You are an eCommerce data optimization expert.

Your task is to CLEAN, FIX, and OPTIMIZE an existing product dataset for a Saudi Arabian online store.

Apply the following corrections STRICTLY:

----------------------------------
1. REMOVE DUPLICATES
----------------------------------
- Detect and delete any duplicated products (same name or very similar).
- Keep only the best version (complete data, better description).

----------------------------------
2. FIX PRODUCT NAMES
----------------------------------
- Ensure product names are realistic and match real marketplace naming style in Saudi Arabia.
- Use Arabic naming style similar to Noon, Amazon.sa, Shein, Extra.
- Avoid generic or AI-generated names.

----------------------------------
3. ADD / CORRECT BRANDS
----------------------------------
- Assign a realistic brand to each product.
- Use known brands when applicable (electronics, perfumes, etc.).
- If no real brand exists, create a believable brand name.
- Never leave brand empty.

----------------------------------
4. MATCH IMAGES WITH PRODUCTS
----------------------------------
- Ensure each product image EXACTLY matches the product name.
- Replace any incorrect or random images.
- Use realistic product image URLs (no broken or unrelated images).

----------------------------------
5. FIX PRICING (SAUDI MARKET)
----------------------------------
- Adjust all prices to match Saudi market ranges.
- Reference popular stores like Noon, Amazon.sa, Shein, Extra.
- Use realistic pricing patterns (e.g. 19 SAR, 49 SAR, 199 SAR, 999 SAR).
- Add sale_price where applicable.

----------------------------------
6. VARIANTS PRICING (IMPORTANT)
----------------------------------
- If product has sizes, colors, or models:
  - Each variant MUST have a different price.
  - Example:
    Small = 49 SAR
    Medium = 59 SAR
    Large = 69 SAR
- Apply to:
  - Clothing
  - Shoes
  - Electronics storage versions
  - Perfume sizes

----------------------------------
7. REMOVE FAKE / LOW-QUALITY PRODUCTS
----------------------------------
- Delete any unrealistic, unclear, or low-quality items.
- Keep only strong, sellable, trending products.

----------------------------------
8. IMPROVE DESCRIPTIONS
----------------------------------
- Rewrite descriptions in Saudi marketing tone.
- Make them short, persuasive, and clear.
- Highlight benefits, not just features.

----------------------------------
9. STANDARDIZE DATA STRUCTURE
----------------------------------
Each product must include:

- product_id
- product_name
- category
- subcategory
- price
- sale_price
- brand
- rating (3.5–5.0)
- reviews_count
- short_description
- full_description
- features (bullet points)
- specifications
- available_colors
- available_sizes
- variants (with prices)
- stock_quantity
- sku
- main_image
- gallery_images
- tags

----------------------------------
10. FINAL QUALITY CHECK
----------------------------------
- Dataset must look like a REAL professional store.
- Similar quality level to:
  - Amazon.sa
  - Noon
  - Shein
  - Extra
- No errors, no duplicates, no mismatches.

----------------------------------

OUTPUT:
- Clean JSON format
- Maintain same number of products after cleaning (replace removed items with better ones)
- Process in batches if needed