config wildcard ignores: ignored

[bpinsard]

I am trying to ignore a set of errors when running the validator on a datalad without any file fetched.

Here is the config I wrote following https://github.com/bids-standard/bids-validator/blob/main/docs/user_guide/command-line.md#configuration-file

{
  "ignore": [
    {"code": "TSV_COLUMN_MISSING", "location": "/**/*_scans.tsv"},
    {"code": "TSV_ADDITIONAL_COLUMNS_UNDEFINED", "location": "/**/*_scans.tsv"},
    {"code": "SCANS_FILENAME_NOT_MATCH_DATASET"}
  ]
}

But the errors keep showing up (except the last ignore SCANS_FILENAME_NOT_MATCH_DATASET ).
I tried different type of globbing patterns, but none seems to work.
Are wildcard patterns supported? If so in which format?
Thanks!

[effigies]

No, the matches are literals. We hadn't really considered the possibility of globs.

[bpinsard]

From my limited understanding of the code, it seems that the config ignores are merged with .bidsignore which can contains globs similar to a .gitignore, so that wouldn't work through that either? IIRC the legacy validator is able to handle globs in .bidsignore.
How complex would that be to implement globs?
Thanks!

[effigies]

Globs are supported in .bidsignore. The config takes partial issues and recategorizes any issue found that matches them into the appropriate bucket. It's a completely different mechanism.

[bpinsard]

Ok, do you agree that it would be a useful feature to support glob?
If I understood the code correctly, it would require this function:

bids-validator/src/issues/datasetIssues.ts

Lines 37 to 47 in da2c6e6

	get(issue: Partial<Issue>): Issue[] {
	let found: Issue[] = this.issues
	for (const key in issue) {
	const value = issue[key as keyof Issue]
	if (!value) {
	continue
	}
	found = found.filter((x) => x[key as keyof Issue] === value)
	}
	return found
	}

to be able to use globs for comparing location with something like:

import type { Ignore } from '@ignore'
.
.
.

  get(issue: Partial<Issue>): Issue[] {
    let found: Issue[] = this.issues
    for (const key in issue) {
      const value = issue[key as keyof Issue]
      if (!value) {
        continue
      }
      if (key === 'location'){
        let key_ignore: Ignore = ignore()
        key_ignore.add(value)
        found = found.filter((x) => key_ignore.ignore(x[key as keyof Issue].slice(1, x[key as keyof Issue].length)))
      } else {
        found = found.filter((x) => x[key as keyof Issue] === value)
      }
    }
    return found
  }

Or we could choose to use the leading slash to distinguish real patterns from globs?
Or prefix globs with a specific character.