refactor: check packs through a single path

scripts/generate_pack.py

@@ -50,7 +50,6 @@ from common import (
     require_yaml,
     resolution_is_hash_exact,
     resolve_local_file,
-    safe_extract_zip,
     yaml_load,
 )
 import region as region_mod
@@ -2821,9 +2820,13 @@ def _pack_output_lock(output_dir: str, exclusive: bool = True):
 
 
 def _run_verify_packs(args):
-    """Extract each pack and verify file paths + hashes."""
+    """Verify each pack in the output directory against its platform.
-    import shutil
 
+    Delegates to verify_pack_against_platform so the command answers the same
+    question the pipeline asks: baseline files, core extras and structure. A
+    second implementation here answered a narrower one and reported OK while
+    the core extras went unchecked.
+    """
     with open(args.db) as f:
         verify_db = json.load(f)
 
@@ -2837,226 +2840,50 @@ def _run_verify_packs(args):
     all_ok = True
     verify_regions = getattr(args, "regions", None)
     verify_profiles = load_emulator_profiles(args.emulators_dir)
-    verify_zip_contents = build_zip_contents_index(verify_db)
     verify_data_registry = load_data_dir_registry(args.platforms_dir)
+
     for platform_name in platforms:
         config = load_platform_config(platform_name, args.platforms_dir)
         display = config.get("platform", platform_name).replace(" ", "_")
-        base_dest = config.get("base_destination", "")
-        mode = config.get("verification_mode", "existence")
-        systems = config.get("systems", {})
 
-        # Find ZIP
         zip_path = None
         if os.path.isdir(args.output_dir):
-            for f in os.listdir(args.output_dir):
+            for entry in sorted(os.listdir(args.output_dir)):
-                if f.endswith("_BIOS_Pack.zip") and display in f:
+                if entry.endswith("_BIOS_Pack.zip") and display in entry:
-                    zip_path = os.path.join(args.output_dir, f)
+                    zip_path = os.path.join(args.output_dir, entry)
                     break
         if not zip_path:
             print(f"  {platform_name}: SKIP (no pack in {args.output_dir})")
             continue
 
-        # Detect source from ZIP filename
+        if _narrows_contents(os.path.basename(zip_path)):
-        pack_source = "full"
+            print(f"  {platform_name}: SKIP (narrowed variant)")
-        if zip_path:
-            bn = os.path.basename(zip_path)
-            if "_Platform_" in bn:
-                pack_source = "platform"
-            elif "_Truth_" in bn:
-                pack_source = "truth"
-
-        if pack_source == "truth":
-            print(f"  {platform_name}: OK (truth pack, verified by hash integrity)")
             continue
 
-        extract_dir = os.path.join("tmp", "verify_packs", platform_name)
+        result = verify_pack_against_platform(
-        os.makedirs(extract_dir, exist_ok=True)
+            zip_path,
-        try:
+            platform_name,
-            # Extract
+            args.platforms_dir,
-            safe_extract_zip(zip_path, extract_dir)
+            db=verify_db,
-
+            emulators_dir=args.emulators_dir,
-            # Auto-detect flat vs nested extraction
+            emu_profiles=verify_profiles,
-            is_flat = bool(base_dest) and not os.path.isdir(
+            regions=verify_regions,
-                os.path.join(extract_dir, base_dest)
+            data_registry=verify_data_registry,
-            )
+        )
-
+        ok, errors = result[0], result[3]
-            region_drops: set[str] = set()
+        bl_checked, bl_present = result[4], result[5]
-            if verify_regions:
+        core_checked, core_present = result[6], result[7]
-                region_index = region_mod.build_region_index(verify_profiles)
+        counts = (
-                region_groups: dict[str, list[tuple[str, str]]] = {}
+            f"{bl_present}/{bl_checked} baseline, "
-                for sys_id, sys_data in systems.items():
+            f"{core_present}/{core_checked} cores"
-                    members = region_groups.setdefault(sys_id, [])
+        )
-                    for fe in sys_data.get("files", []):
+        if ok:
-                        d = _sanitize_path(
+            print(f"  {platform_name}: OK ({counts})")
-                            fe.get("destination", fe.get("name", ""))
+        else:
-                        )
+            print(f"  {platform_name}: FAIL ({counts}, {len(errors)} errors)")
-                        if d:
+            for err in errors[:5]:
-                            members.append((d, fe.get("name", "")))
+                print(f"    {err}")
-                # Core extras join the groups so a platform file dropped in
+            all_ok = False
-                # favour of a better-ranked core file is not reported missing.
-                from verify import find_undeclared_files as _fud
-
-                for u in _fud(
-                    config, args.emulators_dir, verify_db, verify_profiles
-                ):
-                    if not u.get("in_repo"):
-                        continue
-                    raw = u.get("path") or u["name"]
-                    d = _sanitize_path(
-                        f"{raw}{u['name']}" if raw.endswith("/") else raw
-                    )
-                    if not d:
-                        continue
-                    systems_for_extra = (
-                        [u["system"]] if u.get("system") else u.get("systems", [])
-                    ) or ["_extras"]
-                    for sid in systems_for_extra:
-                        variant = u.get("variant_group")
-                        group_id = f"{sid}:variant:{variant}" if variant else sid
-                        region_groups.setdefault(group_id, []).append((d, u["name"]))
-                region_drops = region_mod.resolve_region_drops(
-                    region_groups, region_index, verify_regions
-                )
-
-            missing = []
-            hash_fail = []
-            excluded = []
-            ok = 0
-            for sys_id, sys_data in systems.items():
-                for fe in sys_data.get("files", []):
-                    dest = fe.get("destination", fe.get("name", ""))
-                    if not dest:
-                        continue
-                    if region_drops and _sanitize_path(dest) in region_drops:
-                        continue
-                    fp = (
-                        os.path.join(extract_dir, base_dest, dest)
-                        if base_dest and not is_flat
-                        else os.path.join(extract_dir, dest)
-                    )
-                    # Case-insensitive fallback
-                    if not os.path.exists(fp):
-                        parent = os.path.dirname(fp)
-                        bn = os.path.basename(fp)
-                        if os.path.isdir(parent):
-                            for e in os.listdir(parent):
-                                if e.lower() == bn.lower():
-                                    fp = os.path.join(parent, e)
-                                    break
-                    if not os.path.exists(fp):
-                        # File-vs-directory conflict: upstream declares both
-                        # SGB1.sfc and SGB1.sfc/program.rom; the builder can
-                        # only ship one shape, the other was skipped
-                        ancestor = os.path.dirname(fp)
-                        conflicted = False
-                        while len(ancestor) > len(extract_dir):
-                            if os.path.isfile(ancestor):
-                                conflicted = True
-                                break
-                            ancestor = os.path.dirname(ancestor)
-                        if conflicted or os.path.isdir(fp):
-                            ok += 1
-                            continue
-                        if _intentional_hash_exclusion(
-                            [fe],
-                            verify_db,
-                            args.bios_dir,
-                            verify_zip_contents,
-                            data_dir_registry=verify_data_registry,
-                            verification_mode=mode,
-                        ):
-                            excluded.append(f"{sys_id}: {dest}")
-                            continue
-                        missing.append(f"{sys_id}: {dest}")
-                        continue
-                    if mode == "existence":
-                        ok += 1
-                        continue
-                    if mode == "sha1":
-                        expected = fe.get("sha1", "")
-                        if not expected:
-                            ok += 1
-                            continue
-                        with open(fp, "rb") as source:
-                            actual = hashlib.sha1(source.read()).hexdigest()
-                        if actual == expected.lower():
-                            ok += 1
-                        else:
-                            hash_fail.append(f"{sys_id}: {dest}")
-                        continue
-                    # MD5
-                    expected_md5 = fe.get("md5", "")
-                    if not expected_md5:
-                        ok += 1
-                        continue
-                    md5_list = [
-                        m.strip().lower() for m in expected_md5.split(",") if m.strip()
-                    ]
-                    with open(fp, "rb") as source:
-                        actual_md5 = hashlib.md5(source.read()).hexdigest()
-                    if actual_md5 in md5_list or any(
-                        actual_md5.startswith(m) for m in md5_list if len(m) < 32
-                    ):
-                        ok += 1
-                        continue
-                    # ZIP inner content
-                    if fp.endswith(".zip"):
-                        ok += 1  # inner content verified by verify.py
-                        continue
-                    # Path collision
-                    bn = os.path.basename(dest)
-                    collision = (
-                        sum(
-                            1
-                            for sd in systems.values()
-                            for ff in sd.get("files", [])
-                            if os.path.basename(
-                                ff.get("destination", ff.get("name", "")) or ""
-                            )
-                            == bn
-                        )
-                        > 1
-                    )
-                    if collision:
-                        ok += 1
-                    elif not _repo_satisfies_declaration([fe], verify_db, "md5"):
-                        # No repo file matches the declared hash: shipped file
-                        # is the best effort, the gap is a data issue reported
-                        # by verify.py, not a pack error
-                        ok += 1
-                    else:
-                        hash_fail.append(f"{sys_id}: {dest}")
-
-            total = sum(
-                len(
-                    [
-                        f
-                        for f in s.get("files", [])
-                        if f.get("destination", f.get("name", ""))
-                    ]
-                )
-                for s in systems.values()
-            )
-            if missing or hash_fail:
-                print(
-                    f"  {platform_name}: FAIL ({len(missing)} missing, {len(hash_fail)} hash errors / {total})"
-                )
-                for m in missing[:5]:
-                    print(f"    MISSING: {m}")
-                for h in hash_fail[:5]:
-                    print(f"    HASH: {h}")
-                all_ok = False
-            else:
-                exclusion_note = (
-                    f", {len(excluded)} unsafe excluded" if excluded else ""
-                )
-                print(
-                    f"  {platform_name}: OK ({ok}/{total} verified"
-                    f"{exclusion_note})"
-                )
-        finally:
-            shutil.rmtree(extract_dir, ignore_errors=True)
 
     if not all_ok:
         sys.exit(1)
@@ -4293,7 +4120,7 @@ def verify_pack_against_platform(
     emu_profiles: dict | None = None,
     regions: list[str] | None = None,
     data_registry: dict | None = None,
-) -> tuple[bool, int, int, list[str], int, int, int, int, int, int]:
+) -> tuple[bool, int, int, list[str], int, int, int, int, int]:
     """Verify a pack ZIP against its platform config and core requirements.
 
     A region priority list narrows the expectation to what the builder would

tests/test_e2e.py

@@ -5659,6 +5659,32 @@ struct BurnDriver BurnDrvneogeo = {
         )
         self.assertTrue(checked.issubset(self._names_in(out)))
 
+    def test_verify_pack_return_matches_its_annotation(self):
+        import inspect
+        import typing
+
+        from generate_pack import generate_pack, verify_pack_against_platform
+
+        self._create_region_fixtures()
+        profiles = self._region_profiles()
+        out = generate_pack(
+            "regionplat",
+            self.platforms_dir,
+            self.db,
+            self.bios_dir,
+            self.root,
+            emu_profiles=profiles,
+        )
+        result = verify_pack_against_platform(
+            out, "regionplat", self.platforms_dir, db=self.db, emu_profiles=profiles
+        )
+        annotated = typing.get_args(
+            inspect.signature(
+                verify_pack_against_platform, eval_str=True
+            ).return_annotation
+        )
+        self.assertEqual(len(result), len(annotated))
+
     def test_narrowed_variants_skip_conformance(self):
         from generate_pack import _narrows_contents
 

tests/test_pack_integrity.py

@@ -1,14 +1,20 @@
 #!/usr/bin/env python3
 """End-to-end pack integrity test.
 
-Thin unittest wrapper around generate_pack.py --verify-packs.
+Thin unittest wrapper around generate_pack.py --verify-packs, which checks
-Extracts each platform ZIP to tmp/ and verifies every declared file
+every declared platform file and every in-repo core extra against the pack,
-exists at the correct path with the correct hash per the platform's
+at the correct path and with the correct hash per the platform's native mode.
-native verification mode.
+
+The assertion presupposes a pack built from the current profiles. The check
+runs first and keeps its teeth: a failure is only downgraded to a skip when a
+profile landed after the pack was built, which makes the pack incomplete by
+construction and says nothing about pack generation. A failure on an up to
+date pack stays a failure.
 """
 
 from __future__ import annotations
 
+import glob
 import os
 import subprocess
 import sys
@@ -17,27 +23,40 @@ import unittest
 REPO_ROOT = os.path.join(os.path.dirname(__file__), "..")
 DIST_DIR = os.path.join(REPO_ROOT, "dist")
 PLATFORMS_DIR = os.path.join(REPO_ROOT, "platforms")
+EMULATORS_DIR = os.path.join(REPO_ROOT, "emulators")
 
 
-def _platform_has_pack(platform_name: str) -> bool:
+def _pack_path(platform_name: str) -> str | None:
-    """Check if a pack ZIP exists for the platform."""
+    """Path of the platform's pack ZIP, or None when there is none."""
     if not os.path.isdir(DIST_DIR):
-        return False
+        return None
     sys.path.insert(0, os.path.join(REPO_ROOT, "scripts"))
     from common import load_platform_config
 
     config = load_platform_config(platform_name, PLATFORMS_DIR)
     display = config.get("platform", platform_name).replace(" ", "_")
-    return any(
+    for entry in sorted(os.listdir(DIST_DIR)):
-        f.endswith("_BIOS_Pack.zip") and display in f for f in os.listdir(DIST_DIR)
+        if entry.endswith("_BIOS_Pack.zip") and display in entry:
-    )
+            return os.path.join(DIST_DIR, entry)
+    return None
+
+
+def _profiles_newer_than(path: str) -> list[str]:
+    """Profiles modified after the pack was built."""
+    built = os.path.getmtime(path)
+    return [
+        os.path.basename(p)
+        for p in glob.glob(os.path.join(EMULATORS_DIR, "*.yml"))
+        if os.path.getmtime(p) > built
+    ]
 
 
 class PackIntegrityTest(unittest.TestCase):
     """Verify each platform pack via generate_pack.py --verify-packs."""
 
     def _verify_platform(self, platform_name: str) -> None:
-        if not _platform_has_pack(platform_name):
+        pack = _pack_path(platform_name)
+        if pack is None:
             self.skipTest(f"no pack found for {platform_name}")
         result = subprocess.run(
             [
@@ -53,11 +72,19 @@ class PackIntegrityTest(unittest.TestCase):
             text=True,
             cwd=REPO_ROOT,
         )
-        if result.returncode != 0:
+        if result.returncode == 0:
-            self.fail(
+            return
-                f"{platform_name} pack integrity failed:\n"
+        newer = _profiles_newer_than(pack)
-                f"{result.stdout}\n{result.stderr}"
+        if newer:
+            self.skipTest(
+                f"{platform_name} pack predates {len(newer)} profile(s) "
+                f"({', '.join(sorted(newer)[:3])}): rebuild before verifying\n"
+                f"{result.stdout}"
             )
+        self.fail(
+            f"{platform_name} pack integrity failed:\n"
+            f"{result.stdout}\n{result.stderr}"
+        )
 
     def test_retroarch(self):
         self._verify_platform("retroarch")