import requests
import sqlite3
import re
import base64
from datetime import datetime, timedelta
import os
import json

# ==============================================================================
# --- ⚙️ 配置区域 ---
# ==============================================================================

SEARCH_DAYS_AGO = 7
MIN_STARS = 50
DB_FILE = 'github_projects.db'
README_EXCERPT_LENGTH = 300

# ==============================================================================
# --- 脚本核心代码 ---
# ==============================================================================

# SESSION = requests.Session()
# SESSION.headers.update({'Authorization': f'token {GITHUB_TOKEN}'})

def setup_database():
    print(" [Scheduler] 正在检查并设置数据库...")
    conn = sqlite3.connect(DB_FILE)
    cursor = conn.cursor()
    cursor.execute('''
        CREATE TABLE IF NOT EXISTS projects (
            repo_id INTEGER PRIMARY KEY, name TEXT NOT NULL, url TEXT NOT NULL UNIQUE,
            description TEXT, readme_excerpt TEXT, language TEXT, stars INTEGER, pushed_at TIMESTAMP,
            owner_avatar_url TEXT, topics TEXT
        )
    ''')
    conn.commit()
    conn.close()

def has_chinese(text):
    if not text: return False
    return re.search(r'[\u4e00-\u9fa5]', text) is not None

def get_readme_content_with_excerpt(repo_full_name, session):
    """
    智能两步扫描逻辑
    """
    print(f"  -> 智能扫描 '{repo_full_name}'...")
    
    chinese_readme_filenames = ["README_CN.md", "README_zh.md", "README_zh-CN.md", "README-zh_CN.md"]
    
    for filename in chinese_readme_filenames:
        url = f"https://api.github.com/repos/{repo_full_name}/contents/{filename}"
        try:
            response = session.get(url, timeout=10)
            if response.status_code == 200:
                print(f"  ✨ 找到中文文件: '{filename}'.")
                content = response.json().get('content', '')
                decoded_content = base64.b64decode(content).decode('utf-8', errors='ignore')
                if has_chinese(decoded_content):
                    excerpt = re.sub(r'#+\s*|\!\[.*\]\(.*\)|<.*?>|\[|\]|\(|\)', '', decoded_content).strip().replace('\n', ' ')
                    return excerpt[:README_EXCERPT_LENGTH]
        except requests.RequestException:
            continue

    print("  - 检查默认 README.md...")
    url = f"https://api.github.com/repos/{repo_full_name}/contents/README.md"
    try:
        response = session.get(url, timeout=10)
        if response.status_code == 200:
            content = response.json().get('content', '')
            decoded_content = base64.b64decode(content).decode('utf-8', errors='ignore')
            if has_chinese(decoded_content):
                print(f"  ✔️ 默认 README 中发现中文.")
                excerpt = re.sub(r'#+\s*|\!\[.*\]\(.*\)|<.*?>|\[|\]|\(|\)', '', decoded_content).strip().replace('\n', ' ')
                return excerpt[:README_EXCERPT_LENGTH]
            else:
                print(f"  ❌ 默认 README 中无中文.")
        else:
             print(f"  - 默认 README.md 未找到.")
    except requests.RequestException as e:
        print(f"  ⚠️  抓取 README 出错: {e}")

    return None

def get_new_hot_repos(session):
    search_date = (datetime.now() - timedelta(days=SEARCH_DAYS_AGO)).strftime('%Y-%m-%d')
    query = f'created:>{search_date} stars:>{MIN_STARS}'
    url = f'https://api.github.com/search/repositories?q={query}&sort=stars&order=desc&per_page=100'
    try:
        response = session.get(url, timeout=20)
        response.raise_for_status()
        return response.json().get('items', [])
    except requests.RequestException:
        return []

def discover_and_save_projects(github_token):
    if not github_token or github_token == '111112222233333':
        print(" [Scheduler] 错误: GitHub Token 未设置或无效. 跳过本次抓取。")
        return 0

    session = requests.Session()
    session.headers.update({'Authorization': f'token {github_token}'})
    
    conn = sqlite3.connect(DB_FILE)
    cursor = conn.cursor()
    all_repos = get_new_hot_repos(session)
    if not all_repos:
        print(" [Scheduler] API 未返回任何匹配条件的项目。")
        conn.close()
        return 0

    print(f" [Scheduler] API 返回 {len(all_repos)} 个项目. 开始过滤...")
    
    new_projects_count = 0
    for repo in all_repos:
        repo_id, repo_full_name = repo['id'], repo['full_name']
        cursor.execute("SELECT 1 FROM projects WHERE repo_id = ?", (repo_id,))
        if cursor.fetchone(): continue

        print(f"\n [Scheduler] 处理 [{repo_full_name}]...")
        description = repo.get('description', '')
        
        is_desc_chinese = has_chinese(description)
        # V6.0: 传入 session
        readme_excerpt = get_readme_content_with_excerpt(repo_full_name, session) if not is_desc_chinese else None

        if not is_desc_chinese and not readme_excerpt:
            print(f"  -> 丢弃 '{repo_full_name}' (无中文内容).")
            continue
        
        owner_avatar_url = repo['owner']['avatar_url'] if repo.get('owner') else None
        topics_list = repo.get('topics', [])
        topics_json = json.dumps(topics_list) 

        cursor.execute('''
            INSERT INTO projects (
                repo_id, name, url, description, readme_excerpt, language, 
                stars, pushed_at, owner_avatar_url, topics
            )
            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
        ''', (
            repo['id'], repo['name'], repo['html_url'], description, readme_excerpt,
            repo.get('language', 'N/A'), repo['stargazers_count'], 
            repo['pushed_at'], owner_avatar_url, topics_json
        ))
        conn.commit()
        new_projects_count += 1
        print(f"  💾 已保存 '{repo_full_name}' 到数据库.")
    
    print(f"\n [Scheduler] 本轮抓取完成. 新增 {new_projects_count} 个项目.")
    conn.close()
    return new_projects_count

# ★★★ V6.0: 移除了 main() 和 if __name__ == '__main__' ★★★