extract_page_content

Instructions

Implementation Reference

• src/index.js:299-330 (registration)

  Tool registration for 'extract_page_content' via server.tool() with name, description, schema, and handler.

  server.tool(
    'extract_page_content',
    'Extrae el contenido estructurado de una p�gina.',
    {
      url: z.string(),
      include_tables: z.boolean().default(true),
      include_links: z.boolean().default(true)
    },
    async ({ url, include_tables, include_links }) => {
      try {
        const { html, finalUrl } = await fetchPage(url);
        const extracted = extractReadableContent(html, finalUrl);
  
        return {
          content: [
            {
              type: 'text',
              text: JSON.stringify({
                title: extracted.title,
                url: finalUrl,
                sections: extracted.sections,
                tables: include_tables ? extractTables(html) : [],
                links: include_links ? extracted.links : []
              }, null, 2)
            }
          ]
        };
      } catch (error) {
        return mcpError(error);
      }
    }
  );

• src/index.js:302-306 (schema)

  Input schema for extract_page_content: url (required string), include_tables (boolean, default true), include_links (boolean, default true).

  {
    url: z.string(),
    include_tables: z.boolean().default(true),
    include_links: z.boolean().default(true)
  },

• src/index.js:307-329 (handler)

  Handler function that fetches the page, extracts readable content, and returns title, sections, tables (optional), and links (optional) as JSON.

  async ({ url, include_tables, include_links }) => {
    try {
      const { html, finalUrl } = await fetchPage(url);
      const extracted = extractReadableContent(html, finalUrl);
  
      return {
        content: [
          {
            type: 'text',
            text: JSON.stringify({
              title: extracted.title,
              url: finalUrl,
              sections: extracted.sections,
              tables: include_tables ? extractTables(html) : [],
              links: include_links ? extracted.links : []
            }, null, 2)
          }
        ]
      };
    } catch (error) {
      return mcpError(error);
    }
  }

• src/index.js:114-164 (helper)

  extractReadableContent() helper: parses HTML with cheerio, extracts title, heading-based sections, and deduplicated internal links.

  function extractReadableContent(html, pageUrl) {
    const $ = cheerio.load(html);
    $('script, style, noscript').remove();
  
    const title = $('title').first().text().trim() || $('h1').first().text().trim() || pageUrl;
    const sections = [];
  
    $('h1, h2, h3').each((_, el) => {
      const heading = $(el).text().trim();
      const texts = [];
      let current = $(el).next();
  
      while (current.length && !['h1', 'h2', 'h3'].includes(current.get(0)?.tagName)) {
        const text = current.text().trim();
        if (text) texts.push(text);
        current = current.next();
      }
  
      if (heading || texts.length) {
        sections.push({ heading, text: texts.join('\n\n') });
      }
    });
  
    const links = [];
    $('a[href]').each((_, el) => {
      const href = $(el).attr('href');
      const label = $(el).text().trim();
      if (!href) return;
      try {
        const url = new URL(href, pageUrl).toString();
        if (url.startsWith(BASE_URL)) {
          links.push({ label: label || url, url });
        }
      } catch {
      }
    });
  
    const contentText = $('body')
      .text()
      .replace(/\s+\n/g, '\n')
      .replace(/\n\s+/g, '\n')
      .replace(/\n{3,}/g, '\n\n')
      .trim();
  
    return {
      title,
      contentText,
      sections,
      links: dedupeLinks(links)
    };
  }

• src/index.js:175-196 (helper)

  extractTables() helper: parses HTML tables into headers/rows structure.

  function extractTables(html) {
    const $ = cheerio.load(html);
    const tables = [];
  
    $('table').each((_, table) => {
      const headers = [];
      $(table).find('thead th').each((__, th) => headers.push($(th).text().trim()));
  
      const rows = [];
      $(table).find('tbody tr').each((__, tr) => {
        const row = [];
        $(tr).find('td').each((___, td) => row.push($(td).text().trim()));
        if (row.length) rows.push(row);
      });
  
      if (headers.length || rows.length) {
        tables.push({ title: null, headers, rows });
      }
    });
  
    return tables;
  }