refactor(clew): recreate extracted game tree

src/clew/clew.py

@@ -10,7 +10,7 @@ from pathlib import Path
 from rpycdec.rpa import extract_rpa
 
 LOGGER = logging.getLogger("clew")
-MANAGED_DIRS = ("source", "work", "logs")
+MANAGED_DIRS = ("extracted", "logs")
 
 
 def parse_args(argv: list[str]) -> argparse.Namespace:
@@ -28,7 +28,7 @@ def parse_args(argv: list[str]) -> argparse.Namespace:
         "--output-root",
         type=Path,
         default=Path("output"),
-        help="Preparation output root. Defaults to repository output/.",
+        help="Output root. Defaults to repository output/.",
     )
     parser.add_argument(
         "--force",
@@ -58,76 +58,28 @@ def resolve_output_root(output_root: Path) -> Path:
     return output_root.resolve()
 
 
-def prepare_output(output_root: Path, force: bool) -> tuple[Path, Path, Path]:
+def prepare_output(output_root: Path, force: bool) -> tuple[Path, Path]:
     for name in MANAGED_DIRS:
         path = output_root / name
         if path.exists() and not force:
-            fail(f"refusing to overwrite existing preparation path: {path}")
+            fail(f"refusing to overwrite existing output path: {path}")
 
     if force:
         for name in MANAGED_DIRS:
             shutil.rmtree(output_root / name, ignore_errors=True)
 
-    source_dir = output_root / "source"
-    work_dir = output_root / "work"
+    extracted_dir = output_root / "extracted"
     logs_dir = output_root / "logs"
-    for path in (source_dir, work_dir, logs_dir):
+    for path in (extracted_dir, logs_dir):
         path.mkdir(parents=True, exist_ok=True)
 
-    return source_dir, work_dir, logs_dir
+    return extracted_dir, logs_dir
 
 
 def relative_to(root: Path, path: Path) -> Path:
     return path.resolve().relative_to(root.resolve())
 
 
-def is_renpy_runtime_path(relative: Path) -> bool:
-    parts = relative.parts
-    return bool(parts) and (parts[0] == "renpy" or "renpy" in parts[:-1])
-
-
-def copy_source_file(
-    source_file: Path,
-    input_root: Path,
-    source_dir: Path,
-    label: str,
-    destination_prefix: Path = Path(),
-) -> None:
-    relative = relative_to(input_root, source_file)
-    output_relative = destination_prefix / relative
-    if is_renpy_runtime_path(relative):
-        LOGGER.info("skipped %s Ren'Py runtime source: %s", label, output_relative)
-        return
-
-    destination = source_dir / output_relative
-    destination.parent.mkdir(parents=True, exist_ok=True)
-    shutil.copy2(source_file, destination)
-    LOGGER.info("copied %s source: %s", label, output_relative)
-
-
-def copy_compiled_file(compiled_file: Path, input_root: Path, work_dir: Path) -> None:
-    relative = relative_to(input_root, compiled_file)
-    if is_renpy_runtime_path(relative):
-        LOGGER.info("skipped Ren'Py runtime compiled script: %s", relative)
-        return
-
-    destination = work_dir / "compiled" / relative
-    destination.parent.mkdir(parents=True, exist_ok=True)
-    shutil.copy2(compiled_file, destination)
-    LOGGER.info("copied compiled script: %s", relative)
-
-
-def archive_work_path(work_dir: Path, archive_relative: Path) -> Path:
-    return work_dir / "archives" / archive_relative.with_suffix("")
-
-
-def archive_destination_prefix(archive_relative: Path) -> Path:
-    parent = archive_relative.parent
-    if str(parent) == ".":
-        return Path()
-    return parent
-
-
 def normalize_suffixes(suffixes: list[str] | None) -> list[str] | None:
     if suffixes is None:
         return None
@@ -142,15 +94,22 @@ def normalize_suffixes(suffixes: list[str] | None) -> list[str] | None:
     return normalized
 
 
+def copy_loose_file(file: Path, game_root: Path, extracted_dir: Path) -> None:
+    relative = relative_to(game_root, file)
+    destination = extracted_dir / relative
+    destination.parent.mkdir(parents=True, exist_ok=True)
+    shutil.copy2(file, destination)
+    LOGGER.info("copied loose file: %s", relative)
+
+
 def extract_archive(
     archive: Path,
     game_root: Path,
-    source_dir: Path,
-    work_dir: Path,
+    extracted_dir: Path,
     suffixes: list[str] | None,
 ) -> None:
     archive_relative = relative_to(game_root, archive)
-    output_dir = archive_work_path(work_dir, archive_relative)
+    output_dir = extracted_dir / archive_relative.with_suffix("")
     output_dir.mkdir(parents=True, exist_ok=True)
 
     LOGGER.info("extracting archive: %s", archive_relative)
@@ -158,16 +117,6 @@ def extract_archive(
         extract_rpa(handle, output_dir=str(output_dir), suffixes=suffixes)
     LOGGER.info("extracted archive: %s", archive_relative)
 
-    destination_prefix = archive_destination_prefix(archive_relative)
-    for source_file in sorted(output_dir.rglob("*.rpy")):
-        copy_source_file(
-            source_file,
-            output_dir,
-            source_dir,
-            "archived",
-            destination_prefix,
-        )
-
 
 def configure_logging(logs_dir: Path) -> None:
     log_file = logs_dir / "clew.log"
@@ -190,21 +139,17 @@ def run(args: argparse.Namespace) -> None:
         fail(f"game root is not a directory: {game_root}")
 
     suffixes = normalize_suffixes(args.suffix)
-    source_dir, work_dir, logs_dir = prepare_output(output_root, args.force)
+    extracted_dir, logs_dir = prepare_output(output_root, args.force)
     configure_logging(logs_dir)
 
-    LOGGER.info("scanning loose source files")
-    for source_file in sorted(game_root.rglob("*.rpy")):
-        copy_source_file(source_file, game_root, source_dir, "loose")
+    LOGGER.info("copying loose files")
+    for file in sorted(game_root.rglob("*")):
+        if file.is_file() and file.suffix.lower() != ".rpa":
+            copy_loose_file(file, game_root, extracted_dir)
 
-    LOGGER.info("scanning loose compiled scripts")
-    for suffix in ("*.rpyc", "*.rpymc"):
-        for compiled_file in sorted(game_root.rglob(suffix)):
-            copy_compiled_file(compiled_file, game_root, work_dir)
-
-    LOGGER.info("scanning archives")
+    LOGGER.info("extracting archives")
     for archive in sorted(game_root.rglob("*.rpa")):
-        extract_archive(archive, game_root, source_dir, work_dir, suffixes)
+        extract_archive(archive, game_root, extracted_dir, suffixes)
 
     LOGGER.info("done")