logbook: statistics run on a clean detached snapshot of main ([logbook] snapshot), not the integrator's live checkout, which is often mid-batch; a failed run retries on the next tick instead of waiting for main to move

harbor/logbook_auto.py

@@ -109,6 +109,21 @@ def rebuild_stats(project, repo, stats_repo, head):
     return {"classified": stats.get("classified_percent"), "weighted": stats.get("weighted_percent")}
 
 
+def snapshot_at(config, repo, head):
+    """A clean checkout of `head` for the statistics: [logbook] snapshot, a detached worktree of the decomp repository
+    that only this refresher uses. The live checkout is the integrator's and is often mid-batch with uncommitted
+    files, which the statistics tools rightly refuse. Without a snapshot setting, the live checkout is used."""
+    snapshot = config.get("snapshot")
+    if not snapshot:
+        return repo
+    path = Path(snapshot)
+    if not (path / ".git").exists():
+        _git(repo, "worktree", "add", "--detach", str(path), head)
+    else:
+        _git(str(path), "checkout", "--detach", "--force", head)
+    return str(path)
+
+
 def refresh(project, force=False):
     """Bring the Logbook up to main. Returns what changed."""
     config = getattr(project, "logbook", None) or {}
@@ -127,10 +142,12 @@ def refresh(project, force=False):
             out["log_error"] = str(error)
     if stats_repo:
         try:
-            out["stats"] = rebuild_stats(project, repo, stats_repo, head)
+            out["stats"] = rebuild_stats(project, snapshot_at(config, repo, head), stats_repo, head)
         except (RuntimeError, OSError, ValueError, subprocess.TimeoutExpired) as error:
             out["stats_error"] = str(error)
-    _save_state(project, {**state, "last_head": head, "last_refresh": logbook.now(),
+    # a failed statistics run is retried next time, instead of waiting for main to move again
+    done_head = state.get("last_head") if out.get("stats_error") else head
+    _save_state(project, {**state, "last_head": done_head, "last_refresh": logbook.now(),
                           **({"last_error": out.get("stats_error") or out.get("log_error")}
                              if out.get("stats_error") or out.get("log_error") else {"last_error": None})})
     return out