From 7a03578dd96efa8a55fa3ee6fbb9931fa48e6b4a Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Fri, 6 Nov 2026 08:21:47 +0100
Subject: [PATCH] fix: count name holders with case folded

A hashless profile entry adopted the only exact holder of its name as the file its aliases designate. dosbox's SC-55 ROM2.BIN found Galaksija's ROM2.BIN, and ROM2_ST.BIN was indexed against a 4 KiB BASIC ROM while SC-55's rom2.bin sat under the same folded name. The name now has to designate one file once case is folded, a primary and its own variants counted as one.
---
 database.json            |  5 +----
 scripts/generate_db.py   | 27 +++++++++++++++++++++++----
 tests/test_db_aliases.py | 19 +++++++++++++++++++
 3 files changed, 43 insertions(+), 8 deletions(-)

diff --git a/database.json b/database.json
index bc7d27139..375671581 100644
--- a/database.json
+++ b/database.json
@@ -1,6 +1,6 @@
 {
   "schema_version": 1,
-  "generated_at": "2026-11-05T23:59:37Z",
+  "generated_at": "2026-11-06T07:21:27Z",
   "total_files": 19947,
   "total_size": 16926035621,
   "bios_dir": "bios",
@@ -281813,9 +281813,6 @@
       "TRIO64 (Ver. 1.5-07) [VGA] (S3 Incorporated).bin": [
         "a308b15f93da34961c0141ec95b98c211557e2d5"
       ],
-      "ROM2_ST.BIN": [
-        "5d5ab4313a2d0effe7572bb129193b64cab002c1"
-      ],
       "mt32_ctrl_1_04.rom": [
         "5a5cb5a77d7d55ee69657c2f870416daed52dea7"
       ],
diff --git a/scripts/generate_db.py b/scripts/generate_db.py
index ade383713..32c956171 100644
--- a/scripts/generate_db.py
+++ b/scripts/generate_db.py
@@ -576,12 +576,26 @@ def _collect_all_aliases(
     # naming its content once 64DD_IPL.bin sorted first, and its aliases
     # 64dd_ipl_J.bin and 64DD_IPL_JP.n64 went with it.
     holders: dict[str, set[str]] = {}
+    # The files a name designates once case is folded, as the resolver's
+    # name step folds it, a primary and its own .variants/ counted as one
+    # file. dosbox's SC-55 ROM2.BIN found Galaksija's ROM2.BIN the only
+    # exact holder, and its alias ROM2_ST.BIN went to a 4 KiB BASIC ROM,
+    # while SC-55's rom2.bin sat under the same folded name.
+    families: dict[str, set[str]] = {}
+
+    def _hold(name: str, path: str, sha1: str) -> None:
+        holders.setdefault(name, set()).add(sha1)
+        directory = path.rpartition("/")[0].removesuffix("/.variants")
+        families.setdefault(name.casefold(), set()).add(
+            f"{directory}/{variant_base_name(name)}".casefold()
+        )
+
     for sha1, entry in files.items():
-        holders.setdefault(entry["name"], set()).add(sha1)
+        _hold(entry["name"], entry["path"], sha1)
     for sha1, copy_list in (copies or {}).items():
         if sha1 in files:
             for copy in copy_list:
-                holders.setdefault(copy["name"], set()).add(sha1)
+                _hold(copy["name"], copy["path"], sha1)
 
     aliases = {}
 
@@ -691,11 +705,16 @@ def _collect_all_aliases(
                     ]
                     if taken:
                         shadowed.append((entry_name, set(matched), taken))
-                if not matched and len(holders.get(entry_name, ())) == 1:
+                held = holders.get(entry_name, set())
+                if (
+                    not matched
+                    and len(held) == 1
+                    and len(families[entry_name.casefold()]) == 1
+                ):
                     # A name carried by several contents names none of them:
                     # quasi88's disk.rom aliases went to whichever
                     # disk.rom the scan met last, a Tandy CoCo ROM.
-                    matched = set(holders[entry_name])
+                    matched = set(held)
                 for sha in sorted(matched):
                     for alias_name in entry_aliases:
                         _add_alias(alias_name, sha)
diff --git a/tests/test_db_aliases.py b/tests/test_db_aliases.py
index 91bfd56fa..eff324b5f 100644
--- a/tests/test_db_aliases.py
+++ b/tests/test_db_aliases.py
@@ -56,6 +56,25 @@ class AliasesNeedAnUnambiguousName(unittest.TestCase):
         named = {a["name"] for entries in aliases.values() for a in entries}
         self.assertNotIn("n88sub.rom", named)
 
+    def test_homonyms_differing_in_case_take_no_alias(self):
+        """dosbox's SC-55 ROM2.BIN: Galaksija's ROM2.BIN was the only exact
+        holder, and ROM2_ST.BIN went to it beside SC-55's rom2.bin."""
+        files = {
+            "a": self._file("DISK.ROM", "bios/NEC/PC-98/DISK.ROM", "m1"),
+            "b": self._file("disk.rom", "bios/Tandy/CoCo/disk.rom", "m2"),
+        }
+        aliases = self._aliases(files)
+        named = {a["name"] for entries in aliases.values() for a in entries}
+        self.assertNotIn("n88sub.rom", named)
+
+    def test_a_primary_and_its_variant_are_one_file(self):
+        files = {
+            "a": self._file("disk.rom", "bios/NEC/PC-88/disk.rom", "m1"),
+            "b": self._file("DISK.ROM", "bios/NEC/PC-88/.variants/DISK.ROM.0123abcd", "m2"),
+        }
+        aliases = self._aliases(files)
+        self.assertEqual([a["name"] for a in aliases.get("a", [])], ["n88sub.rom"])
+
     def test_a_unique_name_still_carries_its_aliases(self):
         files = {"a": self._file("disk.rom", "bios/NEC/PC-88/disk.rom", "m1")}
         aliases = self._aliases(files)