Group by Similarity

sortViaSimilarity.py

import os
import shutil
from difflib import SequenceMatcher

def similarity(a, b):
    return SequenceMatcher(None, a, b).ratio()

def main():
    base_dir = '.'  # Set the base directory to search for files
    similarity_threshold = 0.6
    file_groups = {}

    for root, _, files in os.walk(base_dir):
        for file in files:
            file_prefix = file[:10]  # Compare the first 10 characters of the file name
            grouped = False

            for group_prefix, group_files in file_groups.items():
                if similarity(file_prefix, group_prefix) >= similarity_threshold:
                    group_files.append(file)
                    grouped = True
                    break

            if not grouped:
                file_groups[file_prefix] = [file]

    # Move the files into their respective folders based on similarity
    for group_prefix, group_files in file_groups.items():
        new_folder_name = group_prefix + "_group"
        new_folder_path = os.path.join(base_dir, new_folder_name)

        if not os.path.exists(new_folder_path):
            os.makedirs(new_folder_path)

        for file in group_files:
            file_path = os.path.join(root, file)
            new_file_path = os.path.join(new_folder_path, file)
            shutil.move(file_path, new_file_path)

if __name__ == '__main__':
    main()