Enhancement: programmatic hook, or structured output, for downstream CLI tools

[BatmanAoD]

I'd like to be able to generate "possible" lints with relint, and then use other CLI tools to determine whether each lint is a false positive or not.

Specifically, my use-case is to find instances of \bTODO #\d+ and check whether the #\d+ refers to an open ticket. This requires checking the status of tickets in a ticket-tracker, which is well outside the purview of relint. But if relint had a way to feed a particular lint to a downstream program, especially using regex captures, this would be pretty simple:

- name: TODOS with tickets
  pattern: '\bTODO #(\d+)'
  hint: make sure that the ticket is open
  error: false
  filePattern: '^(?!docs/)'
  postprocess: "glab issue show '${1}' | grep '^state:' | grep -q open"

Here, a status code of 1 means that the lint IS valid, and 0 means it's a false positive. (I'm not sure whether there's a better way to do this than using status codes, and it's not obvious whether 0 should mean "valid" or "false positive", but I think those are primarily documentation issues.)

[mroy-seedbox]

Related:

• Secondary regex matching step (does not fulfil the requirements described above).
• Test functions in Python (could be used to address the requirements described above).

Bonus: if test_func is used in combination with pattern, it should only operate on regex matches instead of on the entire file. This would make it the equivalent of the postprocess proposed above, but would make for a better integration than CLI exit codes (assuming that some kind of Python interface would exist in order to integrate the test_funcs with reLint).

Note: from Python, we could then pretty easily integrate with CLI tools if necessary.