From f69bd7cbd50425734f2230e95caaadd095a63c5f Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Thu, 5 Nov 2026 11:44:40 +0100
Subject: [PATCH] fix: read the database size of a declared hash

A platform declaration that names only an md5 left the size check
undecided, and an undecided same-named match overrode a profile's
system:. The size the database holds for that md5 now refutes it: clk's
Oric basic10.rom no longer joins Batocera's BK list.
---
 scripts/truth.py           | 19 ++++++++++++++-----
 tests/test_truth_layout.py | 20 ++++++++++++++++++++
 2 files changed, 34 insertions(+), 5 deletions(-)

diff --git a/scripts/truth.py b/scripts/truth.py
index 302d6562e..bc1e3c98c 100644
--- a/scripts/truth.py
+++ b/scripts/truth.py
@@ -508,14 +508,19 @@ def _sizes_clash(a: dict, b: dict) -> bool:
     return False
 
 
-def _identity_values(entry: dict, db: dict | None) -> set[str]:
-    """Every hash an entry declares, with the siblings the database knows."""
+def _identity(entry: dict, db: dict | None) -> dict:
+    """The entry with the hashes and size the database knows for its hash."""
     probe = {k: entry[k] for k in ("sha1", "md5", "crc32", "size") if entry.get(k)}
     if db and probe and not any(isinstance(v, list) for v in probe.values()):
         _enrich_hashes(probe, db)
+    return {**entry, **{k: v for k, v in probe.items() if not entry.get(k)}}
+
+
+def _identity_values(entry: dict) -> set[str]:
+    """Every hash an identity declares."""
     values: set[str] = set()
     for field in ("sha1", "md5", "sha256", "crc32"):
-        values |= _declared(entry, field) | _declared(probe, field)
+        values |= _declared(entry, field)
     return values
 
 
@@ -525,10 +530,14 @@ def _declaration_verdict(entry: dict, declaration: dict, db: dict | None) -> str
     A shared hash proves it. Sizes no value of which agrees refute it, as
     name_match_size_ok refutes a candidate. Two hashes at one size are two
     revisions of one slot, as slots.py reads them, so they refute nothing.
+    The size the database holds for a declared hash counts as declared:
+    Batocera's BK BASIC10.ROM names only an md5, of a 24576-byte file, and
+    clk's 16 KB Oric basic10.rom was filed under the BK beside it.
     """
-    if _identity_values(entry, db) & _identity_values(declaration, db):
+    ours, theirs = _identity(entry, db), _identity(declaration, db)
+    if _identity_values(ours) & _identity_values(theirs):
         return "same"
-    if _sizes_clash(entry, declaration):
+    if _sizes_clash(ours, theirs):
         return "other"
     return "undecided"
 
diff --git a/tests/test_truth_layout.py b/tests/test_truth_layout.py
index d644350aa..4c187b261 100644
--- a/tests/test_truth_layout.py
+++ b/tests/test_truth_layout.py
@@ -454,6 +454,26 @@ class TheModelIsAFunctionOfItsInputs(unittest.TestCase):
         )
 
 
+class DatabaseSizeRefutesASameNamedDeclaration(unittest.TestCase):
+    """Batocera's BK system declares BASIC10.ROM by md5 alone. clk's Oric
+    basic10.rom, 16 KB with its own system:, was filed under the BK and
+    added to its list: the 24576 bytes the database holds for that md5
+    were never compared."""
+
+    def test_the_size_behind_a_declared_hash_counts(self):
+        bk_md5, oric_md5 = "3" * 32, "e" * 32
+        db = {"files": {"1" * 40: {"sha1": "1" * 40, "md5": bk_md5, "size": 24576}},
+              "indexes": {"by_md5": {bk_md5: "1" * 40}}}
+        profiles = {"clk": {"emulator": "clk", "type": "libretro", "systems": ["oric"],
+                            "files": [{"name": "basic10.rom", "path": "Oric/basic10.rom",
+                                       "system": "oric", "size": 16384, "md5": oric_md5}]}}
+        config = {"cores": ["clk"], "systems": {"elektronika-bk": {"files": [
+            {"name": "BASIC10.ROM", "md5": bk_md5}]}}}
+        truth = generate_platform_truth("p", config, {}, profiles, db=db)
+        self.assertNotIn("basic10.rom", {
+            f["name"] for f in truth["systems"].get("elektronika-bk", {}).get("files", [])})
+
+
 class RegionFollowsEveryReader(unittest.TestCase):
     """The first core to tag a file set its region in the truth: the ST-V
     BIOS kept beetle_saturn's Americas and lost mednafen_stv's Korea, and