xiaoli-cli/plugins/ai_search.py

import json
import time
from urllib.parse import quote
import requests


class Liugin:
    """网页内容工具 — jina.ai 代理 + AI 直连"""

    _UA = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36'
    _HEADERS = {
        'User-Agent': _UA,
        'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
        'Accept-Language': 'zh-CN,zh;q=0.9,en;q=0.8',
    }

    def __init__(self):
        self.usage = """网页内容工具

【jina.ai 代理组】— 通过 jina.ai 服务，返回干净文本:
  ai_url <网址>      - jina 提取网页正文
  ai_search <搜索词> - jina 网络搜索
  ai_web <搜索词>    - jina 网络搜索（同 ai_search）

【AI 直连组】— 直接 HTTP 请求，模型自行处理:
  url <网址>         - 直接获取网页原始内容
  search <搜索词>    - 直接搜索（模型自行选择搜索引擎）
  web <搜索词>       - 同 search

JSON 格式示例:
  {"action": "use_tool", "tool": "ai_search", "args": "ai_url https://www.example.com"}
  {"action": "use_tool", "tool": "ai_search", "args": "ai_search Python教程"}
  {"action": "use_tool", "tool": "ai_search", "args": "url https://api.example.com/data"}

⚠️ 使用原则:
  - 优先使用 AI 直连组 (url/search/web)，自己获取和处理网页内容
  - 仅当直连失败、网页需要 JS 渲染、或需要 jina 的智能提取时才用 jina 代理组
  - 不要滥用 jina，它有速率限制
"""
        self.cli = None
        self.timeout = 30
        self.max_retries = 2

    def set_cli(self, cli):
        self.cli = cli
        self.cli.register_liugin_command('ai_search', self.command_handler)

    def command_handler(self, args):
        parts = args.strip().split(maxsplit=1)
        if len(parts) < 1:
            return "请提供操作: ai_url/ai_search/ai_web/url/search/web <参数>"
        operation = parts[0].lower()
        remaining_args = parts[1] if len(parts) > 1 else ""
        return self.handle(f"{operation} {remaining_args}")

    def get_tool_info(self):
        return {
            "name": "ai_search",
            "description": "网页内容工具。ai_url/ai_search/ai_web 用 jina.ai 代理；url/search/web 直接 HTTP 请求让模型自行处理。优先用直连组，jina 仅作备用。",
            "keywords": ["网址", "url", "网页", "web", "search", "搜索", "fetch", "提取", "jina",
                         "ai_url", "ai_search", "ai_web", "内容", "获取", "联网"],
            "usage": self.usage
        }

    def handle(self, args):
        try:
            parts = args.strip().split(maxsplit=1)
            if len(parts) < 1:
                return "错误：参数不足。可用操作: ai_url/ai_search/ai_web/url/search/web <参数>"

            operation = parts[0].lower()
            param = parts[1].strip() if len(parts) > 1 else ""

            if not param:
                return f"错误：请提供参数。格式: {operation} <网址或搜索词>"

            # ── jina.ai 代理组 ──
            if operation == "ai_url":
                return self._jina_extract(param)

            elif operation == "ai_search":
                return self._jina_search(param)

            elif operation == "ai_web":
                return self._jina_search(param)

            # ── AI 直连组 ──
            elif operation == "url":
                return self._fetch_raw(param)

            elif operation == "search":
                return self._direct_search(param)

            elif operation == "web":
                return self._direct_search(param)

            else:
                return f"不支持的操作 '{operation}'。可用: ai_url/ai_search/ai_web/url/search/web"
        except Exception as e:
            return f"操作错误: {str(e)}"

    # ═══════════════════════════════════════
    #  jina.ai 代理组
    # ═══════════════════════════════════════

    def _jina_extract(self, url):
        """jina.ai 提取网页正文"""
        jina_url = f"https://r.jina.ai/{url}"
        result = self._execute_curl(jina_url)
        if result.startswith("错误："):
            return result
        return f"[jina 提取] 网页内容 (来自 {url}):\n\n{result}"

    def _jina_search(self, query):
        """jina.ai 网络搜索"""
        encoded = quote(query, safe='')
        jina_url = f"https://s.jina.ai/{encoded}"
        result = self._execute_curl(jina_url)
        if result.startswith("错误："):
            return result
        return f"[jina 搜索] 结果 (搜索词: {query}):\n\n{result}"

    # ═══════════════════════════════════════
    #  AI 直连组
    # ═══════════════════════════════════════

    def _fetch_raw(self, url):
        """直接 HTTP GET 获取网页原始内容"""
        if not url.startswith('http://') and not url.startswith('https://'):
            url = 'https://' + url

        try:
            response = requests.get(
                url, timeout=self.timeout, headers=self._HEADERS, allow_redirects=True
            )
            if response.status_code != 200:
                return f"错误：HTTP {response.status_code}"

            if not response.encoding or response.encoding.lower() == 'iso-8859-1':
                response.encoding = response.apparent_encoding or 'utf-8'

            content = response.text.strip()
            if not content:
                return "错误：未获取到内容"

            return f"[直连] 原始网页内容 (来自 {url}):\n\n{content}"
        except requests.exceptions.Timeout:
            return f"错误：请求超时 (>{self.timeout}秒)"
        except requests.exceptions.ConnectionError:
            return "错误：网络连接失败"
        except Exception as e:
            return f"错误：{str(e)}"

    def _direct_search(self, query):
        """直接搜索 — 用搜索引擎获取结果页，模型自行解析"""
        encoded = quote(query, safe='')
        # 用必应搜索，返回结果页 HTML，模型自行提取链接和摘要
        search_url = f"https://www.bing.com/search?q={encoded}"

        try:
            response = requests.get(
                search_url, timeout=self.timeout, headers=self._HEADERS, allow_redirects=True
            )
            if response.status_code != 200:
                return f"错误：HTTP {response.status_code}"

            if not response.encoding or response.encoding.lower() == 'iso-8859-1':
                response.encoding = response.apparent_encoding or 'utf-8'

            content = response.text.strip()
            if not content:
                return "错误：未获取到搜索结果"

            return f"[直连] 搜索结果页 (搜索词: {query}):\n\n{content}"
        except requests.exceptions.Timeout:
            return f"错误：请求超时 (>{self.timeout}秒)"
        except requests.exceptions.ConnectionError:
            return "错误：网络连接失败"
        except Exception as e:
            return f"错误：{str(e)}"

    # ═══════════════════════════════════════
    #  通用请求（jina 用，带重试）
    # ═══════════════════════════════════════

    def _execute_curl(self, url):
        last_error = None
        for attempt in range(self.max_retries + 1):
            try:
                response = requests.get(
                    url, timeout=self.timeout, headers=self._HEADERS, allow_redirects=True
                )
                if response.status_code != 200:
                    return f"错误：HTTP {response.status_code}"

                if not response.encoding or response.encoding.lower() == 'iso-8859-1':
                    response.encoding = response.apparent_encoding or 'utf-8'

                content = response.text.strip()
                if not content:
                    return "错误：未获取到内容"

                return content
            except requests.exceptions.Timeout:
                last_error = f"请求超时 (>{self.timeout}秒)"
                if attempt < self.max_retries:
                    time.sleep(1)
            except requests.exceptions.ConnectionError:
                last_error = "网络连接失败"
                if attempt < self.max_retries:
                    time.sleep(1)
            except requests.exceptions.RequestException as e:
                last_error = str(e)
                break
            except Exception as e:
                last_error = str(e)
                break

        return f"错误：{last_error}"