chore(romhacks): add extraction-cruft cleanup script (keeps archives, audits for loss)

stacks/romhacks/orchestrator/clean_extracted.py

@@ -0,0 +1,95 @@
+#!/usr/bin/env python3
+"""Remove JDownloader auto-extraction cruft from the romhacks library.
+
+For each library/<game>/ folder, list every archive's contents (bsdtar) and
+delete only the loose files whose path exactly matches an archive entry — i.e.
+files that are reproducible by re-extracting an archive that is STILL PRESENT.
+Archives themselves, art, guides/spreadsheets, and anything NOT contained in an
+archive are always kept. If a folder has no archive, nothing is deleted (the
+loose files are the only copy).
+
+Also audits for possible data loss: flags any folder that ends up with neither a
+source archive nor a rom file.
+
+CLEAN_DRY=true (default) previews only.
+"""
+import os
+import pathlib
+import subprocess
+import sys
+
+LIBRARY = pathlib.Path(os.environ.get("LIBRARY_DIR", "/library"))
+DRY = os.environ.get("CLEAN_DRY", "true").lower() != "false"
+ARCHIVE_EXTS = {".zip", ".rar", ".7z"}
+ROM_EXTS = {".gba", ".gb", ".gbc", ".nds", ".3ds", ".cia", ".nsp", ".xci"}
+
+
+def listing(path):
+    try:
+        out = subprocess.run(["bsdtar", "-tf", str(path)], capture_output=True,
+                             text=True, timeout=300)
+        return [l for l in out.stdout.splitlines() if l.strip()]
+    except Exception:
+        return []
+
+
+def main():
+    total_files = 0
+    total_bytes = 0
+    flagged = []
+    for folder in sorted(p for p in LIBRARY.iterdir() if p.is_dir()):
+        archives = [p for p in folder.rglob("*")
+                    if p.is_file() and p.suffix.lower() in ARCHIVE_EXTS]
+        # Resolve the set of paths each archive would extract to (relative to the
+        # archive's own directory) — these are the reproducible files.
+        extracted = set()
+        for a in archives:
+            for entry in listing(a):
+                extracted.add((a.parent / entry).resolve())
+
+        del_files = [p for p in folder.rglob("*")
+                     if p.is_file() and p.suffix.lower() not in ARCHIVE_EXTS
+                     and p.resolve() in extracted]
+        nbytes = sum(p.stat().st_size for p in del_files)
+
+        # Audit: will the folder still hold a recoverable rom (an archive, or a
+        # loose rom we are NOT deleting)?
+        keeps = bool(archives) or any(
+            p.suffix.lower() in ROM_EXTS and p not in del_files
+            for p in folder.rglob("*") if p.is_file())
+        if not keeps:
+            flagged.append(folder.name)
+
+        if del_files:
+            verb = "would delete" if DRY else "deleting"
+            print(f"[clean] {folder.name}: {verb} {len(del_files)} file(s) "
+                  f"{nbytes/1048576:.0f}MB (keeping {len(archives)} archive(s))")
+            total_files += len(del_files)
+            total_bytes += nbytes
+            if not DRY:
+                for p in del_files:
+                    try:
+                        p.unlink()
+                    except Exception as e:
+                        print(f"  unlink fail {p}: {e}", file=sys.stderr)
+        if not DRY:
+            for d in sorted((p for p in folder.rglob("*") if p.is_dir()),
+                            key=lambda x: len(x.parts), reverse=True):
+                try:
+                    d.rmdir()
+                except OSError:
+                    pass
+
+    print(f"\n[clean] {'WOULD reclaim' if DRY else 'reclaimed'} "
+          f"{total_bytes/1073741824:.2f} GB across {total_files} files; DRY={DRY}")
+    if flagged:
+        print(f"[clean] WARNING — {len(flagged)} folder(s) with NO archive and "
+              f"NO rom (possible loss / check these):")
+        for f in flagged:
+            print("   ", f)
+    else:
+        print("[clean] audit OK — every folder retains an archive or rom.")
+
+
+if __name__ == "__main__":
+    main()