@dyyz1993/agent-browser 0.24.0 → 0.25.0

package/scripts/extract_jd_product_detail.js

@@ -0,0 +1,62 @@
+(() => {
+  const result = {
+    url: window.location.href,
+    title: document.title,
+    price: '',
+    shop: '',
+    sales: '',
+    reviews: {
+      good: '',
+      neutral: '',
+      bad: ''
+    },
+    region: ''
+  };
+  
+  const priceMatch = document.body.textContent.match(/[¥￥]\s*(\d+\.?\d*)/);
+  if (priceMatch) {
+    result.price = priceMatch[1];
+  }
+  
+  const shopMatch = document.body.textContent.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+  if (shopMatch) {
+    result.shop = shopMatch[1];
+  }
+  
+  const reviewElements = document.querySelectorAll('[class*="comment"], [class*="review"], [class*="rating"]');
+  
+  reviewElements.forEach(el => {
+    const text = el.textContent || '';
+    
+    if (text.includes('好评') && !result.reviews.good) {
+      const match = text.match(/好评[：:]\s*(\d+)/);
+      if (match) {
+        result.reviews.good = match[1];
+      }
+    }
+    
+    if (text.includes('中评') && !result.reviews.neutral) {
+      const match = text.match(/中评[：:]\s*(\d+)/);
+      if (match) {
+        result.reviews.neutral = match[1];
+      }
+    }
+    
+    if (text.includes('差评') && !result.reviews.bad) {
+      const match = text.match(/差评[：:]\s*(\d+)/);
+      if (match) {
+        result.reviews.bad = match[1];
+      }
+    }
+  });
+  
+  const regionElements = document.querySelectorAll('[class*="address"], [class*="region"], [class*="location"]');
+  regionElements.forEach(el => {
+    const text = el.textContent || '';
+    if (!result.region && text.length > 0 && text.length < 100) {
+      result.region = text.trim();
+    }
+  });
+  
+  return result;
+})()

package/scripts/extract_jd_products_correct_links.js

@@ -0,0 +1,78 @@
+(async () => {
+  const result = {
+    keyword: "生鲜海鲜",
+    crawlTime: new Date().toISOString(),
+    products: []
+  };
+  
+  window.scrollTo(0, 500);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, 2000);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, 4000);
+  await new Promise(resolve => setTimeout(resolve, 3000));
+  
+  const containers = document.querySelectorAll('._goodsContainer_1p2ae_1');
+  const seenLinks = new Set();
+  
+  containers.forEach((container, index) => {
+    const link = container.querySelector('a[href]');
+    if (!link) return;
+    
+    const href = link.href;
+    if (seenLinks.has(href)) return;
+    seenLinks.add(href);
+    
+    const text = container.textContent || '';
+    const product = {
+      id: index + 1,
+      link: href
+    };
+    
+    const img = container.querySelector('img');
+    if (img) {
+      const src = img.getAttribute('data-src') || img.src;
+      if (src) product.image = src.substring(0, 150);
+    }
+    
+    const priceMatch = text.match(/[¥￥]\s*(\d+\.?\d*)/);
+    if (priceMatch) {
+      product.price = priceMatch[1];
+    }
+    
+    const salesMatch = text.match(/已售([\d\+万]+)/);
+    if (salesMatch) {
+      product.sales = salesMatch[1];
+    }
+    
+    const shopMatch = text.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+    if (shopMatch) {
+      product.shop = shopMatch[1];
+    }
+    
+    const adIndex = text.indexOf('广告');
+    const titleStart = adIndex >= 0 ? adIndex + 2 : 0;
+    let title = text.substring(titleStart).trim();
+    
+    title = title.replace(/已售[\d\+万]+/, '');
+    title = title.replace(/[\d\+万]+人加购/, '');
+    title = title.replace(/[\d\+万]+人种草/, '');
+    title = title.replace(/[\d\+万]+人浏览/, '');
+    title = title.replace(/[^\s]{2,10}(?:旗舰店|专营店|京东自营)/, '');
+    title = title.replace(/搜同款|关注|对比/g, '');
+    title = title.trim();
+    
+    if (title.length > 5) {
+      product.title = title.substring(0, 100);
+    }
+    
+    if (product.title || product.price) {
+      result.products.push(product);
+    }
+  });
+  
+  result.total = result.products.length;
+  return result;
+})()

package/scripts/extract_jd_products_final.js

@@ -0,0 +1,80 @@
+(async () => {
+  const result = {
+    keyword: "生鲜海鲜",
+    crawlTime: new Date().toISOString(),
+    products: []
+  };
+  
+  window.scrollTo(0, 100);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, document.body.scrollHeight);
+  await new Promise(resolve => setTimeout(resolve, 3000));
+  
+  const containers = document.querySelectorAll('._goodsContainer_1p2ae_1');
+  
+  const seenLinks = new Set();
+  
+  containers.forEach((container, index) => {
+    const link = container.querySelector('a[href]');
+    if (!link) return;
+    
+    const href = link.href;
+    if (seenLinks.has(href)) return;
+    seenLinks.add(href);
+    
+    const text = container.textContent || '';
+    
+    const product = {
+      id: index + 1,
+      link: href
+    };
+    
+    const img = container.querySelector('img');
+    if (img) {
+      const src = img.getAttribute('data-src') || img.src;
+      if (src) product.image = src.substring(0, 150);
+    }
+    
+    const priceMatch = text.match(/[¥￥]\s*(\d+\.?\d*)/);
+    if (priceMatch) {
+      product.price = priceMatch[1];
+    }
+    
+    const salesMatch = text.match(/已售([\d\+万]+)/);
+    if (salesMatch) {
+      product.sales = salesMatch[1];
+    }
+    
+    const shopMatch = text.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+    if (shopMatch) {
+      product.shop = shopMatch[1];
+    }
+    
+    const adIndex = text.indexOf('广告');
+    const titleStart = adIndex >= 0 ? adIndex + 2 : 0;
+    const titleText = text.substring(titleStart).trim();
+    
+    const titleParts = titleText.split(/[¥￥]/);
+    if (titleParts.length > 0) {
+      let title = titleParts[0].trim();
+      title = title.replace(/已售[\d\+万]+/, '');
+      title = title.replace(/[\d\+万]+人加购/, '');
+      title = title.replace(/[\d\+万]+人种草/, '');
+      title = title.replace(/[\d\+万]+人浏览/, '');
+      title = title.replace(/[^\s]{2,10}(?:旗舰店|专营店|京东自营)/, '');
+      title = title.replace(/搜同款|关注|对比/g, '');
+      title = title.trim();
+      
+      if (title.length > 5) {
+        product.title = title.substring(0, 100);
+      }
+    }
+    
+    if (product.title || product.price) {
+      result.products.push(product);
+    }
+  });
+  
+  return result;
+})()

package/scripts/extract_jd_reviews.js

@@ -0,0 +1,48 @@
+(() => {
+  const result = {
+    reviews: {
+      total: '',
+      good: '',
+      neutral: '',
+      bad: ''
+    },
+    allReviewText: []
+  };
+  
+  const allText = document.body.textContent;
+  
+  const totalMatch = allText.match(/买家评价[（(](\d+[+万])[）)]/);
+  if (totalMatch) {
+    result.reviews.total = totalMatch[1];
+  }
+  
+  const reviewButtons = document.querySelectorAll('[class*="filter"], [class*="tab"]');
+  
+  reviewButtons.forEach(btn => {
+    const text = btn.textContent || '';
+    result.allReviewText.push(text);
+    
+    if (text.includes('好评') || text.includes('positive')) {
+      const match = text.match(/(\d+)/);
+      if (match) result.reviews.good = match[1];
+    }
+    
+    if (text.includes('中评') || text.includes('neutral')) {
+      const match = text.match(/(\d+)/);
+      if (match) result.reviews.neutral = match[1];
+    }
+    
+    if (text.includes('差评') || text.includes('negative')) {
+      const match = text.match(/(\d+)/);
+      if (match) result.reviews.bad = match[1];
+    }
+  });
+  
+  const commentSummary = document.querySelector('[class*="comment-summary"], [class*="summary"]');
+  if (commentSummary) {
+    const summaryText = commentSummary.textContent || '';
+    result.allReviewText.push(summaryText);
+  }
+  
+  return result;
+})()

package/scripts/extract_jd_seafood_final.js

@@ -0,0 +1,78 @@
+(async () => {
+  const result = {
+    keyword: "生鲜海鲜",
+    crawlTime: new Date().toISOString(),
+    products: []
+  };
+  
+  window.scrollTo(0, 500);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, 2000);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, 4000);
+  await new Promise(resolve => setTimeout(resolve, 3000));
+  
+  const containers = document.querySelectorAll('._goodsContainer_1p2ae_1');
+  const seenLinks = new Set();
+  
+  containers.forEach((container, index) => {
+    const link = container.querySelector('a[href]');
+    if (!link) return;
+    
+    const href = link.href;
+    if (seenLinks.has(href)) return;
+    seenLinks.add(href);
+    
+    const text = container.textContent || '';
+    const product = {
+      id: index + 1,
+      link: href
+    };
+    
+    const img = container.querySelector('img');
+    if (img) {
+      const src = img.getAttribute('data-src') || img.src;
+      if (src) product.image = src.substring(0, 150);
+    }
+    
+    const priceMatch = text.match(/[¥￥]\s*(\d+\.?\d*)/);
+    if (priceMatch) {
+      product.price = priceMatch[1];
+    }
+    
+    const salesMatch = text.match(/已售([\d\+万]+)/);
+    if (salesMatch) {
+      product.sales = salesMatch[1];
+    }
+    
+    const shopMatch = text.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+    if (shopMatch) {
+      product.shop = shopMatch[1];
+    }
+    
+    const adIndex = text.indexOf('广告');
+    const titleStart = adIndex >= 0 ? adIndex + 2 : 0;
+    let title = text.substring(titleStart).trim();
+    
+    title = title.replace(/已售[\d\+万]+/, '');
+    title = title.replace(/[\d\+万]+人加购/, '');
+    title = title.replace(/[\d\+万]+人种草/, '');
+    title = title.replace(/[\d\+万]+人浏览/, '');
+    title = title.replace(/[^\s]{2,10}(?:旗舰店|专营店|京东自营)/, '');
+    title = title.replace(/搜同款|关注|对比/g, '');
+    title = title.trim();
+    
+    if (title.length > 5) {
+      product.title = title.substring(0, 100);
+    }
+    
+    if (product.title || product.price) {
+      result.products.push(product);
+    }
+  });
+  
+  result.total = result.products.length;
+  return result;
+})()

package/scripts/extract_multiple_products.js

@@ -0,0 +1,77 @@
+(async () => {
+  const result = {
+    keyword: "生鲜海鲜",
+    crawlTime: new Date().toISOString(),
+    products: []
+  };
+  
+  window.scrollTo(0, 500);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, 2000);
+  await new Promise(resolve => setTimeout(resolve, 2000));
+  
+  window.scrollTo(0, 4000);
+  await new Promise(resolve => setTimeout(resolve, 3000));
+  
+  const containers = document.querySelectorAll('._goodsContainer_1p2ae_1');
+  const seenLinks = new Set();
+  
+  containers.forEach((container, index) => {
+    const link = container.querySelector('a[href]');
+    if (!link) return;
+    
+    const href = link.href;
+    if (seenLinks.has(href)) return;
+    seenLinks.add(href);
+    
+    const text = container.textContent || '';
+    const product = {
+      id: index + 1,
+      link: href
+    };
+    
+    const img = container.querySelector('img');
+    if (img) {
+      const src = img.getAttribute('data-src') || img.src;
+      if (src) product.image = src.substring(0, 150);
+    }
+    
+    const priceMatch = text.match(/[¥￥]\s*(\d+\.?\d*)/);
+    if (priceMatch) {
+      product.price = priceMatch[1];
+    }
+    
+    const salesMatch = text.match(/已售([\d\+万]+)/);
+    if (salesMatch) {
+      product.sales = salesMatch[1];
+    }
+    
+    const shopMatch = text.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+    if (shopMatch) {
+      product.shop = shopMatch[1];
+    }
+    
+    const adIndex = text.indexOf('广告');
+    const titleStart = adIndex >= 0 ? adIndex + 2 : 0;
+    let title = text.substring(titleStart).trim();
+    
+    title = title.replace(/已售[\d\+万]+/, '');
+    title = title.replace(/[\d\+万]+人加购/, '');
+    title = title.replace(/[\d\+万]+人种草/, '');
+    title = title.replace(/[\d\+万]+人浏览/, '');
+    title = title.replace(/[^\s]{2,10}(?:旗舰店|专营店|京东自营)/, '');
+    title = title.replace(/搜同款|关注|对比/g, '');
+    title = title.trim();
+    
+    if (title.length > 5) {
+      product.title = title.substring(0, 100);
+    }
+    
+    if (product.title || product.price) {
+      result.products.push(product);
+    }
+  });
+  
+  return result;
+})()

package/scripts/extract_products_no_scroll.js

@@ -0,0 +1,68 @@
+(() => {
+  const result = {
+    keyword: "生鲜海鲜",
+    crawlTime: new Date().toISOString(),
+    products: []
+  };
+  
+  const containers = document.querySelectorAll('._goodsContainer_1p2ae_1');
+  const seenLinks = new Set();
+  
+  containers.forEach((container, index) => {
+    const link = container.querySelector('a[href]');
+    if (!link) return;
+    
+    const href = link.href;
+    if (seenLinks.has(href)) return;
+    seenLinks.add(href);
+    
+    const text = container.textContent || '';
+    const product = {
+      id: index + 1,
+      link: href
+    };
+    
+    const img = container.querySelector('img');
+    if (img) {
+      const src = img.getAttribute('data-src') || img.src;
+      if (src) product.image = src.substring(0, 150);
+    }
+    
+    const priceMatch = text.match(/[¥￥]\s*(\d+\.?\d*)/);
+    if (priceMatch) {
+      product.price = priceMatch[1];
+    }
+    
+    const salesMatch = text.match(/已售([\d\+万]+)/);
+    if (salesMatch) {
+      product.sales = salesMatch[1];
+    }
+    
+    const shopMatch = text.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+    if (shopMatch) {
+      product.shop = shopMatch[1];
+    }
+    
+    const adIndex = text.indexOf('广告');
+    const titleStart = adIndex >= 0 ? adIndex + 2 : 0;
+    let title = text.substring(titleStart).trim();
+    
+    title = title.replace(/已售[\d\+万]+/, '');
+    title = title.replace(/[\d\+万]+人加购/, '');
+    title = title.replace(/[\d\+万]+人种草/, '');
+    title = title.replace(/[\d\+万]+人浏览/, '');
+    title = title.replace(/[^\s]{2,10}(?:旗舰店|专营店|京东自营)/, '');
+    title = title.replace(/搜同款|关注|对比/g, '');
+    title = title.trim();
+    
+    if (title.length > 5) {
+      product.title = title.substring(0, 100);
+    }
+    
+    if (product.title || product.price) {
+      result.products.push(product);
+    }
+  });
+  
+  return result;
+})()

package/scripts/extract_products_simple.js

@@ -0,0 +1,68 @@
+(() => {
+  const result = {
+    keyword: "生鲜海鲜",
+    crawlTime: new Date().toISOString(),
+    products: []
+  };
+  
+  const containers = document.querySelectorAll('._goodsContainer_1p2ae_1');
+  const seenLinks = new Set();
+  
+  containers.forEach((container, index) => {
+    const link = container.querySelector('a[href]');
+    if (!link) return;
+    
+    const href = link.href;
+    if (seenLinks.has(href)) return;
+    seenLinks.add(href);
+    
+    const text = container.textContent || '';
+    const product = {
+      id: index + 1,
+      link: href
+    };
+    
+    const img = container.querySelector('img');
+    if (img) {
+      const src = img.getAttribute('data-src') || img.src;
+      if (src) product.image = src.substring(0, 150);
+    }
+    
+    const priceMatch = text.match(/[¥￥]\s*(\d+\.?\d*)/);
+    if (priceMatch) {
+      product.price = priceMatch[1];
+    }
+    
+    const salesMatch = text.match(/已售([\d\+万]+)/);
+    if (salesMatch) {
+      product.sales = salesMatch[1];
+    }
+    
+    const shopMatch = text.match(/([^\s]{2,10}(?:旗舰店|专营店|京东自营))/);
+    if (shopMatch) {
+      product.shop = shopMatch[1];
+    }
+    
+    const adIndex = text.indexOf('广告');
+    const titleStart = adIndex >= 0 ? adIndex + 2 : 0;
+    let title = text.substring(titleStart).trim();
+    
+    title = title.replace(/已售[\d\+万]+/, '');
+    title = title.replace(/[\d\+万]+人加购/, '');
+    title = title.replace(/[\d\+万]+人种草/, '');
+    title = title.replace(/[\d\+万]+人浏览/, '');
+    title = title.replace(/[^\s]{2,10}(?:旗舰店|专营店|京东自营)/, '');
+    title = title.replace(/搜同款|关注|对比/g, '');
+    title = title.trim();
+    
+    if (title.length > 5) {
+      product.title = title.substring(0, 100);
+    }
+    
+    if (product.title || product.price) {
+      result.products.push(product);
+    }
+  });
+  
+  return result;
+})()

package/scripts/find_applause_rate.js

@@ -0,0 +1,26 @@
+(() => {
+  const applauseRateEl = document.querySelector('.applause-rate');
+  
+  if (applauseRateEl) {
+    return {
+      elementFound: true,
+      elementText: applauseRateEl.textContent,
+      elementHTML: applauseRateEl.outerHTML.substring(0, 500)
+    };
+  } else {
+    const allElements = document.querySelectorAll('[class*="applause"], [class*="rate"]');
+    const elements = [];
+    allElements.forEach(el => {
+      elements.push({
+        className: el.className,
+        text: el.textContent?.substring(0, 100)
+      });
+    });
+    
+    return {
+      elementFound: false,
+      message: 'applause-rate element not found',
+      similarElements: elements.slice(0, 10)
+    };
+  }
+})()

package/scripts/find_jd_links.js

@@ -0,0 +1,28 @@
+(() => {
+  const result = {
+    allLinks: [],
+    jdLinks: [],
+    productLinks: []
+  };
+  
+  const links = document.querySelectorAll('a[href*="jd"]');
+  
+  links.forEach((link, index) => {
+    if (index < 20) {
+      result.allLinks.push({
+        href: link.href.substring(0, 100),
+        text: link.textContent?.substring(0, 50)
+      });
+    }
+    
+    if (link.href.includes('item.jd') || link.href.includes('product')) {
+      result.jdLinks.push({
+        href: link.href.substring(0, 100),
+        text: link.textContent?.substring(0, 50),
+        parentClass: link.parentElement?.className
+      });
+    }
+  });
+  
+  return result;
+})()

package/scripts/find_main_content.js

@@ -0,0 +1,20 @@
+(() => {
+  const result = {
+    bodyChildren: []
+  };
+  
+  const bodyChildren = document.body.children;
+  
+  for (let i = 0; i < Math.min(bodyChildren.length, 30); i++) {
+    const child = bodyChildren[i];
+    result.bodyChildren.push({
+      index: i,
+      tagName: child.tagName,
+      id: child.id,
+      className: child.className,
+      childrenCount: child.children.length
+    });
+  }
+  
+  return result;
+})()

package/scripts/find_product_cards.js

@@ -0,0 +1,38 @@
+(() => {
+  const result = {
+    bodyClasses: document.body.className,
+    children: []
+  };
+  
+  const mainContent = document.querySelector('main, #J_goodsList, [id*="goods"], [id*="product"]');
+  
+  if (mainContent) {
+    result.mainContainer = {
+      id: mainContent.id,
+      className: mainContent.className,
+      childrenCount: mainContent.children.length
+    };
+    
+    Array.from(mainContent.children).slice(0, 5).forEach((child, index) => {
+      result.children.push({
+        index,
+        tagName: child.tagName,
+        className: child.className,
+        id: child.id,
+        innerHTML: child.innerHTML.substring(0, 200)
+      });
+    });
+  } else {
+    Array.from(document.body.querySelectorAll('div')).slice(0, 20).forEach((div, index) => {
+      result.children.push({
+        index,
+        tagName: div.tagName,
+        className: div.className,
+        id: div.id,
+        innerHTML: div.innerHTML.substring(0, 200)
+      });
+    });
+  }
+  
+  return result;
+})()