Merge pull request #4854 from rommapp/fix/rom-file-category-anchor

fix(scan): anchor rom file category to the folder below the rom root

backend/handler/filesystem/roms_handler.py

@@ -149,12 +149,15 @@ def category_matches(category: str, path_parts: list[str]) -> bool:
 
 
 def category_for_path_parts(path_parts_lower: list[str]) -> RomFileCategory | None:
-    """The file category a folder path implies, from its lowercased parts."""
+    """The file category of a folder, from its lowercased parts below the ROM root."""
+    # Only the top folder counts, so a dump's inner `content/game` never matches.
+    if not path_parts_lower:
+        return None
     return next(
         (
             category
             for category in RomFileCategory
-            if category_matches(category.value, path_parts_lower)
+            if category_matches(category.value, path_parts_lower[:1])
         ),
         None,
     )
@@ -515,6 +518,17 @@ class FSRomsHandler(FSHandler):
 
         return kept_roms
 
+    @staticmethod
+    def _folder_category(rom: Rom, rom_path: Path) -> RomFileCategory | None:
+        """The category a file's folder gives it, relative to the ROM's own root."""
+        rom_root = Path(rom.full_path)
+        rom_relative_parts = (
+            rom_path.relative_to(rom_root).parts
+            if rom_path.is_relative_to(rom_root)
+            else ()
+        )
+        return category_for_path_parts(list(map(str.lower, rom_relative_parts)))
+
     def _build_rom_file(
         self,
         rom: Rom,
@@ -526,10 +540,7 @@ class FSRomsHandler(FSHandler):
         archive_members: list[dict[str, Any]] | None = None,
     ) -> RomFile:
         abs_file_path = Path(self.base_path, rom_path, file_name)
-
-        matching_category = category_for_path_parts(
-            list(map(str.lower, rom_path.parts))
-        )
+        matching_category = self._folder_category(rom, rom_path)
 
         track_meta = None
         if matching_category == RomFileCategory.SOUNDTRACK:
@@ -732,6 +743,8 @@ class FSRomsHandler(FSHandler):
                         hashable=hashable_platform,
                     )
                 ):
+                    # Title id extraction below may settle a content category.
+                    row.category = self._folder_category(rom, rel_dir)
                     rom_files.append(row)
                     _record_title_id_source(abs_file_path, row)
                     continue

backend/tests/handler/filesystem/test_roms_handler.py

@@ -24,6 +24,7 @@ from handler.filesystem.roms_handler import (
     FileHash,
     FSRomsHandler,
     _TitleIdSource,
+    category_for_path_parts,
     category_matches,
     mtime_matches,
 )
@@ -561,7 +562,7 @@ class TestFSRomsHandler:
     def test_build_rom_file_with_category(self, rom_multi: Rom, handler: FSRomsHandler):
         """Test _build_rom_file with category detection"""
         # Test with DLC category
-        rom_path = Path(rom_multi.fs_path, "dlc")
+        rom_path = Path(rom_multi.full_path, "dlc")
         file_name = "test_dlc.n64"
         file_hash = FileHash(
             {
@@ -3482,6 +3483,69 @@ class TestIncrementalRomFiles:
         assert by_name["part1 (Hack).bin"].category == RomFileCategory.HACK
         assert by_name["fix.ips"].category == RomFileCategory.PATCH
 
+    async def test_category_comes_from_the_folder_below_the_rom_root(
+        self, handler: FSRomsHandler, platform: Platform
+    ):
+        rom = self._rom(platform)
+        self._write(handler, f"{self.ROM_DIR}/dlc/foo.bin", b"dlc")
+        self._write(handler, f"{self.ROM_DIR}/update/content/game/x.bin", b"update")
+        self._write(
+            handler,
+            f"{self.ROM_DIR}/Sonic [Update]/content/Game/Stats/archive.bin",
+            b"stats",
+        )
+        self._write(handler, f"{self.ROM_DIR}/content/CMCmn/demo/effect.szs", b"fx")
+
+        parsed = await handler.get_rom_files(rom)
+
+        by_name = {f.file_name: f for f in parsed.rom_files}
+        assert by_name["foo.bin"].category == RomFileCategory.DLC
+        assert by_name["x.bin"].category == RomFileCategory.UPDATE
+        assert by_name["archive.bin"].category is None
+        assert by_name["effect.szs"].category is None
+
+    async def test_reused_row_takes_its_folder_category(
+        self, handler: FSRomsHandler, platform: Platform
+    ):
+        rom = self._rom(platform)
+        rel_dir = f"{self.ROM_DIR}/Sonic [Update]/content/Game"
+        st = self._write(handler, f"{rel_dir}/archive.bin", b"stats")
+        row = self._row(rom, rel_dir, "archive.bin", st)
+        row.category = RomFileCategory.GAME
+
+        parsed = await handler.get_rom_files(rom, existing_files=[row])
+
+        assert parsed.rom_files == [row]
+        assert row.category is None
+
+    async def test_rom_folder_named_like_a_category_is_not_a_category(
+        self, handler: FSRomsHandler
+    ):
+        platform = Platform(name="Nintendo 64", slug="n64", fs_slug="n64")
+        rom = self._rom(platform, fs_name="Demo")
+        self._write(handler, "n64/roms/Demo/game.n64", b"game")
+
+        parsed = await handler.get_rom_files(rom)
+
+        assert parsed.rom_files[0].category is None
+
+    @pytest.mark.parametrize(
+        "parts, expected",
+        [
+            ([], None),
+            (["dlc"], RomFileCategory.DLC),
+            (["update", "x"], RomFileCategory.UPDATE),
+            (["patches", "v2"], RomFileCategory.PATCH),
+            (["update", "content", "game"], RomFileCategory.UPDATE),
+            (["some game [update]", "content", "game", "stats"], None),
+            (["content", "cmcmn", "demo"], None),
+        ],
+    )
+    def test_category_for_path_parts(
+        self, parts: list[str], expected: RomFileCategory | None
+    ):
+        assert category_for_path_parts(parts) == expected
+
     def test_category_matches_plural_forms(self):
         assert category_matches("patch", ["patches"])
         assert category_matches("hack", ["hacks"])

backend/tests/handler/test_rom_files.py

@@ -7,7 +7,7 @@ import pytest
 
 from handler.database import db_rom_handler
 from handler.filesystem import fs_rom_handler
-from handler.filesystem.roms_handler import ParsedRomFiles
+from handler.filesystem.roms_handler import ParsedRomFiles, category_for_path_parts
 from handler.rom_files import refresh_rom_files
 from models.platform import Platform
 from models.rom import (
@@ -64,6 +64,9 @@ def _folder_rom(
                 rom_id=rom.id,
                 file_name=Path(rel).name,
                 file_path=str(Path(rom.fs_path, FOLDER, rel).parent),
+                category=category_for_path_parts(
+                    [part.lower() for part in Path(rel).parent.parts]
+                ),
                 file_size_bytes=st.st_size,
                 last_modified=st.st_mtime,
                 crc_hash="row-crc",