# unity_parts_compare.py # Usage: # python unity_parts_compare.py from pathlib import Path # ===== CONFIGURATION ===== # Set your folder path here (works with Windows, Mac, Linux) # Examples: # r"C:\Users\Vinay\Desktop\unity" # "C:/Users/Vinay/Desktop/unity" # "/Users/vinay/Desktop/unity" INPUT_FOLDER = Path(r"C:\Users\Vinay\Desktop\unity") OUTPUT_FOLDER = INPUT_FOLDER # Put all results back in the same folder OUTPUT_COMMON_NAME = "unity_common_parts.txt" # Parts common to ALL files OUTPUT_UNIQUE_SUFFIX = "_unique.txt" # Deduped list per file OUTPUT_EXCLUSIVE_SUFFIX = "_exclusive.txt" # Only in this file, nowhere else OUTPUT_NONCOMMON_SUFFIX = "_noncommon.txt" # In this file, but not common-to-all OUTPUT_SUMMARY_NAME = "unity_summary.txt" # Summary report # ========================= def read_set(path: Path) -> set: s = set() with path.open("r", encoding="utf-8", errors="ignore") as f: for line in f: part = line.rstrip("\n\r") if part: s.add(part) return s def safe_name(p: Path) -> str: return p.stem.replace(" ", "_") def main(): if not INPUT_FOLDER.exists() or not INPUT_FOLDER.is_dir(): print(f"ERROR: Input folder '{INPUT_FOLDER}' not found.") return OUTPUT_FOLDER.mkdir(parents=True, exist_ok=True) files = sorted(INPUT_FOLDER.glob("*.txt")) if len(files) < 2: print(f"Need at least two .txt files inside '{INPUT_FOLDER}'. Found {len(files)}.") return sets = [(p, read_set(p)) for p in files] # Common to ALL common_all = set.intersection(*(s for _, s in sets)) common_path = OUTPUT_FOLDER / OUTPUT_COMMON_NAME with common_path.open("w", encoding="utf-8") as out: for part in sorted(common_all): out.write(part + "\n") union_all = set.union(*(s for _, s in sets)) summary_lines = [f"Processed folder: {INPUT_FOLDER.resolve()}", f"Files: {len(sets)}", ""] for p, s in sets: base = safe_name(p) # Unique (deduped per file) uniq_path = OUTPUT_FOLDER / f"{base}{OUTPUT_UNIQUE_SUFFIX}" with uniq_path.open("w", encoding="utf-8") as out: for part in sorted(s): out.write(part + "\n") # Exclusive (only in this file) exclusive = s - (union_all - s) excl_path = OUTPUT_FOLDER / f"{base}{OUTPUT_EXCLUSIVE_SUFFIX}" with excl_path.open("w", encoding="utf-8") as out: for part in sorted(exclusive): out.write(part + "\n") # Non-common (remove global common) noncommon = s - common_all noncommon_path = OUTPUT_FOLDER / f"{base}{OUTPUT_NONCOMMON_SUFFIX}" with noncommon_path.open("w", encoding="utf-8") as out: for part in sorted(noncommon): out.write(part + "\n") summary_lines.append( f"{p.name}: {len(s)} unique | {len(exclusive)} exclusive | {len(noncommon)} non-common" ) summary_lines.append("") summary_lines.append( f"Common to ALL {len(sets)} files: {len(common_all)} -> {OUTPUT_COMMON_NAME}" ) # Save summary summary_path = OUTPUT_FOLDER / OUTPUT_SUMMARY_NAME with summary_path.open("w", encoding="utf-8") as f: f.write("\n".join(summary_lines)) print("\n".join(summary_lines)) if __name__ == "__main__": main()