{
  "player": "Caitlin Clark",
  "team": "Indiana Fever",
  "injury_type": "back injury",
  "status": "out (late scratch)",
  "quote_sentiment": "cautiously confident",
  "mention_link": "https://sports.yahoo.com/articles/caitlin-clark-issues-statement-sudden-011510647.html"
}

from pydantic import BaseModel, Field
from typing import Optional

class InjuryInfo(BaseModel):
    player: str = Field(description="球员全名")
    team: str = Field(description="所属球队")
    injury_type: str = Field(description="伤病类型/部位，如back injury, ankle sprain")
    status: str = Field(description="缺阵状态，如out, day-to-day, questionable")
    quote_sentiment: Optional[str] = Field(description="球员/教练发言的情绪倾向，如confident, worried, cautious")
    mention_link: str = Field(description="新闻原文URL")

import os
from openai import OpenAI
import json

client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

def extract_injury_info(text: str, url: str) -> InjuryInfo:
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[
            {"role": "system", "content": "You are an expert sports data analyst. Extract structured injury information from the news."},
            {"role": "user", "content": f"请从以下新闻中提取伤病信息：\n\n{text}"}
        ],
        functions=[
            {
                "name": "extract_injury_info",
                "description": "从体育新闻中提取球员伤病的关键结构化信息",
                "parameters": InjuryInfo.schema()
            }
        ],
        function_call={"name": "extract_injury_info"}
    )
    args = json.loads(response.choices[0].message.function_call.arguments)
    args["mention_link"] = url
    return InjuryInfo(**args)

import requests
from bs4 import BeautifulSoup

def fetch_article_text(url: str) -> str:
    headers = {"User-Agent": "Mozilla/5.0"}
    resp = requests.get(url, headers=headers, timeout=10)
    soup = BeautifulSoup(resp.text, "html.parser")
    # Yahoo Sports 的文章正文通常在 <article> 下的 <p>
    article = soup.find("article")
    if not article:
        # 兜底：取所有<p>前5段
        paragraphs = soup.find_all("p")[:5]
        return "\n".join([p.get_text() for p in paragraphs])
    paragraphs = article.find_all("p")
    return "\n".join([p.get_text() for p in paragraphs])

if __name__ == "__main__":
    url = "https://sports.yahoo.com/articles/caitlin-clark-issues-statement-sudden-011510647.html"
    text = fetch_article_text(url)
    info = extract_injury_info(text, url)
    print(info.json(indent=2))

sports-injury-extractor/
├── main.py          # 爬取+提取入口
├── models.py        # Pydantic模型定义
├── extractor.py     # 封装调用OpenAI的逻辑
├── scraper.py       # 新闻正文爬取
├── .env             # OPENAI_API_KEY
└── requirements.txt

openai>=1.12.0
pydantic>=2.5
requests>=2.31
beautifulsoup4>=4.12

message = response.choices[0].message
if message.function_call:
    args = json.loads(message.function_call.arguments)
else:
    # 兜底：让模型用纯文本输出JSON
    content = message.content
    # 尝试提取```json块
    import re
    match = re.search(r'```json\n(.+?)\n```', content, re.DOTALL)
    if match:
        args = json.loads(match.group(1))
    else:
        raise ValueError("模型没有按要求输出结构化数据")