fix(prune): protect dir iteration, add OSError handling, improve test coverage

paperforge/worker/prune.py

@@ -18,7 +18,12 @@ def _collect_orphan_candidates(lit_dir: Path, fresh_keys: set[str]) -> list[dict
     for domain_dir in sorted(lit_dir.iterdir()):
         if not domain_dir.is_dir():
             continue
-        for sub in sorted(domain_dir.iterdir()):
+        try:
+            entries = sorted(domain_dir.iterdir())
+        except OSError as exc:
+            logger.warning("prune: cannot scan %s: %s", domain_dir, exc)
+            continue
+        for sub in entries:
             if not sub.is_dir():
                 continue
             parts = sub.name.split(" - ", 1)
@@ -89,12 +94,14 @@ def prune_orphan_papers(
             ocr = c["ocr_dir"]
             if ocr and ocr.exists():
                 shutil.rmtree(ocr, ignore_errors=True)
-                counts["ocr"] += 1
+                if not ocr.exists():
+                    counts["ocr"] += 1
 
             ws = c["workspace_dir"]
             if ws.exists():
                 shutil.rmtree(ws, ignore_errors=True)
-                counts["workspace"] += 1
+                if not ws.exists():
+                    counts["workspace"] += 1
 
             try:
                 n = delete_paper_vectors(vault, key)

tests/unit/worker/test_prune.py

@@ -8,6 +8,7 @@ import pytest
 
 from paperforge.worker.prune import (
     _collect_orphan_candidates,
+    _resolve_ocr_dir,
     prune_orphan_papers,
 )
 
@@ -65,6 +66,14 @@ class TestCollectOrphanCandidates:
         assert returned_keys == {"key2", "key3"}
 
 
+class TestResolveOcrDir:
+    """_resolve_ocr_dir(vault, key)"""
+
+    def test_resolves_ocr_dir_name_is_key(self, tmp_path: Path) -> None:
+        result = _resolve_ocr_dir(tmp_path, "testkey123")
+        assert result.name == "testkey123"
+
+
 class TestPruneOrphanPapers:
     """prune_orphan_papers(vault, fresh_index, dry_run)"""
 
@@ -124,7 +133,16 @@ class TestPruneOrphanPapers:
         prune_orphan_papers(tmp_path, fresh_index=fresh_index, dry_run=True)
         assert calls == []
 
-    def test_orphan_not_in_fresh_index_is_skipped(self, tmp_path: Path) -> None:
+    def test_empty_fresh_index_does_not_crash(self, tmp_path: Path) -> None:
+        result = prune_orphan_papers(tmp_path, fresh_index={}, dry_run=True)
+        assert result["preview"] == []
+
+    def test_handles_missing_lit_dir(self, tmp_path: Path) -> None:
+        no_lit = tmp_path / "NoLiterature"
+        result = prune_orphan_papers(no_lit, fresh_index={"schema_version": "3", "items": []}, dry_run=False)
+        assert result["deleted"] == []
+
+    def test_active_papers_are_not_deleted(self, tmp_path: Path) -> None:
         lit = tmp_path / "Resources" / "Literature" / "CS"
         ws = lit / "key1 - Active Paper"
         ws.mkdir(parents=True)