[Ingest][convert-processor] Added MAC type in convert processor

[vinit-chauhan]

This PR here is to add a new mac type in the convert processor. It works similarly to an IP processor which would assert the format of the MAC Address.

It is seen that Elasticsearch encourages users to store the MAC Address in the RFC 7042 format Here. This PR will add the capability to assert and convert the MAC Address to the suggested format and throws an error if MAC Address is invalid.

[github-actions]

Documentation preview:

• ✨ Changed pages

[elasticsearchmachine]

Pinging @elastic/es-data-management (Team:Data Management)

[joegallo]

Thank you for the pull request, @vinit-chauhan!

I did a bit of history spelunking, and the reason there's an ip type for convert at all is a little bit of a journey. #36145 is the original issue for it, and #36145 (comment) in particular is illuminating. #69989 was the pull request that added the functionality.

Long story short, it was added as a convenience, yes, but more importantly it was added for feature parity with the pre-existing ip type on the beats version of convert. That's also the source of our slightly niche behavior here, specifically, that we're only validating the format of the ip address string and not actually doing a data-type conversion (i.e. to a proper java.net.InetAddress).

Another consideration is that Elasticsearch has a proper ip mapping type, but there's no such support for mac addresses in Elasticsearch (they'd just be strings, I suppose).

With the history/origin of the ip type for convert processors and the lack of mac address support elsewhere in Elasticsearch in mind, I don't think this is an enhancement we want to consider at this time.

---

On the bright side, that doesn't mean you can't validate whether Strings look like mac addresses in an ingest pipeline! Here's a script processor that's a reasonably faithful translation of your code:

PUT _ingest/pipeline/test-pipeline-1
{
  "processors" : [
    {
      "script": {
        "description": "Validate that the field 'foo' is a mac address, converting the format slightly along the way",
        "lang": "painless",
        "source": """
        def val = ctx.foo;
        def sb = new StringBuilder();
        val = val.toUpperCase();
        val = val.replaceAll(/[^a-fA-F0-9]/, m -> "");
        if ((val.length() == 12 || val.length() == 16) == false) {
          throw new IllegalArgumentException('[' + val + '] is not a valid MAC Address');
        }
        
        for (int i = 0; i < val.length(); i++) {
          if (i != 0 && i % 2 == 0) {
            sb.append('-');
          }
          sb.append(val.charAt(i));
        }
        
        ctx.foo = sb.toString();
        """
      }
    }
  ]
}

invoking it like this:

POST /_ingest/pipeline/test-pipeline-1/_simulate
{
  "docs": [
    {
      "_index": "index",
      "_id": "id",
      "_source": {
        "foo": "0123456789ab"
      }
    }
  ]
}

gives the following result:

{
  "docs" : [
    {
      "doc" : {
        "_index" : "index",
        "_id" : "id",
        "_version" : "-3",
        "_source" : {
          "foo" : "01-23-45-67-89-AB"
        },
        "_ingest" : {
          "timestamp" : "2023-06-09T19:44:56.809747Z"
        }
      }
    }
  ]
}