restructreLibrary.py

import os
import re
import shutil

# --- Paths ---
entry_point = "/run/media/mohammed28/Ventoy/New folder/"  # messy books
organized_path = "/run/media/mohammed28/Ventoy/Organized_Library"  # organized library
os.makedirs(organized_path, exist_ok=True)

# --- File extensions to consider ---
book_extensions = [".epub", ".pdf", ".mobi"]

# --- Category keywords ---
categories_keywords = {
    'Linux': ['linux'],
    'Rust': ['rust'],
    'Haskell': ['haskell'],
    'Operating System': ['operating system', 'os'],
    'DevOps': ['devops'],
    'C': ['c programming', 'c language', 'learn c'],
    'C++': ['c++', 'cpp'],
    'C#': ['c#'],
    'Git': ['git'],
    'Linear Algebra': ['linear algebra'],
    'Discrete': ['discrete mathematics', 'discrete'],
    'Network': ['network'],
    'Arabic': []  # handled separately
}

# Dictionary to store categorized files
dic_paths_name_category = {key: [] for key in categories_keywords.keys()}
dic_paths_name_category["Others"] = []

# --- Helper functions ---
def clean_name(name):
    # Remove forbidden characters in folder names
    return re.sub(r'[\\/:"*?<>|]', "_", name).strip()

def contains_arabic(text):
    # Check if string contains Arabic letters
    return any('\u0600' <= c <= '\u06FF' for c in text)

# --- Scan files recursively ---
for root, dirs, files in os.walk(entry_point):
    for filename in files:
        ext = os.path.splitext(filename)[1].lower()
        if ext not in book_extensions:
            continue

        file_name = os.path.splitext(filename)[0]
        found_category = False

        # Check Arabic first
        if contains_arabic(file_name):
            dic_paths_name_category['Arabic'].append((filename, root))
            found_category = True
        else:
            # Check other categories using keywords
            for category, keywords in categories_keywords.items():
                if category == 'Arabic':
                    continue
                for kw in keywords:
                    if kw.lower() in file_name.lower():
                        dic_paths_name_category[category].append((filename, root))
                        found_category = True
                        break
                if found_category:
                    break

        # Uncategorized files
        if not found_category:
            dic_paths_name_category['Others'].append((filename, root))

# --- Create folders and copy files ---
for category, files in dic_paths_name_category.items():
    category_folder = os.path.join(organized_path, clean_name(category))
    os.makedirs(category_folder, exist_ok=True)

    for f, src_root in files:
        src_file = os.path.join(src_root, f)
        dst_file = os.path.join(category_folder, f)
        shutil.copy2(src_file, dst_file)  # preserves metadata
        print(f"Copied '{f}' → '{category_folder}/'")