# -*- coding: utf-8 -*-
import sqlite3
import time
import requests
import os
from datetime import datetime

# 数据库所在的真实绝对路径
DB_PATH = '/www/wwwroot/my_api/database.db'
SITEMAP_PATH = '/www/wwwroot/myhtml.asia/sitemap.xml'

# 根域名与推送秘钥
SITE_DOMAIN = 'https://myhtml.asia'
BAIDU_TOKEN = 'COZ4XmrTA1KdhU4C' 

def generate_and_push_sitemap():
    print(f"[{datetime.now().strftime('%Y-%m-%d %H:%M:%S')}] 开始生成 Sitemap 并推送到百度...")
    
    if not os.path.exists(DB_PATH):
        print(f"错误：找不到数据库文件！当前指定的路径是: {DB_PATH}")
        return

    conn = sqlite3.connect(DB_PATH)
    cursor = conn.cursor()
    current_ts = int(time.time())
    
    # 提取有效记录，按生成时间倒序
    cursor.execute('''
        SELECT url, subdomain, time 
        FROM records 
        WHERE (is_legacy = 1 OR expire_time > ?) 
        AND title NOT LIKE '[APP专属内核]%'
        ORDER BY id DESC LIMIT 5000
    ''', (current_ts,))
    
    rows = cursor.fetchall()
    conn.close()
    
    sitemap_urls = []
    push_urls = [] 
    
    # 使用 replace 方法，彻底向宝塔编辑器隐藏 XML 声明符号，告别一切红线误报
    xml_content = '【XML_START】xml version="1.0" encoding="UTF-8"?>\n'.replace('【XML_START】', '<?')
    xml_content += '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n'
    
    # 将主页加进去
    sitemap_urls.append(f"{SITE_DOMAIN}/")
    push_urls.append(f"{SITE_DOMAIN}/")
    
    xml_content += f'''  <url>
    <loc>{SITE_DOMAIN}/</loc>
    <changefreq>daily</changefreq>
    <priority>1.0</priority>
  </url>\n'''
    
    for r in rows:
        raw_url, subdomain, create_time = r[0], r[1], r[2]
        
        if subdomain:
            final_url = f"https://{subdomain}.myhtml.asia"
        else:
            final_url = raw_url
            
        # Sitemap 包含所有链接（包含子域名）
        sitemap_urls.append(final_url)
        
        # 主动推送只发根域名下的链接，避免触发 not_same_site 被百度拒绝
        if not subdomain:
            push_urls.append(final_url)
        
        date_str = create_time.split(' ')[0] if create_time else datetime.now().strftime('%Y-%m-%d')
        xml_content += f'''  <url>
    <loc>{final_url}</loc>
    <lastmod>{date_str}</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
  </url>\n'''

    xml_content += '</urlset>'
    
    # 1. 写入物理 Sitemap 文件
    try:
        with open(SITEMAP_PATH, 'w', encoding='utf-8') as f:
            f.write(xml_content)
        print(f"✅ 成功生成 sitemap.xml，共包含 {len(sitemap_urls)} 条链接。")
    except Exception as e:
        print(f"❌ 写入 sitemap.xml 失败: {str(e)}")
        return

    # 百度单次推送最多接收 2000 条，做安全切割
    push_urls = push_urls[:2000]

    # 2. API 主动推送 (使用 params 传参彻底绕过 & 符号的语法报错)
    base_api = "http://data.zz.baidu.com/urls"
    api_params = {
        "site": SITE_DOMAIN,
        "token": BAIDU_TOKEN
    }
    headers = {'Content-Type': 'text/plain'}
    push_data = "\n".join(push_urls)
    
    try:
        print(f"🚀 正在向百度推送 {len(push_urls)} 条符合主域名的链接...")
        res = requests.post(base_api, params=api_params, headers=headers, data=push_data, timeout=15)
        result = res.json()
        
        if res.status_code == 200:
            print(f"✅ 推送成功！")
            print(f"🎉 百度实际成功接收: {result.get('success', 0)} 条")
            print(f"🔋 今日剩余可推送配额: {result.get('remain', '未知')} 条")
            if 'not_same_site' in result:
                print(f"⚠️ 因域名不匹配被拒绝: {len(result['not_same_site'])} 条")
        else:
            print(f"❌ 推送失败或超额，百度错误信息: {result}")
    except Exception as e:
        print(f"❌ 请求发生网络故障: {str(e)}")

if __name__ == '__main__':
    generate_and_push_sitemap()