/**
 * Dead link detector — tracks broken links discovered during crawl.
 *
 * A "dead link" is a URL that:
 * - Returns 404, 410, or other 4xx/5xx responses
 * - Times out after maxRetries
 * - Redirects to an error page (detected by title heuristic)
 *
 * Results: list of broken links with source page, status code, and anchor text.
 * Used for: SEO audit export, test report enrichment.
 */

export interface DeadLink {
  url: string;
  statusCode: number;
  sourceUrl?: string;
  anchorText?: string;
  checkedAt: string;
  reason: 'http-error' | 'timeout' | 'dns-error';
}

export class DeadLinkTracker {
  private deadLinks: DeadLink[] = [];
  private checked = new Map<string, number>(); // url → statusCode

  record(
    url: string,
    statusCode: number,
    source?: { sourceUrl?: string; anchorText?: string },
  ): void {
    if (this.checked.has(url)) return;
    this.checked.set(url, statusCode);
    if (statusCode >= 400) {
      this.deadLinks.push({
        url,
        statusCode,
        sourceUrl: source?.sourceUrl,
        anchorText: source?.anchorText,
        checkedAt: new Date().toISOString(),
        reason: statusCode === 0 ? 'timeout' : 'http-error',
      });
    }
  }

  recordTimeout(url: string, sourceUrl?: string): void {
    if (this.checked.has(url)) return;
    this.checked.set(url, 0);
    this.deadLinks.push({ url, statusCode: 0, sourceUrl, checkedAt: new Date().toISOString(), reason: 'timeout' });
  }

  getDeadLinks(): DeadLink[] {
    return [...this.deadLinks];
  }

  isKnownBroken(url: string): boolean {
    const code = this.checked.get(url);
    return code !== undefined && code >= 400;
  }

  stats(): { total: number; by404: number; by5xx: number; byTimeout: number } {
    return {
      total: this.deadLinks.length,
      by404: this.deadLinks.filter(l => l.statusCode === 404).length,
      by5xx: this.deadLinks.filter(l => l.statusCode >= 500).length,
      byTimeout: this.deadLinks.filter(l => l.reason === 'timeout').length,
    };
  }

  toCsv(): string {
    const rows = [['URL', 'Status', 'Source URL', 'Anchor Text', 'Checked At']];
    for (const l of this.deadLinks) {
      rows.push([l.url, String(l.statusCode), l.sourceUrl ?? '', l.anchorText ?? '', l.checkedAt]);
    }
    return rows.map(r => r.map(c => `"${c.replace(/"/g, '""')}"`).join(',')).join('\n');
  }
}