基于Python开发一个自动写作工具

import polars as pl
from bs4 import BeautifulSoup
import httpx

async def fetch_hot_topics():
    # 使用Polars并行处理多个数据源
    sources = [
        "https://trends.baidu.com",
        "https://weibo.com/hot",
        "https://toutiao.com/hot"
    ]
    
    async with httpx.AsyncClient() as client:
        tasks = [client.get(url) for url in sources]
        responses = await asyncio.gather(*tasks)
    
    # 使用Polars进行数据清洗和分析
    df = pl.DataFrame({
        "source": [r.url.host for r in responses],
        "content": [BeautifulSoup(r.text).get_text() for r in responses]
    })
    
    # 提取关键词并计算热度
    hot_topics = (df
                 .with_columns(pl.col("content").str.extract_all(r"\#(.+?)\#").alias("topics"))
                 .explode("topics")
                 .groupby("topics")
                 .agg(pl.count().alias("frequency"))
                 .sort("frequency", descending=True)
                 .head(20))
    
    return hot_topics.to_dicts()

import openai
from transformers import GPT2Tokenizer

class ArticleGenerator:
    def __init__(self):
        self.tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
        openai.api_key = "your_api_key_here"
        
    def generate_article(self, topic, style="professional"):
        prompt = f"""根据以下要求撰写一篇专业文章：
        标题：{topic}
        风格：{style}
        字数：1500字左右
        结构：引言→3个核心观点→结论
        要求：每段不超过200字，包含数据支持"""
        
        response = openai.ChatCompletion.create(
            model="gpt-4-turbo-2025",
            messages=[{"role": "user", "content": prompt}],
            temperature=0.7,
            max_tokens=3000
        )
        
        return response.choices[0].message.content

from rouge import Rouge 
import numpy as np

class QualityOptimizer:
    def __init__(self):
        self.rouge = Rouge()
        self.min_acceptable_score = 0.65
        
    def optimize_content(self, text):
        # 分句处理
        sentences = [s for s in text.split("。") if len(s) > 10]
        
        # 计算句子间相似度
        scores = []
        for i in range(len(sentences)-1):
            score = self.rouge.get_scores(sentences[i], sentences[i+1])[0]['rouge-l']['f']
            scores.append(score)
        
        # 移除冗余内容
        optimized = []
        for i, s in enumerate(sentences):
            if i == 0 or scores[i-1] < self.min_acceptable_score:
                optimized.append(s)
                
        return "。".join(optimized) + "。"