Draft v2.0.0 - Error in new build system prevents correct schema generation

[jh-RLI]

Description of the issue

The new build system to generate the full chain of: schema.json -> template.json / example.json is currently broken. There is an error that occurs during the schema resolve step. In this step the structure is read and the $refs are resolved. Based on the structure, the resources list element is generated. This element contains n objects. The jsonschema property is missing some properties:

type": "object",
properties": {

This leads to a malformed jsonschema.

Additionally, the group keys from the structure files are copied to the the generated schema.json. This is not expected as they are only used to better group the $ref elements. Thy should be replaced with the content form the reference file.

Steps to Reproduce

1. Start the schema generation script
2. Validate the resulting schema.json
3. try to generate an example or template json file

Ideas of solution

Fix the scripts to properly resolve the refs and replace the ref group field names in the resulting schema.json.

Workflow checklist

• I am aware of the workflow in CONTRIBUTING.md

[Ludee]

Running the scripts doesn't work for me (Windows 10).
FileNotFoundError: [Errno 2] No such file or directory: 'metadata\\v200_draft\\schema.json'

[jh-RLI]

Okay, will adapt this to run on Windows. I run everything using Linux in WSL (on Windows).

[jh-RLI]

Another thing the "metaMetadata" field is not added to the example metadata.

[Ludee]

There is a duplication of the table name in the general keys and resource keys.

resource/Name: A resource MUST contain a name unique to amongst all resources in this data package. To comply with the data package standard it must consist of only lowercase alphanumeric character plus ".", "-" and "_". It may not start with a number. In a database this will be the name of the table within its containing schema. It would be usual for the name to correspond to the file name (minus the file-extension) of the data file the resource describes. Example:
general/Name: A filename or database conform table name.

resource/path: A URL that should be a permanent http(s) address or other path directly linking to the resource.
general/id: A unique identifier (URI/UUID/DOI) for the table or file.

[jh-RLI]

The restriction of the table name when creating tables on the oep is slightly different. Only alphanumeric characters plus “_” are permitted. But since there is no field to save the schema name, this is fine.
Off-topic: Now that I think about it ... maybe we should introduce a field for the data tpics? At some point in the future we will resolve the physical database schemas that are currently associated with the data themes. Once this is done, it is only important to know whether the data is published or not. Storing the table type tobic as part of the table name would then no longer make sense.

[Ludee]

Yes, that is good and sustainable solution. It is somehow duplicate to the subject field.

[jh-RLI]

Subject IMO is also or maybe more duplicated with keywords. It is meant for ontological annotation for specific topics like "co2 emission"but more detailed than the data topics on the oep. But keywords are harmonized with OEP Tags ...