From b5c1e4fc032335d49a16afd554ea295ab9c5f771 Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Sun, 4 Oct 2026 23:47:21 +0200
Subject: [PATCH] fix: count engine and publisher trees as game data

---
 scripts/common.py               | 25 ++++++++++++++++++++++---
 scripts/generate_site.py        |  3 ++-
 tests/test_audit_regressions.py | 16 ++++++++++++++++
 3 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/scripts/common.py b/scripts/common.py
index f290d0f4..6d6b1a95 100644
--- a/scripts/common.py
+++ b/scripts/common.py
@@ -1002,7 +1002,26 @@ def unique_emulator_profiles(profiles: dict[str, dict]) -> dict[str, dict]:
     }
 
 
-GAME_DATA_TOPS = ("RPG Maker", "ScummVM")
+# Trees that hold game and engine data rather than the system files of a
+# machine: the engine trees, and the publishers whose games an engine runs.
+GAME_DATA_TOPS = (
+    "RPG Maker",
+    "ScummVM",
+    "Game Engines",
+    "sdlpal",
+    "Dinothawr",
+    "3D Realms",
+    "Apogee",
+    "Blizzard",
+    "Crack dot Com",
+    "Electronic Arts",
+    "Epic MegaGames",
+    "Id Software",
+    "Interplay",
+    "Monolith",
+    "Softdisk",
+    "Valve",
+)
 
 
 def composition_tier(path: str) -> str:
@@ -1020,8 +1039,8 @@ def compute_composition(db: dict) -> dict:
     """File and byte counts by tree area.
 
     Three buckets, each re-derivable from paths alone: arcade ROM sets
-    (Arcade/), game and engine data (the RPG Maker/ and ScummVM/ trees),
-    and console or computer system files (everything else).
+    (Arcade/), game and engine data (the trees of GAME_DATA_TOPS), and
+    console or computer system files (everything else).
     """
     buckets = {
         "systems": {"files": 0, "size_bytes": 0},
diff --git a/scripts/generate_site.py b/scripts/generate_site.py
index fcda2dd4..39e2b2f4 100644
--- a/scripts/generate_site.py
+++ b/scripts/generate_site.py
@@ -428,7 +428,8 @@ def composition_sentence(db: dict) -> str:
         f"The collection holds {comp['systems']['files']:,} console and "
         f"computer system files, {comp['arcade']['files']:,} arcade ROM sets "
         f"(`Arcade/`), and {comp['game_data']['files']:,} game and engine "
-        "data files (the `RPG Maker/` and `ScummVM/` trees). No pack holds "
+        "data files (`Game Engines/`, `RPG Maker/`, `ScummVM/` and the game "
+        "publishers' trees). No pack holds "
         "them all: each carries what its platform's emulators load."
     )
 
diff --git a/tests/test_audit_regressions.py b/tests/test_audit_regressions.py
index bfd454df..b448945b 100644
--- a/tests/test_audit_regressions.py
+++ b/tests/test_audit_regressions.py
@@ -388,6 +388,22 @@ class CatalogRatioRegressions(unittest.TestCase):
         self.assertEqual(count_catalog_matched(db), 1)
         self.assertEqual(compute_composition(db)["systems"]["files"], 2)
 
+    def test_engine_and_publisher_trees_are_game_data(self):
+        """Engine data collected under `Game Engines/` and under a game
+        publisher counted as console and computer system files: 7 697 of
+        them made the README announce 13 564 system files."""
+        from scripts.common import composition_tier
+
+        for path in (
+            "bios/Game Engines/C-Dogs SDL/cdogs/data/guns.json",
+            "bios/Id Software/Quake III Arena/baseq3/pak1.pk3",
+            "bios/Epic MegaGames/Jazz Jackrabbit 2/jazz2/Source/share.j2e",
+            "bios/RPG Maker/easyrpg/rtp/2000/Backdrop/Bridge.png",
+        ):
+            self.assertEqual(composition_tier(path), "game_data", path)
+        self.assertEqual(composition_tier("bios/Sony/PlayStation/scph5501.bin"), "systems")
+        self.assertEqual(composition_tier("bios/Microsoft/MSX/MSX2.ROM"), "systems")
+
     def test_readme_ratio_matches_the_database(self):
         from scripts.common import compute_composition, count_catalog_matched, load_database