compare2xxh3.py

import os
import difflib
from collections import defaultdict

def find_xxh3_files(directory):
    xxh3_files = []
    for root, _, files in os.walk(directory):
        for file in files:
            if file.endswith('.xxh3'):
                xxh3_files.append(os.path.join(root, file))
    return xxh3_files

def read_file_content(file_path):
    with open(file_path, 'r') as file:
        return file.readlines()

def find_closest_match(file_content, comparison_files):
    closest_match = None
    highest_similarity = 0
    
    for comp_file, comp_content in comparison_files.items():
        similarity = difflib.SequenceMatcher(None, file_content, comp_content).ratio()
        if similarity > highest_similarity:
            highest_similarity = similarity
            closest_match = comp_file
    
    return closest_match, highest_similarity

def print_diff(file1, file2):
    file1_content = read_file_content(file1)
    file2_content = read_file_content(file2)
    
    diff = difflib.unified_diff(file1_content, file2_content, fromfile=file1, tofile=file2)
    print("\n".join(diff))

def main(dir1, dir2):
    files_dir1 = find_xxh3_files(dir1)
    files_dir2 = find_xxh3_files(dir2)
    
    if not files_dir1 or not files_dir2:
        print("No .xxh3 files found in one or both directories.")
        return
    
    dir2_contents = {file: read_file_content(file) for file in files_dir2}
    
    for file1 in files_dir1:
        file1_content = read_file_content(file1)
        closest_match, similarity = find_closest_match(file1_content, dir2_contents)

        print("###################################\n"*3)
        
        if closest_match:
            print(f"\nComparing {file1} to {closest_match} (Similarity: {similarity:.2f}):")
            print_diff(file1, closest_match)
        else:
            print(f"No match found for {file1}.")

if __name__ == "__main__":
    dir1 = ""
    dir2 = ""
    main(dir1, dir2)