit conversion of true into String (TypeError)
	from /Users/dharsha/repos/json_schemer/lib/json_schemer/schema/base.rb:583:in `block (2 levels) in validate_object'

There's also the issue of referencing non-string keys in schemas. JSON schemas don't allow non-string properties (because JSON itself doesn't), so if we implicitly to_s keys they can't be referenced or validated. For example, how would these keys be differentiated?

data = {
  true => 'value',
  'true' => 'value'
}

A valid JSON schema can only target the string version.

That said, this is a change in behavior and calling to_s in escape_json_pointer_token would be more like the previous behavior. I opened an issue for more discussion, but for now I think I'm going to leave this as it is. Let me know in that issue if you have additional thoughts/concerns.