Add support for regex in resources filter about aws-nuke HOT 9 CLOSED

Probably most of the requirements could be fulfilled by simple glob style matching, right?

I would go for the first solution (type: regex).

from aws-nuke.

Probably most of the requirements could be fulfilled by simple glob style matching, right?

Do you mean like this?

accounts:
  "00000000000":
    filters:
      S3Bucket:
      - "s3://my-bucket/b*sh.txt"

This would work, but I do not see any guarantee that the wildcard * will never be part of an resource identifier.

And if you mean it like this:

accounts:
  "00000000000":
    filters:
      S3Bucket:
      - type: glob
        id: "s3://my-bucket/b*sh.txt"

This would be easy to implement, if we have the basic code foundation.

from aws-nuke.

I meant it like this:

accounts:
  "00000000000":
    filters:
      S3Bucket:
      - type: glob
        id: "s3://my-bucket/b*sh.txt"

from aws-nuke.

Probably we should implement both (which is easy), because globs are easier to write and regex has the full functionality.

from aws-nuke.

from aws-nuke.

@svenwltr this should be available to more than S3 buckets, it a highly useful update

from aws-nuke.

@tomvachon Yes, it is intended to be available to all resources. Actually, implementing it for S3 Buckets only would be even harder.

from aws-nuke.

I very much like this proposal. Regex / glob would greatly simplify things. The old format would become the shorthand for the new type: exact format (this is just rephrasing it from a conceptual view).

This map should contain a type and optionally the data for the filter.

Why do you want to make the data field for the filter optional? What would be the default value?

Being rather nitpicky:

 accounts:
   "00000000000":
     filters:
       S3Bucket:
       - regex: "s3://my-bucket/b[iao]sh\\.txt"

How come we would need two \ to escape the .?

from aws-nuke.

This map should contain a type and optionally the data for the filter.

Why do you want to make the data field for the filter optional? What would be the default value?

Probably depending on the type, but for exact, glob and regex they should be mandatory. Ok it is unlikely that this will be optional.

How come we would need two \ to escape the .?
One for escaping the \ in YAML and the next one for actually escaping the . in the regex.

from aws-nuke.