ciaovolo/flight-comparator/scan_discovered_routes.py

#!/usr/bin/env python3
"""
Automated Daily Scans for Discovered Routes

Reads discovered_routes.json and runs targeted daily scans.
Much faster than scanning all airports because it only queries known routes.

Usage:
    # First, discover routes
    python discover_routes.py --from BDS --to-country DE --window 3

    # Then, run targeted daily scans
    python scan_discovered_routes.py discovered_routes.json --daily-scan
"""

import json
import subprocess
import sys
import click
from datetime import datetime


@click.command()
@click.argument('routes_file', type=click.Path(exists=True))
@click.option('--daily-scan', is_flag=True, help='Run daily scans (vs seasonal)')
@click.option('--start-date', help='Start date for daily scan (YYYY-MM-DD)')
@click.option('--end-date', help='End date for daily scan (YYYY-MM-DD)')
@click.option('--window', type=int, help='Override window months from discovery')
@click.option('--workers', default=5, type=int, help='Concurrency level (default: 5)')
@click.option('--output-dir', default='results', help='Directory to save results (default: results)')
@click.option('--dry-run', is_flag=True, help='Show what would be scanned without executing')
def scan_discovered(routes_file, daily_scan, start_date, end_date, window, workers, output_dir, dry_run):
    """
    Run targeted scans on discovered routes.

    Example:
        python scan_discovered_routes.py discovered_routes.json --daily-scan
    """
    # Load discovered routes
    with open(routes_file, 'r') as f:
        data = json.load(f)

    origin = data['origin']
    routes = data['routes']
    default_window = data.get('window_months', 3)

    if window is None:
        window = default_window

    print()
    print("=" * 70)
    print("TARGETED SCAN OF DISCOVERED ROUTES")
    print("=" * 70)
    print(f"Origin: {origin}")
    print(f"Discovered routes: {len(routes)}")
    print(f"Mode: {'Daily scan' if daily_scan else 'Seasonal scan'}")
    if daily_scan and start_date and end_date:
        print(f"Date range: {start_date} to {end_date}")
    else:
        print(f"Window: {window} months")
    print(f"Workers: {workers}")
    print()

    if not routes:
        print("⚠️  No routes to scan!")
        print(f"Discovery file {routes_file} contains no routes with flights.")
        sys.exit(1)

    # Display routes to scan
    print("Routes to scan:")
    for i, route in enumerate(routes, 1):
        dest = route['destination']
        city = route['destination_city']
        airlines = ', '.join(route['airlines'][:2])
        if len(route['airlines']) > 2:
            airlines += f" +{len(route['airlines']) - 2}"
        print(f"  {i}. {origin} → {dest} ({city}) - {airlines}")

    print()

    if dry_run:
        print("=" * 70)
        print("DRY RUN - Commands that would be executed:")
        print("=" * 70)
        print()

    # Build and execute commands
    results_summary = {
        "scan_date": datetime.now().isoformat(),
        "origin": origin,
        "routes_scanned": len(routes),
        "mode": "daily" if daily_scan else "seasonal",
        "results": []
    }

    for i, route in enumerate(routes, 1):
        dest = route['destination']
        city = route['destination_city']

        # Build command
        cmd_parts = [
            "python", "main.py",
            "--from", origin,
            "--to", dest,
        ]

        if daily_scan:
            cmd_parts.append("--daily-scan")
            if start_date:
                cmd_parts.extend(["--start-date", start_date])
            if end_date:
                cmd_parts.extend(["--end-date", end_date])
            if not start_date and not end_date:
                cmd_parts.extend(["--window", str(window)])
        else:
            cmd_parts.extend(["--window", str(window)])

        cmd_parts.extend(["--workers", str(workers)])

        # Add output file
        output_file = f"{output_dir}/{origin}_{dest}_{'daily' if daily_scan else 'seasonal'}.json"
        cmd_parts.extend(["--output", "json"])

        command = " ".join(cmd_parts)

        print(f"[{i}/{len(routes)}] Scanning {origin} → {dest} ({city})")

        if dry_run:
            print(f"  Command: {command}")
            print()
            continue

        try:
            # Execute command
            result = subprocess.run(
                command,
                shell=True,
                capture_output=True,
                text=True,
                timeout=600  # 10 minute timeout per route
            )

            output = result.stdout + result.stderr

            # Parse results (look for flight count)
            import re
            flights_match = re.search(r'Flights Found:\s+(\d+)', output)
            flights_found = int(flights_match.group(1)) if flights_match else 0

            # Save output to file
            import os
            os.makedirs(output_dir, exist_ok=True)
            with open(output_file, 'w') as f:
                f.write(output)

            results_summary["results"].append({
                "destination": dest,
                "destination_city": city,
                "flights_found": flights_found,
                "output_file": output_file,
                "success": result.returncode == 0
            })

            print(f"  ✅ Complete - {flights_found} flights found")
            print(f"  📄 Saved to: {output_file}")

        except subprocess.TimeoutExpired:
            print(f"  ⏱️  Timeout - scan took too long")
            results_summary["results"].append({
                "destination": dest,
                "destination_city": city,
                "error": "timeout",
                "success": False
            })
        except Exception as e:
            print(f"  ❌ Error: {e}")
            results_summary["results"].append({
                "destination": dest,
                "destination_city": city,
                "error": str(e),
                "success": False
            })

        print()

    if not dry_run:
        # Save summary
        summary_file = f"{output_dir}/scan_summary.json"
        with open(summary_file, 'w') as f:
            json.dump(results_summary, f, indent=2)

        # Display summary
        print("=" * 70)
        print("SCAN SUMMARY")
        print("=" * 70)
        total_scanned = len(routes)
        successful = sum(1 for r in results_summary["results"] if r.get("success", False))
        total_flights = sum(r.get("flights_found", 0) for r in results_summary["results"])

        print(f"Routes scanned: {total_scanned}")
        print(f"Successful: {successful}/{total_scanned}")
        print(f"Total flights found: {total_flights}")
        print()
        print(f"Results saved to: {output_dir}/")
        print(f"Summary: {summary_file}")
        print()

        # Show top routes by flight count
        sorted_results = sorted(
            results_summary["results"],
            key=lambda x: x.get("flights_found", 0),
            reverse=True
        )

        print("Top routes by flight count:")
        for route in sorted_results[:5]:
            if route.get("flights_found", 0) > 0:
                print(f"  {origin} → {route['destination']}: {route['flights_found']} flights")

        print()


if __name__ == '__main__':
    scan_discovered()