webbanalys/webbanalys.py

#!/usr/bin/env python3
"""
╔═══════════════════════════════════════════════════════════════════════╗
║                     WEBBPLATSANALYS RAPPORTGENERATOR                  ║
║                                                                       ║
║   Analysera webbplatser och generera professionella PDF-rapporter     ║
╚═══════════════════════════════════════════════════════════════════════╝

Usage:
    python webbanalys.py scan <url>             # Run analysis
    python webbanalys.py generate               # Generate PDF report
    python webbanalys.py run <url>              # Scan + generate in one step

Requirements (NixOS):
    nix-shell shell.nix

Requirements (pip):
    pip install -r requirements.txt
"""

import argparse
import json
import subprocess
import shutil
import sys
from pathlib import Path
from datetime import datetime

# Ensure lib is importable
sys.path.insert(0, str(Path(__file__).parent))

from lib.analyzer import load_ci_result, load_all_audits
from lib.pdf_report import generate_pdf


BANNER = """
╔═══════════════════════════════════════════════════════════════════════╗
║                     WEBBPLATSANALYS RAPPORTGENERATOR                  ║
╚═══════════════════════════════════════════════════════════════════════╝
"""


def load_config(config_path: Path) -> dict:
    """Load project configuration from JSON file."""
    if config_path.exists():
        with open(config_path, "r", encoding="utf-8") as f:
            return json.load(f)
    return {}


def save_config(config_path: Path, config: dict):
    """Save project configuration to JSON file."""
    with open(config_path, "w", encoding="utf-8") as f:
        json.dump(config, f, indent=2, ensure_ascii=False)


def find_data_files(base_path: Path) -> tuple[Path, Path]:
    """Find ci-result.json and reports directory."""
    search_paths = [
        base_path / "rapport",
        base_path / ".unlighthouse",
        base_path,
    ]
    
    ci_result = None
    reports_dir = None
    
    for path in search_paths:
        if not path.exists():
            continue
        
        candidate = path / "ci-result.json"
        if candidate.exists():
            ci_result = candidate
        
        candidate = path / "reports"
        if candidate.exists() and candidate.is_dir():
            reports_dir = candidate
        
        if ci_result and reports_dir:
            break
    
    return ci_result, reports_dir


def find_logo(base_path: Path) -> Path:
    """Find logo file in project."""
    # Prefer PNG/JPG, then SVG
    for pattern in ["*logo*.png", "*logo*.jpg", "*.png", "*.jpg"]:
        for f in base_path.glob(pattern):
            if f.is_file() and f.suffix.lower() in [".png", ".jpg", ".jpeg"]:
                return f
    # Check for SVG (will be converted)
    for pattern in ["*logo*.svg", "*.svg"]:
        for f in base_path.glob(pattern):
            if f.is_file():
                return f
    return None


def convert_svg_to_png(svg_path: Path) -> Path:
    """Convert SVG to PNG for PDF embedding."""
    try:
        import cairosvg
        png_path = svg_path.with_suffix(".png")
        if not png_path.exists() or png_path.stat().st_mtime < svg_path.stat().st_mtime:
            cairosvg.svg2png(url=str(svg_path), write_to=str(png_path), output_width=200)
        return png_path
    except ImportError:
        return None
    except Exception:
        return None


def cmd_generate(args, base_path: Path):
    """Generate PDF report."""
    config_path = base_path / "webbanalys.json"
    config = load_config(config_path)
    
    # Find data files
    print("\n🔍 Söker efter analysdata...")
    
    ci_result_path, reports_dir = find_data_files(base_path)
    
    if not ci_result_path:
        print("❌ Kunde inte hitta ci-result.json")
        print("   Kör först: unlighthouse-ci --site <url>")
        print(f"   Sökte i: {base_path}")
        sys.exit(1)
    
    if not reports_dir:
        print("❌ Kunde inte hitta reports/ katalog")
        sys.exit(1)
    
    print(f"   ✓ Hittade data i: {ci_result_path.parent}")
    
    # Determine parameters (CLI > config > default)
    customer = args.customer or config.get("customer", "")
    site_url = args.url or config.get("site_url", "")
    consultant = args.consultant or config.get("consultant", "")
    
    # Find logo
    logo_path = None
    if args.logo:
        logo_path = Path(args.logo)
    elif config.get("logo"):
        logo_path = base_path / config["logo"]
    else:
        logo_path = find_logo(base_path)
    
    # Convert SVG to PNG if needed
    if logo_path and logo_path.exists():
        if logo_path.suffix.lower() == ".svg":
            png_path = convert_svg_to_png(logo_path)
            if png_path:
                logo_path = png_path
                print(f"   ✓ Logotyp: {logo_path.name} (konverterad från SVG)")
            else:
                print(f"   ⚠ Kunde inte konvertera SVG-logotyp")
                logo_path = None
        else:
            print(f"   ✓ Logotyp: {logo_path.name}")
    
    # Load data
    print("\n📊 Laddar analysdata...")
    data = load_ci_result(ci_result_path)
    
    # Extract site URL from data if not provided
    if not site_url:
        routes = data.get("routes", [])
        if routes:
            first_path = routes[0].get("path", "")
            # Try to construct from common patterns
            for route in routes[:5]:
                path = route.get("path", "")
                if path.startswith("http"):
                    site_url = "/".join(path.split("/")[:3])
                    break
    
    print(f"   Webbplats: {site_url or '(okänd)'}")
    if customer:
        print(f"   Kund: {customer}")
    
    # Analyze
    print("\n🔬 Analyserar sidor...")
    all_audits, resource_issues, element_issues, page_count = load_all_audits(
        reports_dir, site_url
    )
    
    print(f"   Sidor analyserade:    {page_count}")
    print(f"   Problemtyper:         {len(all_audits)}")
    print(f"   Resursproblem:        {len(resource_issues)}")
    print(f"   Elementproblem:       {len(element_issues)}")
    
    # Overall score
    overall = data["summary"]["score"] * 100
    if overall >= 90:
        score_emoji = "🟢"
    elif overall >= 50:
        score_emoji = "🟡"
    else:
        score_emoji = "🔴"
    print(f"\n   Övergripande betyg:   {score_emoji} {overall:.0f}/100")
    
    # Determine output path
    if args.output:
        output_path = Path(args.output)
    else:
        date_str = datetime.now().strftime("%Y%m%d")
        site_slug = site_url.replace("https://", "").replace("http://", "").replace("/", "_").replace(".", "_")
        output_path = base_path / f"rapport_{site_slug}_{date_str}.pdf"
    
    # Generate report
    print(f"\n📝 Genererar rapport...")
    generate_pdf(
        output_path=output_path,
        data=data,
        all_audits=all_audits,
        resource_issues=resource_issues,
        element_issues=element_issues,
        page_count=page_count,
        customer_name=customer,
        site_url=site_url,
        logo_path=logo_path if logo_path and logo_path.exists() else None,
        consultant_name=consultant
    )
    
    file_size = output_path.stat().st_size / 1024
    print(f"\n✅ Rapport genererad!")
    print(f"   📄 {output_path}")
    print(f"   📦 {file_size:.0f} KB")
    
    # Offer to save config
    if not config and (customer or site_url):
        print("\n💾 Spara inställningar för framtida rapporter? (y/n): ", end="")
        try:
            if input().lower() == "y":
                new_config = {}
                if customer:
                    new_config["customer"] = customer
                if site_url:
                    new_config["site_url"] = site_url
                if consultant:
                    new_config["consultant"] = consultant
                if logo_path:
                    new_config["logo"] = str(logo_path.relative_to(base_path))
                save_config(config_path, new_config)
                print(f"   ✓ Sparad till {config_path.name}")
        except (EOFError, KeyboardInterrupt):
            pass


def cmd_init(args, base_path: Path):
    """Initialize project configuration."""
    config_path = base_path / "webbanalys.json"
    
    print("\n🔧 Konfigurera projekt")
    print("   (tryck Enter för att hoppa över)\n")
    
    config = load_config(config_path)
    
    try:
        customer = input(f"   Kundnamn [{config.get('customer', '')}]: ").strip()
        if customer:
            config["customer"] = customer
        
        site_url = input(f"   Webbplats URL [{config.get('site_url', '')}]: ").strip()
        if site_url:
            config["site_url"] = site_url
        
        consultant = input(f"   Konsultnamn [{config.get('consultant', '')}]: ").strip()
        if consultant:
            config["consultant"] = consultant
        
        save_config(config_path, config)
        print(f"\n✅ Konfiguration sparad till {config_path.name}")
        
    except (EOFError, KeyboardInterrupt):
        print("\n   Avbrutet.")


def cmd_info(args, base_path: Path):
    """Show project information."""
    config_path = base_path / "webbanalys.json"
    config = load_config(config_path)
    
    print("\n📋 Projektinformation")
    print(f"   Mapp: {base_path}")
    
    if config:
        print(f"\n   Konfiguration ({config_path.name}):")
        for key, value in config.items():
            print(f"     {key}: {value}")
    else:
        print(f"\n   ⚠ Ingen konfiguration hittad")
        print(f"     Kör: python webbanalys.py init")
    
    # Check for data
    ci_result, reports_dir = find_data_files(base_path)
    print(f"\n   Analysdata:")
    if ci_result:
        print(f"     ✓ ci-result.json: {ci_result}")
    else:
        print(f"     ✗ ci-result.json saknas")
    
    if reports_dir:
        report_count = len(list(reports_dir.glob("*/lighthouse.json")))
        print(f"     ✓ reports/: {report_count} sidor")
    else:
        print(f"     ✗ reports/ saknas")
    
    # Check for logo
    logo = find_logo(base_path)
    if logo:
        print(f"     ✓ Logotyp: {logo.name}")


def cmd_scan(args, base_path: Path):
    """Run Unlighthouse scan."""
    config_path = base_path / "webbanalys.json"
    config = load_config(config_path)
    
    # Determine URL
    site_url = args.url or config.get("site_url", "")
    if not site_url:
        print("❌ Ingen URL angiven.")
        print("   Använd: python webbanalys.py scan https://example.com")
        print("   Eller kör 'python webbanalys.py init' först")
        sys.exit(1)
    
    # Ensure URL has protocol
    if not site_url.startswith("http"):
        site_url = "https://" + site_url
    
    samples = args.samples or config.get("samples", 3)
    output_path = base_path / "rapport"
    
    print(f"\n🔍 Startar webbanalys...")
    print(f"   URL: {site_url}")
    print(f"   Samples: {samples}")
    print(f"   Output: {output_path}")
    
    # Check for pnpm
    pnpm_path = shutil.which("pnpm")
    npx_path = shutil.which("npx")
    
    if pnpm_path:
        cmd = [
            "pnpm", "dlx", "-p", "@unlighthouse/cli",
            "unlighthouse-ci",
            "--site", site_url,
            "--samples", str(samples),
            "--output-path", str(output_path),
            "--reporter", "jsonExpanded"
        ]
    elif npx_path:
        cmd = [
            "npx", "-y", "@unlighthouse/cli",
            "unlighthouse-ci",
            "--site", site_url,
            "--samples", str(samples),
            "--output-path", str(output_path),
            "--reporter", "jsonExpanded"
        ]
    else:
        print("❌ Varken pnpm eller npx hittades.")
        print("   Installera Node.js och pnpm/npm först.")
        sys.exit(1)
    
    print(f"\n   Kör: {' '.join(cmd)}\n")
    print("─" * 70)
    
    try:
        result = subprocess.run(cmd, cwd=base_path)
        print("─" * 70)
        
        if result.returncode == 0:
            print("\n✅ Analys klar!")
            
            # Update config with URL
            if site_url and site_url != config.get("site_url"):
                config["site_url"] = site_url
                config["samples"] = samples
                save_config(config_path, config)
            
            return True
        else:
            print(f"\n❌ Analys misslyckades (kod {result.returncode})")
            return False
            
    except KeyboardInterrupt:
        print("\n\n⚠ Avbruten av användaren")
        return False
    except Exception as e:
        print(f"\n❌ Fel: {e}")
        return False


def cmd_run(args, base_path: Path):
    """Run scan + generate in one step."""
    # First scan
    if cmd_scan(args, base_path):
        print("\n" + "═" * 70 + "\n")
        # Then generate
        cmd_generate(args, base_path)


def main():
    parser = argparse.ArgumentParser(
        description="Analysera webbplatser och generera PDF-rapporter",
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog="""
Kommandon:
  scan        Kör Unlighthouse-analys
  generate    Generera PDF-rapport från befintlig data
  run         Kör analys + generera rapport (allt-i-ett)
  init        Konfigurera projekt
  info        Visa projektinformation

Exempel:
  python webbanalys.py scan https://example.com
  python webbanalys.py scan -u https://example.com -s 5
  python webbanalys.py generate -c "Företag AB"
  python webbanalys.py run https://example.com -c "Företag AB"
        """
    )
    
    subparsers = parser.add_subparsers(dest="command", help="Kommando")
    
    # Scan command
    scan_parser = subparsers.add_parser("scan", help="Kör Unlighthouse-analys")
    scan_parser.add_argument("url", nargs="?", help="Webbplats-URL")
    scan_parser.add_argument("--samples", "-s", type=int, default=3, help="Antal samples per sida (default: 3)")
    
    # Generate command
    gen_parser = subparsers.add_parser("generate", help="Generera PDF-rapport")
    gen_parser.add_argument("--customer", "-c", help="Kundnamn")
    gen_parser.add_argument("--url", "-u", help="Webbplats-URL")
    gen_parser.add_argument("--consultant", help="Konsultnamn")
    gen_parser.add_argument("--logo", "-l", help="Sökväg till logotyp (PNG/SVG)")
    gen_parser.add_argument("--output", "-o", help="Utdatafil")
    
    # Run command (scan + generate)
    run_parser = subparsers.add_parser("run", help="Analysera + generera rapport")
    run_parser.add_argument("url", nargs="?", help="Webbplats-URL")
    run_parser.add_argument("--samples", "-s", type=int, default=3, help="Antal samples per sida (default: 3)")
    run_parser.add_argument("--customer", "-c", help="Kundnamn")
    run_parser.add_argument("--consultant", help="Konsultnamn")
    run_parser.add_argument("--logo", "-l", help="Sökväg till logotyp (PNG/SVG)")
    run_parser.add_argument("--output", "-o", help="Utdatafil")
    
    # Init command
    subparsers.add_parser("init", help="Konfigurera projekt")
    
    # Info command
    subparsers.add_parser("info", help="Visa projektinformation")
    
    args = parser.parse_args()
    base_path = Path.cwd()
    
    print(BANNER)
    
    if args.command == "scan":
        cmd_scan(args, base_path)
    elif args.command == "generate":
        cmd_generate(args, base_path)
    elif args.command == "run":
        cmd_run(args, base_path)
    elif args.command == "init":
        cmd_init(args, base_path)
    elif args.command == "info":
        cmd_info(args, base_path)
    else:
        # Default: show help
        parser.print_help()
        print("\n💡 Snabbstart:")
        print("   python webbanalys.py run https://example.com -c 'Företag AB'")


if __name__ == "__main__":
    main()