gistfile1.txt

#!/usr/bin/env python3
"""
Test script to browser-use.

This script attempts to:
1. Search for flights to Singapore on Google Flights
2. Set departure date to September 17th and return date to September 24th  
3. Extract top 4 flight options in JSON format with keys:
   airline, departure_time, arrival_time, duration, total_amount, route
"""

import asyncio
import os
import json
from datetime import datetime
from browser_use import Agent, ChatOpenAI, BrowserProfile
import logging

from dotenv import load_dotenv
load_dotenv()

async def test_flight_search():
    # Enable detailed logging
    logging.basicConfig(
        level=logging.INFO,
        format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
    )
        
    # Check if we have API keys
    openai_key = os.getenv('OPENAI_API_KEY')
    assert openai_key
    
    task = """
    Search for flights from 'Seoul' to "Singapore" on Google Flights (google.com/travel/flights) for 1 adult.
    Set departure date to September 17th and return date to September 24th.
    Extract top 4 flight options and return them in JSON format with these exact keys:
    airline, departure_time, arrival_time, duration, total_amount, route

    Return only the JSON array, nothing else.
    """
    
    try:
        print("🛫 Starting flight search automation with browser-use...")
        
        # Initialize ChatOpenAI with Claude Sonnet 4 via OpenRouter
        llm = ChatOpenAI(
            model="openai/gpt-4.1-mini",
            api_key=openai_key,
            base_url="https://openrouter.ai/api/v1"
        )
        
        # Create browser profile
        browser_profile = BrowserProfile(
            headless=True,
            highlight_elements=True,  # Enable boxes and numbers around clickable elements  
            disable_security=True,
            args=[
                "--no-sandbox",
                "--disable-dev-shm-usage",
                "--enable-logging",
                "--log-level=0",
                "--v=1"
            ]
        )
        
        print("🔧 Creating agent...")
        
        # Create browser-use agent with GIF generation
        agent = Agent(
            task=task,
            llm=llm,
            browser_profile=browser_profile,
            generate_gif="flight_search.gif"  # Generate GIF with visual indicators
        )
        
        print("🔍 Running agent...")
        result = await agent.run()
        
        print("✅ Agent completed!")
        print(f"📊 Result type: {type(result)}")
        print(f"📊 Result: {result}")
                
        # Try to extract any JSON from the result
        if hasattr(result, 'all_results') and result.all_results:
            print(f"📋 Found {len(result.all_results)} results")
            for i, res in enumerate(result.all_results):
                print(f"  Result {i+1}: {res}")
                if hasattr(res, 'extracted_content') and res.extracted_content:
                    print(f"    📊 Extracted content: {res.extracted_content}")
        
        # Try to parse final result as JSON if possible
        result_str = str(result)
        if '{' in result_str and '}' in result_str:
            try:
                # Extract JSON-like content
                start = result_str.find('{')
                end = result_str.rfind('}') + 1
                json_candidate = result_str[start:end]
                json_result = json.loads(json_candidate)
                print(f"📊 Extracted JSON: {json.dumps(json_result, indent=2)}")
            except json.JSONDecodeError:
                print("⚠️ Could not parse JSON from result")
        
        return result
        
    except Exception as e:
        print(f"❌ Error during flight search: {str(e)}")
        return None


if __name__ == "__main__":
    asyncio.run(test_flight_search())