Chinese Data Scraper — 中国网站数据爬取专家

⚡ INSTANT VALUE — Install This If You:

• Need to scrape Baidu/Taobao/Douyin/Zhihu/WeChat/1688 but keep hitting anti-crawl walls
• Want platform-specific bypass recipes — not generic "use Selenium" advice, but tested configs for each Chinese site
• Are using Scrapling but need Chinese site configurations (Baidu's cookie walls, Taobao's login gates, Douyin's dynamic rendering)
• Want to know what's legal — China's data scraping legal boundaries (Criminal Law 285/286, Data Security Law, PIPL)

🎯 Why this over generic scraping skills? Generic scraping skills give you BeautifulSoup/Selenium tutorials. We give you tested anti-crawl configs for 10+ Chinese websites, legal compliance boundaries (avoid Criminal Law 285!), and Scrapling integration with Chinese site presets. Tutorials vs Recipes — you decide.

🔗 Based on Scrapling (31K+ GitHub Stars) — the fastest Python scraping framework with adaptive selectors and Cloudflare bypass. We add the China layer on top.

You are a Chinese website data scraping expert. You help AI agents and developers scrape data from Chinese websites that are notoriously difficult to crawl — Baidu, Taobao, Douyin, Zhihu, WeChat, 1688, and more.

Core Philosophy

Scrapling solves the general scraping problem. We solve the China-specific problem.

Chinese websites have unique anti-crawl mechanisms that generic tools can't handle:

• Baidu's multi-layer cookie verification + JS encryption
• Taobao's mandatory login gates + dynamic pricing
• Douyin's signature algorithm + device fingerprinting
• Zhihu's X-Zse-93 encryption + rate limiting
• WeChat's closed ecosystem + no public API
• 1688's anti-bot + business data protection

This skill provides tested recipes for each platform, not generic advice.

🏗️ Architecture: Scrapling + China Layer

┌─────────────────────────────────────┐
│         AI Agent / User             │
├─────────────────────────────────────┤
│      cn-data-scraper Skill          │
│  ┌─────────────┐ ┌───────────────┐  │
│  │ Platform    │ │ Legal         │  │
│  │ Recipes     │ │ Compliance    │  │
│  │ (10+ sites) │ │ Boundaries    │  │
│  └──────┬──────┘ └───────┬───────┘  │
│         │                │          │
│  ┌──────▼────────────────▼───────┐  │
│  │    Scrapling Framework        │  │
│  │  (Adaptive Selectors +        │  │
│  │   StealthyFetcher +           │  │
│  │   Camoufox Engine)            │  │
│  └──────────────────────────────┘  │
└─────────────────────────────────────┘

📋 Platform-Specific Anti-Crawl Recipes

1. 百度 (Baidu) — 搜索结果 + 百度知道 + 百科

Anti-crawl mechanisms:

• Multi-layer cookie verification (BDORZ + BAIDUID + BIDUPSID)
• JS-encrypted search parameters
• Rate limiting by IP (100 requests/hour for unauthenticated)
• CAPTCHA after threshold

Scrapling configuration:

from scrapling import StealthyFetcher

# Baidu search with anti-crawl bypass
page = StealthyFetcher.fetch(
    'https://www.baidu.com/s?wd=关键词',
    headless=True,
    network_idle=True,  # Wait for JS execution
    timeout=30000,
)

# Extract search results — use adaptive selectors
# Baidu frequently changes class names, so use structural selectors
results = page.css('div.c-container')  # More stable than class-based
for result in results:
    title = result.css_first('h3 a')
    snippet = result.css_first('span.content-right_8Zs40')
    # Fallback: adaptive selector if structure changed
    if not snippet:
        snippet = result.css_first('[class*="content"]')

Key tips:

• Add User-Agent with Baidu app identifier for mobile results
• Use network_idle=True — Baidu loads results via AJAX
• Rotate cookies every 50 requests
• Avoid scraping between 8:00-10:00 AM (peak monitoring hours)

2. 淘宝/天猫 (Taobao/Tmall) — 商品数据 + 评价

Anti-crawl mechanisms:

• Mandatory login gate for search
• Dynamic pricing based on user profile
• Anti-bot via mtop API signature
• Device fingerprinting (umid token)

Scrapling configuration:

# Taobao requires login — use cookie injection
page = StealthyFetcher.fetch(
    'https://s.taobao.com/search?q=关键词',
    headless=True,
    network_idle=True,
    # Must inject login cookies
    cookies={
        '_m_h5_tk': 'your_token_here',
        '_m_h5_tk_enc': 'your_enc_token_here',
        'cookie2': 'your_cookie2',
        'sgcookie': 'your_sgcookie',
    }
)

# Extract product data
products = page.css('div.Card--doubleCardWrapper')
for product in products:
    title = product.css_first('span.Title--titleSpan')
    price = product.css_first('span.Price--priceInt')
    sales = product.css_first('span.Sales--sales')

Key tips:

• Login cookies expire every 24 hours — need refresh mechanism
• Mobile API (h5api.m.taobao.com) is easier to scrape than desktop
• Use mtop.taobao.searchapi.search API endpoint for structured data
• Price data differs by user profile — use anonymous cookies for baseline

3. 抖音 (Douyin/TikTok CN) — 视频数据 + 用户信息

Anti-crawl mechanisms:

• X-Bogus signature algorithm (proprietary)
• Device fingerprinting (aid + version + platform)
• Rate limiting by device ID
• CAPTCHA for suspicious patterns

Scrapling configuration:

# Douyin web version — easier than app API
page = StealthyFetcher.fetch(
    'https://www.douyin.com/search/关键词',
    headless=True,
    network_idle=True,
    wait_selector='div.video-card',  # Wait for video cards to load
)

# Extract video data
videos = page.css('div.video-card')
for video in videos:
    title = video.css_first('p.title')
    author = video.css_first('span.author-card-user-name')
    likes = video.css_first('span.video-like-count')

Key tips:

• Web version is easier than app API — start with web
• X-Bogus algorithm changes every 2-4 weeks — use adaptive selectors
• Rate limit: 30 requests/minute for web, 10/minute for API
• Video download URLs expire in 1 hour
• ⚠️ DO NOT attempt to scrape user private data — Criminal Law 253

4. 知乎 (Zhihu) — 问答 + 文章 + 评论

Anti-crawl mechanisms:

• X-Zse-93/X-Zse-96 encryption headers
• Rate limiting by IP + cookie
• Login wall for full content
• Anti-bot via behavioral analysis

Scrapling configuration:

# Zhihu search — use API endpoint for structured data
page = StealthyFetcher.fetch(
    'https://www.zhihu.com/search?type=content&q=关键词',
    headless=True,
    network_idle=True,
    # Zhihu requires specific headers
    headers={
        'Referer': 'https://www.zhihu.com/',
    }
)

# Extract search results
results = page.css('div.SearchResult-Card')
for result in results:
    title = result.css_first('h2.ContentItem-title a')
    excerpt = result.css_first('span.RichText')
    author = result.css_first('meta[itemprop="name"]')

Key tips:

• API endpoint api.zhihu.com/search_v3 returns JSON — easier to parse
• X-Zse-93 encryption requires reverse engineering — use browser automation instead
• Anonymous access limited to 5 pages per search
• Login cookies extend to 20 pages
• ⚠️ Respect Zhihu's robots.txt — do not scrape user profiles in bulk

5. 微信公众号 (WeChat Official Accounts) — 文章内容

Anti-crawl mechanisms:

• Closed ecosystem — no public search API
• Article URLs expire and are non-guessable
• Anti-bot via WeChat referrer check
• Rate limiting by IP

Scrapling configuration:

# WeChat article — direct URL access
page = StealthyFetcher.fetch(
    'https://mp.weixin.qq.com/s/ARTICLE_ID',
    headless=True,
    network_idle=True,
)

# Extract article content
title = page.css_first('h1.rich_media_title')
content = page.css_first('div.rich_media_content')
author = page.css_first('a.rich_media_meta_link')
publish_time = page.css_first('em#publish_time')

Key tips:

• Article URLs can only be found via Sogou WeChat search or account history
• Sogou WeChat search: weixin.sogou.com — but has aggressive anti-crawl
• Use mp.weixin.qq.com direct access for known URLs
• Content is server-rendered — no need for JS execution
• ⚠️ DO NOT redistribute scraped articles — copyright infringement risk

6. 1688 (Alibaba Wholesale) — 商品数据 + 供应商

Anti-crawl mechanisms:

• Anti-bot via Alibaba's security system
• Login requirement for contact info
• Dynamic pricing based on buyer profile
• Image watermarking

Scrapling configuration:

# 1688 search
page = StealthyFetcher.fetch(
    'https://s.1688.com/selloffer/offer_search.htm?keywords=关键词',
    headless=True,
    network_idle=True,
)

# Extract product listings
products = page.css('div.offer-item')
for product in products:
    title = product.css_first('a.title')
    price = product.css_first('span.price')
    min_order = product.css_first('span.min-order')
    supplier = product.css_first('a.company-name')

7. 小红书 (Xiaohongshu/RED) — 笔记 + 评论

Anti-crawl mechanisms:

• X-s/X-t signature algorithm
• Device fingerprinting
• Aggressive rate limiting (10 requests/minute)
• CAPTCHA after threshold

Scrapling configuration:

# Xiaohongshu web version
page = StealthyFetcher.fetch(
    'https://www.xiaohongshu.com/search_result?keyword=关键词',
    headless=True,
    network_idle=True,
    wait_selector='section.note-item',
)

# Extract notes
notes = page.css('section.note-item')
for note in notes:
    title = note.css_first('div.title')
    author = note.css_first('span.name')
    likes = note.css_first('span.like-wrapper .count')

Key tips:

• Web version has limited results — mobile API has more data
• X-s signature changes frequently — use browser automation as fallback
• Rate limit is very aggressive — 10 req/min for anonymous
• ⚠️ Scraping user-generated content may violate platform ToS

8. 微博 (Weibo) — 热搜 + 用户帖子 + 评论

Anti-crawl mechanisms:

• Login requirement for search
• Rate limiting by cookie
• Anti-bot via behavioral analysis
• Content censorship (some posts invisible without login)

Scrapling configuration:

# Weibo search
page = StealthyFetcher.fetch(
    'https://s.weibo.com/weibo?q=关键词',
    headless=True,
    network_idle=True,
    cookies={
        'SUB': 'your_sub_cookie',  # Required for search
    }
)

# Extract posts
posts = page.css('div.card-wrap[action-type="feed_list_item"]')
for post in posts:
    author = post.css_first('a.name')
    content = post.css_first('p.txt')
    reposts = post.css_first('a[action-type="fl_forward"] em')
    comments = post.css_first('a[action-type="flcomment"] em')
    likes = post.css_first('a[action-type="fl_like"] em')

9. CSDN — 技术博客 + 资源

Anti-crawl mechanisms:

• Login wall for full article content
• Anti-copy protection (JS overlay)
• Rate limiting by IP
• VIP content paywall

Scrapling configuration:

# CSDN article
page = StealthyFetcher.fetch(
    'https://blog.csdn.net/author/article/ID',
    headless=True,
    network_idle=True,
)

# Remove anti-copy overlay
content = page.css_first('article.baidu_pl')
# Content is in HTML, anti-copy is just a CSS overlay

10. Boss直聘 (Boss Zhipin) — 招聘数据

Anti-crawl mechanisms:

• Mandatory login
• Device fingerprinting
• Anti-bot via zpData encryption
• Rate limiting by cookie + IP

Scrapling configuration:

# Boss Zhipin search
page = StealthyFetcher.fetch(
    'https://www.zhipin.com/web/geek/job?query=关键词',
    headless=True,
    network_idle=True,
    cookies={
        'geek_zp_token': 'your_token',
    }
)

# Extract job listings
jobs = page.css('li.job-card-wrapper')
for job in jobs:
    title = job.css_first('span.job-name')
    salary = job.css_first('span.salary')
    company = job.css_first('h3.company-name')
    location = job.css_first('span.job-area')

⚖️ Legal Compliance Boundaries — 法律合规红线

This is NOT optional. Violating these can result in criminal prosecution.

🟢 Generally Safe (Legal)

• Scraping publicly accessible data without login
• Collecting data for personal research (non-commercial)
• Scraping with explicit API access (platform-approved)
• Academic research with anonymized data
• Price monitoring of publicly listed prices

🟡 Gray Area (Proceed with Caution)

• Scraping data behind login walls (may violate ToS)
• Collecting user-generated content at scale (copyright risk)
• Rate limiting bypass (may be considered "unauthorized access")
• Scraping for commercial competitive intelligence
• Using scraped data for AI training (copyright unclear)

🔴 Illegal (Criminal Liability)

• Scraping personal private data (Criminal Law Article 253: 侵犯公民个人信息罪)
  • Penalties: Up to 7 years imprisonment + fine
  • Examples: ID numbers, phone numbers, home addresses, financial data
• Bypassing computer system security measures (Criminal Law Article 285: 非法侵入计算机信息系统罪)
  • Penalties: Up to 7 years imprisonment
  • Examples: Breaking encryption, forging authentication tokens
• Disrupting computer systems (Criminal Law Article 286: 破坏计算机信息系统罪)
  • Penalties: Up to 15 years imprisonment
  • Examples: DDoS-level scraping, crashing servers
• Scraping state secrets or classified data
• Scraping financial data without proper licensing
• Reselling scraped personal data (PIPL violations)

Key Legal References

Real Cases (China)

1. 2023: 某数据公司爬取1.2亿条个人信息 — Criminal Law 253, CEO sentenced to 4 years
2. 2022: 爬取大众点评数据案 — Anti-Unfair Competition Law, ¥500K fine + injunction
3. 2021: 爬取微博数据案 — Court ruled scraping behind login = unauthorized access
4. 2020: 脉脉爬取LinkedIn数据 — Anti-Unfair Competition Law, ¥2M fine

🛠️ Scrapling Quick Start for Chinese Sites

Installation

# Install Scrapling with all features
pip install scrapling[all]

# Or minimal install
pip install scrapling

Basic Usage Pattern

from scrapling import StealthyFetcher, Fetcher

# 1. Simple HTTP fetch (fast, no JS rendering)
page = Fetcher.get('https://example.com')

# 2. Stealthy browser fetch (bypasses anti-bot)
page = StealthyFetcher.fetch(
    'https://www.baidu.com/s?wd=test',
    headless=True,
    network_idle=True,
)

# 3. Adaptive selectors — survive site redesigns
element = page.find_by_text('价格')  # Find by text content
element = page.css_first('[class*="price"]')  # Partial class match

Adaptive Selector Strategies for Chinese Sites

Chinese websites redesign frequently. Use these strategies to make your selectors resilient:

# ❌ BAD: Exact class names (break on redesign)
title = page.css_first('span.title_3wVZ1')

# ✅ GOOD: Structural selectors
title = page.css_first('h2 a')  # Semantic HTML

# ✅ GOOD: Partial class match
title = page.css_first('[class*="title"]')

# ✅ GOOD: Text-based selection
title = page.find_by_text('价格')

# ✅ GOOD: Attribute-based
title = page.css_first('[data-type="title"]')

# ✅ BEST: Scrapling's adaptive selector
# Scrapling remembers element characteristics and re-finds after changes
element = page.css_first('div.product-title')
# If class changes, Scrapling's smart locator adapts automatically

Rate Limiting Best Practices

import time
import random

def polite_scrape(urls, min_delay=2, max_delay=5):
    """Scrape with polite rate limiting"""
    results = []
    for url in urls:
        page = Fetcher.get(url)
        results.append(page)
        # Random delay to appear human
        delay = random.uniform(min_delay, max_delay)
        time.sleep(delay)
    return results

# Platform-specific rate limits
RATE_LIMITS = {
    'baidu': {'min': 3, 'max': 8, 'max_per_hour': 80},
    'taobao': {'min': 5, 'max': 12, 'max_per_hour': 30},
    'douyin': {'min': 5, 'max': 15, 'max_per_hour': 20},
    'zhihu': {'min': 3, 'max': 8, 'max_per_hour': 40},
    'wechat': {'min': 2, 'max': 5, 'max_per_hour': 60},
    '1688': {'min': 5, 'max': 10, 'max_per_hour': 30},
    'xiaohongshu': {'min': 8, 'max': 20, 'max_per_hour': 10},
    'weibo': {'min': 3, 'max': 8, 'max_per_hour': 40},
}

📊 Data Extraction Templates

E-commerce Product Data

def extract_product(page, platform='generic'):
    """Universal product data extractor"""
    templates = {
        'taobao': {
            'title': 'span.Title--titleSpan',
            'price': 'span.Price--priceInt',
            'sales': 'span.Sales--sales',
            'shop': 'a.ShopName--shopName',
        },
        '1688': {
            'title': 'a.title',
            'price': 'span.price',
            'min_order': 'span.min-order',
            'supplier': 'a.company-name',
        },
        'jd': {
            'title': 'div.sku-name',
            'price': 'span.price',
            'comments': 'a.comment-count',
        },
    }
    
    selector = templates.get(platform, templates['taobao'])
    return {
        field: page.css_first(sel).text() if page.css_first(sel) else None
        for field, sel in selector.items()
    }

Social Media Content

def extract_social_post(page, platform='generic'):
    """Universal social media post extractor"""
    templates = {
        'weibo': {
            'author': 'a.name',
            'content': 'p.txt',
            'reposts': 'a[action-type="fl_forward"] em',
            'comments': 'a[action-type="flcomment"] em',
            'likes': 'a[action-type="fl_like"] em',
        },
        'xiaohongshu': {
            'author': 'span.name',
            'content': 'span.note-text',
            'likes': 'span.like-wrapper .count',
            'collects': 'span.collect-wrapper .count',
        },
        'douyin': {
            'author': 'span.author-card-user-name',
            'content': 'p.title',
            'likes': 'span.video-like-count',
        },
    }
    
    selector = templates.get(platform, templates['weibo'])
    return {
        field: page.css_first(sel).text() if page.css_first(sel) else None
        for field, sel in selector.items()
    }

🔧 Executable Scripts

scripts/scrape.sh — Quick CLI Scraper

#!/bin/bash
# cn-data-scraper CLI tool
# Usage: ./scripts/scrape.sh \x3Cplatform> \x3Ckeyword> [options]

PLATFORM=$1
KEYWORD=$2
OUTPUT=${3:-/tmp/scrape_result.json}

if [ -z "$PLATFORM" ] || [ -z "$KEYWORD" ]; then
    echo "Usage: ./scripts/scrape.sh \x3Cplatform> \x3Ckeyword> [output_file]"
    echo "Platforms: baidu taobao douyin zhihu wechat 1688 xiaohongshu weibo csdn boss"
    exit 1
fi

python3 -c "
from scrapling import StealthyFetcher, Fetcher
import json

platform = '$PLATFORM'
keyword = '$KEYWORD'
output = '$OUTPUT'

URLS = {
    'baidu': f'https://www.baidu.com/s?wd={keyword}',
    'zhihu': f'https://www.zhihu.com/search?type=content&q={keyword}',
    'weibo': f'https://s.weibo.com/weibo?q={keyword}',
    'csdn': f'https://so.csdn.net/so/search?q={keyword}',
}

url = URLS.get(platform)
if not url:
    print(json.dumps({'error': f'Platform {platform} not supported for CLI scraping. Use Python API for full features.'}))
    exit(0)

try:
    if platform in ['baidu', 'zhihu', 'weibo']:
        page = StealthyFetcher.fetch(url, headless=True, network_idle=True, timeout=30000)
    else:
        page = Fetcher.get(url)
    
    # Extract all text content
    texts = [el.text() for el in page.css('p, span, h1, h2, h3, h4, h5, h6') if el.text()]
    
    result = {
        'platform': platform,
        'keyword': keyword,
        'url': url,
        'content_count': len(texts),
        'preview': texts[:20],
    }
    
    with open(output, 'w') as f:
        json.dump(result, f, ensure_ascii=False, indent=2)
    print(json.dumps(result, ensure_ascii=False, indent=2))
except Exception as e:
    print(json.dumps({'error': str(e)}))
"

🔗 MCP Server Integration

For AI agent workflows, this skill can be used with MCP servers:

# Example: MCP tool for scraping Chinese websites
from mcp.server import Server

server = Server("cn-data-scraper")

@server.tool()
def scrape_chinese_site(platform: str, keyword: str, max_results: int = 10) -> dict:
    """Scrape data from Chinese websites with anti-crawl bypass.
    
    Args:
        platform: Target platform (baidu/taobao/douyin/zhihu/wechat/1688/xiaohongshu/weibo)
        keyword: Search keyword
        max_results: Maximum number of results to return
    
    Returns:
        Dictionary with scraped data and metadata
    """
    # Implementation using Scrapling + platform recipes
    pass

@server.tool()
def check_legal_compliance(scraping_plan: str) -> dict:
    """Check if a scraping plan complies with Chinese law.
    
    Args:
        scraping_plan: Description of what data you plan to scrape
    
    Returns:
        Compliance assessment with risk level and legal references
    """
    pass

📈 Scrapling vs Other Frameworks (China Context)

Our value-add: Scrapling handles the technical scraping. We add the China layer (platform recipes + legal compliance + adaptive selectors for Chinese sites).

🚨 Common Pitfalls for Chinese Website Scraping

1. Don't use English-language scraping tutorials for Chinese sites — anti-crawl mechanisms are completely different
2. Don't ignore rate limits — Chinese platforms are MORE aggressive than Western ones (小红书 10/min vs Twitter 900/15min)
3. Don't scrape personal data — China's PIPL is stricter than GDPR in some aspects (criminal liability, not just fines)
4. Don't assume HTTPS = safe — Chinese ISPs can see your traffic; use proxy for sensitive scraping
5. Don't hardcode selectors — Chinese sites redesign every 2-4 weeks; use adaptive selectors
6. Don't scrape during peak hours — 8:00-10:00 AM and 7:00-9:00 PM have the most aggressive monitoring
7. Don't ignore robots.txt — While not legally binding in China, ignoring it increases legal risk
8. Don't store raw personal data — Anonymize immediately; PIPL requires "最小必要原则" (minimum necessary)

📚 Resources

• Scrapling GitHub: https://github.com/D4Vinci/Scrapling (31K+ Stars)
• Scrapling 中文文档: https://github.com/D4Vinci/Scrapling/blob/main/docs/README_CN.md
• Camoufox (反指纹引擎): https://github.com/nicegamer7/camoufox
• 中国爬虫法律案例: https://www.chinacourt.org (搜索"非法获取计算机信息系统数据")

Important Notes

• This skill teaches scraping techniques for legitimate data collection only
• Always check robots.txt before scraping any website
• Always comply with China's Criminal Law, Data Security Law, and PIPL
• When in doubt, use official APIs instead of scraping
• Scraping personal data without consent is a CRIMINAL OFFENSE in China
• Free tier: This skill is free. Platform-specific advanced recipes may require API access.