andersoncardoso
commented
7 years ago Hi, probably it's better to use small files for each sub-schema to be reused. This can be achieve by referencing the definitions by url, but I have to check how we can generate new schemas for a community on the plataform.
About the validation problems. You have a little mistake on the schema. notice you defined the ctdl schema on the "definitions", but those are never used unless you point to them (using the json pointer notation "#/definitinos/name_of_the_defintion").
For example:
{
"context": { ... }
"definitions": {
"definition_A": { .... },
"definition_B": { .... },
"definition_C": { .... }
},
allOf: [
{"$ref": "#/definitions/definition_A"},
{"$ref": "#/definitions/defintion_B}
] // notice that in this case "C" is never used, so it wont be used for validation.
}you don't have an "allOf" clause (there is others, like "anyOf", etc).
just adding this: "allOf": [{"$ref": "#/definitions/ctdl"}] should do the trick.
If you don't want to split your schema into definitions you can just to this way:
{
"context": {...}
"properties": { ... place properties directly here, instead using a definitions pointer }
}After fixing the schema, I tested with http://www.jsonschemavalidator.net/ and it shows the errors properly :
Found 3 error(s)
Message:
JSON does not match all schemas from 'allOf'. Invalid schema indexes: 0.
Schema path:
#/allOf
Message:
String 'urnX:ctid:25843D95-13E6-4CF8-B3F3-E4B9DBBEDB10' does not match regex pattern '^urn:ctid:.*'.
Schema path:
#/definitions/ctdl/properties/ctdl:ctid/pattern
Message:
Required properties are missing from object: ctdl:name, ctdl:url.
Schema path:
#/definitions/ctdl/requiredI hope this helps. best regards
Anderson
ghost
@andersoncardoso , @science As noted previously, the updated schema (http://credreg.net/ctdl/Schema/20161117/validation/json is quite different from that currently in the registry. It is currently generated from the properties spreadsheet at : https://docs.google.com/spreadsheets/d/1Bw94UrAWwno9gFg7yOZjnTGq9XUlsyYrgEi_FwgZ_tY/edit#gid=0 The previous version of the spreadsheet is a little easier to read at: https://docs.google.com/spreadsheets/d/1Bw94UrAWwno9gFg7yOZjnTGq9XUlsyYrgEi_FwgZ_tY/edit#gid=0
As noted the schema is generated, and so is in one file. I would think that it would be more difficult to work with, and that we should split it into multiple files. I have already made this request to @siuc-nate. @andersoncardoso would you agree?
There are lots of classes that are referenced from the 'top' level classes. So rather than duplicating these, should these all be placed in a separate 'common' schema file? We would then have at least:
Will this make sense, or would additional common files be preferred, perhaps to ease management (with smaller files), or to logically group related classes?
I see that there already an external schema file called json_ld, that has some common definitions. It doesn't appear that these are used yet by the ce registry. This looks like it is intended to use with graphs and probably would not be modified at all related to the ce-registry.
Testing I have tried to use the site: http://www.jsonschemavalidator.net/ to test modifying a schema, and testing. I have not been able to properly get expected errors. Following is the schema (abbreviated) and test json.
Schema Test file { "$schema": "http://json-schema.org/draft-04/schema#", "description": "CE/Registry Credential metadata", "type": "object", "definitions": { "ctdl": { "properties": { "@type": { "enum": [ "ctdl:Credential", "ctdl:Badge", "ctdl:DigitalBadge", "ctdl:OpenBadge", "ctdl:Certificate", "ctdl:ApprenticeshipCertificate", "ctdl:JourneymanCertificate", "ctdl:MasterCertificate", "ctdl:Certification", "ctdl:Degree", "ctdl:GeneralEducationDevelopment", "ctdl:SecondarySchoolDiploma", "ctdl:License", "ctdl:MicroCredential", "ctdl:QualityAssuranceCredential" ] }, "ctdl:accreditedBy": { "description": "An agent that accredits the described resource.", "enum": [ "ctdl:Organization", "ctdl:CredentialOrganization", "ctdl:Person" ], "error": "Must be one of the values: 'ctdl:Organization', 'ctdl:CredentialOrganization', 'ctdl:Person'" }, "ctdl:ctid": { "description": "A globally unique Credential Transparency Identifier (CTID) issued by the Credential Registry Service (CRS) by which the creator/owner/provider of a credential recognizes the credential in transactions with the external environment (e.g., in verifiable claims involving the credential).", "type": "string", "pattern": "^urn:ctid:.*" },
}
Test Json I tried forcing required property errors by using:
{ "@context": { "schema": "http://schema.org/", "dc": "http://purl.org/dc/elements/1.1/", "dct": "http://dublincore.org/terms/", "rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#", "rdfs": "http://www.w3.org/2000/01/rdf-schema#", "ctdl": "[CTI Namespace Not Determined Yet]" }, "@type": "ctdl:Credential", "ctdl:credentialLevel": [ "Postsecondary (Less than 1 year)", "Postsecondary (1-3 years)", "Postsecondary (3-6 years)", "Postsecondary (6+ years)" ], "ctdl:credentialType": [ "Badge" ], "ctdl:ctid": "urnX:ctid:25843D95-13E6-4CF8-B3F3-E4B9DBBEDB10", "ctdl:dateEffective": "2016-05-04T00:00:00", "ctdl:estimatedTimeToEarn": [ ], "ctdl:description": "This is the description of the credential.\n0710a", "ctdl:image": "http://myimageurl.com", "ctdl:nameX": "zzz Credential Name Example- 0823a", "schema:url": "http://credreg.net/0710a" }
NewtonSoft has a package for testing json. validation: NewtonSoft.Json.Schema. I have downloaded it done some testing. I have been able to get errors but not to the level that I had hoped. I want to ensure that I am not wasting time by not using the right tools. So: