fix resolving tag to hash

Author: PiotrKorkus

scripts/tooling/cli/misc/html_report.py

@@ -22,7 +22,7 @@
 
 from jinja2 import Environment, FileSystemLoader, select_autoescape
 
-from scripts.tooling.lib.github import fetch_compare
+from scripts.tooling.lib.github import fetch_compare, resolve_ref_sha
 from scripts.tooling.lib.known_good import KnownGood, load_known_good
 
 _LOG = logging.getLogger(__name__)
@@ -73,9 +73,22 @@ def _collect_entries(known_good: KnownGood) -> list[dict[str, Any]]:
 
 def _enrich_with_compare_data(entries: list[dict[str, Any]], token: str) -> None:
     for entry in entries:
-        if not entry.get("owner_repo") or not entry.get("hash") or entry.get("version"):
+        if not entry.get("owner_repo"):
             continue
-        result = fetch_compare(entry["owner_repo"], entry["hash"], entry["branch"], token)
+
+        # Modules are pinned either by commit hash or by release version. A version
+        # pin (e.g. "0.2.9") is resolved to its tag's commit so it can be compared
+        # against the branch HEAD just like a hash pin.
+        base_ref = entry.get("hash")
+        if not base_ref and entry.get("version"):
+            base_ref = resolve_ref_sha(entry["owner_repo"], entry["version"], token)
+            if base_ref:
+                # Surface the resolved commit as the pinned hash for display/linking.
+                entry["hash"] = base_ref
+        if not base_ref:
+            continue
+
+        result = fetch_compare(entry["owner_repo"], base_ref, entry["branch"], token)
         if result:
             entry["current_hash"] = result.head_sha
             entry["behind_by"] = result.ahead_by

scripts/tooling/lib/github.py

@@ -23,6 +23,12 @@
 _LOG = logging.getLogger(__name__)
 
 
+def _get_repo(owner_repo: str, token: Optional[str]):
+    """Return the PyGithub ``Repository`` for *owner_repo*, authenticated if *token* is set."""
+    gh = Github(token) if token else Github()
+    return gh.get_repo(owner_repo)
+
+
 @dataclass
 class CompareResult:
     """Result of comparing a pinned commit against a branch HEAD."""
@@ -54,8 +60,7 @@ def fetch_compare(
                Without a token requests are unauthenticated (60 req/h rate limit).
     """
     try:
-        gh = Github(token) if token else Github()
-        repo = gh.get_repo(owner_repo)
+        repo = _get_repo(owner_repo, token)
         comparison = repo.compare(base_hash, branch)
         commits = list(comparison.commits)
         head_sha = commits[-1].sha if commits else base_hash
@@ -69,3 +74,38 @@ def fetch_compare(
     except Exception as exc:  # noqa: BLE001
         _LOG.debug("Unexpected error comparing %s: %s", owner_repo, exc)
     return None
+
+
+def resolve_ref_sha(
+    owner_repo: str,
+    ref: str,
+    token: Optional[str] = None,
+) -> Optional[str]:
+    """Resolve a git ref (tag, branch or SHA) to a commit SHA.
+
+    Modules pinned by ``version`` store the bare release string (e.g. ``"0.2.9"``)
+    while the git tag is usually prefixed with ``v`` (``"v0.2.9"``). Both spellings
+    are tried so a version pin resolves to the tag's commit.
+
+    Args:
+        owner_repo: GitHub ``owner/repo`` slug (e.g. ``"eclipse-score/baselibs"``).
+        ref: The tag/branch/version string to resolve.
+        token: Optional GitHub PAT or ``GITHUB_TOKEN``.
+
+    Returns:
+        The resolved commit SHA, or ``None`` if no candidate ref could be resolved.
+    """
+    candidates = [ref]
+    if not ref.startswith("v"):
+        candidates.append(f"v{ref}")
+    try:
+        repo = _get_repo(owner_repo, token)
+        for candidate in candidates:
+            try:
+                return repo.get_commit(candidate).sha
+            except GithubException:
+                continue
+        _LOG.debug("Could not resolve ref %r for %s (tried %s)", ref, owner_repo, candidates)
+    except Exception as exc:  # noqa: BLE001
+        _LOG.debug("Unexpected error resolving ref %s for %s: %s", ref, owner_repo, exc)
+    return None