BUILD-10739 Cache cleanup with list and delete modes

[mikolaj-matuszny-ext-sonarsource]

Summary

Rewrites the cache cleanup script to use a two-phase approach with list and delete modes:

• List mode (no branch): Lists all cache entries for the repo, optionally filtered by key prefix. Helps users discover what's cached before deciding what to delete.
• Delete mode (branch provided): Deletes matching cache entries with optional dry-run preview.

Changes

• scripts/cleanup-cache.sh: Replaced aws s3 rm --include/--exclude with aws s3api list-objects-v2 + grep filtering + aws s3api delete-objects batch deletion (up to 1000 objects per API call)
• cleanup/action.yml: Made branch input optional to support list mode
• .github/workflows/cleanup-cache.yml: Updated inputs to match (branch optional, descriptions updated)
• README.md: Rewrote Cache Cleanup section with usage table showing all scenarios
• .github/workflows/test-action.yml: Removed test-cleanup-dry-run job (tested manually, requires IAM changes to work in CI)

How users interact

Scenario	Branch	Key	Dry-run
List all cache entries	(empty)	(empty)	n/a
List entries matching a key	(empty)	sccache-Linux-	n/a
Preview what would be deleted	feature/my-branch	(optional)	true
Delete cache for a branch	feature/my-branch	(optional)	false

Dependencies

Requires IAM policy changes in github-runners-infra PR to grant s3:ListBucket and s3:DeleteObject for workflow_dispatch from protected branches.

Test plan

• Test list mode locally: S3_BUCKET=sonarsource-s3-cache-dev-bucket GITHUB_REPOSITORY=SonarSource/gh-action_cache bash scripts/cleanup-cache.sh
• Test list with key filter: add CLEANUP_KEY=git-clean-test
• Test dry-run delete: add CLEANUP_BRANCH=feat/... DRY_RUN=true
• Test actual delete after IAM changes are merged
• Verify CI passes (existing cache tests unaffected)

🤖 Generated with Claude Code

[sonar-review-alpha]

Summary

Rewrites cache cleanup from simple aws s3 rm --include/--exclude to a two-phase API approach: list all objects with pagination via aws s3api list-objects-v2, filter by branch/key using jq and grep, then batch-delete up to 1000 objects per call with aws s3api delete-objects. Adds list mode (no branch) for cache discovery. Updates action inputs and workflow to reflect the two modes.

What reviewers should know

Start with: scripts/cleanup-cache.sh (core logic rewrite) — the filtering logic (grep escaping via escape_grep()), pagination handling, and batch deletion are new.

Key changes to review:

• The jq operations for filtering and building delete payloads — verify correctness of array manipulation and JSON structure
• The escape_grep() function ensures grep patterns don't fail on special characters in branch/key names
• Pagination loop handles NextContinuationToken to fetch >1000 objects
• Batch deletion: script accumulates matched keys into a temp file, deletes in 1000-object chunks, and tracks progress
• Output now shows size (human-readable via format_size()) and date for each object

Gotchas:

• Branch matching still searches for both bare (feature/x) and full ref (refs/heads/feature/x) forms to cover PR and push events
• DRY_RUN only applies in delete mode; list mode always shows what exists without deleting
• Size/date come from S3 object metadata (Size, LastModified fields)
• The test-cleanup-dry-run job was removed (author notes it needs manual testing post-IAM merge)

---

• Generate Walkthrough
• Generate Diagram

🗣️ Give feedback

[sonar-review-alpha]

Conclusion: The script rewrite is solid and the new list mode is a useful addition. One logic duplication worth cleaning up, but nothing blocking.

🗣️ Give feedback

[sonar-review-alpha]

The pagination bug is the main concern — it could cause silent data loss in delete mode on busy repos. The table duplication is straightforward to fix with a helper function.

Conclusion: Two issues need fixing before merge: a real bug (truncated S3 listing) and new logic duplication introduced by this commit. The key-filter duplication flagged in the previous review (comment #2953033147) also remains unaddressed.

🗣️ Give feedback

[sonarqube-cloud-us]

SonarQube reviewer guide

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[sonarqubecloud]

SonarQube reviewer guide

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud