real-estate-ai-flutter-app

Nawy App - AI-Powered Real Estate Search

🤖 AI-powered real estate search app built with Flutter & Firebase AI. Features natural language property search using Gemini 2.5 Pro, advanced filtering, favorites, and responsive web deployment. Built with Clean Architecture, BLoC pattern, and comprehensive testing.

🚀 Try the app live →

✨ Features

• 🤖 AI Property Assistant – Natural language property search powered by Firebase AI & Gemini 2.5 Pro:
  • "Find me a 3-bedroom villa in New Cairo under 5M EGP"
• 🔧 Function Calling – AI can intelligently call property search functions with proper parameters
• 💬 Custom Chat UI – Beautiful chat interface with message bubbles, loading states, and auto-scrolling
• 🔍 Smart Search – Real-time text search across properties, areas, and compounds
• ⚡ Advanced Filtering – Filter by area, compound, price, bedrooms, and property type
• ❤️ Favorites – Save properties locally for offline access
• 📱 Modern UI – Clean, responsive design with iPhone frame simulation on web
• 🌐 Web Compatible – Full web deployment with responsive wrapper for desktop/mobile

📱 Screenshots

Property Search & Results

Advanced Filtering Options

AI-Powered Assistant

🏗️ Architecture

Built with modern Flutter practices and feature-first folder structure:

• Feature-first approach: Each feature is isolated in its own module
• State Management: BLoC pattern with improved state management architecture
• Local Storage: Hive CE database for favorites only (properties and compounds)
• Networking: Dio for HTTP requests with interceptors
• Dependency Injection: get_it with injectable for code generation
• Logging: Professional logging framework with AppLogger service

Tech Stack

• Framework: Flutter with Dart
• State Management: flutter_bloc + bloc_test
• Local Database: Hive CE (for favorites only) - Web Compatible
• Networking: Dio with interceptors
• Serialization: json_annotation + build_runner
• Dependency Injection: get_it + injectable (code generation)
• Testing: mocktail + bloc_test for comprehensive testing
• Logging: logger package with production-safe implementation

🔄 Database Migration: ObjectBox → Hive CE

Migrated from ObjectBox to Hive CE for full web platform support, enabling Firebase AI integration deployment without requiring reviewers to set up their own Firebase projects.

Key Benefits:

• ✅ Full web compatibility via IndexedDB
• ✅ Cross-platform support (mobile/desktop/web)
• ✅ Seamless Firebase AI integration on web

📁 Project Structure

lib/
├── main.dart               # Application entry point
├── app/
│   ├── core/               # Shared core functionality
│   │   ├── constants/      # App-wide constants
│   │   ├── utils/          # Utility functions & extensions
│   │   ├── widgets/        # Shared, reusable widgets
│   │   └── common/         # Shared types (e.g., Status<T>)
│   ├── features/           # Application features (modular)
│   │   ├── property_search/ # Property search feature
│   │   ├── favorites/      # Favorites management
│   │   ├── ai_assistant/   # AI assistant with Firebase AI
│   │   └── [feature]/      # Each feature module
│   │       ├── domain/     # Business logic (entities, repositories)
│   │       ├── data/       # Data layer (sources + models)
│   │       │   ├── remote/ # API sources and DTOs
│   │       │   └── local/  # Local sources and Hive entities
│   │       └── presentation/ # UI layer (bloc/pages/widgets)
│   └── widgets/            # Shared widgets

🚀 Quick Start

# Install dependencies
flutter pub get

# Generate code
dart run build_runner build --delete-conflicting-outputs

# Run the app
flutter run

🧪 Testing

Comprehensive test suite with 285 automated tests achieving 100% pass rate:

Test Architecture

• Unit Tests: Core utilities (DioClient, AppLogger, HiveService)
• DTO Tests: JSON serialization/deserialization for all models
• Integration Tests: API client integration with mocking
• Repository Tests: Domain layer business logic
• BLoC Tests: Business logic with bloc_test (17 tests)
• API Contract Tests: Live endpoint validation

Test Coverage Breakdown

• Core Utils: Error handling, DioClient, AppLogger, HiveService
• DTO Models: JSON serialization/deserialization for all models
• Remote Sources: API integration + text search functionality
• Domain Layer: Repository pattern and business logic
• BLoC Layer: State management with bloc_test
• AI Assistant: Firebase AI service and BLoC testing
• API Contract: Live endpoint validation
• Favorites: Local storage and repository testing

# Run all tests
flutter test

# Run specific test file
flutter test test/core/utils/dio_client_test.dart

# Run with coverage
flutter test --coverage

# Run tests with verbose output
flutter test --reporter=expanded

🤖 AI Assistant

Core Features

• 🧠 Natural Language Processing: Ask questions in plain English
• 🔧 Function Calling: AI intelligently calls property search functions
• 💬 Custom Chat UI: Beautiful message bubbles with loading states and auto-scrolling
• 🎯 Contextual Responses: AI understands property search context and provides relevant results
• ⚡ Real-time Interaction: Instant responses with proper error handling

Example Queries

Try these natural language searches with the AI assistant:

• Property Search: "Find me a 3-bedroom villa in New Cairo under 5 million EGP"
• Discover Areas: "What areas do you have available?" or "Show me all locations"
• Explore Compounds: "What compounds are available in Sheikh Zayed?"

🔍 Text Search Implementation

Comprehensive text search functionality allows users to find properties by typing search terms:

Search Capabilities

• Property Names: Search within property titles (e.g., "Luxury Apartment", "Modern Villa")
• Area Names: Search by location names (e.g., "El Sheikh Zayed", "New Cairo")
• Compound Names: Search by development names (e.g., "ZED West", "Bloomfields")

Search Features

• Case-Insensitive: "VILLA" and "villa" return identical results
• Partial Matching: "zayed" matches "El Sheikh Zayed"
• Real-time Filtering: Results update as you type with 500ms debounce
• Combined Search: Text search works alongside area, price, and bedroom filters
• Smart Query Handling: Empty/whitespace queries return all properties

⚠️ API Limitations Discovered

• Properties Search: Always returns the same 12 static properties regardless of any search filters
• No Server-Side Filtering: All query parameters (area_ids, compound_ids, price ranges, bedrooms, etc.) are completely ignored
• Static JSON Files: The API serves static JSON files instead of dynamic server responses
• No Pagination: No server-side pagination functionality is implemented

Workaround Implementation

To demonstrate proper search and filtering functionality that would work with a real API:

• Local Data Source: Created properties.json with 100+ realistic properties matching the exact API response structure
• Client-Side Filtering: Implemented comprehensive search and filtering logic in PropertyRemoteSource
• Preserved Network Code: All original API integration code is preserved and commented out with detailed explanations
• Production Ready: The architecture is ready to switch back to network calls when a properly functioning API is available

Files Modified for Workaround

• lib/app/features/property_search/data/sources/remote/property_remote_source.dart - Contains both commented network implementation and working local implementation
• properties.json - 100+ properties in exact API format for demonstration
• All original network integration code is preserved and can be easily restored

This approach demonstrates both proper API integration patterns and functional search/filtering capabilities that would work with a real backend API.