test(infra): bind historical evidence to its original snapshot

Keep historical RC03 hashes independent of later source changes while verifying every current manifest item against its committed blob.

yuqei-infra/tests/test_release_candidate_manifest_tools.py

@@ -42,9 +42,9 @@ def _git(repo: Path, *args: str) -> str:
     return completed.stdout.strip()
 
 
-def _git_blob(repo: Path, path: str) -> bytes:
+def _git_blob(repo: Path, path: str, revision: str = "HEAD") -> bytes:
     completed = subprocess.run(
-        ["git", "-C", str(repo), "cat-file", "blob", f"HEAD:{path}"],
+        ["git", "-C", str(repo), "cat-file", "blob", f"{revision}:{path}"],
         check=True,
         capture_output=True,
     )
@@ -411,53 +411,6 @@ def test_manifest_is_deterministic_and_summarizes_complete_committed_graph(tmp_p
         "**/*.sqlite3",
     } <= dockerignore_rules
 
-    current_index = json.loads(
-        (
-            REPO_ROOT
-            / "docs/qa/multiscene-closeout-20260910/rc03-current-candidate-evidence-index-v3.json"
-        ).read_text(encoding="utf-8")
-    )
-    assert current_index["validation"]["self_contained"] is True
-    assert current_index["validation"]["predecessor_required_for_interpretation"] is False
-    assert current_index["validation"]["listed_file_count"] == 54
-    assert current_index["validation"]["accepted_current_evidence_count"] == 50
-    assert current_index["validation"]["excluded_invalid_lineage_count"] == 4
-    assert all(item["normative_for_v3"] is False for item in current_index["lineage"])
-    indexed_paths = [item["path"] for item in current_index["files"]]
-    assert len(indexed_paths) == len(set(indexed_paths)) == 54
-    for item in current_index["files"]:
-        evidence = REPO_ROOT / item["path"]
-        category, _, _ = ownership.classify_path(item["path"], size=item["size_bytes"])
-        if category == "generated_evidence_retain_exclude":
-            # Raw and invalid-lineage QA outputs are intentionally retained outside
-            # the immutable source candidate.  When they remain in a development
-            # workspace, still verify that the compact index recorded their bytes
-            # correctly; a clean release checkout must not require them to exist.
-            if not evidence.exists():
-                continue
-            evidence_bytes = evidence.read_bytes()
-        else:
-            assert category in {"release_source_include", "release_test_include"}
-            assert evidence.is_file()
-            # Verify the immutable Git bytes instead of checkout bytes: Windows may
-            # materialize text files with CRLF.  The RC-03 index was created from
-            # worktree bytes, so accept only an exact LF/CRLF rendering of the same
-            # immutable blob; the candidate manifest/archive bind the raw blob.
-            evidence_bytes = _git_blob(REPO_ROOT, item["path"])
-            variants = _text_eol_variants(evidence_bytes)
-            assert any(
-                len(candidate) == item["size_bytes"]
-                and hashlib.sha256(candidate).hexdigest() == item["sha256"]
-                for candidate in variants
-            )
-            if item["path"].endswith(".json"):
-                json.loads(evidence_bytes)
-            continue
-        assert len(evidence_bytes) == item["size_bytes"]
-        assert hashlib.sha256(evidence_bytes).hexdigest() == item["sha256"]
-        if item["path"].endswith(".json"):
-            json.loads(evidence_bytes)
-
     repo, commit = _candidate_repo(tmp_path)
     output = tmp_path / "manifest.json"
 
@@ -474,6 +427,10 @@ def test_manifest_is_deterministic_and_summarizes_complete_committed_graph(tmp_p
     assert first["summaries"]["migrations"][0]["heads"] == ["contract-head"]
     assert first["summaries"]["migrations"][1]["heads"] == ["ai-head"]
     recorded_paths = {item["path"] for item in first["files"]["items"]}
+    for item in first["files"]["items"]:
+        content = _git_blob(repo, item["path"], commit)
+        assert item["size"] == len(content)
+        assert item["sha256"] == hashlib.sha256(content).hexdigest()
     assert set(manifest.RELEASE_CONTROL_PATHS) <= recorded_paths
     configuration_paths = {item["path"] for item in first["summaries"]["configuration"]}
     assert set(manifest.RELEASE_CONTROL_PATHS) <= configuration_paths
@@ -490,6 +447,49 @@ def test_manifest_is_deterministic_and_summarizes_complete_committed_graph(tmp_p
     assert "No image was built or assigned a digest" in first["claims_not_made"][0]
 
 
+def test_historical_rc03_index_matches_original_committed_snapshot() -> None:
+    if _git(REPO_ROOT, "rev-parse", "--is-shallow-repository") == "true":
+        pytest.skip("historical RC03 byte verification requires complete Git history")
+    index_path = "docs/qa/multiscene-closeout-20260910/rc03-current-candidate-evidence-index-v3.json"
+    index_commit = _git(REPO_ROOT, "log", "-1", "--diff-filter=A", "--format=%H", "--", index_path)
+    assert index_commit
+    historical_index = json.loads((REPO_ROOT / index_path).read_text(encoding="utf-8"))
+    assert historical_index == json.loads(_git_blob(REPO_ROOT, index_path, index_commit))
+    assert historical_index["release_claim"] is False
+    assert historical_index["validation"]["self_contained"] is True
+    assert historical_index["validation"]["predecessor_required_for_interpretation"] is False
+    assert historical_index["validation"]["listed_file_count"] == 54
+    assert historical_index["validation"]["accepted_current_evidence_count"] == 50
+    assert historical_index["validation"]["excluded_invalid_lineage_count"] == 4
+    assert all(item["normative_for_v3"] is False for item in historical_index["lineage"])
+    indexed_paths = [item["path"] for item in historical_index["files"]]
+    assert len(indexed_paths) == len(set(indexed_paths)) == 54
+    for item in historical_index["files"]:
+        evidence = REPO_ROOT / item["path"]
+        category, _, _ = ownership.classify_path(item["path"], size=item["size_bytes"])
+        if category == "generated_evidence_retain_exclude":
+            if not evidence.exists():
+                continue
+            evidence_bytes = evidence.read_bytes()
+        else:
+            assert category in {"release_source_include", "release_test_include"}
+            evidence_bytes = _git_blob(REPO_ROOT, item["path"], index_commit)
+            # The historical index recorded worktree bytes; bind its own source snapshot, not today's HEAD.
+            variants = _text_eol_variants(evidence_bytes)
+            assert any(
+                len(candidate) == item["size_bytes"]
+                and hashlib.sha256(candidate).hexdigest() == item["sha256"]
+                for candidate in variants
+            )
+            if item["path"].endswith(".json"):
+                json.loads(evidence_bytes)
+            continue
+        assert len(evidence_bytes) == item["size_bytes"]
+        assert hashlib.sha256(evidence_bytes).hexdigest() == item["sha256"]
+        if item["path"].endswith(".json"):
+            json.loads(evidence_bytes)
+
+
 def test_manifest_refuses_ambiguous_dirty_missing_and_secret_candidates(tmp_path: Path) -> None:
     repo, commit = _candidate_repo(tmp_path)
     external_output = tmp_path / "manifest.json"