content.js

/**
 * SatyaCheck — Content Script
 * सत्य की जाँच | India's AI Fake News Detector
 *
 * This script runs on EVERY web page you visit (invisibly, in the background).
 * When you click "Check This Page" in the SatyaCheck popup, this script:
 *   1. Extracts the article title, body text, images, and URL
 *   2. Sends it back to the popup
 *   3. The popup sends it to the SatyaCheck backend for AI analysis
 *
 * File: dist/content.js
 * Manifest V3 — runs at document_idle
 */

'use strict';

// ─── Article Content Extraction ───────────────────────────────────────────────

/**
 * Extracts article content from the current page.
 * Tries multiple selectors to find the article body — works on most Indian
 * and international news websites.
 *
 * @returns {Object} { url, title, body_text, image_urls, domain }
 */
function extractArticleContent() {
  // ── Title ──────────────────────────────────────────────────────────────────
  const title =
    document.querySelector('h1')?.innerText?.trim() ||
    document.querySelector('meta[property="og:title"]')?.content?.trim() ||
    document.querySelector('meta[name="twitter:title"]')?.content?.trim() ||
    document.querySelector('title')?.innerText?.trim() ||
    'Unknown Title';

  // ── Article Body ───────────────────────────────────────────────────────────
  // Try known article body selectors (ordered by specificity)
  const bodySelectors = [
    'article',
    '[class*="article-body"]',
    '[class*="story-body"]',
    '[class*="story-content"]',
    '[class*="post-content"]',
    '[class*="entry-content"]',
    '[class*="article-content"]',
    '[class*="news-content"]',
    '[class*="content-body"]',
    '[itemprop="articleBody"]',
    '[data-testid="article-body"]',
    '.article__body',
    '.story__content',
    '.article-detail',
    'main',
    '.main-content',
    '#article-body',
    '#story-body',
  ];

  let bodyText = '';
  for (const selector of bodySelectors) {
    const el = document.querySelector(selector);
    if (el && el.innerText && el.innerText.trim().length > 200) {
      bodyText = el.innerText.trim();
      break;
    }
  }

  // Fallback: use the page body but strip nav/header/footer
  if (!bodyText || bodyText.length < 100) {
    const clone = document.body.cloneNode(true);
    // Remove non-content elements
    ['nav', 'header', 'footer', 'aside', 'script', 'style', '.ad', '.advertisement'].forEach(sel => {
      clone.querySelectorAll(sel).forEach(el => el.remove());
    });
    bodyText = clone.innerText?.trim() || '';
  }

  // Trim to 10,000 chars (enough for AI analysis, not too large for the API)
  bodyText = bodyText.substring(0, 10000);

  // ── Images ─────────────────────────────────────────────────────────────────
  const allImages = Array.from(document.querySelectorAll('img'));
  const imageUrls = allImages
    .map(img => img.src || img.dataset.src || img.dataset.lazySrc || '')
    .filter(src =>
      src.startsWith('http') &&
      !src.includes('icon') &&
      !src.includes('logo') &&
      !src.includes('avatar') &&
      !src.includes('1x1') &&
      !src.includes('pixel') &&
      !src.includes('ad') &&
      (src.includes('.jpg') || src.includes('.jpeg') || src.includes('.png') || src.includes('.webp'))
    )
    .slice(0, 5); // Max 5 images

  // ── Open Graph image (usually the hero/featured image) ────────────────────
  const ogImage = document.querySelector('meta[property="og:image"]')?.content;
  if (ogImage && !imageUrls.includes(ogImage)) {
    imageUrls.unshift(ogImage);
  }

  // ── Domain ─────────────────────────────────────────────────────────────────
  const domain = window.location.hostname.replace(/^www\./, '');

  return {
    url: window.location.href,
    title: title.substring(0, 300),
    body_text: bodyText,
    image_urls: imageUrls.slice(0, 5),
    domain: domain,
  };
}


// ─── Message Listener ─────────────────────────────────────────────────────────

/**
 * Listen for messages from the SatyaCheck popup.
 * When the popup asks for article content, extract and return it.
 */
chrome.runtime.onMessage.addListener((request, sender, sendResponse) => {
  if (request.action === 'extractContent') {
    try {
      const content = extractArticleContent();
      sendResponse({ success: true, data: content });
    } catch (error) {
      sendResponse({
        success: false,
        error: error.message,
        data: {
          url: window.location.href,
          title: document.title || 'Unknown',
          body_text: document.body?.innerText?.substring(0, 5000) || '',
          image_urls: [],
          domain: window.location.hostname.replace(/^www\./, ''),
        }
      });
    }
    return true; // Keep message channel open for async response
  }

  if (request.action === 'ping') {
    sendResponse({ success: true, message: 'SatyaCheck content script active' });
    return true;
  }
});


// ─── Page Type Detection ──────────────────────────────────────────────────────

/**
 * Detect if the current page is a news article (vs. homepage, category, etc.)
 * Used to show/hide the "Check This Page" button in the popup.
 */
function isArticlePage() {
  const url = window.location.href;

  // URL-based detection
  const articleUrlPatterns = [
    /\/news\//i, /\/article\//i, /\/story\//i, /\/post\//i,
    /\/\d{4}\/\d{2}\//i,  // Date-based URLs: /2024/01/
    /\/politics\//i, /\/sports\//i, /\/business\//i,
  ];
  if (articleUrlPatterns.some(p => p.test(url))) return true;

  // Content-based detection
  const hasArticleTag = !!document.querySelector('article');
  const hasH1 = !!document.querySelector('h1');
  const bodyLength = document.body?.innerText?.length || 0;

  return hasArticleTag && hasH1 && bodyLength > 500;
}

// Notify popup whether this page looks like an article
chrome.runtime.sendMessage({
  action: 'pageInfo',
  isArticle: isArticlePage(),
  domain: window.location.hostname.replace(/^www\./, ''),
  url: window.location.href,
}).catch(() => {
  // Popup not open — ignore
});