parse tags

[LittleWat]

Reason/Context

Hello! I want to annotate PII tag to avro schemas
https://issues.apache.org/jira/browse/AVRO-3026

But this is ignored in AsyncAPI now.

Description

I created the folloiwng avro schema but the tag property is ignored in AsyncAPI.

{
  "type": "record",
  "name": "Greeting",
  "tags": [
    "PI",
    "PII"
  ],
  "fields": [
    {
      "name": "Sender",
      "type": "string",
      "doc": "the person who sent the message",
      "tags": {
        "<PII-KEY>": "<PII-VALUE>"
      },
      "default": "Api Curio"
    }
  ]
}

I'm glad if this is supported 🙏 Thank you!

[github-actions]

Welcome to AsyncAPI. Thanks a lot for reporting your first issue. Please check out our contributors guide and the instructions about a basic recommended setup useful for opening a pull request.
Keep in mind there are also other channels you can use to interact with AsyncAPI community. For more details check out this issue.

[derberg]

@dalelane @M3lkior please have a look

[dalelane]

@LittleWat I haven't seen the tags property in Avro schemas before. I don't see any mention of it in the Avro specification.

The JIRA issue you linked to was written by someone who was adding a tags property to their Avro schemas, saying that:

Avro AVSC files, being valid json, can easily be modified to add tags that will be used by downstream processors, and also wont interfere with Avro itself

Is that what you were proposing? That we adopt a convention where we look for additional properties that aren't part of the Avro specification, that we then copy into the AsyncAPI document?

[LittleWat]

@dalelane Thank you for your reply! I didn't know that the tag property is not officially supported by the Avro specification. 🤦

I just want to show that the field contains PII in AsyncAPI.
All we can do is just use the doc property...? (Or any other choices...?)

Thank you!

[dalelane]

You can always modify an AsyncAPI doc to add additional information that isn't in the Avro schema.

If you prefer to keep everything within the Avro schema, then yes - you could use doc properties.

The only other idea I can think of just now is to adopt some naming convention - e.g. where fields with PII data all have names starting with pii_ - however that would make it difficult to change whether a field is considered PII, so I'd probably avoid that myself.

[M3lkior]

I share @dalelane answer. Because the tag field is not part of the AVRO spec, i'm not comfortable with the idea to add a support of this root level field in the avro parser ;

If you link your avro file into the AsyncAPI spec, maybe you can use the tag attribute directly in your asyncapi document like

components:
  messages:
    ShareholderOrder:
       tags:
        - name:  "PI"
        - name: "PII"
      payload:
       $ref: "Greeting.avsc"

This will give you the following rendering: