app.py

import json
import zipfile
import os
import requests
import re
import asyncio
import tempfile
import urllib3

# 關閉忽略 SSL 驗證時產生的警告，保持 Log 乾淨
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)

from datetime import datetime
from contextlib import asynccontextmanager
from fastapi import FastAPI, HTTPException
from fastapi.responses import HTMLResponse
from pydantic import BaseModel
import httpx

# --- 初始化與設定 ---
# 根據 2026 最新 Swagger 文件更新路徑
MOJ_API_URLS = {
    "chlaw": "https://law.moj.gov.tw/api/ch/law/json",
    "chorder": "https://law.moj.gov.tw/api/ch/order/json"
}

LOCAL_FILES = ["chlaw.json", "chorder.json"]
MAPPING_FILE = "fund_mapping.json"

CACHE = {
    "data": [],
    "last_update": "初始化中...",
    "source": "等待載入...",
    "is_updating": False
}

UPDATE_LOCK = asyncio.Lock()

class AnalyzeRequest(BaseModel):
    model_provider: str
    model_name: str
    api_key: str
    prompt: str
    context: str

# --- 工具函數 ---

def load_fund_mapping():
    if os.path.exists(MAPPING_FILE):
        try:
            with open(MAPPING_FILE, 'r', encoding='utf-8') as f:
                return json.load(f)
        except: return {}
    return {}

def parse_taiwan_date(date_str):
    if not date_str: return "未找到"
    date_str = str(date_str).strip()
    try:
        # 處理格式如 20240101 或 1130101
        if len(date_str) == 8:
            return f"民國 {int(date_str[:4]) - 1911} 年 {date_str[4:6]} 月 {date_str[6:]} 日"
        elif len(date_str) == 7:
            return f"民國 {int(date_str[:3])} 年 {date_str[3:5]} 月 {date_str[5:]} 日"
        return date_str
    except: return date_str

def clean_text(text):
    if not text: return ""
    return re.sub(r'\s+', ' ', re.sub(r'<[^>]+>', '', str(text))).strip()

def clean_article_no(art_no):
    if not art_no: return ""
    cleaned = str(art_no).replace("第", "").replace("條", "").strip().replace(".", "-")
    return f"第 {cleaned} 條"

def clean_authority(auth_str):
    """
    配合行政院組織改造，清洗並更新各主管機關名稱為現行版本
    """
    if not auth_str: return "未找到"
    auth_str = str(auth_str).strip()
    
    # 1. 金管會 (2012年更名)
    auth_str = auth_str.replace("行政院金融監督管理委員會", "金融監督管理委員會")
    
    # 2. 其他常見部會更名
    auth_str = auth_str.replace("行政院勞工委員會", "勞動部")
    auth_str = auth_str.replace("行政院衛生署", "衛生福利部")
    auth_str = auth_str.replace("行政院環境保護署", "環境部")
    auth_str = auth_str.replace("行政院農業委員會", "農業部")
    auth_str = auth_str.replace("科技部", "國家科學及技術委員會") 
    
    return auth_str

def process_raw_data(laws_list):
    processed = []
    target_keyword = "收支保管及運用辦法"
    mapping = load_fund_mapping()
    
    for law in laws_list:
        law_name = law.get('LawName', '')
        # 篩選特定基金法規
        if target_keyword in law_name:
            fund_name = law_name.replace("收支保管及運用辦法", "").replace("收支保管運用辦法", "")
            modified_date = parse_taiwan_date(law.get('LawModifiedDate'))
            law_history = clean_text(law.get('LawHistory', ''))
            
            is_abolished = (law.get('LawAbandon') == '廢')
            law_status = "廢止法規" if is_abolished else "現行法規"
            sort_weight = mapping.get(fund_name, 999999)
            
            for art in law.get('LawArticles', []):
                processed.append({
                    "基金名稱": fund_name,
                    "規定名稱": law_name,
                    "最近發布日期": modified_date,
                    "條次": clean_article_no(art.get('ArticleNo', '')),
                    "規定": clean_text(art.get('ArticleContent', '')),
                    "立法理由": clean_text(art.get('ArticleReason', '')),
                    "法規沿革": law_history,
                    "法規狀態": law_status,
                    "主管機關": clean_authority(law.get('LawCategory', '未找到')),
                    "sort_weight": sort_weight
                })
    return processed

# --- 資料載入與 API 更新 ---

def load_local_data():
    print(">>> [啟動] 正在檢查本地 JSON 檔案...")
    all_local_data = []
    has_file = False
    
    for filename in LOCAL_FILES:
        if os.path.exists(filename):
            try:
                with open(filename, 'r', encoding='utf-8-sig') as f:
                    raw_json = json.load(f)
                    laws = raw_json.get('Laws', [])
                    all_local_data.extend(process_raw_data(laws))
                    has_file = True
            except Exception as e:
                print(f"    - 讀取 {filename} 失敗: {e}")
                
    if has_file and all_local_data:
        all_local_data.sort(key=lambda x: (x['法規狀態'] == '廢止法規', x['sort_weight'], x['基金名稱'], x['條次']))
        CACHE["data"] = all_local_data
        CACHE["last_update"] = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        CACHE["source"] = "本地備份 (快速啟動)"
        print(f">>> [啟動] 本地資料載入完成，共 {len(all_local_data)} 筆。")
    else:
        print(">>> [啟動] 本地無可用資料，等待 API 更新。")

def download_file_with_retry(url, retries=3):
    headers = {
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36'
    }
    for i in range(retries):
        try:
            # 加入 verify=False 忽略 SSL 憑證問題
            with requests.get(url, stream=True, timeout=180, headers=headers, verify=False) as r:
                r.raise_for_status()
                with tempfile.NamedTemporaryFile(delete=False) as tmp_file:
                    for chunk in r.iter_content(chunk_size=8192):
                        if chunk: tmp_file.write(chunk)
                    return tmp_file.name
        except Exception as e:
            print(f"    - 下載嘗試 {i+1} 失敗: {e}")
            if i == retries - 1: return None
    return None

async def update_data_from_api():
    async with UPDATE_LOCK:
        if CACHE["is_updating"]: return
        print(">>> [背景] 開始請求全台法規最新資料...")
        CACHE["is_updating"] = True
        new_api_data = []
        download_success = False
        
        for key, url in MOJ_API_URLS.items():
            tmp_path = None
            try:
                tmp_path = download_file_with_retry(url)
                if tmp_path:
                    with zipfile.ZipFile(tmp_path, 'r') as z:
                        # 自動尋找壓縮檔內的 json 檔案
                        json_files = [n for n in z.namelist() if n.endswith('.json')]
                        if not json_files: continue
                        
                        with z.open(json_files[0]) as f:
                            raw_json = json.load(f)
                            processed = process_raw_data(raw_json.get('Laws', []))
                            new_api_data.extend(processed)
                            
                            # 更新本地備份
                            with open(f"{key}.json", 'w', encoding='utf-8') as local_f:
                                json.dump(raw_json, local_f, ensure_ascii=False)
                    download_success = True
            except Exception as e:
                print(f"    - API 處理錯誤 [{key}]: {e}")
            finally:
                if tmp_path and os.path.exists(tmp_path):
                    os.remove(tmp_path)
        
        if download_success and new_api_data:
            new_api_data.sort(key=lambda x: (x['法規狀態'] == '廢止法規', x['sort_weight'], x['基金名稱'], x['條次']))
            CACHE["data"] = new_api_data
            CACHE["last_update"] = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
            CACHE["source"] = "全國法規資料庫 API (最新)"
            print(f">>> [背景] API 更新完成，共 {len(new_api_data)} 筆。")
        
        CACHE["is_updating"] = False

# --- FastAPI 生命週期與路由 ---

@asynccontextmanager
async def lifespan(app: FastAPI):
    # 啟動時執行
    load_local_data()
    asyncio.create_task(update_data_from_api())
    yield
    # 關閉時可在此釋放資源

# 初始化 app 並帶入 lifespan
app = FastAPI(lifespan=lifespan)

@app.get("/api/get_funds")
async def get_funds_data():
    return {
        "status": "updating" if CACHE["is_updating"] else "ready",
        "source": CACHE["source"],
        "update_time": CACHE["last_update"],
        "count": len(CACHE["data"]),
        "data": CACHE["data"]
    }

@app.get("/", response_class=HTMLResponse)
def index():
    if os.path.exists("index.html"):
        with open("index.html", "r", encoding="utf-8") as f:
            return f.read()
    return "<h3>Error: index.html 檔案未找到，請確保該檔案位於程式碼相同目錄。</h3>"

@app.post("/api/analyze")
async def analyze_data(req: AnalyzeRequest):
    if not req.api_key:
        raise HTTPException(status_code=400, detail="請提供 API Key")
    
    provider = req.model_provider.lower()
    
    # 組合完整的 Prompt
    full_prompt = f"""
以下是特種基金的相關法條作為參考知識：
<fund_rules>
{req.context}
</fund_rules>
請根據上述法條，回答以下使用者問題：
{req.prompt}
"""

    if provider == "openai":
        url = "https://api.openai.com/v1/chat/completions"
        headers = {
            "Authorization": f"Bearer {req.api_key}",
            "Content-Type": "application/json"
        }
        payload = {
            "model": req.model_name or "gpt-4o",
            "messages": [
                {"role": "system", "content": "你是一個專業的台灣法規分析助手，請根據提供的條文內容給出精確、客觀的分析。"},
                {"role": "user", "content": full_prompt}
            ],
            "temperature": 0.2
        }
        
        async with httpx.AsyncClient() as client:
            try:
                response = await client.post(url, headers=headers, json=payload, timeout=60.0)
                response.raise_for_status()
                data = response.json()
                return {"result": data["choices"][0]["message"]["content"]}
            except Exception as e:
                raise HTTPException(status_code=500, detail=f"OpenAI API 請求失敗: {str(e)}")
                
    elif provider == "gemini":
        # Google Gemini API REST call
        model = req.model_name or "gemini-1.5-pro"
        url = f"https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent?key={req.api_key}"
        headers = {
            "Content-Type": "application/json"
        }
        payload = {
            "contents": [{
                "parts": [{"text": full_prompt}]
            }],
            "systemInstruction": {
                "parts": [{"text": "你是一個專業的台灣法規分析助手，請根據提供的條文內容給出精確、客觀的分析。"}]
            },
             "generationConfig": {
                "temperature": 0.2
            }
        }
        
        async with httpx.AsyncClient() as client:
            try:
                response = await client.post(url, headers=headers, json=payload, timeout=60.0)
                response.raise_for_status()
                data = response.json()
                text = data.get("candidates", [{}])[0].get("content", {}).get("parts", [{}])[0].get("text", "")
                if not text:
                    raise Exception("API 返回格式無法解析或被安全機制阻擋")
                return {"result": text}
            except Exception as e:
                raise HTTPException(status_code=500, detail=f"Gemini API 請求失敗: {str(e)}")
    
    else:
        raise HTTPException(status_code=400, detail="不支援的模型提供者 (目前僅支援 openai 或 gemini)")

if __name__ == "__main__":
    import uvicorn
    # 配合 Hugging Face Space 的預設 Port 通常為 7860
    uvicorn.run(app, host="0.0.0.0", port=7860)