Custom types not matching as expected

[Indemnity83]

Describe the bug

I defined custom types that include an emoji, but when I run the action the validation fails even though the PR title uses one of the type options and it is unclear why the workflow doesn't think it matches.

To reproduce

Setup custom types with emoji's:

with:
  types: |
    ✨ feat
    🐛 fix
    📝 docs
    💎 style
    ♻️ refactor
    📈 perf
    🧪 test
    🏗️ build
    📦 ci
    🧹 chore
    ⏪️ revert

Generate a PR with one of these types in the title "📦 ci: run tests"

Expected behavior

I'd expect this to pass validation, but it fails with the following message:

No release type found in pull request title "📦 ci: run tests". Add a prefix to indicate what kind of release this pull request corresponds to. For reference, see https://www.conventionalcommits.org/

Available types:
 - ✨ feat
 - 🐛 fix
 - 📝 docs
 - 💎 style
 - ♻️ refactor
 - 📈 perf
 - 🧪 test
 - 🏗️ build
 - 📦 ci
 - 🧹 chore
 - ⏪️ revert

[amannn]

This might be a duplicate of #257

[Indemnity83]

Digging a bit deeper, it looks like an issue with the regex used by conventional-commit-parser. It expects a single word.

/^(\w*)(?:(([\w$.-* ])))?: (.)$/

That library also appears to be 8 years without an update so not much point in trying to fix that.

Possibly, would need to replace that library with something like https://commitlint.js.org/api/lint

[amannn]

That library also appears to be 8 years without an update so not much point in trying to fix that.

conventional-commits-parser? It looks actively maintained to me, no?

[buglavecz]

Hi.
The latest version 6.0.0, but you used the 3.2.x version.
I have a same problem, i try use gitmoji icons before commit types, but I try header regexp config change, it is doesn't works.

[amannn]

I've just updated all dependencies for the v6 release of this action. Is this issue still relevant?

[gabriel-pragin-clearscale]

@amannn: this issue happened for the first time for me on v6 (24e6f01), and although I have custom types, the type it failed to find was default.

Edit: It started working again when I removed the headerPattern and headerPatternCorrespondence lines from my config. Might be due to a change in regex parsing from v5 -> v6, but you can disregard my comment at this point. Leaving it here in case someone else runs into this.

[amannn]

@gabriel-pragin-clearscale Thanks for the hint! Can you share the headerPattern and headerPatternCorrespondence you had before?

[gabriel-pragin-clearscale]

@amannn: I was using the values from the README.md here.

[amannn]

Hmm, interesting. I tried to reproduce this in #293 but didn't have luck …

Is there anything I misconfigured in the test I added there?

[gabriel-pragin-clearscale]

Maybe the custom types? I added it into this issue just in case that might have been linked in some way. Here's my full config which failed:

---
name: Lint PR
on:
  pull_request_target:
    types: [opened, edited, reopened, synchronize]
jobs:
  main:
    name: Validate PR title
    runs-on: ubuntu-latest
    permissions:
      pull-requests: read
    steps:
      - uses: amannn/action-semantic-pull-request@v6
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        with:
          # Configure which types are allowed (newline-delimited).
          # Default: https://github.com/commitizen/conventional-commit-types
          types: |
            feat
            fix
            docs
            style
            refactor
            perf
            test
            build
            ci
            chore
            revert
            release
          # Configure which scopes are allowed (newline-delimited).
          # These are regex patterns auto-wrapped in `^ $`.
          scopes: |
            main
            deps
            except
            scp
          # Configure that a scope must always be provided.
          requireScope: false
          # Configure additional validation for the subject based on a regex.
          # This example ensures the subject doesn't start with an uppercase character.
          subjectPattern: ^(?![A-Z]).+$
          # If `subjectPattern` is configured, you can use this property to override
          # the default error message that is shown when the pattern doesn't match.
          # The variables `subject` and `title` can be used within the message.
          subjectPatternError: |
            The subject "{subject}" found in the pull request title "{title}"
            didn't match the configured pattern. Please ensure that the subject
            doesn't start with an uppercase character.
          # If the PR contains one of these newline-delimited labels, the
          # validation is skipped. If you want to rerun the validation when
          # labels change, you might want to use the `labeled` and `unlabeled`
          # event triggers in your workflow.
          ignoreLabels: |
            bot
            ignore-semantic-pull-request
          # If you're using a format for the PR title that differs from the traditional Conventional
          # Commits spec, you can use these options to customize the parsing of the type, scope and
          # subject. The `headerPattern` should contain a regex where the capturing groups in parentheses
          # correspond to the parts listed in `headerPatternCorrespondence`.
          # See: https://github.com/conventional-changelog/conventional-changelog/tree/master/packages/conventional-commits-parser#headerpattern
          headerPattern: '^(\w*)(?:\(([\w$.\-*/ ]*)\))?: (.*)$'
          headerPatternCorrespondence: type, scope, subject

[amannn]

Thanks for sharing the remaining config! Somehow surprisingly, even with this I don't see an issue in my test suite in #293.

I'll keep an eye on this topic in case it comes up again!

[amannn]

I think there's been a regression here, let's continue in this thread: #294