Added the shopping automation logic in a new tab

README.md

@@ -8,11 +8,33 @@ sdk_version: 5.42.0
 app_file: app.py
 pinned: false
 ---
-
-# 🤖 AI-Powered Virtual Try-On
+# AI Virtual Try-On & Fashion Advisor (Hugging Face Space)
 
 A portfolio-ready, full-stack AI fashion assistant and virtual try-on system. Combines computer vision, conversational AI, and web automation to deliver style advice, outfit analysis, and shopping recommendations—all in one interactive web app.
 
+## Features
+
+### 1. Try-On & Analysis
+- **Upload Clothing Image & Avatar Image:**
+	- Upload two images: one of a clothing item and one of an avatar/person.
+- **AI Try-On Generation:**
+	- Uses the [try-on-diffusion API](https://rapidapi.com/hlbq/api/try-on-diffusion/) to generate a virtual try-on result.
+- **BLIP/CLIP AI Analysis:**
+	- Automatically analyzes the generated try-on image using BLIP for captioning and fashion prompts.
+- **LLM Fashion Advice:**
+	- Provides detailed, context-aware fashion advice using TinyLlama LLM, based on the AI analysis.
+- **All results (image, analysis, advice) are displayed together.**
+
+### 2. Chatbot (TinyLlama LLM)
+- **Fashion Chatbot:**
+	- Chat with an AI stylist powered by TinyLlama.
+	- The chatbot always has context from your latest try-on and analysis, so you can ask about "this outfit" or "these colors" and get relevant answers.
+
+### 3. Find Similar
+- **Upload any try-on or fashion image.**
+- **Get recommendations for similar outfits, complementary accessories, and alternative colors.**
+- **Suggested stores are included in the recommendations.**
+
 ## 🚀 Features
 - **Virtual Try-On:** Upload clothing and avatar images to generate realistic try-on results.
 - **AI Fashion Analysis:** Get style advice and outfit analysis using BLIP and CLIP models.
@@ -26,24 +48,43 @@ A portfolio-ready, full-stack AI fashion assistant and virtual try-on system. Co
 - **AI Models:** BLIP, CLIP, TinyLlama (all open-source, loaded on demand)
 - **Automation:** Playwright for web scraping and product search
 
+## How It Works
+- All AI models (BLIP, CLIP, TinyLlama) are loaded and run in the backend.
+- The try-on image is generated via an external API (RapidAPI key required).
+- The app is organized into tabs for a clean, modern user experience.
+
+## Requirements
+- Python 3.8+
+- `gradio`, `transformers`, `torch`, `Pillow`, `requests`
+- A valid RapidAPI key for try-on-diffusion (replace in `app.py` if needed)
+
+## Usage
+1. **Clone this repo and open in Hugging Face Spaces or run locally.**
+2. **Install requirements:**
+	 ```bash
+	 pip install -r requirements.txt
+	 ```
+3. **Run the app:**
+	 ```bash
+	 python app.py
+	 ```
+4. **Use the web UI:**
+	 - Try-On & Analysis: Upload images, generate try-on, get analysis/advice.
+	 - Chatbot: Ask fashion questions with context.
+	 - Find Similar: Get recommendations for similar outfits.
+
 ## 🛠️ Setup & Deployment
 1. All dependencies are listed in `requirements.txt`.
 2. The backend Flask server is started automatically by `app.py`.
 3. The Gradio interface is the entry point for Hugging Face Spaces.
 
-## 💡 Usage
-- Upload a fashion image to get instant AI-powered analysis and advice.
-- Chat with the AI assistant for style tips and outfit suggestions.
-- Use the shopping automation to discover and compare similar products online.
-
-## 🌐 Public Demo
-Deploy this project on [Hugging Face Spaces](https://huggingface.co/spaces) for a live, shareable demo. Free CPU and GPU options available.
-
-## 📄 License
-Open-source, for educational and portfolio use. See project root for details.
-
----
+## Credits
+- BLIP: [Salesforce BLIP](https://huggingface.co/Salesforce/blip-image-captioning-base)
+- CLIP: [OpenAI CLIP](https://huggingface.co/openai/clip-vit-base-patch32)
+- LLM: [TinyLlama](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0)
+- Try-On API: [try-on-diffusion](https://rapidapi.com/hlbq/api/try-on-diffusion/)
 
-**For more details, see the full documentation and quick start guides in the main project repository.**
+## License
+MIT
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+---

app.py

@@ -130,7 +130,6 @@ def analyze_fashion(image):
     advice = llm_advise(sys_prompt, user_msg)
     return advice
 
-
 # --- Try-On API Integration ---
 import requests
 def tryon_api(clothing_img, avatar_img):
@@ -153,6 +152,115 @@ def tryon_api(clothing_img, avatar_img):
         return Image.open(io.BytesIO(response.content))
     else:
         raise Exception('Try-on generation failed: ' + response.text)
+    
+import asyncio
+import re
+try:
+    from playwright.async_api import async_playwright
+except ImportError:
+    async_playwright = None
+
+FASHION_SITES = {
+    'zara': {
+        'url': 'https://www.zara.com/us/en/search',
+        'search_param': 'searchTerm',
+        'selectors': {
+            'products': '.product-item',
+            'title': '.product-link',
+            'price': '.price',
+            'image': '.media-image img',
+            'link': '.product-link'
+        }
+    },
+    'hm': {
+        'url': 'https://www2.hm.com/en_us/search-results.html',
+        'search_param': 'q',
+        'selectors': {
+            'products': '.item-link',
+            'title': '.item-heading',
+            'price': '.item-price',
+            'image': '.item-image img',
+            'link': '.item-link'
+        }
+    },
+    'asos': {
+        'url': 'https://www.asos.com/us/search/',
+        'search_param': 'q',
+        'selectors': {
+            'products': '[data-testid="product-tile"]',
+            'title': '[data-testid="product-title"]',
+            'price': '[data-testid="current-price"]',
+            'image': 'img',
+            'link': 'a'
+        }
+    }
+}
+
+async def search_fashion_items(query, max_results=3):
+    results = []
+    if async_playwright is None:
+        return results
+    try:
+        async with async_playwright() as p:
+            browser = await p.chromium.launch(headless=True)
+            context = await browser.new_context(
+                user_agent='Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
+            )
+            for site_name, site_config in FASHION_SITES.items():
+                try:
+                    page = await context.new_page()
+                    search_url = f"{site_config['url']}?{site_config['search_param']}={query}"
+                    await page.goto(search_url, wait_until='networkidle', timeout=10000)
+                    await page.wait_for_timeout(2000)
+                    products = await page.query_selector_all(site_config['selectors']['products'])
+                    site_results = []
+                    for product in products[:max_results]:
+                        try:
+                            title_elem = await product.query_selector(site_config['selectors']['title'])
+                            price_elem = await product.query_selector(site_config['selectors']['price'])
+                            image_elem = await product.query_selector(site_config['selectors']['image'])
+                            link_elem = await product.query_selector(site_config['selectors']['link'])
+                            title = await title_elem.inner_text() if title_elem else 'N/A'
+                            price = await price_elem.inner_text() if price_elem else 'N/A'
+                            image_src = await image_elem.get_attribute('src') if image_elem else ''
+                            link_href = await link_elem.get_attribute('href') if link_elem else ''
+                            title = title.strip()[:100]
+                            price = re.sub(r'[^\\d.,$€£]', '', price) if price != 'N/A' else 'N/A'
+                            if link_href and not link_href.startswith('http'):
+                                base_url = f"https://{site_name}.com" if site_name != 'hm' else 'https://www2.hm.com'
+                                link_href = base_url + link_href
+                            site_results.append({
+                                'title': title,
+                                'price': price,
+                                'image': image_src,
+                                'link': link_href,
+                                'store': site_name.upper(),
+                                'query': query
+                            })
+                        except Exception:
+                            pass
+                    results.extend(site_results)
+                    await page.close()
+                except Exception:
+                    pass
+            await browser.close()
+    except Exception:
+        pass
+    return results
+
+def shopping_search_sync(query):
+    loop = asyncio.new_event_loop()
+    asyncio.set_event_loop(loop)
+    try:
+        results = loop.run_until_complete(search_fashion_items(query, max_results=3))
+    finally:
+        loop.close()
+    if not results:
+        return "No products found. Try a different keyword."
+    out = ""
+    for r in results:
+        out += f"\n**{r['title']}**\nStore: {r['store']}\nPrice: {r['price']}\n[View Product]({r['link']})\n\n"
+    return out
 
 # --- Gradio UI: Try-On Only ---
 with gr.Blocks() as demo:
@@ -232,5 +340,11 @@ with gr.Blocks() as demo:
         sim_btn = gr.Button("Find Similar Outfits")
         sim_output = gr.Textbox(label="Similar Outfits & Recommendations")
         sim_btn.click(recommend_similar, inputs=sim_input, outputs=sim_output)
+    
+    with gr.Tab("Shopping Automation"):
+        shop_query = gr.Textbox(label="Describe the item or keywords to search (e.g. 'black dress', 'summer shirt')")
+        shop_btn = gr.Button("Search Fashion Stores")
+        shop_results = gr.Markdown(label="Shopping Results")
+        shop_btn.click(shopping_search_sync, inputs=shop_query, outputs=shop_results)
 
 demo.launch()

requirements.txt

@@ -46,3 +46,4 @@ typing-inspection==0.4.1
 typing_extensions==4.14.0
 urllib3==2.5.0
 Werkzeug==3.1.3
+playwright