用 Python 做一个社交媒体内容生成器，一篇文章自动拆成 5 个平台文案

pip install openai beautifulsoup4 requests pyyaml

platforms:
  xiaohongshu:
    name: "小红书"
    style: "活泼亲切，多用emoji，短句分段，每段不超过2行"
    max_length: 1000
    include_hashtags: true
    hashtag_count: 5
    tone: "种草风格，像在跟闺蜜分享好物"
  
  weibo:
    name: "微博"
    style: "短平快，观点鲜明，带话题标签"
    max_length: 280
    include_hashtags: true
    hashtag_count: 3
    tone: "犀利点评风格"
  
  zhihu:
    name: "知乎"
    style: "专业严谨，逻辑清晰，适当引用数据"
    max_length: 3000
    include_hashtags: false
    tone: "理性分析风格"
  
  weixin:
    name: "公众号"
    style: "深度长文，结构完整，有引言和总结"
    max_length: 5000
    include_hashtags: false
    tone: "专业科普风格"
  
  moments:
    name: "朋友圈"
    style: "口语化，像日常分享，不要太正式"
    max_length: 200
    include_hashtags: false
    tone: "朋友间闲聊风格"

import yaml
import os
from openai import OpenAI
from bs4 import BeautifulSoup
from pathlib import Path

# 加载配置
with open("platforms.yaml", "r", encoding="utf-8") as f:
    platforms_config = yaml.safe_load(f)

# 初始化客户端（支持 OpenAI 兼容接口）
client = OpenAI(
    api_key=os.getenv("OPENAI_API_KEY", "YOUR_API_KEY"),
    base_url=os.getenv("OPENAI_BASE_URL", "https://api.openai.com/v1"),
)

OUTPUT_DIR = Path(__file__).parent / "outputs"
OUTPUT_DIR.mkdir(exist_ok=True)


def extract_article(text):
    """提取文章核心内容，去掉多余格式"""
    # 如果输入是 HTML，用 BeautifulSoup 提取正文
    if "<html" in text.lower():
        soup = BeautifulSoup(text, "html.parser")
        # 移除 script 和 style 标签
        for tag in soup(["script", "style", "nav", "footer"]):
            tag.decompose()
        text = soup.get_text(separator="\n")
    
    # 按段落分割，取前 N 段作为核心内容
    paragraphs = [p.strip() for p in text.split("\n\n") if len(p.strip()) > 20]
    return "\n\n".join(paragraphs[:20])


def generate_for_platform(article_text, platform_key):
    """针对单个平台生成文案"""
    plat = platforms_config["platforms"][platform_key]
    
    prompt = f"""
你是一个专业的社交媒体内容编辑。请把下面的文章改写成{plat['name']}风格的文案。

平台要求：
- 风格：{plat['style']}
- 语气：{plat['tone']}
- 最大字数：{plat['max_length']}字

规则：
1. 保留原文核心观点和信息
2. 按照平台风格调整表达方式和长度
3. {'需要添加 ' + str(plat['hashtag_count']) + ' 个相关话题标签' if plat.get('include_hashtags') else '不需要添加话题标签'}
4. 直接输出文案，不要加任何解释

原文：
{article_text[:3000]}
"""
    
    response = client.chat.completions.create(
        model=os.getenv("MODEL_NAME", "gpt-3.5-turbo"),
        messages=[{"role": "user", "content": prompt}],
        temperature=0.8,
        max_tokens=plat["max_length"],
    )
    
    return response.choices[0].message.content


def generate_all(article_text):
    """为所有平台生成文案"""
    cleaned = extract_article(article_text)
    results = {}
    
    for platform_key in platforms_config["platforms"]:
        print(f"正在生成 {platforms_config['platforms'][platform_key]['name']} 文案...")
        try:
            content = generate_for_platform(cleaned, platform_key)
            results[platform_key] = content
        except Exception as e:
            results[platform_key] = f"生成失败: {e}"
    
    return results


def save_output(results, source_title=""):
    """保存生成结果到文件"""
    timestamp = Path(__file__).parent / "outputs" / f"{source_title or 'post'}_{os.popen('date +%Y%m%d_%H%M%S').read().strip()}.md"
    
    with open(timestamp, "w", encoding="utf-8") as f:
        if source_title:
            f.write(f"# {source_title}\n\n")
        
        for platform_key, content in results.items():
            plat_name = platforms_config["platforms"][platform_key]["name"]
            f.write(f"---\n\n")
            f.write(f"## 📱 {plat_name}\n\n")
            f.write(f"{content}\n\n")
    
    print(f"✅ 已保存到: {timestamp}")
    return timestamp


if __name__ == "__main__":
    import sys
    
    # 支持命令行输入或读取文件
    if len(sys.argv) > 1:
        article_path = sys.argv[1]
        with open(article_path, "r", encoding="utf-8") as f:
            article_text = f.read()
        source = os.path.splitext(os.path.basename(article_path))[0]
    else:
        print("请输入文章内容（输入空行结束）:")
        lines = []
        while True:
            line = input()
            if not line.strip():
                break
            lines.append(line)
        article_text = "\n".join(lines)
        source = "manual_input"
    
    results = generate_all(article_text)
    save_output(results, source)
    
    # 同时在终端打印结果，方便直接复制
    print("\n" + "=" * 50)
    for platform_key, content in results.items():
        plat_name = platforms_config["platforms"][platform_key]["name"]
        print(f"\n📱 {plat_name}:\n{content}\n{'-' * 40}")

# 方式一：从文件读取
python social_poster.py article.md

# 方式二：交互式输入
python social_poster.py

# 方式三：配合环境变量
export OPENAI_API_KEY=sk-xxx
export OPENAI_BASE_URL=https://dashscope.aliyuncs.com/compatible-mode/v1
export MODEL_NAME=qwen-plus
python social_poster.py article.md

from apscheduler.schedulers.blocking import BlockingScheduler

def daily_generate():
    """每天早上9点检查 articles/ 目录下最新的文章"""
    articles_dir = Path(__file__).parent / "articles"
    if not articles_dir.exists():
        return
    
    latest = max(articles_dir.glob("*.md"), key=os.path.getmtime)
    with open(latest, "r", encoding="utf-8") as f:
        content = f.read()
    
    results = generate_all(content)
    save_output(results, latest.stem)

# 每天早上9点执行
scheduler = BlockingScheduler()
scheduler.add_job(daily_generate, "cron", hour=9, minute=0)
scheduler.start()