import { Logger } from "../utils/Logger";

export interface ExtractedJsonLd {
  headline?: string;
  articleBody?: string;
  description?: string;
  author?: string;
  datePublished?: string;
  dateModified?: string;
  publisher?: string;
}

/**
 * Extracts author string from a JSON-LD author property value (string, object, or array).
 */
function extractAuthorString(authorVal: unknown): string | undefined {
  if (!authorVal) return undefined;

  if (typeof authorVal === "string" && authorVal.trim()) {
    return authorVal.trim();
  }

  if (Array.isArray(authorVal)) {
    const authors = authorVal
      .map((a) => extractAuthorString(a))
      .filter((a): a is string => Boolean(a));
    return authors.length > 0 ? authors.join(", ") : undefined;
  }

  if (typeof authorVal === "object" && authorVal !== null) {
    const obj = authorVal as Record<string, unknown>;
    if (typeof obj.name === "string" && obj.name.trim()) {
      return obj.name.trim();
    }
  }

  return undefined;
}

/**
 * Extracts publisher string from a JSON-LD publisher property value (string or object).
 */
function extractPublisherString(publisherVal: unknown): string | undefined {
  if (!publisherVal) return undefined;

  if (typeof publisherVal === "string" && publisherVal.trim()) {
    return publisherVal.trim();
  }

  if (typeof publisherVal === "object" && publisherVal !== null) {
    const obj = publisherVal as Record<string, unknown>;
    if (typeof obj.name === "string" && obj.name.trim()) {
      return obj.name.trim();
    }
  }

  return undefined;
}

/**
 * Normalizes raw JSON-LD script text by stripping HTML comment and CDATA wrappers.
 * Does not perform entity decoding or structural JSON repair.
 */
export function normalizeJsonLdText(rawText: string): string {
  let text = rawText.trim();

  // Strip HTML comment wrapper: <!-- ... -->
  if (text.startsWith("<!--") && text.endsWith("-->")) {
    text = text.slice(4, -3).trim();
  }

  // Strip CDATA wrappers (single-line comments, multi-line comments, or bare tags)
  text = text
    .replace(/^\s*\/\/\s*<!\[CDATA\[\s*/i, "")
    .replace(/\s*\/\/\s*\]\]>\s*$/i, "")
    .replace(/^\s*\/\*\s*<!\[CDATA\[\s*\*\/\s*/i, "")
    .replace(/\s*\/\*\s*\]\]>\s*\*\/\s*$/i, "")
    .replace(/^\s*<!\[CDATA\[\s*/i, "")
    .replace(/\s*\]\]>\s*$/i, "")
    .trim();

  return text;
}

/**
 * Parses all <script type="application/ld+json"> elements in the document
 * and extracts relevant structured metadata fields.
 */
export function extractJsonLd(doc: Document): ExtractedJsonLd {
  const result: ExtractedJsonLd = {};

  const scripts = doc.querySelectorAll('script[type="application/ld+json"]');
  if (!scripts || scripts.length === 0) {
    return result;
  }

  const articleTypes = new Set([
    "article",
    "newsarticle",
    "blogposting",
    "techarticle",
    "webpage",
    "report",
    "scholarlyarticle",
    "socialmediaposting",
  ]);

  for (const script of Array.from(scripts)) {
    const rawText = normalizeJsonLdText(script.textContent || "");
    if (!rawText) continue;

    try {
      const parsed = JSON.parse(rawText);
      const itemsToProcess: Record<string, unknown>[] = [];

      const collectItems = (node: unknown) => {
        if (!node || typeof node !== "object") return;
        if (Array.isArray(node)) {
          node.forEach(collectItems);
        } else {
          const obj = node as Record<string, unknown>;
          itemsToProcess.push(obj);
          if (Array.isArray(obj["@graph"])) {
            obj["@graph"].forEach(collectItems);
          }
        }
      };

      collectItems(parsed);

      for (const item of itemsToProcess) {
        const rawType = item["@type"];
        const types = Array.isArray(rawType)
          ? rawType.map((t) => String(t).toLowerCase())
          : typeof rawType === "string"
          ? [rawType.toLowerCase()]
          : [];

        const isArticle = types.some((t) => articleTypes.has(t)) || Boolean(item.articleBody);

        if (isArticle || (!result.headline && !result.articleBody)) {
          if (typeof item.headline === "string" && item.headline.trim()) {
            result.headline = result.headline || item.headline.trim();
          } else if (typeof item.name === "string" && item.name.trim()) {
            result.headline = result.headline || item.name.trim();
          }

          if (typeof item.articleBody === "string" && item.articleBody.trim()) {
            result.articleBody = result.articleBody || item.articleBody.trim();
          }

          if (typeof item.description === "string" && item.description.trim()) {
            result.description = result.description || item.description.trim();
          }

          const authorStr = extractAuthorString(item.author);
          if (authorStr) {
            result.author = result.author || authorStr;
          }

          if (typeof item.datePublished === "string" && item.datePublished.trim()) {
            result.datePublished = result.datePublished || item.datePublished.trim();
          } else if (typeof item.dateCreated === "string" && item.dateCreated.trim()) {
            result.datePublished = result.datePublished || item.dateCreated.trim();
          }

          if (typeof item.dateModified === "string" && item.dateModified.trim()) {
            result.dateModified = result.dateModified || item.dateModified.trim();
          }

          const publisherStr = extractPublisherString(item.publisher);
          if (publisherStr) {
            result.publisher = result.publisher || publisherStr;
          }
        }
      }
    } catch (err) {
      const errMsg = err instanceof Error ? err.message : String(err);
      Logger.debug("[HTMLExtractor]", `Failed to parse JSON-LD script block: ${errMsg}`);
    }
  }

  return result;
}

import { Logger } from "../utils/Logger";

export interface ExtractedJsonLd {
  headline?: string;
  articleBody?: string;
  description?: string;
  author?: string;
  datePublished?: string;
  dateModified?: string;
  publisher?: string;
}

/**
 * Extracts author string from a JSON-LD author property value (string, object, or array).
 */
function extractAuthorString(authorVal: unknown): string | undefined {
  if (!authorVal) return undefined;

  if (typeof authorVal === "string" && authorVal.trim()) {
    return authorVal.trim();
  }

  if (Array.isArray(authorVal)) {
    const authors = authorVal
      .map((a) => extractAuthorString(a))
      .filter((a): a is string => Boolean(a));
    return authors.length > 0 ? authors.join(", ") : undefined;
  }

  if (typeof authorVal === "object" && authorVal !== null) {
    const obj = authorVal as Record<string, unknown>;
    if (typeof obj.name === "string" && obj.name.trim()) {
      return obj.name.trim();
    }
  }

  return undefined;
}

/**
 * Extracts publisher string from a JSON-LD publisher property value (string or object).
 */
function extractPublisherString(publisherVal: unknown): string | undefined {
  if (!publisherVal) return undefined;

  if (typeof publisherVal === "string" && publisherVal.trim()) {
    return publisherVal.trim();
  }

  if (typeof publisherVal === "object" && publisherVal !== null) {
    const obj = publisherVal as Record<string, unknown>;
    if (typeof obj.name === "string" && obj.name.trim()) {
      return obj.name.trim();
    }
  }

  return undefined;
}

/**
 * Normalizes raw JSON-LD script text by stripping HTML comment and CDATA wrappers.
 * Does not perform entity decoding or structural JSON repair.
 */
export function normalizeJsonLdText(rawText: string): string {
  let text = rawText.trim();

  // Strip HTML comment wrapper: <!-- ... -->
  if (text.startsWith("<!--") && text.endsWith("-->")) {
    text = text.slice(4, -3).trim();
  }

  // Strip CDATA wrappers (single-line comments, multi-line comments, or bare tags)
  text = text
    .replace(/^\s*\/\/\s*<!\[CDATA\[\s*/i, "")
    .replace(/\s*\/\/\s*\]\]>\s*$/i, "")
    .replace(/^\s*\/\*\s*<!\[CDATA\[\s*\*\/\s*/i, "")
    .replace(/\s*\/\*\s*\]\]>\s*\*\/\s*$/i, "")
    .replace(/^\s*<!\[CDATA\[\s*/i, "")
    .replace(/\s*\]\]>\s*$/i, "")
    .trim();

  return text;
}

/**
 * Parses all <script type="application/ld+json"> elements in the document
 * and extracts relevant structured metadata fields.
 */
export function extractJsonLd(doc: Document): ExtractedJsonLd {
  const result: ExtractedJsonLd = {};

  const scripts = doc.querySelectorAll('script[type="application/ld+json"]');
  if (!scripts || scripts.length === 0) {
    return result;
  }

  const articleTypes = new Set([
    "article",
    "newsarticle",
    "blogposting",
    "techarticle",
    "webpage",
    "report",
    "scholarlyarticle",
    "socialmediaposting",
  ]);

  for (const script of Array.from(scripts)) {
    const rawText = normalizeJsonLdText(script.textContent || "");
    if (!rawText) continue;

    try {
      const parsed = JSON.parse(rawText);
      const itemsToProcess: Record<string, unknown>[] = [];

      const collectItems = (node: unknown) => {
        if (!node || typeof node !== "object") return;
        if (Array.isArray(node)) {
          node.forEach(collectItems);
        } else {
          const obj = node as Record<string, unknown>;
          itemsToProcess.push(obj);
          if (Array.isArray(obj["@graph"])) {
            obj["@graph"].forEach(collectItems);
          }
        }
      };

      collectItems(parsed);

      for (const item of itemsToProcess) {
        const rawType = item["@type"];
        const types = Array.isArray(rawType)
          ? rawType.map((t) => String(t).toLowerCase())
          : typeof rawType === "string"
          ? [rawType.toLowerCase()]
          : [];

        const isArticle = types.some((t) => articleTypes.has(t)) || Boolean(item.articleBody);

        if (isArticle || (!result.headline && !result.articleBody)) {
          if (typeof item.headline === "string" && item.headline.trim()) {
            result.headline = result.headline || item.headline.trim();
          } else if (typeof item.name === "string" && item.name.trim()) {
            result.headline = result.headline || item.name.trim();
          }

          if (typeof item.articleBody === "string" && item.articleBody.trim()) {
            result.articleBody = result.articleBody || item.articleBody.trim();
          }

          if (typeof item.description === "string" && item.description.trim()) {
            result.description = result.description || item.description.trim();
          }

          const authorStr = extractAuthorString(item.author);
          if (authorStr) {
            result.author = result.author || authorStr;
          }

          if (typeof item.datePublished === "string" && item.datePublished.trim()) {
            result.datePublished = result.datePublished || item.datePublished.trim();
          } else if (typeof item.dateCreated === "string" && item.dateCreated.trim()) {
            result.datePublished = result.datePublished || item.dateCreated.trim();
          }

          if (typeof item.dateModified === "string" && item.dateModified.trim()) {
            result.dateModified = result.dateModified || item.dateModified.trim();
          }

          const publisherStr = extractPublisherString(item.publisher);
          if (publisherStr) {
            result.publisher = result.publisher || publisherStr;
          }
        }
      }
    } catch (err) {
      const errMsg = err instanceof Error ? err.message : String(err);
      Logger.debug("[HTMLExtractor]", `Failed to parse JSON-LD script block: ${errMsg}`);
    }
  }

  return result;
}