From ce48dee42c42cc6486846853f3a9d735415c6429 Mon Sep 17 00:00:00 2001
From: sladro <sladro@163.com>
Date: Thu, 2 Apr 2026 14:07:59 +0800
Subject: [PATCH] Handle stray artifact files during restore

---
 engine/artifact_manager.py     | 51 ++++++++++++++++++++++++++++++++--
 tests/test_artifact_manager.py | 28 +++++++++++++++++++
 2 files changed, 76 insertions(+), 3 deletions(-)

diff --git a/engine/artifact_manager.py b/engine/artifact_manager.py
index 7034458..e344cd1 100644
--- a/engine/artifact_manager.py
+++ b/engine/artifact_manager.py
@@ -13,6 +13,51 @@ from engine.models import BaselineSnapshot, TaskSpec
 class ArtifactManager:
     task: TaskSpec
 
+    def _is_excluded(self, relative_path: str) -> bool:
+        return any(fnmatch(relative_path, exclude) for exclude in self.task.artifacts.exclude)
+
+    def _artifact_roots(self) -> list[Path]:
+        root_dir = self.task.root_dir
+        roots: list[Path] = []
+        for pattern in self.task.artifacts.include:
+            normalized_pattern = pattern.replace("\\", "/")
+            segments = [segment for segment in normalized_pattern.split("/") if segment and segment != "."]
+            prefix: list[str] = []
+            for segment in segments:
+                if any(char in segment for char in "*?["):
+                    break
+                prefix.append(segment)
+            if prefix:
+                roots.append(root_dir.joinpath(*prefix))
+                continue
+            roots.append(root_dir)
+        minimal_roots: list[Path] = []
+        for root in sorted(set(roots), key=lambda candidate: (len(candidate.parts), candidate.as_posix())):
+            if any(existing == root or existing in root.parents for existing in minimal_roots):
+                continue
+            minimal_roots.append(root)
+        return minimal_roots
+
+    def _current_paths_within_roots(self) -> list[Path]:
+        root_dir = self.task.root_dir
+        discovered: set[Path] = set()
+        for root in self._artifact_roots():
+            if not root.exists():
+                continue
+            if root.is_file():
+                relative_path = root.relative_to(root_dir).as_posix()
+                if not self._is_excluded(relative_path):
+                    discovered.add(root)
+                continue
+            for path in root.rglob("*"):
+                if not path.is_file():
+                    continue
+                relative_path = path.relative_to(root_dir).as_posix()
+                if self._is_excluded(relative_path):
+                    continue
+                discovered.add(path)
+        return sorted(discovered)
+
     def resolve_paths(self) -> list[Path]:
         root_dir = self.task.root_dir
         resolved: set[Path] = set()
@@ -21,7 +66,7 @@ class ArtifactManager:
                 if not path.is_file():
                     continue
                 relative_path = path.relative_to(root_dir).as_posix()
-                if any(fnmatch(relative_path, exclude) for exclude in self.task.artifacts.exclude):
+                if self._is_excluded(relative_path):
                     continue
                 resolved.add(path)
         return sorted(resolved)
@@ -37,7 +82,7 @@ class ArtifactManager:
         return BaselineSnapshot(file_contents=file_contents, file_hashes=file_hashes)
 
     def restore(self, snapshot: BaselineSnapshot) -> None:
-        current_paths = set(self.resolve_paths())
+        current_paths = set(self._current_paths_within_roots())
         snapshot_paths = set(snapshot.file_contents)
         for path in current_paths - snapshot_paths:
             path.unlink()
@@ -49,7 +94,7 @@ class ArtifactManager:
     def diff_summary(self, snapshot: BaselineSnapshot) -> str:
         lines: list[str] = []
         current_contents: dict[Path, str] = {}
-        for path in self.resolve_paths():
+        for path in self._current_paths_within_roots():
             with path.open("r", encoding="utf-8", newline="") as handle:
                 current_contents[path] = handle.read()
         all_paths = sorted(set(snapshot.file_contents) | set(current_contents))
diff --git a/tests/test_artifact_manager.py b/tests/test_artifact_manager.py
index a7ca848..c4be45d 100644
--- a/tests/test_artifact_manager.py
+++ b/tests/test_artifact_manager.py
@@ -107,6 +107,34 @@ class ArtifactManagerTest(unittest.TestCase):
             self.assertIn("-before", summary)
             self.assertIn("+after", summary)
 
+    def test_diff_summary_and_restore_handle_stray_file_outside_include_glob(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            artifact_dir = root / "artifacts"
+            artifact_dir.mkdir()
+            target = artifact_dir / "sample.md"
+            target.write_text("baseline\n", encoding="utf-8")
+            manager = ArtifactManager(make_task(root))
+            snapshot = manager.snapshot()
+
+            target.unlink()
+            stray = artifact_dir / "archive" / "sample.txt"
+            stray.parent.mkdir()
+            stray.write_text("renamed\n", encoding="utf-8")
+
+            summary = manager.diff_summary(snapshot)
+
+            self.assertIn("artifacts/sample.md (before)", summary)
+            self.assertIn("artifacts/archive/sample.txt (after)", summary)
+            self.assertIn("-baseline", summary)
+            self.assertIn("+renamed", summary)
+
+            manager.restore(snapshot)
+
+            self.assertFalse(stray.exists())
+            self.assertTrue(target.exists())
+            self.assertEqual(target.read_text(encoding="utf-8"), "baseline\n")
+
 
 if __name__ == "__main__":
     unittest.main()