Add gc_disc.py

src/retro_data_structures/asset_manager.py

@@ -7,6 +7,7 @@
 import typing
 from collections import defaultdict
 
+import construct
 import nod
 
 from retro_data_structures import formats
@@ -24,13 +25,12 @@
 from retro_data_structures.formats import Dgrp, dependency_cheating
 from retro_data_structures.formats.audio_group import Agsc, Atbl
 from retro_data_structures.formats.pak import Pak
+from retro_data_structures.game_disc import GameDisc
 
 if typing.TYPE_CHECKING:
     from collections.abc import Iterator
     from pathlib import Path
 
-    import construct
-
     from retro_data_structures.formats.ancs import Ancs
     from retro_data_structures.game_check import Game
 
@@ -48,6 +48,9 @@ def rglob(self, pattern: str) -> Iterator[str]:
     def open_binary(self, name: str) -> typing.BinaryIO:
         raise NotImplementedError
 
+    def read_binary(self, name: str) -> bytes:
+        raise NotImplementedError
+
     def get_dol(self) -> bytes:
         raise NotImplementedError
 
@@ -71,24 +74,40 @@ def rglob(self, name: str) -> Iterator[str]:
     def open_binary(self, name: str) -> typing.BinaryIO:
         return self.root.joinpath(name).open("rb")
 
+    def read_binary(self, name: str) -> bytes:
+        with self.open_binary(name) as f:
+            return f.read()
+
     def get_dol(self) -> bytes:
         with self.open_binary("sys/main.dol") as f:
             return f.read()
 
 
 class IsoFileProvider(FileProvider):
+    game_disc: GameDisc | None
+
     def __init__(self, iso_path: Path):
-        result = nod.open_disc_from_image(iso_path)
-        if result is None:
-            raise ValueError(f"{iso_path} is not a GC/Wii ISO")
+        self.iso_path = iso_path
 
-        self.disc = result[0]
-        self.data = self.disc.get_data_partition()
-        if self.data is None:
-            raise ValueError(f"{iso_path} does not have data")
+        self.game_disc = None
 
-        self.all_files = self.data.files()
-        self.iso_path = iso_path
+        try:
+            self.game_disc = GameDisc.parse(iso_path)
+            self.all_files = self.game_disc.files()
+
+        except construct.ConstError:
+            # Fallback to nod, likely a Wii ISO
+
+            result = nod.open_disc_from_image(iso_path)
+            if result is None:
+                raise ValueError(f"{iso_path} is not a GC/Wii ISO")
+
+            self.disc = result[0]
+            self.data = self.disc.get_data_partition()
+            if self.data is None:
+                raise ValueError(f"{iso_path} does not have data")
+
+            self.all_files = self.data.files()
 
     def __repr__(self):
         return f"<IsoFileProvider {self.iso_path}>"
@@ -102,7 +121,17 @@ def rglob(self, pattern: str) -> Iterator[str]:
                 yield it
 
     def open_binary(self, name: str):
-        return self.data.read_file(name)
+        if self.game_disc is None:
+            return self.data.read_file(name)
+        else:
+            return self.game_disc.open_binary(name)
+
+    def read_binary(self, name: str) -> bytes:
+        if self.game_disc is None:
+            with self.open_binary(name) as f:
+                return f.read()
+        else:
+            return self.game_disc.read_binary(name)
 
     def get_dol(self) -> bytes:
         return self.data.get_dol()
@@ -154,8 +183,7 @@ def _update_headers(self):
 
         self._custom_asset_ids = {}
         if self.provider.is_file("custom_names.json"):
-            with self.provider.open_binary("custom_names.json") as f:
-                custom_names_text = f.read().decode("utf-8")
+            custom_names_text = self.provider.read_binary("custom_names.json").decode("utf-8")
 
             self._custom_asset_ids.update(dict(json.loads(custom_names_text).items()))
 
@@ -380,8 +408,7 @@ def get_pak(self, pak_name: str) -> Pak:
 
         if pak_name not in self._in_memory_paks:
             logger.info("Reading %s", pak_name)
-            with self.provider.open_binary(pak_name) as f:
-                data = f.read()
+            data = self.provider.read_binary(pak_name)
 
             self._in_memory_paks[pak_name] = Pak.parse(data, target_game=self.target_game)
 

src/retro_data_structures/game_disc.py

@@ -0,0 +1,121 @@
+from __future__ import annotations
+
+import collections
+import dataclasses
+import io
+import typing
+from pathlib import Path
+
+import construct
+
+from retro_data_structures.gc_disc import GcDisc
+
+
+@dataclasses.dataclass
+class FileEntry:
+    offset: int
+    size: int
+
+
+FileTree: typing.TypeAlias = dict[str, typing.Union[FileEntry, "FileTree"]]
+
+
+class GameDisc:
+    _file_path: Path
+    _raw: construct.Container
+    _file_tree: FileTree
+
+    def __init__(self, file_path: Path, raw: construct.Container, file_tree: FileTree):
+        self._file_path = file_path
+        self._raw = raw
+        self._file_tree = file_tree
+
+    @classmethod
+    def parse(cls, file_path: Path) -> GameDisc:
+        with file_path.open("rb") as source:
+            data = GcDisc.parse_stream(source)
+
+        file_tree: dict = {}
+        current_dir = file_tree
+
+        end_folder = collections.defaultdict(list)
+
+        names_stream = io.BytesIO(data.fst.names)
+        for i, file in enumerate(data.fst.file_entries):
+            if i == 0:
+                continue
+
+            if i in end_folder:
+                current_dir = end_folder.pop(i)[0]
+
+            names_stream.seek(file.file_name)
+            name = construct.CString("ascii").parse_stream(names_stream)
+            if file.is_directory:
+                new_dir = {}
+                end_folder[file.param].append(current_dir)
+                current_dir[name] = new_dir
+                current_dir = new_dir
+            else:
+                current_dir[name] = FileEntry(
+                    offset=file.offset,
+                    size=file.param,
+                )
+
+        return GameDisc(file_path, data, file_tree)
+
+    def _get_file_entry(self, name: str) -> FileEntry:
+        file_entry = self._file_tree
+        for segment in name.split("/"):
+            file_entry = file_entry[segment]
+
+        if isinstance(file_entry, FileEntry):
+            return file_entry
+        else:
+            raise OSError(f"{name} is a directory")
+
+    def files(self) -> list[str]:
+        result = []
+
+        def recurse(parent: str, tree: FileTree) -> None:
+            for key, item in tree.items():
+                name = f"{parent}/{key}" if parent else key
+
+                if isinstance(item, FileEntry):
+                    result.append(name)
+                else:
+                    recurse(name, item)
+
+        recurse("", self._file_tree)
+        return result
+
+    def open_binary(self, name: str) -> typing.BinaryIO:
+        entry = self._get_file_entry(name)
+        file = self._file_path.open("rb")
+        file.seek(entry.offset)
+        return file
+
+    def read_binary(self, name: str) -> bytes:
+        entry = self._get_file_entry(name)
+        with self._file_path.open("rb") as file:
+            file.seek(entry.offset)
+            return file.read(entry.size)
+
+
+def main():
+    # file = r"F:\Hauzer\GameCube\Metroid Prime 2 - Echoes (USA).iso"
+    file = r"F:\Hauzer\GameCube\Kirby Air Ride.iso"
+    file = r"F:\Hauzer\Wii\Metroid Prime 3 - Corruption [RM3E01]\Metroid Prime 3.iso"
+
+    GameDisc.parse(Path(file))
+
+    # data = GcDisc.parse_file(file)
+    # print(data)
+    #
+    # names_stream = io.BytesIO(data.fst.names)
+    # for file in data.fst.file_entries:
+    #     names_stream.seek(file.file_name)
+    #     print(file.is_directory, file.param, construct.CString("ascii").parse_stream(names_stream))
+
+
+if __name__ == "__main__":
+    main()

src/retro_data_structures/gc_disc.py

@@ -0,0 +1,79 @@
+from __future__ import annotations
+
+import construct
+
+# boot.bin
+DiscHeader = construct.Struct(
+    game_code=construct.Bytes(4),
+    maker_code=construct.Bytes(2),
+    disc_id=construct.Int8ub,  # for multi-disc games
+    version=construct.Int8ub,
+    audio_streaming=construct.Int8ub,
+    stream_buffer_size=construct.Int8ub,
+    _unused_a=construct.Const(b"\x00" * 14),
+    _wii_magic_word=construct.Const(0, construct.Int32ub),  # 0x5D1C9EA3
+    _gc_magic_word=construct.Const(0xC2339F3D, construct.Int32ub),
+    game_name=construct.PaddedString(0x3E0, "utf8"),
+    debug_monitor_offset=construct.Int32ub,
+    debug_monitor_load_address=construct.Int32ub,
+    _unused_b=construct.Const(b"\x00" * 24),
+    main_executable_offset=construct.Int32ub,
+    fst_offset=construct.Int32ub,
+    fst_size=construct.Int32ub,
+    fst_maximum_size=construct.Int32ub,
+    user_position=construct.Int32ub,
+    user_length=construct.Int32ub,
+    unknown=construct.Int32ub,
+    _unused_c=construct.Const(b"\x00" * 4),  # construct.Bytes(0x4),
+)
+assert DiscHeader.sizeof() == 0x0440
+
+DiscHeaderInformation = construct.Struct(
+    debug_monitor_size=construct.Int32ub,
+    simulated_memory_size=construct.Int32ub,
+    argument_offset=construct.Int32ub,
+    debug_flag=construct.Int32ub,
+    track_address=construct.Int32ub,
+    track_size=construct.Int32ub,
+    country_code=construct.Int32ub,
+    unknown=construct.Int32ub,
+    padding=construct.Bytes(8160),
+)
+assert DiscHeaderInformation.sizeof() == 0x2000
+
+AppLoader = construct.Struct(
+    date=construct.Aligned(16, construct.Bytes(10)),
+    entry_point=construct.Hex(construct.Int32ub),
+    _size=construct.Rebuild(construct.Int32ub, construct.len_(construct.this.code)),
+    trailer_size=construct.Int32ub,
+    code=construct.Bytes(construct.this._size),
+)
+
+FileEntry = construct.Struct(
+    is_directory=construct.Flag,
+    file_name=construct.Int24ub,
+    offset=construct.Int32ub,
+    param=construct.Int32ub,
+)
+RootFileEntry = construct.Struct(
+    is_directory=construct.Const(True, construct.Flag),
+    file_name=construct.Const(0, construct.Int24ub),
+    _offset=construct.Const(0, construct.Int32ub),
+    num_entries=construct.Int32ub,
+)
+
+GcDisc = construct.Struct(
+    header=DiscHeader,
+    header_information=DiscHeaderInformation,
+    app_loader=AppLoader,
+    root_offset=construct.Tell,
+    _fst_seek=construct.Seek(construct.this.header.fst_offset),
+    fst=construct.FixedSized(
+        construct.this.header.fst_size,
+        construct.Struct(
+            root_entry=construct.Peek(RootFileEntry),
+            file_entries=FileEntry[construct.this.root_entry.num_entries],
+            names=construct.GreedyBytes,
+        ),
+    ),
+)

tests/formats/test_pak_gc.py

@@ -171,8 +171,7 @@ def test_echoes_resource_encode_decode(compressed_resource):
 
 def test_identical_when_keep_data(prime2_iso_provider):
     game = Game.ECHOES
-    with prime2_iso_provider.open_binary("GGuiSys.pak") as f:
-        raw = f.read()
+    raw = prime2_iso_provider.read_binary("GGuiSys.pak")
 
     decoded = Pak.parse(raw, target_game=game)
     encoded = decoded.build()
@@ -182,8 +181,7 @@ def test_identical_when_keep_data(prime2_iso_provider):
 
 def test_compare_header_keep_data(prime2_iso_provider):
     game = Game.ECHOES
-    with prime2_iso_provider.open_binary("GGuiSys.pak") as f:
-        raw = f.read()
+    raw = prime2_iso_provider.read_binary("GGuiSys.pak")
 
     raw_header = PAKNoData.parse(raw, target_game=game)
     raw_sizes = [(r.compressed, r.offset, r.size) for r in raw_header.resources]