fix(pm): accept digests recorded before metadata was skipped

Facts written by earlier releases hashed desktop metadata as payload. Verification now also accepts the include-everything digest, so an unchanged installed tree is not rejected after upgrading.

pm/build_operations.py

@@ -158,7 +158,7 @@ def verified_tools(names: Sequence[str], *, source_store: Path, target: str,
     from pm.lock import Facts
     from pm.package import StatePackage
     from pm.registry import walk
-    from pm.store import tree_digest
+    from pm.store import digest_matches
 
     if isinstance(names, str):
         raise TypeError("names must be a sequence, not a string")
@@ -186,7 +186,7 @@ def verified_tools(names: Sequence[str], *, source_store: Path, target: str,
         if (not fact or fact.get("entry") != expected
                 or not entry.is_dir() or not entry.resolve().is_relative_to(source_store)
                 or not facts.installed(package.name, version, source_store, identity)
-                or tree_digest(entry) != fact.get("digest")):
+                or not digest_matches(entry, fact.get("digest"))):
             raise InstallError(package.name, "tool source failed verification", "run preparation again")
         _copy_links(package, entry)
         binary = package.binary(entry, target)

pm/cli.py

@@ -425,7 +425,7 @@ def cmd_env(args) -> int:
 
 def cmd_doctor(args) -> int:
     from pm.install import _identity, _installed_location
-    from pm.store import tree_digest
+    from pm.store import digest_matches
 
     lockfile = _lockfile()
     facts = _facts()
@@ -464,7 +464,7 @@ def cmd_doctor(args) -> int:
             bad += 1
             continue
         recorded = fact.get("digest")
-        if recorded is not None and tree_digest(entry) != recorded:
+        if recorded is not None and not digest_matches(entry, recorded):
             # Doctor is the expensive-path tool: re-hash the realized
             # bytes. Boot checks stay O(1) json compares.
             print(f"✗ {name}: realized bytes do not match recorded digest")

pm/install.py

@@ -20,7 +20,7 @@ from pm.lock import Facts, Lockfile
 from pm.package import InstallError, Package, Runner, StatePackage, compose_env
 from pm.plugin_inputs import Candidates, Members, PluginInput, Selection, StagedUpdate
 from pm.registry import get_package, walk
-from pm.store import Store, current_target, merge_tree, tree_digest
+from pm.store import Store, current_target, digest_matches, merge_tree, tree_digest
 
 LOG = logging.getLogger(__name__)
 
@@ -300,7 +300,7 @@ def _entry_verified(package: Package, fact: dict, store: Store, target: str) ->
     """Explicit installs re-check realized bytes; startup keeps its cheap facts check."""
     entry = store.entry(fact["entry"])
     try:
-        return not package.verify(entry, target) and fact.get("digest") == tree_digest(entry)
+        return not package.verify(entry, target) and digest_matches(entry, fact.get("digest"))
     except OSError:
         return False
 
@@ -416,7 +416,7 @@ def _copy_verified_source(package, lockfile, copy_from, staged, version, target)
             or not _entry_verified(package, source, source_store, target)):
         raise InstallError(package.name, "bundled copy source failed verification")
     shutil.copytree(source_store.entry(source["entry"]), staged, symlinks=True)
-    if tree_digest(staged) != source["digest"]:
+    if not digest_matches(staged, source["digest"]):
         raise InstallError(package.name, "copied bytes do not match the bundled source")
 
 

pm/store.py

@@ -326,6 +326,16 @@ def merge_tree(src: Path, dst: Path) -> None:
 
 
 def tree_digest(root: Path) -> str:
+    return _tree_digest(root, skip_metadata=True)
+
+
+def digest_matches(root: Path, recorded: str | None) -> bool:
+    """Facts recorded before desktop metadata was skipped hashed it as payload."""
+    return recorded is not None and (
+        tree_digest(root) == recorded or _tree_digest(root, skip_metadata=False) == recorded)
+
+
+def _tree_digest(root: Path, *, skip_metadata: bool) -> str:
     """Deterministic sha256 over a directory tree: walk every file, sort
     by posix relpath, hash `relpath\\0<content>` per entry. No mtimes, no
     mode bits. Symlinks contribute their LINK TARGET TEXT (os.readlink),
@@ -350,7 +360,7 @@ def tree_digest(root: Path) -> str:
         dirnames[:] = descend
         for fname in filenames:
             path = Path(dirpath) / fname
-            if _is_os_metadata(path):
+            if skip_metadata and _is_os_metadata(path):
                 continue
             files.append((path.relative_to(root).as_posix(), path))
     files.sort(key=lambda item: item[0])

tests/pm/test_store.py

@@ -1,5 +1,6 @@
 """Desktop metadata must not hide package payloads or obstruct extraction (#128588)."""
 
+import hashlib
 import io
 import tarfile
 from pathlib import Path
@@ -65,6 +66,38 @@ def test_metadata_named_directories_links_and_executables_are_hashed(tmp_path: P
     assert (wrapper / "bin").is_dir()
 
 
+@pytest.mark.platforms("any")
+def test_facts_recorded_with_metadata_in_the_digest_still_verify(tmp_path: Path):
+    from pm import build_operations
+    from pm.lock import Facts, Lockfile
+    from pm.package import InstallError
+    from pm.registry import get_package
+    from pm.store import current_target
+
+    target, store = current_target(), tmp_path / "tools"
+    lock = Lockfile(tmp_path / "lock.json")
+    lock.set_pin("python", "1.0", {target: {"url": "https://invalid.test/python", "sha256": "a" * 64}})
+    package = get_package("python")
+    entry = store / package.store_entry("1.0", target)
+    binary = package.binary(entry, target)
+    assert binary is not None
+    binary.parent.mkdir(parents=True)
+    binary.write_bytes(b"runtime")
+    (entry / ".DS_Store").write_bytes(b"desktop metadata")
+    # Earlier releases hashed every regular file, metadata included.
+    legacy = hashlib.sha256()
+    for path in sorted((p for p in entry.rglob("*") if p.is_file()), key=lambda p: p.relative_to(entry).as_posix()):
+        legacy.update(path.relative_to(entry).as_posix().encode() + b"\0" + path.read_bytes())
+    Facts(store / "facts.json").record("python", "1.0", entry.name, package.env(entry, target), store,
+                                       target=target, artifacts=["a" * 64], digest=legacy.hexdigest())
+
+    selected = build_operations.verified_tools(["python"], source_store=store, target=target, lock=lock)
+    assert selected.entries["python"].binary == binary
+    binary.write_bytes(b"changed runtime")
+    with pytest.raises(InstallError, match="source failed verification"):
+        build_operations.verified_tools(["python"], source_store=store, target=target, lock=lock)
+
+
 @pytest.mark.platforms("any")
 def test_metadata_named_payload_collision_refuses_before_any_move(tmp_path: Path):
     tree = tmp_path / "tree"