Initial version

.gitignore

@@ -0,0 +1,37 @@
+# Environment variables
+.env
+
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+
+# Virtual Environment
+venv/
+ENV/
+env/
+.env/
+
+# IDE
+.idea/
+.vscode/
+*.swp
+*.swo
+.DS_Store

app.py

@@ -0,0 +1,164 @@
+import os
+import json
+import base64
+from typing import List, Tuple, Dict
+import gradio as gr
+import httpx
+from sqlalchemy import create_engine, text
+from dotenv import load_dotenv
+import google.generativeai as genai
+
+def get_secret(secret_name, service="", username=""):
+    try:
+        from google.colab import userdata
+        return userdata.get(secret_name)
+    except:
+        try:
+            return os.environ[secret_name]
+        except:
+            import keyring
+            return keyring.get_password(service, username)
+
+# Load environment variables
+load_dotenv()
+
+# Database configuration
+DB_NAME = "kroyscappingdb"
+DB_USER = "read_only"
+DB_PASSWORD = get_secret('FASHION_PG_PASS')
+DB_HOST = "rc1d-vbh2dw5ha0gpsazk.mdb.yandexcloud.net"
+DB_PORT = "6432"
+
+DATABASE_URL = f"postgresql://{DB_USER}:{DB_PASSWORD}@{DB_HOST}:{DB_PORT}/{DB_NAME}"
+
+# Create the SQLAlchemy engine
+db_conn = create_engine(DATABASE_URL)
+
+# Configure Gemini API
+genai.configure(api_key=get_secret("GEMINI_API_KEY"))
+
+def get_marketplace_and_main_image(id_product_money: str) -> Tuple[str, str]:
+    """Get marketplace and main image URL for a product."""
+    query = text("""
+        select mp, image as main_image_url
+        from public.products
+        where id_product_money = :id_product_money
+    """)
+    
+    with db_conn.connect() as connection:
+        result = connection.execute(query, {"id_product_money": id_product_money}).first()
+        if result is None:
+            raise ValueError(f"No product found with id_product_money: {id_product_money}")
+        return result.mp, result.main_image_url
+
+def get_additional_images(id_product_money: str, marketplace: str) -> List[str]:
+    """Get additional images based on marketplace."""
+    if marketplace == 'lamoda':
+        query = text("""
+            select info_chrc->'gallery' as more_images
+            from public.lamoda_chrc_and_reviews
+            where id_product_money = :id_product_money
+            limit 1
+        """)
+        with db_conn.connect() as connection:
+            result = connection.execute(query, {"id_product_money": id_product_money}).first()
+            if result and result.more_images:
+                paths = json.loads(result.more_images)
+                return [f"https://a.lmcdn.ru/product{path}" for path in paths]
+    
+    elif marketplace == 'wildberries':
+        query = text("""
+            select features->'images' as more_images
+            from public.wb_chrc
+            where id_product_money = :id_product_money
+            limit 1
+        """)
+        with db_conn.connect() as connection:
+            result = connection.execute(query, {"id_product_money": id_product_money}).first()
+            if result and result.more_images:
+                return json.loads(result.more_images)
+    
+    return []
+
+def download_and_encode_images(image_urls: List[str]) -> List[Dict]:
+    """Download images and convert them to base64 format for Gemini."""
+    encoded_images = []
+    with httpx.Client() as client:
+        for url in image_urls:
+            try:
+                response = client.get(url)
+                response.raise_for_status()
+                encoded_image = base64.b64encode(response.content).decode('utf-8')
+                encoded_images.append({
+                    'mime_type': 'image/jpeg',  # Assuming JPEG format
+                    'data': encoded_image
+                })
+            except Exception as e:
+                print(f"Error downloading image {url}: {str(e)}")
+    return encoded_images
+
+def get_gemini_response(model_name: str, encoded_images: List[Dict], prompt: str) -> str:
+    """Get response from a Gemini model."""
+    try:
+        model = genai.GenerativeModel(model_name)
+        response = model.generate_content(encoded_images + [prompt])
+        return response.text
+    except Exception as e:
+        return f"Error with {model_name}: {str(e)}"
+
+def process_input(id_product_money: str, prompt: str) -> Tuple[List[str], str, str]:
+    """Main processing function."""
+    try:
+        # Get marketplace and main image
+        marketplace, main_image = get_marketplace_and_main_image(id_product_money)
+        
+        # Get additional images
+        additional_images = get_additional_images(id_product_money, marketplace)
+        
+        # Combine all images
+        all_image_urls = [main_image] + additional_images
+        
+        # Download and encode images
+        encoded_images = download_and_encode_images(all_image_urls)
+        
+        if not encoded_images:
+            raise ValueError("No images could be downloaded")
+        
+        # Get responses from both models
+        gemini_1_5_response = get_gemini_response("gemini-1.5-pro", encoded_images, prompt)
+        gemini_2_0_response = get_gemini_response("gemini-pro-vision", encoded_images, prompt)
+        
+        return all_image_urls, gemini_1_5_response, gemini_2_0_response
+    
+    except Exception as e:
+        return [], f"Error: {str(e)}", f"Error: {str(e)}"
+
+# Create Gradio interface
+with gr.Blocks() as demo:
+    gr.Markdown("# Product Image Analysis with Gemini Models")
+    
+    with gr.Row():
+        id_input = gr.Textbox(label="Product ID (id_product_money)")
+        prompt_input = gr.Textbox(label="Prompt for VLMs")
+    
+    submit_btn = gr.Button("Analyze")
+    
+    with gr.Row():
+        image_gallery = gr.Gallery(label="Product Images")
+    
+    with gr.Row():
+        with gr.Column():
+            gr.Markdown("### Gemini 1.5 Pro Response")
+            gemini_1_5_output = gr.Textbox(label="")
+        with gr.Column():
+            gr.Markdown("### Gemini Pro Vision Response")
+            gemini_2_0_output = gr.Textbox(label="")
+    
+    submit_btn.click(
+        fn=process_input,
+        inputs=[id_input, prompt_input],
+        outputs=[image_gallery, gemini_1_5_output, gemini_2_0_output]
+    )
+
+if __name__ == "__main__":
+    demo.launch(share=True)

requirements.txt

@@ -0,0 +1,7 @@
+gradio
+sqlalchemy
+psycopg2-binary
+google-generativeai
+python-dotenv
+httpx
+grpcio