feat: add by_sha256 index, fix reporting attribution

generate_db: add by_sha256 index for O(1) variant lookup.
verify: _find_best_variant uses indexed sha256 instead of O(n) scan.
validation: check_file_validation returns (reason, emulators) tuple,
attributing mismatch only to emulators whose check actually failed.
beetle_psx: remove incorrect size field for ps1_rom.bin (code does
not validate size, swanstation is sole size authority).

emulators/beetle_psx.yml

@@ -92,10 +92,9 @@ files:
     note: "override_bios=1. Falls back to region BIOS if not found."
 
   - name: "ps1_rom.bin"
-    description: "PS3 embedded PS1 BIOS (region-free override)"
+    description: "PS3 embedded PS1 BIOS, first 512KB extracted (region-free override)"
     region: "Auto"
     required: false
-    size: 524288
     sha1: "c40146361eb8cf670b19fdc9759190257803cab7"
     md5: "81bbe60ba7a3d1cea1d48c14cbcc647b"
     validation: [sha1]

scripts/generate_db.py

@@ -168,6 +168,7 @@ def build_indexes(files: dict, aliases: dict) -> dict:
     by_md5 = {}
     by_name = {}
     by_crc32 = {}
+    by_sha256 = {}
     by_path_suffix = {}
 
     for sha1, entry in files.items():
@@ -179,6 +180,7 @@ def build_indexes(files: dict, aliases: dict) -> dict:
         by_name[name].append(sha1)
 
         by_crc32[entry["crc32"]] = sha1
+        by_sha256[entry["sha256"]] = sha1
 
         # Path suffix index for regional variant resolution
         suffix = _path_suffix(entry["path"])
@@ -208,6 +210,7 @@ def build_indexes(files: dict, aliases: dict) -> dict:
         "by_md5": by_md5,
         "by_name": by_name,
         "by_crc32": by_crc32,
+        "by_sha256": by_sha256,
         "by_path_suffix": by_path_suffix,
     }
 

scripts/generate_pack.py

@@ -1308,10 +1308,11 @@ def generate_pack(
                     and validation_index
                 ):
                     fname = file_entry.get("name", "")
-                    reason = check_file_validation(
+                    check = check_file_validation(
                         local_path, fname, validation_index, bios_dir
                     )
-                    if reason:
+                    if check:
+                        reason, emus_list = check
                         better = _find_candidate_satisfying_both(
                             file_entry,
                             db,
@@ -1322,8 +1323,7 @@ def generate_pack(
                         if better:
                             local_path = better
                         else:
-                            ventry = validation_index.get(fname, {})
-                            emus = ", ".join(ventry.get("emulators", []))
+                            emus = ", ".join(emus_list)
                             file_reasons.setdefault(
                                 dedup_key,
                                 f"{platform_display} says OK but {emus} says {reason}",

scripts/validation.py

@@ -204,35 +204,56 @@ def build_ground_truth(filename: str, validation_index: dict[str, dict]) -> list
     return result
 
 
+def _emulators_for_check(
+    check_type: str, per_emulator: dict[str, dict],
+) -> list[str]:
+    """Return emulator names that validate a specific check type."""
+    result = []
+    for emu, detail in per_emulator.items():
+        emu_checks = detail.get("checks", [])
+        if check_type in emu_checks:
+            result.append(emu)
+        # adler32 is stored as known_hash, not always in validation list
+        if check_type == "adler32" and detail.get("expected", {}).get("adler32"):
+            if emu not in result:
+                result.append(emu)
+    return sorted(result)
+
+
 def check_file_validation(
     local_path: str,
     filename: str,
     validation_index: dict[str, dict],
     bios_dir: str = "bios",
-) -> str | None:
+) -> tuple[str, list[str]] | None:
     """Check emulator-level validation on a resolved file.
 
     Supports: size (exact/min/max), crc32, md5, sha1, adler32,
     signature (RSA-2048 PKCS1v15 SHA256), crypto (AES-128-CBC + SHA256).
 
     Returns None if all checks pass or no validation applies.
-    Returns a reason string if a check fails.
+    Returns (reason, emulators) tuple on failure, where *emulators*
+    lists only those cores whose check actually failed.
     """
     entry = validation_index.get(filename)
     if not entry:
         return None
     checks = entry["checks"]
+    pe = entry.get("per_emulator", {})
 
     # Size checks -sizes is a set of accepted values
     if "size" in checks:
         actual_size = os.path.getsize(local_path)
         if entry["sizes"] and actual_size not in entry["sizes"]:
             expected = ",".join(str(s) for s in sorted(entry["sizes"]))
-            return f"size mismatch: got {actual_size}, accepted [{expected}]"
+            emus = _emulators_for_check("size", pe)
+            return f"size mismatch: got {actual_size}, accepted [{expected}]", emus
         if entry["min_size"] is not None and actual_size < entry["min_size"]:
-            return f"size too small: min {entry['min_size']}, got {actual_size}"
+            emus = _emulators_for_check("size", pe)
+            return f"size too small: min {entry['min_size']}, got {actual_size}", emus
         if entry["max_size"] is not None and actual_size > entry["max_size"]:
-            return f"size too large: max {entry['max_size']}, got {actual_size}"
+            emus = _emulators_for_check("size", pe)
+            return f"size too large: max {entry['max_size']}, got {actual_size}", emus
 
     # Hash checks -compute once, reuse for all hash types.
     # Each hash field is a set of accepted values (multiple valid ROM versions).
@@ -245,14 +266,23 @@ def check_file_validation(
             if hash_type in checks and entry[hash_type]:
                 if hashes[hash_type].lower() not in entry[hash_type]:
                     expected = ",".join(sorted(entry[hash_type]))
-                    return f"{hash_type} mismatch: got {hashes[hash_type]}, accepted [{expected}]"
+                    emus = _emulators_for_check(hash_type, pe)
+                    return (
+                        f"{hash_type} mismatch: got {hashes[hash_type]}, "
+                        f"accepted [{expected}]",
+                        emus,
+                    )
         if entry["adler32"]:
             actual_adler = hashes["adler32"].lower()
             if entry.get("adler32_byteswap"):
                 actual_adler = _adler32_byteswapped(local_path)
             if actual_adler not in entry["adler32"]:
                 expected = ",".join(sorted(entry["adler32"]))
-                return f"adler32 mismatch: got 0x{actual_adler}, accepted [{expected}]"
+                emus = _emulators_for_check("adler32", pe)
+                return (
+                    f"adler32 mismatch: got 0x{actual_adler}, accepted [{expected}]",
+                    emus,
+                )
 
     # Signature/crypto checks (3DS RSA, AES)
     if entry["crypto_only"]:
@@ -260,7 +290,8 @@ def check_file_validation(
 
         crypto_reason = check_crypto_validation(local_path, filename, bios_dir)
         if crypto_reason:
-            return crypto_reason
+            emus = sorted(entry.get("emulators", []))
+            return crypto_reason, emus
 
     return None
 

scripts/verify.py

@@ -103,8 +103,9 @@ def verify_entry_existence(
         return {"name": name, "status": Status.MISSING, "required": required}
     result = {"name": name, "status": Status.OK, "required": required}
     if validation_index:
-        reason = check_file_validation(local_path, name, validation_index)
-        if reason:
+        check = check_file_validation(local_path, name, validation_index)
+        if check:
+            reason, emus_list = check
             suppressed = False
             if db:
                 better = _find_best_variant(
@@ -113,8 +114,7 @@ def verify_entry_existence(
                 if better:
                     suppressed = True
             if not suppressed:
-                ventry = validation_index.get(name, {})
-                emus = ", ".join(ventry.get("emulators", []))
+                emus = ", ".join(emus_list)
                 result["discrepancy"] = (
                     f"file present (OK) but {emus} says {reason}"
                 )
@@ -608,24 +608,25 @@ def _find_best_variant(
 
     # Pass 1: hash-based lookup from emulator expected values
     ventry = validation_index[fname]
+    indexes = db.get("indexes", {})
     for hash_type, db_index_key in (
-        ("sha1", "by_sha1"),
+        ("sha1", None),
         ("md5", "by_md5"),
         ("crc32", "by_crc32"),
+        ("sha256", "by_sha256"),
     ):
         expected = ventry.get(hash_type)
         if not expected:
             continue
-        db_index = db.get("indexes", {}).get(db_index_key, {})
-        if not db_index:
-            # by_sha1 is the files dict itself (sha1 = primary key)
-            if hash_type == "sha1":
-                for h in expected:
-                    if h in files_db:
-                        result = _try_candidate(h)
-                        if result:
-                            return result
+        if db_index_key is None:
+            # SHA1 is the primary key of files_db
+            for h in expected:
+                if h in files_db:
+                    result = _try_candidate(h)
+                    if result:
+                        return result
             continue
+        db_index = indexes.get(db_index_key, {})
         for h in expected:
             entries = db_index.get(h)
             if not entries:
@@ -640,16 +641,7 @@ def _find_best_variant(
                 if result:
                     return result
 
-    # Pass 2: SHA256 scan (no index, but emulators like ares validate by sha256)
-    expected_sha256 = ventry.get("sha256")
-    if expected_sha256:
-        for sha1, entry in files_db.items():
-            if entry.get("sha256", "").lower() in expected_sha256:
-                result = _try_candidate(sha1)
-                if result:
-                    return result
-
-    # Pass 3: name-based lookup (aliases, .variants/ with same filename)
+    # Pass 2: name-based lookup (aliases, .variants/ with same filename)
     by_name = db.get("indexes", {}).get("by_name", {})
     for sha1 in by_name.get(fname, []):
         result = _try_candidate(sha1)
@@ -732,8 +724,11 @@ def verify_platform(
                 # mismatches are reported as discrepancies, not failures.
                 if result["status"] == Status.OK and local_path and validation_index:
                     fname = file_entry.get("name", "")
-                    reason = check_file_validation(local_path, fname, validation_index)
-                    if reason:
+                    check = check_file_validation(
+                        local_path, fname, validation_index,
+                    )
+                    if check:
+                        reason, emus_list = check
                         better = _find_best_variant(
                             file_entry,
                             db,
@@ -741,8 +736,7 @@ def verify_platform(
                             validation_index,
                         )
                         if not better:
-                            ventry = validation_index.get(fname, {})
-                            emus = ", ".join(ventry.get("emulators", []))
+                            emus = ", ".join(emus_list)
                             result["discrepancy"] = (
                                 f"{platform} says OK but {emus} says {reason}"
                             )
@@ -1233,8 +1227,9 @@ def verify_emulator(
                 result = {"name": name, "status": Status.MISSING, "required": required}
             else:
                 # Apply emulator validation
-                reason = check_file_validation(local_path, name, validation_index)
-                if reason:
+                check = check_file_validation(local_path, name, validation_index)
+                if check:
+                    reason, _emus = check
                     result = {
                         "name": name,
                         "status": Status.UNTESTED,

tests/test_e2e.py

@@ -1137,9 +1137,11 @@ class TestE2E(unittest.TestCase):
         profiles = load_emulator_profiles(self.emulators_dir)
         index = _build_validation_index(profiles)
         path = self.files["present_opt.bin"]["path"]
-        reason = check_file_validation(path, "present_opt.bin", index)
-        self.assertIsNotNone(reason)
+        result = check_file_validation(path, "present_opt.bin", index)
+        self.assertIsNotNone(result)
+        reason, emus = result
         self.assertIn("size mismatch", reason)
+        self.assertIsInstance(emus, list)
 
     def test_73_validation_crc32_pass(self):
         """File with correct CRC32 passes validation."""
@@ -1154,9 +1156,11 @@ class TestE2E(unittest.TestCase):
         profiles = load_emulator_profiles(self.emulators_dir)
         index = _build_validation_index(profiles)
         path = self.files["no_md5.bin"]["path"]
-        reason = check_file_validation(path, "no_md5.bin", index)
-        self.assertIsNotNone(reason)
+        result = check_file_validation(path, "no_md5.bin", index)
+        self.assertIsNotNone(result)
+        reason, emus = result
         self.assertIn("crc32 mismatch", reason)
+        self.assertIsInstance(emus, list)
 
     def test_75_validation_applied_in_existence_mode(self):
         """Existence mode reports discrepancy when validation fails, keeps OK."""
@@ -1212,9 +1216,11 @@ class TestE2E(unittest.TestCase):
         profiles = load_emulator_profiles(self.emulators_dir)
         index = _build_validation_index(profiles)
         path = self.files["alias_target.bin"]["path"]
-        reason = check_file_validation(path, "alias_target.bin", index)
-        self.assertIsNotNone(reason)
+        result = check_file_validation(path, "alias_target.bin", index)
+        self.assertIsNotNone(result)
+        reason, emus = result
         self.assertIn("md5 mismatch", reason)
+        self.assertIsInstance(emus, list)
 
     def test_81_validation_index_has_md5_sha1(self):
         """Validation index stores md5 and sha1 when declared."""