/**
 * URL Classification Guard — Harness 1
 *
 * Separates URL classification into a testable pure function with
 * an embedded test corpus. Prevents misclassification bugs like
 * `/about/meet-our-doctors.php` being tagged as an "about" page.
 *
 * Priority: doctor > surgery > about (specific → general)
 */

// ─── Pattern Definitions ───

const PAGE_PATTERNS = {
  doctor: {
    positive: [
      "/doctor", "/doctors", "/team", "/staff", "/specialist", "/professor",
      "/의료진", "/원장", "meet-our-doctor", "/physicians", "/surgeon",
      "our-team", "our-doctors",
    ],
  },
  surgery: {
    positive: [
      "/surgery", "/service", "/procedure", "/treatment", "/시술", "/수술",
      "/procedures", "/treatments", "/services", "/진료",
    ],
  },
  about: {
    positive: [
      "/about", "/intro", "/소개", "/greeting", "/인사말", "/history", "/연혁",
      "/company", "/clinic-info",
    ],
    negative: [
      "/doctor", "/doctors", "/procedure", "/surgery", "/service",
      "/treatment", "meet-our", "/team", "/staff", "/specialist",
      "/의료진", "/원장", "/시술", "/수술", "/진료",
      "our-team", "our-doctors",
    ],
  },
} as const;

// ─── Classification Function ───

/**
 * Classify a URL into one of: 'doctor' | 'surgery' | 'about' | null.
 * Uses priority ordering (doctor > surgery > about) so that
 * compound URLs like `/about/meet-our-doctors.php` correctly resolve
 * to the more specific category.
 */
export function classifyPageUrl(url: string): "doctor" | "surgery" | "about" | null {
  const lower = url.toLowerCase();

  // Priority 1: Doctor pages
  if (PAGE_PATTERNS.doctor.positive.some((p) => lower.includes(p))) {
    return "doctor";
  }

  // Priority 2: Surgery/service pages
  if (PAGE_PATTERNS.surgery.positive.some((p) => lower.includes(p))) {
    return "surgery";
  }

  // Priority 3: About pages (with negative exclusion)
  if (PAGE_PATTERNS.about.positive.some((p) => lower.includes(p))) {
    if (PAGE_PATTERNS.about.negative.some((p) => lower.includes(p))) {
      return null; // Contains about + doctor/surgery keyword → ambiguous, skip
    }
    return "about";
  }

  return null;
}

// ─── Test Corpus ───

/**
 * Known failure cases and expected classifications.
 * Used by validateClassifier() for regression prevention.
 */
export const CLASSIFICATION_TEST_CORPUS: ReadonlyArray<readonly [string, "doctor" | "surgery" | "about" | null]> = [
  // Bug cases (historically misclassified)
  ["/about/meet-our-doctors.php", "doctor"],     // 그랜드성형외과 bug
  ["/about/procedures.php", "surgery"],           // 그랜드성형외과
  // Standard doctor pages
  ["/doctors", "doctor"],
  ["/의료진", "doctor"],
  ["/team/professor-kim", "doctor"],
  ["/about/our-team", "doctor"],
  // Standard surgery pages
  ["/surgery/rhinoplasty", "surgery"],
  ["/시술안내", "surgery"],
  ["/services/breast", "surgery"],
  // Standard about pages
  ["/about/", "about"],
  ["/about/greeting.php", "about"],
  ["/about/intro", "about"],
  ["/about/history", "about"],
  ["/소개", "about"],
  ["/인사말", "about"],
  // Null cases
  ["/gallery", null],
  ["/contact", null],
  ["/blog/post-123", null],
];

// ─── Self-Test ───

/**
 * Run classification against the test corpus.
 * Call during Edge Function cold-start for automatic regression detection.
 */
export function validateClassifier(): { pass: boolean; failures: string[] } {
  const failures: string[] = [];
  for (const [url, expected] of CLASSIFICATION_TEST_CORPUS) {
    const result = classifyPageUrl(url);
    if (result !== expected) {
      failures.push(`${url}: expected '${expected}', got '${result}'`);
    }
  }
  return { pass: failures.length === 0, failures };
}