src/captcha-solver.js

/**
 * captcha-solver.js — Sistema de Resolución de CAPTCHAs con IA
 * ==============================================================
 * Sin servicios de pago. Todo con IA disponible.
 *
 * ESTRATEGIA POR TIPO DE CAPTCHA:
 *
 * ① reCAPTCHA v2 (imagen) → Gemini Vision analiza cada tile
 *   Research: "YOLOv8 fine-tuned achieves 100% success on reCAPTCHAv2 image challenges"
 *   Nosotros usamos Gemini Vision (gratis) con el mismo enfoque de clasificación
 *
 * ② reCAPTCHA v2 Audio → Whisper/SpeechRecognition gratis
 *   Research: "Audio transcription with wit.ai: 70-80% success, FREE"
 *   GitHub: njraladdin/recaptcha-v2-solver — método audio
 *
 * ③ reCAPTCHA v3 (behavioral) → NO hay que resolverlo: se evita con buen stealth
 *   Research: "reCAPTCHA v3 uses behavioral scoring. Good behavior simulation
 *              achieves 0.9 trust score without solving any challenge"
 *
 * ④ hCaptcha → Mismo enfoque que reCAPTCHA v2 con Gemini Vision
 *
 * ⑤ Cloudflare Turnstile → JavaScript challenge, se resuelve solo con buen stealth
 *
 * ⑥ Text CAPTCHAs → OCR via Gemini Vision
 *
 * ⑦ CAPTCHAs irresolubles → Notificar al owner para resolución manual
 */

import { readConfig } from './utils.js';
const config = readConfig();

// ── Gemini Vision helper ──────────────────────────────────────────────────────
async function callGeminiVision(base64Image, prompt, mimeType = 'image/jpeg') {
  const keys = config.ai?.gemini?.keys ?? [config.ai?.gemini?.apiKey].filter(Boolean);
  const key  = keys[0];
  if (!key) return null;

  const res = await fetch(
    `https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=${key}`,
    {
      method : 'POST',
      headers: { 'Content-Type': 'application/json' },
      body   : JSON.stringify({
        contents: [{ parts: [
          { text: prompt },
          { inline_data: { mime_type: mimeType, data: base64Image } },
        ]}],
        generationConfig: { maxOutputTokens: 200, temperature: 0.1 },
      }),
      signal: AbortSignal.timeout(15000),
    }
  );
  if (!res.ok) return null;
  return (await res.json()).candidates?.[0]?.content?.parts?.[0]?.text?.trim() ?? null;
}

// ── Audio download y transcripción ──────────────────────────────────────────
async function transcribeAudio(audioUrl) {
  try {
    // Descargar el audio del CAPTCHA
    const audioRes  = await fetch(audioUrl, { signal: AbortSignal.timeout(15000) });
    const audioData = Buffer.from(await audioRes.arrayBuffer());
    const base64    = audioData.toString('base64');

    // Usar Gemini para transcribir (soporta audio nativo)
    const keys = config.ai?.gemini?.keys ?? [config.ai?.gemini?.apiKey].filter(Boolean);
    const key  = keys[0];
    if (!key) return null;

    const res = await fetch(
      `https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=${key}`,
      {
        method : 'POST',
        headers: { 'Content-Type': 'application/json' },
        body   : JSON.stringify({
          contents: [{ parts: [
            { text: 'This is a CAPTCHA audio challenge. Transcribe the spoken digits or words exactly. Reply with ONLY the numbers/words, nothing else.' },
            { inline_data: { mime_type: 'audio/mp3', data: base64 } },
          ]}],
          generationConfig: { maxOutputTokens: 50, temperature: 0 },
        }),
        signal: AbortSignal.timeout(20000),
      }
    );
    if (!res.ok) return null;
    const text = (await res.json()).candidates?.[0]?.content?.parts?.[0]?.text?.trim();
    // Limpiar respuesta: solo números y letras
    return text?.replace(/[^a-zA-Z0-9\s]/g, '').trim() ?? null;
  } catch {
    return null;
  }
}

function sleep(ms) { return new Promise(r => setTimeout(r, Math.round(ms))); }

// ═══════════════════════════════════════════════════════════════════════════════
// SOLUCIONADORES POR TIPO
// ═══════════════════════════════════════════════════════════════════════════════

/**
 * Resolver reCAPTCHA v2 con imágenes
 * Estrategia: screenshot de cada tile → Gemini Vision clasifica → clic en los correctos
 */
export async function solveRecaptchaV2Images(page) {
  console.log('[CAPTCHA] 🔍 Intentando resolver reCAPTCHA v2 por imágenes...');

  try {
    // Intentar primero el checkbox "I'm not a robot" (a veces pasa solo con buen stealth)
    const checkbox = await page.$('.recaptcha-checkbox-border, #recaptcha-anchor');
    if (checkbox) {
      await checkbox.click();
      await sleep(2000 + Math.random() * 1000);

      // Verificar si el checkbox se marcó sin challenge
      const solved = await page.evaluate(() => {
        const resp = document.querySelector('textarea#g-recaptcha-response, [name="g-recaptcha-response"]');
        return resp && resp.value.length > 10;
      }).catch(() => false);

      if (solved) {
        console.log('[CAPTCHA] ✅ reCAPTCHA v2 resuelto sin challenge (buen stealth)');
        return { success: true, method: 'stealth_no_challenge' };
      }
    }

    // Hay challenge de imágenes — resolver con visión
    // Obtener el iframe del CAPTCHA
    const frame = page.frames().find(f => f.url().includes('recaptcha'));
    if (!frame) return { success: false, method: 'no_frame' };

    // Obtener la instrucción del challenge
    const instruction = await frame.$eval('.rc-imageselect-desc-wrapper', el => el.textContent?.trim() ?? '').catch(() => '');
    console.log('[CAPTCHA] Instrucción:', instruction.slice(0, 80));

    // Hacer screenshot del área de imágenes
    const grid = await frame.$('.rc-imageselect-table');
    if (!grid) return { success: false, method: 'no_grid' };

    const gridScreenshot = await grid.screenshot({ type: 'jpeg', quality: 90 });
    const base64Grid     = gridScreenshot.toString('base64');

    // Gemini Vision: analizar qué tiles seleccionar
    const prompt = `This is a reCAPTCHA image grid challenge. The instruction says: "${instruction}"

Analyze each tile in the grid (numbered left-to-right, top-to-bottom starting from 1).
Reply with ONLY the tile numbers that match the instruction, separated by commas.
Example: "1,3,7" or "2,5,6,9"
If no tiles match, reply "none".
Do not explain, just give the numbers.`;

    const answer = await callGeminiVision(base64Grid, prompt);
    if (!answer || answer.toLowerCase() === 'none') {
      console.log('[CAPTCHA] Gemini no encontró tiles para seleccionar');
      return { success: false, method: 'vision_no_match' };
    }

    // Parsear números de tiles
    const tileNumbers = answer.split(',').map(n => parseInt(n.trim())).filter(n => !isNaN(n) && n >= 1);
    console.log('[CAPTCHA] Tiles a seleccionar:', tileNumbers);

    // Hacer clic en los tiles indicados con comportamiento humano
    for (const tileNum of tileNumbers) {
      const tile = await frame.$(`.rc-imageselect-tile:nth-child(${tileNum}), .rc-image-tile-target:nth-child(${tileNum})`);
      if (tile) {
        const box = await tile.boundingBox();
        if (box) {
          const x = box.x + box.width * (0.2 + Math.random() * 0.6);
          const y = box.y + box.height * (0.2 + Math.random() * 0.6);
          await page.mouse.move(x, y);
          await sleep(200 + Math.random() * 300);
          await page.mouse.click(x, y);
          await sleep(300 + Math.random() * 400);
        }
      }
    }

    // Hacer clic en Verify
    await sleep(1000 + Math.random() * 500);
    const verifyBtn = await frame.$('#recaptcha-verify-button, .rc-button-default');
    if (verifyBtn) {
      await verifyBtn.click();
      await sleep(2000 + Math.random() * 1000);
    }

    // Verificar resultado
    const token = await page.evaluate(() => {
      const t = document.querySelector('textarea[name="g-recaptcha-response"]');
      return t?.value ?? '';
    }).catch(() => '');

    if (token.length > 20) {
      console.log('[CAPTCHA] ✅ reCAPTCHA v2 resuelto por visión');
      return { success: true, method: 'vision', token };
    }

    return { success: false, method: 'vision_unverified' };

  } catch (e) {
    console.warn('[CAPTCHA] Error en reCAPTCHA v2:', e.message);
    return { success: false, error: e.message };
  }
}

/**
 * Resolver reCAPTCHA v2 por audio (más fiable que imágenes)
 * Estrategia: solicitar versión audio → descargar → transcribir con Gemini → escribir
 */
export async function solveRecaptchaV2Audio(page) {
  console.log('[CAPTCHA] 🎵 Intentando reCAPTCHA v2 por audio...');

  try {
    const frame = page.frames().find(f => f.url().includes('recaptcha'));
    if (!frame) return { success: false, method: 'no_frame' };

    // Cambiar a modo audio
    const audioBtn = await frame.$('#recaptcha-audio-button');
    if (!audioBtn) {
      // Intentar via checkbox primero
      const checkbox = await frame.$('#recaptcha-anchor');
      if (checkbox) { await checkbox.click(); await sleep(1500); }
    }

    const audioBtnRetry = await frame.$('#recaptcha-audio-button');
    if (!audioBtnRetry) return { success: false, method: 'no_audio_button' };

    await audioBtnRetry.click();
    await sleep(2000);

    // Obtener la URL del audio
    const audioUrl = await frame.$eval('.rc-audiochallenge-tdownload-link', el => el.href).catch(() => null);
    if (!audioUrl) return { success: false, method: 'no_audio_url' };

    // Transcribir con Gemini
    const transcription = await transcribeAudio(audioUrl);
    if (!transcription) return { success: false, method: 'transcription_failed' };

    console.log('[CAPTCHA] Transcripción:', transcription);

    // Escribir la respuesta en el campo
    const input = await frame.$('#audio-response');
    if (!input) return { success: false, method: 'no_input' };

    await input.click();
    await sleep(200);
    await page.keyboard.type(transcription.toLowerCase().replace(/\s+/g, ' '), { delay: 80 + Math.random() * 40 });
    await sleep(500);

    // Verificar
    const verifyBtn = await frame.$('#recaptcha-verify-button');
    if (verifyBtn) { await verifyBtn.click(); await sleep(2000); }

    const token = await page.evaluate(() =>
      document.querySelector('textarea[name="g-recaptcha-response"]')?.value ?? ''
    ).catch(() => '');

    if (token.length > 20) {
      console.log('[CAPTCHA] ✅ reCAPTCHA v2 resuelto por audio');
      return { success: true, method: 'audio', token };
    }

    return { success: false, method: 'audio_unverified' };

  } catch (e) {
    return { success: false, error: e.message };
  }
}

/**
 * Resolver text CAPTCHA (imagen con texto/números distorsionados)
 */
export async function solveTextCaptcha(page, imageSelector) {
  try {
    const imgEl = await page.$(imageSelector);
    if (!imgEl) return null;

    const imgBase64 = (await imgEl.screenshot({ type: 'jpeg', quality: 95 })).toString('base64');
    const answer    = await callGeminiVision(imgBase64,
      'This is a text CAPTCHA image with distorted characters. Read the characters carefully and reply with ONLY the text shown, no explanation.'
    );

    return answer?.replace(/\s/g, '') ?? null;
  } catch { return null; }
}

/**
 * Detector principal de CAPTCHA + resolutor automático
 */
export async function detectAndSolve(page, notifyOwner = null) {
  console.log('[CAPTCHA] Escaneando página...');

  // Detectar tipo de CAPTCHA presente
  const captchaInfo = await page.evaluate(() => {
    const body = document.body.innerHTML.toLowerCase();
    const text  = document.body.textContent.toLowerCase();
    return {
      hasRecaptchaV2  : !!document.querySelector('iframe[src*="recaptcha"]') || body.includes('g-recaptcha'),
      hasHcaptcha     : body.includes('hcaptcha') || !!document.querySelector('iframe[src*="hcaptcha"]'),
      hasCloudflare   : text.includes('checking your browser') || text.includes('verifying you are') || body.includes('cf-challenge'),
      hasTurnstile    : body.includes('turnstile') || !!document.querySelector('[data-sitekey]'),
      hasTextCaptcha  : !!document.querySelector('img[src*="captcha"], img[alt*="captcha"]'),
    };
  }).catch(() => ({}));

  console.log('[CAPTCHA] Detectado:', JSON.stringify(captchaInfo));

  // 1. Cloudflare — solo stealth, no hay challenge que resolver directamente
  if (captchaInfo.hasCloudflare) {
    console.log('[CAPTCHA] Cloudflare detectado — esperando resolución automática...');
    await sleep(5000 + Math.random() * 3000);
    // Si sigue ahí después de 8s, notificar
    const stillCloudflare = await page.evaluate(() =>
      document.body.textContent.toLowerCase().includes('checking your browser')
    ).catch(() => false);

    if (stillCloudflare && notifyOwner) {
      await notifyOwner('⚠️ Cloudflare no se resolvió automáticamente. Puede necesitar proxy de mayor calidad.');
    }
    return { type: 'cloudflare', solved: !stillCloudflare };
  }

  // 2. reCAPTCHA v2 — intentar audio primero (más fiable), luego imagen
  if (captchaInfo.hasRecaptchaV2) {
    // Intentar método audio
    let result = await solveRecaptchaV2Audio(page);
    if (!result.success) {
      // Fallback a imágenes
      result = await solveRecaptchaV2Images(page);
    }

    if (!result.success && notifyOwner) {
      await notifyOwner('⚠️ reCAPTCHA v2 no pudo resolverse automáticamente. Requiere intervención manual.');
    }
    return { type: 'recaptcha_v2', ...result };
  }

  // 3. hCaptcha — mismo enfoque que reCAPTCHA v2
  if (captchaInfo.hasHcaptcha) {
    console.log('[CAPTCHA] hCaptcha detectado — intentando resolver...');
    // hCaptcha tiene estructura similar a reCAPTCHA v2
    const result = await solveRecaptchaV2Images(page); // funciona por similitud
    if (!result.success && notifyOwner) {
      await notifyOwner('⚠️ hCaptcha detectado. Requiere intervención manual.');
    }
    return { type: 'hcaptcha', ...result };
  }

  // 4. Text CAPTCHA
  if (captchaInfo.hasTextCaptcha) {
    const answer = await solveTextCaptcha(page, 'img[alt*="captcha"], img[src*="captcha"]');
    return { type: 'text_captcha', solved: !!answer, answer };
  }

  return { type: 'none', solved: true };
}

export async function getCaptchaSolverStats() {
  const geminiAvailable = !!(config.ai?.gemini?.keys?.length || config.ai?.gemini?.apiKey);
  return {
    geminiVision: geminiAvailable,
    audioSolver : geminiAvailable,
    supportedTypes: ['recaptcha_v2_image', 'recaptcha_v2_audio', 'hcaptcha', 'text_captcha', 'cloudflare_basic'],
    notSupported  : ['recaptcha_v3_enterprise', 'funcaptcha_advanced', 'geetest_v4'],
  };
}