allow passing ignore_emails and ignore_logins

Author: vuillaut

README.md

@@ -7,9 +7,8 @@ Why this is useful
 - Helps projects maintain reproducible credit and citation information
 
 What this repository provides
-- A Python script at `.github/scripts/check_contributors.py` that performs the check
-- A GitHub Actions workflow at `.github/workflows/metadata-check.yml` that runs the script on PR events
-- A sample configuration `.github/contrib-metadata-check.yml` to control behavior
+- A Python script at `check_contributors.py` that performs the check
+- A GitHub Actions bot at `action.yml` that runs the script on PR events
 
 How it works
 - The action runs on PR events. It runs `git log --use-mailmap --format='%aN <%aE>' BASE..HEAD` to collect commit authors, so ensure `.mailmap` is present if you need to unify multiple emails.
@@ -21,17 +20,24 @@ How it works
 
 1. Ensure your repository has `CITATION.cff` and/or `codemeta.json` with author/contributor entries.
 2. Add a `.mailmap` at the repository root if you need to unify alternate emails or names from the git history.
-3. Add this action (or copy the workflow) into your repo; the included workflow triggers on pull requests.
+3. Add this action (or copy the workflow) into your repo in `.github/workflows/contrib-check.yml`; the included workflow triggers on pull requests.
 
 
-### Example `contrib-metadata-check.yml`
+### Example `.github/workflows/contrib-check.yml`
 
 ```yaml
-mode: warn        # "warn" (default) or "fail"
-ignore_emails:
-  - dependabot[bot]@users.noreply.github.com
-ignore_logins:
-  - dependabot[bot]
+jobs:
+  contrib-check:
+    runs-on: ubuntu-latest
+    
+    steps:
+      - name: Contrib metadata check
+        uses: vuillaut/contrib-checker@main
+        with:
+          github_token: ${{ secrets.GITHUB_TOKEN }}
+          mode: warn  # or 'fail' to make the workflow fail when contributors are missing
+          ignore_emails: "dependabot[bot]@users.noreply.github.com,[email protected],[email protected]"
+          ignore_logins: "dependabot[bot],github-actions[bot],ci-bot"
 ```
 
 ## Requirements

action.yml

@@ -7,6 +7,14 @@ inputs:
     description: "Behavior mode: 'warn' (default) or 'fail'. 'warn' posts a comment but doesn't fail the CI, while 'fail' posts a comment and fails the CI if contributors are missing."
     required: false
     default: warn
+  ignore_emails:
+    description: "Comma-separated list of email addresses to ignore when checking contributors (e.g., '[email protected],[email protected]')"
+    required: false
+    default: ""
+  ignore_logins:
+    description: "Comma-separated list of login names to ignore when checking contributors (e.g., 'dependabot[bot],ci-bot')"
+    required: false
+    default: ""
 
 
 runs:
@@ -28,4 +36,8 @@ runs:
 
     - name: Run check script
       run: python check_contributors.py
-      shell: bash
+      shell: bash
+      env:
+        ACTION_MODE: ${{ inputs.mode }}
+        ACTION_IGNORE_EMAILS: ${{ inputs.ignore_emails }}
+        ACTION_IGNORE_LOGINS: ${{ inputs.ignore_logins }}

check_contributors.py

@@ -48,6 +48,22 @@ def _load_config(self):
 					default.update(data)
 			except Exception as e:
 				print(f"Warning loading config: {e}")
+		
+		# Override with action inputs if provided
+		action_mode = os.environ.get('ACTION_MODE')
+		if action_mode:
+			default['mode'] = action_mode
+			
+		action_ignore_emails = os.environ.get('ACTION_IGNORE_EMAILS', '').strip()
+		if action_ignore_emails:
+			emails = [email.strip() for email in action_ignore_emails.split(',') if email.strip()]
+			default['ignore_emails'] = emails
+			
+		action_ignore_logins = os.environ.get('ACTION_IGNORE_LOGINS', '').strip()
+		if action_ignore_logins:
+			logins = [login.strip() for login in action_ignore_logins.split(',') if login.strip()]
+			default['ignore_logins'] = logins
+			
 		return default
 
 	def _run_git(self, args: list) -> str:
@@ -62,6 +78,13 @@ def should_include_contributor(self, contributor: str) -> bool:
 		m = re.search(r'<([^>]+)>', contributor)
 		if m and m.group(1) in self.config.get('ignore_emails', []):
 			return False
+		
+		# Check ignore_logins - extract potential login from contributor string
+		ignore_logins = self.config.get('ignore_logins', [])
+		for login in ignore_logins:
+			if login.lower() in contributor.lower():
+				return False
+		
 		lower = contributor.lower()
 		if 'bot' in lower or 'dependabot' in lower:
 			return False
@@ -177,23 +200,94 @@ def check_contributors(self) -> bool:
 		print(f'Found {len(pr_contribs)} contributors in PR commits')
 		for c in sorted(pr_contribs):
 			print(f'  - {c}')
-		metadata = self.parse_citation_cff() | self.parse_codemeta_json()
+		
+		# Check each metadata file separately
+		citation_cff = self.parse_citation_cff()
+		codemeta_json = self.parse_codemeta_json()
+		
+		print(f'\nChecking CITATION.cff:')
+		if citation_cff:
+			print(f'  Found {len(citation_cff)} contributors in CITATION.cff')
+			for c in sorted(citation_cff):
+				print(f'    - {c}')
+			missing_citation = self.find_missing_contributors(pr_contribs, citation_cff)
+			if missing_citation:
+				print(f'  Missing from CITATION.cff: {sorted(missing_citation)}')
+			else:
+				print('  All PR contributors present in CITATION.cff')
+		else:
+			print('  CITATION.cff not found or empty')
+		
+		print(f'\nChecking codemeta.json:')
+		if codemeta_json:
+			print(f'  Found {len(codemeta_json)} contributors in codemeta.json')
+			for c in sorted(codemeta_json):
+				print(f'    - {c}')
+			missing_codemeta = self.find_missing_contributors(pr_contribs, codemeta_json)
+			if missing_codemeta:
+				print(f'  Missing from codemeta.json: {sorted(missing_codemeta)}')
+			else:
+				print('  All PR contributors present in codemeta.json')
+		else:
+			print('  codemeta.json not found or empty')
+		
+		# Overall check (union of both files)
+		metadata = citation_cff | codemeta_json
 		missing = self.find_missing_contributors(pr_contribs, metadata)
+		
+		print(f'\nOverall result:')
 		if missing:
-			print(f'Missing contributors: {sorted(missing)}')
+			print(f'Missing contributors (not in any metadata file): {sorted(missing)}')
 			self.post_pr_comment(missing)
 			return False if self.config.get('mode') == 'fail' else True
-		print('All PR contributors present in metadata')
+		print('All PR contributors present in at least one metadata file')
 		return True
 
 	def check_all_contributors_in_metadata(self) -> bool:
 		allc = self.get_all_contributors()
-		metadata = self.parse_citation_cff() | self.parse_codemeta_json()
+		print(f'Found {len(allc)} total contributors in repository')
+		for c in sorted(allc):
+			print(f'  - {c}')
+		
+		# Check each metadata file separately
+		citation_cff = self.parse_citation_cff()
+		codemeta_json = self.parse_codemeta_json()
+		
+		print('\nChecking CITATION.cff:')
+		if citation_cff:
+			print(f'  Found {len(citation_cff)} contributors in CITATION.cff')
+			for c in sorted(citation_cff):
+				print(f'    - {c}')
+			missing_citation = self.find_missing_contributors(allc, citation_cff)
+			if missing_citation:
+				print(f'  Missing from CITATION.cff: {sorted(missing_citation)}')
+			else:
+				print('  All repository contributors present in CITATION.cff')
+		else:
+			print('  CITATION.cff not found or empty')
+		
+		print('\nChecking codemeta.json:')
+		if codemeta_json:
+			print(f'  Found {len(codemeta_json)} contributors in codemeta.json')
+			for c in sorted(codemeta_json):
+				print(f'    - {c}')
+			missing_codemeta = self.find_missing_contributors(allc, codemeta_json)
+			if missing_codemeta:
+				print(f'  Missing from codemeta.json: {sorted(missing_codemeta)}')
+			else:
+				print('  All repository contributors present in codemeta.json')
+		else:
+			print('  codemeta.json not found or empty')
+		
+		# Overall check (union of both files)
+		metadata = citation_cff | codemeta_json
 		missing = self.find_missing_contributors(allc, metadata)
+		
+		print('\nOverall result:')
 		if missing:
-			print(f'Missing contributors in repo: {sorted(missing)}')
+			print(f'Missing contributors (not in any metadata file): {sorted(missing)}')
 			return False
-		print('All contributors present in metadata files')
+		print('All contributors present in at least one metadata file')
 		return True