From e5420a9dbaa95981db53efaf26aea8dc9ab5c0eb Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Tue, 6 Oct 2026 09:39:07 +0200
Subject: [PATCH] fix: keep platform hashes the truth does not contradict

---
 scripts/exporter/baseline.py | 16 +++++++++++-----
 tests/test_export_counts.py  | 12 ++++++++++++
 2 files changed, 23 insertions(+), 5 deletions(-)

diff --git a/scripts/exporter/baseline.py b/scripts/exporter/baseline.py
index 5bad4103..42415f93 100644
--- a/scripts/exporter/baseline.py
+++ b/scripts/exporter/baseline.py
@@ -61,9 +61,11 @@ class NativeFile:
         """Which side's content fields describe the file: one side, or both.
 
         Mixing them field by field wrote the truth's size and crc beside the
-        platform's sha1, a record no file satisfies. Both sides are merged
-        only where they agree on some hash; where they contradict, the truth
-        decides alone; where the truth declares none, the platform does.
+        platform's sha1, a record no file satisfies. The truth decides alone
+        only on a contradiction: a hash both sides declare without a value in
+        common, or two declared sizes. Different fields prove nothing: the
+        truth's crc32 beside Batocera's md5 for bios7.bin is one dump, and
+        reading it as a contradiction emptied the md5 the frontend checks.
         """
         truth = {f: set(_hash_values(self.truth or {}, f)) for f in HASH_FIELDS}
         plat = {f: set(_hash_values(self.platform or {}, f)) for f in HASH_FIELDS}
@@ -71,9 +73,13 @@ class NativeFile:
             return "platform"
         if not any(plat.values()):
             return "truth"
-        if any(truth[f] & plat[f] for f in HASH_FIELDS):
+        shared = [f for f in HASH_FIELDS if truth[f] and plat[f]]
+        if any(truth[f] & plat[f] for f in shared):
             return "both"
-        return "truth"
+        t_size = (self.truth or {}).get("size")
+        p_size = (self.platform or {}).get("size")
+        sizes_differ = isinstance(t_size, int) and isinstance(p_size, int) and t_size != p_size
+        return "truth" if shared or sizes_differ else "both"
 
     def hashes(self, field_name: str) -> list[str]:
         """Accepted values for a hash, truth first when it has an opinion.
diff --git a/tests/test_export_counts.py b/tests/test_export_counts.py
index 72e99cec..75792d1e 100644
--- a/tests/test_export_counts.py
+++ b/tests/test_export_counts.py
@@ -155,6 +155,18 @@ class OneFileOneIdentity(unittest.TestCase):
         self.assertEqual(fe.hashes("crc32"), ["bbbbbbbb"])
         self.assertEqual(fe.size(), 480)
 
+    def test_different_fields_are_not_a_contradiction(self):
+        """bios7.bin: the truth's crc32 beside Batocera's md5 is one dump."""
+        fe = NativeFile("bios7.bin", "bios/bios7.bin", "nds",
+                        platform={"md5": A}, truth={"crc32": "1280f0d5"})
+        self.assertEqual(fe.hashes("md5"), [A])
+        self.assertEqual(fe.hashes("crc32"), ["1280f0d5"])
+
+    def test_different_sizes_still_contradict(self):
+        fe = NativeFile("boot.bin", "dc/boot.bin", "dc",
+                        platform={"size": 2097152, "md5": A}, truth={"size": 480, "crc32": "bbbbbbbb"})
+        self.assertEqual(fe.hashes("md5"), [])
+
     def test_agreeing_sides_still_merge(self):
         fe = NativeFile("x.bin", "x.bin", "s",
                         platform={"md5": A, "sha1": "1" * 40}, truth={"md5": A})