const BLOCKED_ACADEMIC_HOSTS =
  /\b(academia\.edu|researchgate\.net|sciencedirect\.com|jstor\.org|springer\.com|tandfonline\.com|wiley\.com)\b/i;

/**
 * Intercepts unsolvable/Cloudflare-locked academic URLs (including Wayback mirrors of academia.edu)
 * and resolves their peer-reviewed abstract + DOI directly via OpenAlex JSON API in <800ms.
 */
export async function resolveBlockedAcademicUrl(
  url: string,
  serpTitle: string,
  log?: (msg: string) => void
): Promise<{ title: string; content: string; resolvedUrl: string } | null> {
  if (!BLOCKED_ACADEMIC_HOSTS.test(url)) return null;

  // Clean title or extract slug from URL if title is generic
  let searchTitle = serpTitle
    .replace(/\s*[-|]\s*(Academia\.edu|ResearchGate|ScienceDirect).*$/i, "")
    .replace(/^\(PDF\)\s*/i, "")
    .trim();

  if (searchTitle.length < 12) {
    const slugMatch = url.match(/\/([a-zA-Z0-9_-]{15,})(?:\/|$)/);
    if (slugMatch) {
      searchTitle = slugMatch[1].replace(/[_-]+/g, " ");
    }
  }

  if (searchTitle.length < 10) return null;

  const apiUrl = `https://api.openalex.org/works?search=${encodeURIComponent(
    searchTitle
  )}&per-page=1&select=id,doi,title,publication_year,abstract_inverted_index,primary_location`;

  try {
    const res = await fetch(apiUrl, {
      headers: { Accept: "application/json", "User-Agent": "DeepSwarmResearch/5.9.2" },
      signal: AbortSignal.timeout(2500),
    });
    if (!res.ok) return null;

    const data: any = await res.json();
    const work = data?.results?.[0];
    if (!work?.title || !work?.abstract_inverted_index) return null;

    const words: string[] = [];
    for (const [word, positions] of Object.entries(
      work.abstract_inverted_index as Record<string, number[]>
    )) {
      for (const pos of positions) words[pos] = word;
    }
    const abstract = words.filter(Boolean).join(" ");
    if (abstract.length < 100) return null;

    const resolvedUrl = work.doi || work.primary_location?.landing_page_url || work.id;
    log?.(
      `[ACADEMIC RESOLVER] Replaced blocked ${new URL(url).hostname} URL with OpenAlex record: "${work.title.slice(0, 55)}..."`
    );

    return {
      title: `${work.title} (${work.publication_year ?? "n.d."})`,
      content: `[OpenAlex Resolved Paper]\nTitle: ${work.title}\nYear: ${work.publication_year}\nDOI: ${work.doi ?? "N/A"}\nAbstract: ${abstract}`,
      resolvedUrl: String(resolvedUrl),
    };
  } catch {
    return null;
  }
}

const BLOCKED_ACADEMIC_HOSTS =
  /\b(academia\.edu|researchgate\.net|sciencedirect\.com|jstor\.org|springer\.com|tandfonline\.com|wiley\.com)\b/i;

/**
 * Intercepts unsolvable/Cloudflare-locked academic URLs (including Wayback mirrors of academia.edu)
 * and resolves their peer-reviewed abstract + DOI directly via OpenAlex JSON API in <800ms.
 */
export async function resolveBlockedAcademicUrl(
  url: string,
  serpTitle: string,
  log?: (msg: string) => void
): Promise<{ title: string; content: string; resolvedUrl: string } | null> {
  if (!BLOCKED_ACADEMIC_HOSTS.test(url)) return null;

  // Clean title or extract slug from URL if title is generic
  let searchTitle = serpTitle
    .replace(/\s*[-|]\s*(Academia\.edu|ResearchGate|ScienceDirect).*$/i, "")
    .replace(/^\(PDF\)\s*/i, "")
    .trim();

  if (searchTitle.length < 12) {
    const slugMatch = url.match(/\/([a-zA-Z0-9_-]{15,})(?:\/|$)/);
    if (slugMatch) {
      searchTitle = slugMatch[1].replace(/[_-]+/g, " ");
    }
  }

  if (searchTitle.length < 10) return null;

  const apiUrl = `https://api.openalex.org/works?search=${encodeURIComponent(
    searchTitle
  )}&per-page=1&select=id,doi,title,publication_year,abstract_inverted_index,primary_location`;

  try {
    const res = await fetch(apiUrl, {
      headers: { Accept: "application/json", "User-Agent": "DeepSwarmResearch/5.9.2" },
      signal: AbortSignal.timeout(2500),
    });
    if (!res.ok) return null;

    const data: any = await res.json();
    const work = data?.results?.[0];
    if (!work?.title || !work?.abstract_inverted_index) return null;

    const words: string[] = [];
    for (const [word, positions] of Object.entries(
      work.abstract_inverted_index as Record<string, number[]>
    )) {
      for (const pos of positions) words[pos] = word;
    }
    const abstract = words.filter(Boolean).join(" ");
    if (abstract.length < 100) return null;

    const resolvedUrl = work.doi || work.primary_location?.landing_page_url || work.id;
    log?.(
      `[ACADEMIC RESOLVER] Replaced blocked ${new URL(url).hostname} URL with OpenAlex record: "${work.title.slice(0, 55)}..."`
    );

    return {
      title: `${work.title} (${work.publication_year ?? "n.d."})`,
      content: `[OpenAlex Resolved Paper]\nTitle: ${work.title}\nYear: ${work.publication_year}\nDOI: ${work.doi ?? "N/A"}\nAbstract: ${abstract}`,
      resolvedUrl: String(resolvedUrl),
    };
  } catch {
    return null;
  }
}