import streamlit as st import pandas as pd from collections import Counter import json import os import uuid from datetime import datetime import base64 import glob import re from typing import List, Dict, Tuple # AI/ML imports - lightweight transformers try: from sentence_transformers import SentenceTransformer from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification import torch import numpy as np from sklearn.metrics.pairwise import cosine_similarity AI_AVAILABLE = True except ImportError: AI_AVAILABLE = False st.warning("⚠️ AI features require additional packages. Install with: pip install sentence-transformers transformers torch scikit-learn") # --- Page Configuration --- st.set_page_config( page_title="రుచి చూడు (Ruchi Chudu)", page_icon="🍲", layout="wide" ) # --- AI Model Management --- @st.cache_resource def load_ai_models(): """Load AI models with caching to prevent reloading""" if not AI_AVAILABLE: return None, None, None try: # Lightweight sentence transformer for recipe similarity similarity_model = SentenceTransformer('all-MiniLM-L6-v2') # Sentiment analysis for recipe stories sentiment_analyzer = pipeline( "sentiment-analysis", model="cardiffnlp/twitter-roberta-base-sentiment-latest", return_all_scores=True ) # Text classification for cuisine type - using a lighter model cuisine_classifier = pipeline( "text-classification", model="cardiffnlp/twitter-roberta-base-sentiment-latest" # Reuse sentiment model ) return similarity_model, sentiment_analyzer, cuisine_classifier except Exception as e: st.error(f"Error loading AI models: {e}") return None, None, None # --- Enhanced AI Functions --- def analyze_recipe_sentiment(story: str, sentiment_analyzer) -> Dict: """Analyze the emotional tone of recipe stories""" if not story or not sentiment_analyzer: return {"sentiment": "neutral", "confidence": 0.5, "emotion": "nostalgic"} try: results = sentiment_analyzer(story[:500]) # Limit text length # Map sentiment labels to readable format sentiment_map = { 'LABEL_0': 'negative', 'LABEL_1': 'neutral', 'LABEL_2': 'positive', 'NEGATIVE': 'negative', 'NEUTRAL': 'neutral', 'POSITIVE': 'positive' } top_result = max(results[0], key=lambda x: x['score']) sentiment = sentiment_map.get(top_result['label'], 'neutral') confidence = top_result['score'] # Add emotion context for Telugu recipes emotion_keywords = { 'joyful': ['celebration', 'festival', 'happy', 'joy', 'smile', 'laugh'], 'nostalgic': ['grandmother', 'childhood', 'memory', 'tradition', 'old', 'ammamma'], 'proud': ['special', 'unique', 'best', 'famous', 'perfect', 'excellent'], 'loving': ['family', 'mother', 'love', 'care', 'warmth', 'together'] } story_lower = story.lower() emotion = 'nostalgic' # default for recipe stories for emotion_type, keywords in emotion_keywords.items(): if any(keyword in story_lower for keyword in keywords): emotion = emotion_type break return { "sentiment": sentiment, "confidence": confidence, "emotion": emotion } except Exception as e: return {"sentiment": "neutral", "confidence": 0.5, "emotion": "nostalgic"} def extract_smart_tags(recipe_data: Dict, similarity_model) -> List[str]: """Extract intelligent tags using NLP and pattern matching""" tags = set() # Combine text for analysis text_content = f"{recipe_data.get('dish_name', '')} {recipe_data.get('ingredients', '')} {recipe_data.get('instructions', '')} {recipe_data.get('story', '')}" # Traditional keyword-based tagging (enhanced) keyword_categories = { 'spice_level': { 'mild': ['mild', 'less spicy', 'children', 'sweet'], 'medium': ['medium', 'moderate', 'balanced'], 'hot': ['spicy', 'hot', 'chili', 'pepper', 'karam', 'guntur'], 'very_hot': ['very spicy', 'extra hot', 'burning', 'fire'] }, 'cooking_time': { 'quick': ['quick', 'fast', 'minutes', '15 min', '20 min', 'instant'], 'medium': ['30 min', '45 min', '1 hour', 'moderate time'], 'slow': ['hours', 'slow cook', 'overnight', 'patience'] }, 'meal_type': { 'breakfast': ['breakfast', 'morning', 'tiffin', 'idli', 'dosa'], 'lunch': ['lunch', 'meal', 'rice', 'curry', 'sambar'], 'dinner': ['dinner', 'night', 'heavy'], 'snack': ['snack', 'evening', 'tea time', 'biscuit'] }, 'health': { 'healthy': ['healthy', 'nutritious', 'vitamin', 'protein', 'fiber'], 'diabetic_friendly': ['sugar free', 'diabetic', 'low sugar', 'jaggery'], 'high_protein': ['protein', 'dal', 'lentils', 'sprouts'], 'low_fat': ['oil free', 'steamed', 'boiled', 'grilled'] }, 'occasion': { 'festival': ['festival', 'celebration', 'ugadi', 'sankranti', 'diwali'], 'wedding': ['wedding', 'marriage', 'function', 'ceremony'], 'everyday': ['daily', 'regular', 'simple', 'routine'], 'special': ['special', 'guest', 'party', 'occasion'] } } text_lower = text_content.lower() for category, subcategories in keyword_categories.items(): for tag, keywords in subcategories.items(): if any(keyword in text_lower for keyword in keywords): tags.add(tag.replace('_', ' ').title()) # Telugu-specific patterns telugu_patterns = { 'Regional': ['Andhra', 'Telangana', 'Hyderabad', 'Vijayawada', 'Guntur'], 'Traditional': ['traditional', 'authentic', 'original', 'sampradayam'], 'Modern': ['fusion', 'modern', 'new style', 'innovative'], 'Street Food': ['street', 'vendor', 'roadside', 'chat'] } for tag, patterns in telugu_patterns.items(): if any(pattern.lower() in text_lower for pattern in patterns): tags.add(tag) # Ingredient-based classification ingredient_text = recipe_data.get('ingredients', '').lower() # Detect vegetarian/non-vegetarian non_veg_ingredients = ['chicken', 'mutton', 'fish', 'egg', 'meat', 'prawn', 'crab'] if any(ingredient in ingredient_text for ingredient in non_veg_ingredients): tags.add('Non-Vegetarian') else: tags.add('Vegetarian') # Detect main ingredients main_ingredients = { 'Rice Based': ['rice', 'biryani', 'pulao', 'annam'], 'Dal Based': ['dal', 'lentil', 'pappu', 'sambar'], 'Vegetable': ['vegetable', 'curry', 'fry', 'kura'], 'Sweet': ['sweet', 'sugar', 'jaggery', 'dessert', 'halwa'], 'Pickle': ['pickle', 'achar', 'pachadi'] } for tag, ingredients in main_ingredients.items(): if any(ingredient in ingredient_text for ingredient in ingredients): tags.add(tag) return list(tags) def find_similar_recipes(current_recipe: Dict, all_recipes: List[Dict], similarity_model, top_k=3) -> List[Tuple[Dict, float]]: """Find similar recipes using semantic similarity""" if not similarity_model or len(all_recipes) < 2: return [] try: # Create text representations current_text = f"{current_recipe.get('dish_name', '')} {current_recipe.get('ingredients', '')} {current_recipe.get('recipe_type', '')}" recipe_texts = [] valid_recipes = [] for recipe in all_recipes: if recipe.get('id') != current_recipe.get('id'): # Exclude current recipe recipe_text = f"{recipe.get('dish_name', '')} {recipe.get('ingredients', '')} {recipe.get('recipe_type', '')}" recipe_texts.append(recipe_text) valid_recipes.append(recipe) if not recipe_texts: return [] # Generate embeddings current_embedding = similarity_model.encode([current_text]) recipe_embeddings = similarity_model.encode(recipe_texts) # Calculate similarities similarities = cosine_similarity(current_embedding, recipe_embeddings)[0] # Get top similar recipes similar_indices = np.argsort(similarities)[::-1][:top_k] similar_recipes = [] for idx in similar_indices: if similarities[idx] > 0.3: # Minimum similarity threshold similar_recipes.append((valid_recipes[idx], similarities[idx])) return similar_recipes except Exception as e: st.error(f"Error finding similar recipes: {e}") return [] def generate_cooking_tips(recipe_data: Dict) -> List[str]: """Generate contextual cooking tips based on recipe content""" tips = [] ingredients = recipe_data.get('ingredients', '').lower() instructions = recipe_data.get('instructions', '').lower() dish_name = recipe_data.get('dish_name', '').lower() # Ingredient-specific tips if 'oil' in ingredients: tips.append("🔥 Tip: Heat oil until it shimmers but doesn't smoke for best results") if 'onion' in ingredients: tips.append("🧅 Tip: Soak sliced onions in cold water for 10 minutes to reduce tears while cutting") if any(spice in ingredients for spice in ['turmeric', 'chili', 'coriander']): tips.append("🌶️ Tip: Dry roast spices lightly before grinding for enhanced flavor") if 'rice' in ingredients: tips.append("🍚 Tip: Soak rice for 20-30 minutes before cooking for better texture") # Cooking method tips if 'fry' in instructions or 'deep fry' in instructions: tips.append("🍳 Tip: Maintain oil temperature at 350°F (175°C) for perfect frying") if 'boil' in instructions: tips.append("💧 Tip: Add salt to boiling water for vegetables to retain color and nutrients") # Dish-specific tips if 'curry' in dish_name: tips.append("🍛 Tip: Let curry rest for 10 minutes after cooking to allow flavors to meld") if 'pickle' in dish_name: tips.append("🥒 Tip: Store pickle in airtight containers and use dry spoons to prevent spoilage") # Regional tips tips.append("🏠 Traditional Tip: Taste and adjust seasoning as family preferences vary by region") return tips[:3] # Return top 3 tips # --- Data Storage Functions --- RECIPES_FOLDER = "recipes" IMAGES_FOLDER = "recipe_images" def ensure_folders_exist(): """Create necessary folders if they don't exist""" os.makedirs(RECIPES_FOLDER, exist_ok=True) os.makedirs(IMAGES_FOLDER, exist_ok=True) def generate_recipe_id(): """Generate a unique ID for each recipe""" return str(uuid.uuid4())[:8] def save_recipe(recipe_data): """Save individual recipe to its own JSON file""" try: ensure_folders_exist() recipe_id = recipe_data['id'] filename = f"recipe_{recipe_id}.json" filepath = os.path.join(RECIPES_FOLDER, filename) with open(filepath, 'w', encoding='utf-8') as f: json.dump(recipe_data, f, ensure_ascii=False, indent=2) return True except Exception as e: st.error(f"Error saving recipe: {e}") return False def load_all_recipes(): """Load all recipes from individual JSON files""" recipes = [] try: ensure_folders_exist() recipe_files = glob.glob(os.path.join(RECIPES_FOLDER, "recipe_*.json")) for filepath in recipe_files: try: with open(filepath, 'r', encoding='utf-8') as f: recipe = json.load(f) recipes.append(recipe) except Exception as e: continue recipes.sort(key=lambda x: x.get('submitted_date', ''), reverse=True) except Exception as e: st.error(f"Error loading recipes: {e}") return recipes # --- Language and Text --- TEXT = { "en": { "title": "Ruchi Chudu", "subtitle": "AI-Powered Community Cookbook to Preserve Telugu Cuisine", "language_select": "Choose Language", "sidebar_header": "Contribute Your Recipe!", "tab_submit": "📝 Submit Your Recipe", "tab_gallery": "🍲 Community Gallery", "tab_ai_insights": "🤖 AI Recipe Insights", "tab_analytics": "📊 Recipe Analytics", "form_header": "Tell us about your dish", "dish_name": "Dish Name", "your_name": "Your Name", "district": "Your District", "select_district": "--- Select a District ---", "recipe_type": "Recipe Type (e.g., Curry, Pickle, Snack)", "ingredients": "Ingredients (one per line)", "instructions": "Instructions", "story": "The Story Behind Your Dish (your memories, family traditions, etc.)", "image_upload": "Upload a photo of the dish (optional)", "submit_button": "Submit Recipe", "success_message": "Thank you! Your recipe has been submitted successfully.", "gallery_header": "Explore Recipes from the Community", "ai_analysis": "🤖 AI Analysis", "similar_recipes": "👥 Similar Recipes", "cooking_tips": "💡 Smart Cooking Tips", "recipe_sentiment": "😊 Recipe Story Mood", "search_placeholder": "Search recipes or ask AI...", "ai_search_help": "Try: 'spicy vegetarian curry' or 'quick breakfast recipe'", }, "te": { "title": "రుచి చూడు", "subtitle": "AI శక్తితో మన తెలుగు వంటల సంస్కృతిని కాపాడడం", "language_select": "భాషను ఎంచుకోండి", "sidebar_header": "మీ వంటకాన్ని పంపండి!", "tab_submit": "📝 మీ వంటకాన్ని పంపండి", "tab_gallery": "🍲 కమ్యూనిటీ గ్యాలరీ", "tab_ai_insights": "🤖 AI వంటకాల విశ్లేషణ", "tab_analytics": "📊 వంటకాల విశ్లేషణ", "form_header": "మీ వంటకం గురించి చెప్పండి", "dish_name": "వంటకం పేరు", "your_name": "మీ పేరు", "district": "మీ జిల్లా", "select_district": "--- జిల్లాను ఎంచుకోండి ---", "recipe_type": "వంటకం రకం (ఉదా. కూర, పచ్చడి, స్నాక్)", "ingredients": "కావలసినవి (ఒకదానికి ఒకటి)", "instructions": "తయారీ విధానం", "story": "ఈ వంటతో మీ కథ (మీ జ్ఞాపకాలు, కుటుంబ సంప్రదాయాలు మొదలైనవి)", "image_upload": "వంటకం ఫోటోను అప్‌లోడ్ చేయండి (ఐచ్ఛికం)", "submit_button": "వంటకాన్ని పంపండి", "success_message": "ధన్యవాదాలు! మీ వంటకం విజయవంతంగా సమర్పించబడింది.", "gallery_header": "కమ్యూనిటీ నుంది వంటకాలను అన్వేషించండి", "ai_analysis": "🤖 AI విశ్లేషణ", "similar_recipes": "👥 సమాన వంటకాలు", "cooking_tips": "💡 తెలివైన వంట చిట్కాలు", "recipe_sentiment": "😊 వంటకం కథ భావం", "search_placeholder": "వంటకాలను వెతకండి లేదా AI ని అడగండి...", "ai_search_help": "ప్రయత్నించండి: 'కారంగా వెజిటేరియన్ కూర' లేదా 'తొందర అల్పాహారం'", } } # --- Data for Dropdowns --- DISTRICTS = [ "Adilabad", "Bhadradri Kothagudem", "Hanumakonda", "Hyderabad", "Jagtial", "Jangaon", "Jayashankar Bhupalpally", "Jogulamba Gadwal", "Kamareddy", "Karimnagar", "Khammam", "Komaram Bheem", "Mahabubabad", "Mahbubnagar", "Mancherial", "Medak", "Medchal-Malkajgiri", "Mulugu", "Nagarkurnool", "Nalgonda", "Narayanpet", "Nirmal", "Nizamabad", "Peddapalli", "Rajanna Sircilla", "Ranga Reddy", "Sangareddy", "Siddipet", "Suryapet", "Vikarabad", "Wanaparthy", "Warangal", "Yadadri Bhuvanagiri", "Alluri Sitharama Raju", "Anakapalli", "Anantapur", "Annamayya", "Bapatla", "Chittoor", "East Godavari", "Eluru", "Guntur", "Kakinada", "Konaseema", "Krishna", "Kurnool", "Nandyal", "NTR", "Palnadu", "Parvathipuram Manyam", "Prakasam", "Sri Potti Sriramulu Nellore", "Sri Sathya Sai", "Srikakulam", "Tirupati", "Visakhapatnam", "Vizianagaram", "West Godavari", "YSR Kadapa" ] DISTRICTS.sort() RECIPE_TYPES = ["Curry (కూర)", "Fry (వేపుడు)", "Pickle (పచ్చడి)", "Chutney (చట్నీ)", "Pulusu (పులుసు)", "Snack (చిరుతిండి)", "Sweet (తీపి)", "Breakfast (అల్పాహారం)", "Rice Dish (అన్నం రకం)", "Other (ఇతర)"] # --- Session State Initialization --- if 'language' not in st.session_state: st.session_state['language'] = 'en' if 'recipes' not in st.session_state: st.session_state['recipes'] = load_all_recipes() if 'ai_models_loaded' not in st.session_state: st.session_state['ai_models_loaded'] = False # --- Helper Functions --- def get_text(key): """Fetches text from the dictionary based on the selected language.""" return TEXT[st.session_state['language']][key] # --- Load AI Models --- similarity_model, sentiment_analyzer, cuisine_classifier = None, None, None if AI_AVAILABLE and not st.session_state.get('ai_models_loaded', False): with st.spinner("🤖 Loading AI models for the first time... This may take a moment."): similarity_model, sentiment_analyzer, cuisine_classifier = load_ai_models() if similarity_model: st.session_state['ai_models_loaded'] = True st.success("✅ AI models loaded successfully!") else: similarity_model, sentiment_analyzer, cuisine_classifier = load_ai_models() # --- Sidebar --- with st.sidebar: st.title(f"🍲 {get_text('title')}") st.markdown(get_text('subtitle')) if AI_AVAILABLE and similarity_model: st.success("🤖 AI Features: Active") else: st.warning("🤖 AI Features: Limited") st.markdown("---") # Language selector lang_choice = st.radio( get_text('language_select'), ('English', 'తెలుగు'), horizontal=True, key='lang_radio' ) st.session_state['language'] = 'te' if lang_choice == 'తెలుగు' else 'en' st.info(get_text('sidebar_header')) # --- Main App Layout --- st.title(get_text('title')) # Create tabs including the new AI insights tab tab_submit, tab_gallery, tab_ai, tab_analytics = st.tabs([ get_text('tab_submit'), get_text('tab_gallery'), get_text('tab_ai_insights'), get_text('tab_analytics') ]) # --- Submission Form Tab (Enhanced with AI) --- with tab_submit: st.header(get_text('form_header')) with st.form(key="recipe_form"): col1, col2 = st.columns(2) with col1: dish_name = st.text_input(label=get_text('dish_name')) your_name = st.text_input(label=get_text('your_name')) image_file = st.file_uploader(get_text('image_upload'), type=['jpg', 'jpeg', 'png']) with col2: recipe_type = st.selectbox(label=get_text('recipe_type'), options=RECIPE_TYPES) district = st.selectbox( label=get_text('district'), options=[get_text('select_district')] + DISTRICTS ) ingredients = st.text_area(label=get_text('ingredients'), height=150) instructions = st.text_area(label=get_text('instructions'), height=200) story = st.text_area(label=get_text('story'), height=150) submitted = st.form_submit_button(get_text('submit_button')) if submitted: if not dish_name: st.error("Please enter a dish name before submitting.") else: recipe_id = generate_recipe_id() # Create recipe data for AI analysis recipe_data = { "id": recipe_id, "dish_name": dish_name, "your_name": your_name, "district": district if district != get_text('select_district') else '', "recipe_type": recipe_type, "ingredients": ingredients, "instructions": instructions, "story": story, "submitted_date": datetime.now().strftime("%Y-%m-%d %H:%M:%S") } # AI Analysis if AI_AVAILABLE and sentiment_analyzer: with st.spinner("🤖 AI is analyzing your recipe..."): # Sentiment analysis sentiment_data = analyze_recipe_sentiment(story, sentiment_analyzer) # Smart tag extraction ai_tags = extract_smart_tags(recipe_data, similarity_model) # Add AI analysis to recipe data recipe_data.update({ "ai_tags": ai_tags, "sentiment_analysis": sentiment_data, "cooking_tips": generate_cooking_tips(recipe_data) }) # Show AI insights st.subheader("🤖 AI Analysis Results") col1, col2 = st.columns(2) with col1: st.metric("Recipe Mood", sentiment_data['emotion'].title(), f"{sentiment_data['confidence']:.0%} confidence") st.write("**AI Tags:**", ", ".join([f"`{tag}`" for tag in ai_tags])) with col2: st.write("**Smart Cooking Tips:**") for tip in recipe_data['cooking_tips']: st.write(f"• {tip}") else: # Fallback basic tagging recipe_data["ai_tags"] = ["Community Recipe"] recipe_data["sentiment_analysis"] = {"sentiment": "positive", "emotion": "nostalgic"} recipe_data["cooking_tips"] = ["Follow the recipe step by step for best results."] # Save recipe if save_recipe(recipe_data): st.session_state.recipes.insert(0, recipe_data) st.success(f"{get_text('success_message')} (ID: {recipe_id})") st.balloons() else: st.error("Failed to save recipe. Please try again.") # --- Community Gallery Tab (Enhanced with AI) --- with tab_gallery: st.header(get_text('gallery_header')) # AI-powered search if AI_AVAILABLE and similarity_model: st.subheader("🔍 AI-Powered Recipe Search") search_query = st.text_input( "Search recipes with natural language:", placeholder=get_text('ai_search_help'), help="Ask in natural language like 'show me spicy vegetarian curries' or 'quick breakfast ideas'" ) if search_query and st.session_state.recipes: with st.spinner("🤖 AI is searching..."): # Use semantic search query_embedding = similarity_model.encode([search_query]) recipe_texts = [] valid_recipes = [] for recipe in st.session_state.recipes: recipe_text = f"{recipe.get('dish_name', '')} {recipe.get('ingredients', '')} {recipe.get('recipe_type', '')} {' '.join(recipe.get('ai_tags', []))}" recipe_texts.append(recipe_text) valid_recipes.append(recipe) if recipe_texts: recipe_embeddings = similarity_model.encode(recipe_texts) similarities = cosine_similarity(query_embedding, recipe_embeddings)[0] # Get top matches top_indices = np.argsort(similarities)[::-1][:10] matching_recipes = [] for idx in top_indices: if similarities[idx] > 0.2: # Minimum similarity matching_recipes.append((valid_recipes[idx], similarities[idx])) if matching_recipes: st.success(f"🎯 Found {len(matching_recipes)} recipes matching your search") display_recipes = [recipe for recipe, _ in matching_recipes] else: st.info("No recipes found matching your search. Try different keywords.") display_recipes = st.session_state.recipes[:5] else: display_recipes = st.session_state.recipes else: display_recipes = st.session_state.recipes else: # Fallback to basic search search_term = st.text_input("Search recipes:", placeholder=get_text('search_placeholder')) if search_term: display_recipes = [r for r in st.session_state.recipes if search_term.lower() in r['dish_name'].lower() or search_term.lower() in r.get('ingredients', '').lower()] else: display_recipes = st.session_state.recipes # Display recipes with AI insights for recipe in display_recipes[:10]: # Show top 10 with st.expander(f"🍽️ {recipe['dish_name']} (ID: {recipe.get('id', 'N/A')})"): # Basic recipe info col1, col2, col3 = st.columns(3) with col1: st.write(f"**By:** {recipe.get('your_name', 'Anonymous')}") st.write(f"**District:** {recipe.get('district', 'Not specified')}") st.write(f"**Type:** {recipe.get('recipe_type', 'Not specified')}") with col2: # AI Tags if recipe.get('ai_tags'): st.write("**AI Tags:**") for tag in recipe['ai_tags'][:3]: # Show top 3 tags st.badge(tag) # Sentiment analysis if recipe.get('sentiment_analysis'): sentiment = recipe['sentiment_analysis'] st.write(f"**Story Mood:** {sentiment['emotion'].title()}") with col3: st.write(f"**Submitted:** {recipe.get('submitted_date', 'Unknown')}") if recipe.get('cooking_tips'): with st.popover("💡 Cooking Tips"): for tip in recipe['cooking_tips']: st.write(f"• {tip}") # Recipe content st.markdown("### Ingredients:") st.text(recipe.get('ingredients', 'No ingredients provided')) st.markdown("### Instructions:") st.text(recipe.get('instructions', 'No instructions provided')) if recipe.get('story'): st.markdown("### Story:") st.text(recipe['story']) # AI-powered similar recipes if AI_AVAILABLE and similarity_model and len(st.session_state.recipes) > 1: similar_recipes = find_similar_recipes(recipe, st.session_state.recipes, similarity_model) if similar_recipes: st.markdown("### 👥 Similar Recipes:") sim_cols = st.columns(len(similar_recipes)) for idx, (sim_recipe, similarity) in enumerate(similar_recipes): with sim_cols[idx]: st.write(f"**{sim_recipe['dish_name']}**") st.write(f"Similarity: {similarity:.0%}") st.caption(f"By {sim_recipe.get('your_name', 'Anonymous')}") # --- AI Insights Tab --- with tab_ai: st.header("🤖 AI Recipe Insights") if not AI_AVAILABLE or not similarity_model: st.warning("AI features are not available. Please install required packages to enable AI insights.") st.code("pip install sentence-transformers transformers torch scikit-learn") st.stop() if not st.session_state.recipes: st.info("No recipes available for analysis. Submit some recipes first!") st.stop() # AI Analysis Dashboard st.subheader("📊 Community Recipe Analysis") # Sentiment distribution sentiments = [recipe.get('sentiment_analysis', {}).get('emotion', 'nostalgic') for recipe in st.session_state.recipes] sentiment_counts = Counter(sentiments) col1, col2 = st.columns(2) with col1: st.markdown("**Recipe Story Moods:**") for emotion, count in sentiment_counts.most_common(): percentage = (count / len(st.session_state.recipes)) * 100 st.write(f"😊 {emotion.title()}: {count} recipes ({percentage:.1f}%)") with col2: # Most common AI tags all_tags = [] for recipe in st.session_state.recipes: all_tags.extend(recipe.get('ai_tags', [])) if all_tags: tag_counts = Counter(all_tags) st.markdown("**Most Popular Recipe Types:**") for tag, count in tag_counts.most_common(5): st.write(f"🏷️ {tag}: {count} recipes") st.markdown("---") # Recipe Recommendation Engine st.subheader("🎯 AI Recipe Recommendations") col1, col2 = st.columns(2) with col1: preference = st.selectbox( "What are you in the mood for?", ["Spicy dishes", "Sweet treats", "Quick meals", "Traditional recipes", "Healthy options", "Festival specials", "Comfort food"] ) with col2: dietary_pref = st.selectbox( "Dietary preference:", ["Any", "Vegetarian only", "Non-vegetarian", "Healthy options"] ) if st.button("🤖 Get AI Recommendations"): with st.spinner("🤖 AI is finding perfect recipes for you..."): # Create recommendation query based on preferences query_map = { "Spicy dishes": "spicy hot chili pepper karam", "Sweet treats": "sweet jaggery sugar dessert halwa", "Quick meals": "quick fast instant easy", "Traditional recipes": "traditional authentic grandmother ammamma", "Healthy options": "healthy nutritious protein vitamin", "Festival specials": "festival celebration ugadi sankranti", "Comfort food": "comfort home family love" } search_query = query_map.get(preference, preference) # Filter by dietary preference filtered_recipes = st.session_state.recipes if dietary_pref == "Vegetarian only": filtered_recipes = [r for r in filtered_recipes if 'Vegetarian' in r.get('ai_tags', [])] elif dietary_pref == "Non-vegetarian": filtered_recipes = [r for r in filtered_recipes if 'Non-Vegetarian' in r.get('ai_tags', [])] elif dietary_pref == "Healthy options": filtered_recipes = [r for r in filtered_recipes if any(tag in ['Healthy', 'High Protein', 'Low Fat'] for tag in r.get('ai_tags', []))] if filtered_recipes: # Use AI to find best matches query_embedding = similarity_model.encode([search_query]) recipe_texts = [] for recipe in filtered_recipes: recipe_text = f"{recipe.get('dish_name', '')} {recipe.get('ingredients', '')} {' '.join(recipe.get('ai_tags', []))}" recipe_texts.append(recipe_text) if recipe_texts: recipe_embeddings = similarity_model.encode(recipe_texts) similarities = cosine_similarity(query_embedding, recipe_embeddings)[0] # Get top recommendations top_indices = np.argsort(similarities)[::-1][:3] st.success(f"🎯 Here are your personalized recommendations:") for idx in top_indices: if similarities[idx] > 0.1: # Minimum relevance recipe = filtered_recipes[idx] match_score = similarities[idx] * 100 with st.container(): st.markdown(f"### 🍽️ {recipe['dish_name']}") col1, col2, col3 = st.columns(3) with col1: st.write(f"**Match Score:** {match_score:.0f}%") st.write(f"**By:** {recipe.get('your_name', 'Anonymous')}") with col2: st.write(f"**Type:** {recipe.get('recipe_type', 'N/A')}") st.write(f"**District:** {recipe.get('district', 'N/A')}") with col3: if recipe.get('ai_tags'): st.write("**Tags:**") for tag in recipe['ai_tags'][:2]: st.badge(tag) if recipe.get('story'): st.write(f"*{recipe['story'][:150]}...*") st.markdown("---") else: st.info("No recipes found matching your preferences. Try different criteria!") st.markdown("---") # Recipe Insights st.subheader("🔍 Deep Recipe Analysis") if st.session_state.recipes: selected_recipe = st.selectbox( "Select a recipe to analyze:", options=[f"{r['dish_name']} (by {r.get('your_name', 'Anonymous')})" for r in st.session_state.recipes], index=0 ) if selected_recipe: # Find the selected recipe recipe_idx = next(i for i, r in enumerate(st.session_state.recipes) if f"{r['dish_name']} (by {r.get('your_name', 'Anonymous')})" == selected_recipe) recipe = st.session_state.recipes[recipe_idx] col1, col2 = st.columns(2) with col1: st.markdown("**🤖 AI Analysis:**") # Sentiment analysis details if recipe.get('sentiment_analysis'): sentiment = recipe['sentiment_analysis'] st.metric( "Story Emotion", sentiment['emotion'].title(), f"{sentiment['confidence']:.0%} confidence" ) # AI tags if recipe.get('ai_tags'): st.write("**AI-Detected Categories:**") for tag in recipe['ai_tags']: st.badge(tag) with col2: st.markdown("**💡 Smart Cooking Tips:**") if recipe.get('cooking_tips'): for tip in recipe['cooking_tips']: st.info(tip) else: # Generate tips on demand tips = generate_cooking_tips(recipe) for tip in tips: st.info(tip) # Similar recipes st.markdown("**👥 Similar Recipes:**") similar_recipes = find_similar_recipes(recipe, st.session_state.recipes, similarity_model, top_k=5) if similar_recipes: for sim_recipe, similarity in similar_recipes: with st.expander(f"{sim_recipe['dish_name']} - {similarity:.0%} similar"): st.write(f"**By:** {sim_recipe.get('your_name', 'Anonymous')}") st.write(f"**Type:** {sim_recipe.get('recipe_type', 'N/A')}") if sim_recipe.get('story'): st.write(f"**Story:** {sim_recipe['story'][:200]}...") else: st.info("No similar recipes found in the current database.") # --- Analytics Tab --- with tab_analytics: st.header("📊 Recipe Analytics Dashboard") if not st.session_state.recipes: st.info("No recipes available for analysis. Submit some recipes first!") st.stop() # Basic Statistics total_recipes = len(st.session_state.recipes) st.metric("Total Recipes", total_recipes) # Create dataframe for analysis df_data = [] for recipe in st.session_state.recipes: df_data.append({ 'dish_name': recipe.get('dish_name', ''), 'your_name': recipe.get('your_name', 'Anonymous'), 'district': recipe.get('district', 'Unknown'), 'recipe_type': recipe.get('recipe_type', 'Other'), 'submitted_date': recipe.get('submitted_date', ''), 'ai_tags': recipe.get('ai_tags', []), 'emotion': recipe.get('sentiment_analysis', {}).get('emotion', 'neutral') }) df = pd.DataFrame(df_data) # District-wise distribution col1, col2 = st.columns(2) with col1: st.subheader("📍 Recipes by District") district_counts = df['district'].value_counts().head(10) if not district_counts.empty: st.bar_chart(district_counts) else: st.info("No district data available") with col2: st.subheader("🍽️ Recipe Types Distribution") type_counts = df['recipe_type'].value_counts() if not type_counts.empty: st.bar_chart(type_counts) else: st.info("No recipe type data available") # Contributors st.subheader("👥 Top Contributors") contributor_counts = df['your_name'].value_counts().head(5) for name, count in contributor_counts.items(): st.write(f"🏆 {name}: {count} recipe(s)") # Emotional analysis if AI_AVAILABLE: st.subheader("😊 Recipe Story Emotions") emotion_counts = df['emotion'].value_counts() emotion_colors = { 'joyful': '#FFD700', 'nostalgic': '#DDA0DD', 'proud': '#FF6347', 'loving': '#FF69B4', 'neutral': '#87CEEB' } emotion_data = [] for emotion, count in emotion_counts.items(): emotion_data.append({ 'Emotion': emotion.title(), 'Count': count, 'Percentage': f"{(count/total_recipes)*100:.1f}%" }) emotion_df = pd.DataFrame(emotion_data) st.dataframe(emotion_df, use_container_width=True) # Recent activity st.subheader("📅 Recent Recipe Submissions") recent_recipes = st.session_state.recipes[:5] # Show 5 most recent for recipe in recent_recipes: with st.container(): col1, col2, col3 = st.columns([3, 2, 1]) with col1: st.write(f"**{recipe['dish_name']}**") st.caption(f"by {recipe.get('your_name', 'Anonymous')}") with col2: st.write(f"{recipe.get('recipe_type', 'N/A')}") st.caption(f"{recipe.get('district', 'Unknown')}") with col3: st.write(recipe.get('submitted_date', 'Unknown')[:10]) # Show date only st.markdown("---") # Export functionality st.subheader("📤 Export Data") if st.button("Download Recipe Database as CSV"): # Flatten the data for CSV export export_data = [] for recipe in st.session_state.recipes: export_data.append({ 'ID': recipe.get('id', ''), 'Dish Name': recipe.get('dish_name', ''), 'Contributor': recipe.get('your_name', ''), 'District': recipe.get('district', ''), 'Recipe Type': recipe.get('recipe_type', ''), 'Ingredients': recipe.get('ingredients', ''), 'Instructions': recipe.get('instructions', ''), 'Story': recipe.get('story', ''), 'AI Tags': ', '.join(recipe.get('ai_tags', [])), 'Emotion': recipe.get('sentiment_analysis', {}).get('emotion', ''), 'Submitted Date': recipe.get('submitted_date', '') }) export_df = pd.DataFrame(export_data) csv_data = export_df.to_csv(index=False) st.download_button( label="📥 Download CSV", data=csv_data, file_name=f"ruchi_chudu_recipes_{datetime.now().strftime('%Y%m%d')}.csv", mime="text/csv" ) # --- Footer --- st.markdown("---") st.markdown("""

🍲 రుచి చూడు (Ruchi Chudu) - Preserving Telugu Cuisine Through Community & AI

Made with ❤️ for Telugu food lovers everywhere

AI Features powered by Hugging Face Transformers | Data stored locally

""", unsafe_allow_html=True)