multi-platform-media-manage/server/python/platforms/xiaohongshu.py

# -*- coding: utf-8 -*-
"""
小红书视频发布器
参考: matrix/xhs_uploader/main.py
使用 xhs SDK API 方式发布，更稳定
"""

import asyncio
import os
import sys
import time
from pathlib import Path
from typing import List
from .base import (
    BasePublisher, PublishParams, PublishResult,
    WorkItem, WorksResult, CommentItem, CommentsResult
)

# 添加 matrix 项目路径，用于导入签名脚本
MATRIX_PATH = Path(__file__).parent.parent.parent.parent / "matrix"
sys.path.insert(0, str(MATRIX_PATH))

# 尝试导入 xhs SDK
try:
    from xhs import XhsClient
    XHS_SDK_AVAILABLE = True
except ImportError:
    print("[Warning] xhs 库未安装，请运行: pip install xhs")
    XhsClient = None
    XHS_SDK_AVAILABLE = False

# 签名脚本路径
STEALTH_JS_PATH = MATRIX_PATH / "xhs-api" / "js" / "stealth.min.js"


class XiaohongshuPublisher(BasePublisher):
    """
    小红书视频发布器
    优先使用 xhs SDK API 方式发布
    """
    
    platform_name = "xiaohongshu"
    login_url = "https://creator.xiaohongshu.com/"
    publish_url = "https://creator.xiaohongshu.com/publish/publish"
    cookie_domain = ".xiaohongshu.com"
    
    async def get_sign(self, uri: str, data=None, a1: str = "", web_session: str = ""):
        """获取小红书 API 签名"""
        from playwright.async_api import async_playwright
        
        try:
            async with async_playwright() as playwright:
                browser = await playwright.chromium.launch(headless=True)
                browser_context = await browser.new_context()
                
                if STEALTH_JS_PATH.exists():
                    await browser_context.add_init_script(path=str(STEALTH_JS_PATH))
                
                page = await browser_context.new_page()
                await page.goto("https://www.xiaohongshu.com")
                await asyncio.sleep(1)
                await page.reload()
                await asyncio.sleep(1)
                
                if a1:
                    await browser_context.add_cookies([
                        {'name': 'a1', 'value': a1, 'domain': ".xiaohongshu.com", 'path': "/"}
                    ])
                    await page.reload()
                    await asyncio.sleep(0.5)
                
                encrypt_params = await page.evaluate(
                    "([url, data]) => window._webmsxyw(url, data)", 
                    [uri, data]
                )
                
                await browser_context.close()
                await browser.close()
                
                return {
                    "x-s": encrypt_params["X-s"],
                    "x-t": str(encrypt_params["X-t"])
                }
        except Exception as e:
            import traceback
            traceback.print_exc()
            raise Exception(f"签名失败: {e}")
    
    def sign_sync(self, uri, data=None, a1="", web_session=""):
        """
        同步签名函数，供 XhsClient 使用。
        
        注意：发布流程运行在 asyncio 事件循环中（通过 asyncio.run 启动），
        这里如果再调用 asyncio.run 会触发 “asyncio.run() cannot be called from a running event loop”。
        因此改为使用 sync_playwright 的同步实现（参考 matrix/xhs_uploader）。
        """
        try:
            from playwright.sync_api import sync_playwright
        except Exception as e:
            raise Exception(f"缺少 playwright 同步接口支持: {e}")
        
        last_exc: Exception | None = None
        for attempt in range(1, 6):
            try:
                with sync_playwright() as playwright:
                    browser = playwright.chromium.launch(headless=True)
                    context = browser.new_context()
                    
                    if STEALTH_JS_PATH.exists():
                        context.add_init_script(path=str(STEALTH_JS_PATH))
                    
                    page = context.new_page()
                    page.goto("https://www.xiaohongshu.com", wait_until="domcontentloaded", timeout=60000)
                    
                    if a1:
                        context.add_cookies([
                            {'name': 'a1', 'value': a1, 'domain': ".xiaohongshu.com", 'path': "/"}
                        ])
                        page.reload(wait_until="domcontentloaded")
                    
                    # 参考 matrix：设置完 cookie 后需要稍等，否则可能出现 window._webmsxyw 不存在
                    time.sleep(1.5)
                    
                    encrypt_params = page.evaluate(
                        "([url, data]) => window._webmsxyw(url, data)",
                        [uri, data]
                    )
                    
                    context.close()
                    browser.close()
                    
                    return {
                        "x-s": encrypt_params["X-s"],
                        "x-t": str(encrypt_params["X-t"])
                    }
            except Exception as e:
                last_exc = e
                # 轻微退避重试
                time.sleep(0.4 * attempt)
        
        raise Exception(f"签名失败: {last_exc}")
    
    async def publish_via_api(self, cookies: str, params: PublishParams) -> PublishResult:
        """通过 API 发布视频"""
        if not XHS_SDK_AVAILABLE:
            raise Exception("xhs SDK 未安装，请运行: pip install xhs")
        
        self.report_progress(10, "正在通过 API 发布...")
        print(f"[{self.platform_name}] 使用 XHS SDK API 发布...")
        print(f"[{self.platform_name}] 视频路径: {params.video_path}")
        print(f"[{self.platform_name}] 标题: {params.title}")
        
        # 转换 cookie 格式
        cookie_list = self.parse_cookies(cookies)
        cookie_string = self.cookies_to_string(cookie_list) if cookie_list else cookies
        print(f"[{self.platform_name}] Cookie 长度: {len(cookie_string)}")
        
        self.report_progress(20, "正在上传视频...")
        
        # 创建客户端
        xhs_client = XhsClient(cookie_string, sign=self.sign_sync)
        
        print(f"[{self.platform_name}] 开始调用 create_video_note...")
        
        # 发布视频
        try:
            result = xhs_client.create_video_note(
                title=params.title,
                desc=params.description or params.title,
                topics=params.tags or [],
                post_time=params.publish_date.strftime("%Y-%m-%d %H:%M:%S") if params.publish_date else None,
                video_path=params.video_path,
                cover_path=params.cover_path if params.cover_path and os.path.exists(params.cover_path) else None
            )
            print(f"[{self.platform_name}] SDK 返回结果: {result}")
        except Exception as e:
            import traceback
            traceback.print_exc()
            print(f"[{self.platform_name}] SDK 调用失败: {e}")
            raise Exception(f"XHS SDK 发布失败: {e}")
        
        # 验证返回结果
        if not result:
            raise Exception("XHS SDK 返回空结果")
        
        # 检查是否有错误
        if isinstance(result, dict):
            if result.get("code") and result.get("code") != 0:
                raise Exception(f"发布失败: {result.get('msg', '未知错误')}")
            if result.get("success") == False:
                raise Exception(f"发布失败: {result.get('msg', result.get('error', '未知错误'))}")
        
        note_id = result.get("note_id", "") if isinstance(result, dict) else ""
        video_url = result.get("url", "") if isinstance(result, dict) else ""
        
        if not note_id:
            print(f"[{self.platform_name}] 警告: 未获取到 note_id，返回结果: {result}")
        
        self.report_progress(100, "发布成功")
        print(f"[{self.platform_name}] 发布成功! note_id={note_id}, url={video_url}")
        
        return PublishResult(
            success=True,
            platform=self.platform_name,
            video_id=note_id,
            video_url=video_url,
            message="发布成功"
        )
    
    async def publish(self, cookies: str, params: PublishParams) -> PublishResult:
        """发布视频到小红书 - 参考 matrix/xhs_uploader/main.py"""
        print(f"\n{'='*60}")
        print(f"[{self.platform_name}] 开始发布视频")
        print(f"[{self.platform_name}] 视频路径: {params.video_path}")
        print(f"[{self.platform_name}] 标题: {params.title}")
        print(f"[{self.platform_name}] Headless: {self.headless}")
        print(f"[{self.platform_name}] XHS SDK 可用: {XHS_SDK_AVAILABLE}")
        print(f"{'='*60}")
        
        # 检查视频文件
        if not os.path.exists(params.video_path):
            raise Exception(f"视频文件不存在: {params.video_path}")
        
        print(f"[{self.platform_name}] 视频文件存在，大小: {os.path.getsize(params.video_path)} bytes")
        
        self.report_progress(5, "正在准备发布...")
        
        # 参考 matrix: 优先使用 XHS SDK API 方式发布（更稳定）
        if XHS_SDK_AVAILABLE:
            try:
                print(f"[{self.platform_name}] 尝试使用 XHS SDK API 发布...")
                result = await self.publish_via_api(cookies, params)
                print(f"[{self.platform_name}] API 发布完成: success={result.success}")
                
                # 如果 API 返回成功，直接返回
                if result.success:
                    return result
                
                # 如果 API 返回失败但有具体错误，也返回
                if result.error and "请刷新" not in result.error:
                    return result
                    
                # 其他情况尝试 Playwright 方式
                print(f"[{self.platform_name}] API 方式未成功，尝试 Playwright...")
            except Exception as e:
                import traceback
                traceback.print_exc()
                print(f"[{self.platform_name}] API 发布失败: {e}")
                print(f"[{self.platform_name}] 尝试使用 Playwright 方式...")
        
        # 使用 Playwright 方式发布
        print(f"[{self.platform_name}] 使用 Playwright 方式发布...")
        return await self.publish_via_playwright(cookies, params)
    
    async def publish_via_playwright(self, cookies: str, params: PublishParams) -> PublishResult:
        """通过 Playwright 发布视频"""
        self.report_progress(10, "正在初始化浏览器...")
        print(f"[{self.platform_name}] Playwright 方式开始...")
        
        await self.init_browser()
        
        cookie_list = self.parse_cookies(cookies)
        print(f"[{self.platform_name}] 设置 {len(cookie_list)} 个 cookies")
        await self.set_cookies(cookie_list)
        
        if not self.page:
            raise Exception("Page not initialized")
        
        self.report_progress(15, "正在打开发布页面...")
        
        # 直接访问视频发布页面
        publish_url = "https://creator.xiaohongshu.com/publish/publish?source=official"
        print(f"[{self.platform_name}] 打开页面: {publish_url}")
        await self.page.goto(publish_url)
        await asyncio.sleep(3)
        
        current_url = self.page.url
        print(f"[{self.platform_name}] 当前 URL: {current_url}")
        
        # 检查登录状态
        if "login" in current_url or "passport" in current_url:
            screenshot_base64 = await self.capture_screenshot()
            return PublishResult(
                success=False,
                platform=self.platform_name,
                error="登录已过期，请重新登录",
                screenshot_base64=screenshot_base64,
                page_url=current_url,
                status='need_captcha',
                need_captcha=True,
                captcha_type='login'
            )
        
        # 使用 AI 检查验证码
        ai_captcha = await self.ai_check_captcha()
        if ai_captcha['has_captcha']:
            print(f"[{self.platform_name}] AI检测到验证码: {ai_captcha['captcha_type']}", flush=True)
            screenshot_base64 = await self.capture_screenshot()
            return PublishResult(
                success=False,
                platform=self.platform_name,
                error=f"检测到{ai_captcha['captcha_type']}验证码，需要使用有头浏览器完成验证",
                screenshot_base64=screenshot_base64,
                page_url=current_url,
                status='need_captcha',
                need_captcha=True,
                captcha_type=ai_captcha['captcha_type']
            )
        
        self.report_progress(20, "正在上传视频...")
        
        # 等待页面加载
        await asyncio.sleep(2)
        
        # 上传视频
        upload_triggered = False
        
        # 方法1: 直接设置隐藏的 file input
        print(f"[{self.platform_name}] 尝试方法1: 设置 file input")
        file_inputs = self.page.locator('input[type="file"]')
        input_count = await file_inputs.count()
        print(f"[{self.platform_name}] 找到 {input_count} 个 file input")
        
        if input_count > 0:
            # 找到接受视频的 input
            for i in range(input_count):
                input_el = file_inputs.nth(i)
                accept = await input_el.get_attribute('accept') or ''
                print(f"[{self.platform_name}] Input {i} accept: {accept}")
                if 'video' in accept or '*' in accept or not accept:
                    await input_el.set_input_files(params.video_path)
                    upload_triggered = True
                    print(f"[{self.platform_name}] 视频文件已设置到 input {i}")
                    break
        
        # 方法2: 点击上传区域触发文件选择器
        if not upload_triggered:
            print(f"[{self.platform_name}] 尝试方法2: 点击上传区域")
            try:
                upload_area = self.page.locator('[class*="upload-wrapper"], [class*="upload-area"], .upload-input').first
                if await upload_area.count() > 0:
                    async with self.page.expect_file_chooser(timeout=5000) as fc_info:
                        await upload_area.click()
                    file_chooser = await fc_info.value
                    await file_chooser.set_files(params.video_path)
                    upload_triggered = True
                    print(f"[{self.platform_name}] 通过点击上传区域上传成功")
            except Exception as e:
                print(f"[{self.platform_name}] 方法2失败: {e}")
        
        if not upload_triggered:
            screenshot_base64 = await self.capture_screenshot()
            page_url = await self.get_page_url()
            return PublishResult(
                success=False,
                platform=self.platform_name,
                error="无法上传视频文件",
                screenshot_base64=screenshot_base64,
                page_url=page_url,
                status='need_action'
            )
        
        self.report_progress(40, "等待视频上传完成...")
        print(f"[{self.platform_name}] 等待视频上传和处理...")
        
        # 等待上传完成（检测页面变化）
        upload_complete = False
        for i in range(60):  # 最多等待3分钟
            await asyncio.sleep(3)
            
            # 检查是否有标题输入框（上传完成后出现）
            title_input_count = await self.page.locator('input[placeholder*="标题"], input[placeholder*="填写标题"]').count()
            # 或者检查编辑器区域
            editor_count = await self.page.locator('[class*="ql-editor"], [contenteditable="true"]').count()
            # 检查发布按钮是否可见
            publish_btn_count = await self.page.locator('.publishBtn, button:has-text("发布")').count()
            
            print(f"[{self.platform_name}] 检测 {i+1}: 标题框={title_input_count}, 编辑器={editor_count}, 发布按钮={publish_btn_count}")
            
            if title_input_count > 0 or (editor_count > 0 and publish_btn_count > 0):
                upload_complete = True
                print(f"[{self.platform_name}] 视频上传完成!")
                break
        
        if not upload_complete:
            screenshot_base64 = await self.capture_screenshot()
            page_url = await self.get_page_url()
            return PublishResult(
                success=False,
                platform=self.platform_name,
                error="视频上传超时",
                screenshot_base64=screenshot_base64,
                page_url=page_url,
                status='need_action'
            )
        
        await asyncio.sleep(2)
        
        self.report_progress(60, "正在填写笔记信息...")
        print(f"[{self.platform_name}] 填写标题: {params.title[:20]}")
        
        # 填写标题
        title_filled = False
        title_selectors = [
            'input[placeholder*="标题"]',
            'input[placeholder*="填写标题"]',
            '[class*="title"] input',
            '.c-input_inner',
        ]
        for selector in title_selectors:
            title_input = self.page.locator(selector).first
            if await title_input.count() > 0:
                await title_input.click()
                await title_input.fill('')  # 先清空
                await title_input.fill(params.title[:20])
                title_filled = True
                print(f"[{self.platform_name}] 标题已填写，使用选择器: {selector}")
                break
        
        if not title_filled:
            print(f"[{self.platform_name}] 警告: 未找到标题输入框")
        
        # 填写描述和标签
        if params.description or params.tags:
            desc_filled = False
            desc_selectors = [
                '[class*="ql-editor"]',
                '[class*="content-input"] [contenteditable="true"]',
                '[class*="editor"] [contenteditable="true"]',
                '.ql-editor',
            ]
            for selector in desc_selectors:
                desc_input = self.page.locator(selector).first
                if await desc_input.count() > 0:
                    await desc_input.click()
                    await asyncio.sleep(0.5)
                    
                    if params.description:
                        await self.page.keyboard.type(params.description, delay=20)
                        print(f"[{self.platform_name}] 描述已填写")
                    
                    if params.tags:
                        # 添加标签
                        await self.page.keyboard.press("Enter")
                        for tag in params.tags[:5]:  # 最多5个标签
                            await self.page.keyboard.type(f"#{tag}", delay=20)
                            await asyncio.sleep(0.3)
                            await self.page.keyboard.press("Space")
                        print(f"[{self.platform_name}] 标签已填写: {params.tags[:5]}")
                    
                    desc_filled = True
                    break
            
            if not desc_filled:
                print(f"[{self.platform_name}] 警告: 未找到描述输入框")
        
        await asyncio.sleep(2)
        self.report_progress(80, "正在发布...")
        
        await asyncio.sleep(2)
        
        # 滚动到页面底部确保发布按钮可见
        await self.page.evaluate("window.scrollTo(0, document.body.scrollHeight)")
        await asyncio.sleep(1)
        
        print(f"[{self.platform_name}] 查找发布按钮...")
        
        # 点击发布
        publish_selectors = [
            'button.publishBtn',
            '.publishBtn',
            'button.d-button.red',
            'button:has-text("发布"):not(:has-text("定时发布"))',
            '[class*="publish"][class*="btn"]',
        ]
        
        publish_clicked = False
        for selector in publish_selectors:
            try:
                btn = self.page.locator(selector).first
                if await btn.count() > 0:
                    is_visible = await btn.is_visible()
                    is_enabled = await btn.is_enabled()
                    print(f"[{self.platform_name}] 按钮 {selector}: visible={is_visible}, enabled={is_enabled}")
                    
                    if is_visible and is_enabled:
                        box = await btn.bounding_box()
                        if box:
                            print(f"[{self.platform_name}] 点击发布按钮: {selector}, 位置: ({box['x']}, {box['y']})")
                            # 使用真实鼠标点击
                            await self.page.mouse.click(box['x'] + box['width']/2, box['y'] + box['height']/2)
                            publish_clicked = True
                            break
            except Exception as e:
                print(f"[{self.platform_name}] 选择器 {selector} 错误: {e}")
        
        if not publish_clicked:
            # 保存截图用于调试
            screenshot_path = f"debug_publish_failed_{self.platform_name}.png"
            await self.page.screenshot(path=screenshot_path, full_page=True)
            print(f"[{self.platform_name}] 未找到发布按钮，截图保存到: {screenshot_path}")
            
            # 打印页面 HTML 结构用于调试
            buttons = await self.page.query_selector_all('button')
            print(f"[{self.platform_name}] 页面上共有 {len(buttons)} 个按钮")
            for i, btn in enumerate(buttons[:10]):
                text = await btn.text_content() or ''
                cls = await btn.get_attribute('class') or ''
                print(f"  按钮 {i}: text='{text.strip()[:30]}', class='{cls[:50]}'")
            
            raise Exception("未找到发布按钮")
        
        print(f"[{self.platform_name}] 已点击发布按钮，等待发布完成...")
        self.report_progress(90, "等待发布结果...")
        
        # 等待发布完成（检测 URL 变化或成功提示）
        publish_success = False
        for i in range(20):  # 最多等待 20 秒
            await asyncio.sleep(1)
            
            current_url = self.page.url
            
            # 检查是否跳转到发布成功页面或内容管理页面
            if "published=true" in current_url or "success" in current_url or "content" in current_url:
                publish_success = True
                print(f"[{self.platform_name}] 发布成功! 跳转到: {current_url}")
                break
            
            # 检查是否有成功提示
            try:
                success_msg = await self.page.locator('[class*="success"], .toast-success, [class*="Toast"]').first.is_visible()
                if success_msg:
                    publish_success = True
                    print(f"[{self.platform_name}] 检测到成功提示!")
                    break
            except:
                pass
            
            # 检查是否有错误提示
            try:
                error_elements = self.page.locator('[class*="error"], .toast-error, [class*="fail"]')
                if await error_elements.count() > 0:
                    error_text = await error_elements.first.text_content()
                    if error_text and len(error_text.strip()) > 0:
                        raise Exception(f"发布失败: {error_text.strip()}")
            except Exception as e:
                if "发布失败" in str(e):
                    raise
        
        # 如果没有明确的成功标志，返回截图供 AI 分析
        if not publish_success:
            final_url = self.page.url
            print(f"[{self.platform_name}] 发布结果不确定，当前 URL: {final_url}")
            screenshot_base64 = await self.capture_screenshot()
            print(f"[{self.platform_name}] 已获取截图供 AI 分析")
            
            # 如果 URL 还是发布页面，可能需要继续操作
            if "publish/publish" in final_url:
                return PublishResult(
                    success=False,
                    platform=self.platform_name,
                    error="发布结果待确认，请查看截图",
                    screenshot_base64=screenshot_base64,
                    page_url=final_url,
                    status='need_action'
                )
        
        self.report_progress(100, "发布完成")
        print(f"[{self.platform_name}] Playwright 方式发布完成!")
        screenshot_base64 = await self.capture_screenshot()
        page_url = await self.get_page_url()
        
        return PublishResult(
            success=True,
            platform=self.platform_name,
            message="发布完成",
            screenshot_base64=screenshot_base64,
            page_url=page_url,
            status='success'
        )
    
    async def get_works(self, cookies: str, page: int = 0, page_size: int = 20) -> WorksResult:
        """获取小红书作品列表 - 通过监听页面网络响应获取数据"""
        print(f"\n{'='*60}", flush=True)
        print(f"[{self.platform_name}] 获取作品列表", flush=True)
        print(f"[{self.platform_name}] page={page}, page_size={page_size}", flush=True)
        print(f"{'='*60}", flush=True)
        
        works: List[WorkItem] = []
        total = 0
        has_more = False
        captured_data = {}
        
        try:
            await self.init_browser()
            cookie_list = self.parse_cookies(cookies)
            
            # 打印 cookies 信息用于调试
            print(f"[{self.platform_name}] 解析到 {len(cookie_list)} 个 cookies", flush=True)
            
            await self.set_cookies(cookie_list)
            
            if not self.page:
                raise Exception("Page not initialized")
            
            # 定义响应监听器 - 捕获页面自动发起的 API 请求
            async def handle_response(response):
                nonlocal captured_data
                url = response.url
                # 监听作品列表 API
                if 'creator/note/user/posted' in url or 'creator/note_list' in url:
                    try:
                        json_data = await response.json()
                        print(f"[{self.platform_name}] 捕获到 API 响应: {url[:80]}...", flush=True)
                        if json_data.get('success') or json_data.get('code') == 0:
                            captured_data = json_data
                            print(f"[{self.platform_name}] API 响应成功，data keys: {list(json_data.get('data', {}).keys())}", flush=True)
                    except Exception as e:
                        print(f"[{self.platform_name}] 解析响应失败: {e}", flush=True)
            
            # 注册响应监听器
            self.page.on('response', handle_response)
            print(f"[{self.platform_name}] 已注册 API 响应监听器", flush=True)
            
            # 访问笔记管理页面 - 页面会自动发起 API 请求
            print(f"[{self.platform_name}] 访问笔记管理页面...", flush=True)
            
            try:
                await self.page.goto("https://creator.xiaohongshu.com/new/note-manager", wait_until="domcontentloaded", timeout=30000)
            except Exception as nav_error:
                print(f"[{self.platform_name}] 导航超时，但继续尝试: {nav_error}", flush=True)
            
            # 等待 API 响应被捕获
            await asyncio.sleep(5)
            
            # 检查登录状态
            current_url = self.page.url
            print(f"[{self.platform_name}] 当前页面: {current_url}", flush=True)
            if "login" in current_url:
                raise Exception("Cookie 已过期，请重新登录")
            
            # 如果还没有捕获到数据，等待更长时间
            if not captured_data:
                print(f"[{self.platform_name}] 等待 API 响应...", flush=True)
                await asyncio.sleep(5)
            
            # 移除监听器
            self.page.remove_listener('response', handle_response)
            
            # 处理捕获到的数据（增加分页抓取，避免仅第一页）
            def parse_notes(notes_list):
                parsed = []
                for note in notes_list:
                    note_id = note.get('id', '')
                    if not note_id:
                        continue
                    
                    cover_url = ''
                    images_list = note.get('images_list', [])
                    if images_list:
                        cover_url = images_list[0].get('url', '')
                        if cover_url.startswith('http://'):
                            cover_url = cover_url.replace('http://', 'https://')
                    
                    duration = note.get('video_info', {}).get('duration', 0)
                    
                    status = 'published'
                    tab_status = note.get('tab_status', 1)
                    if tab_status == 0:
                        status = 'draft'
                    elif tab_status == 2:
                        status = 'reviewing'
                    elif tab_status == 3:
                        status = 'rejected'
                    
                    parsed.append(WorkItem(
                        work_id=note_id,
                        title=note.get('display_title', '') or '无标题',
                        cover_url=cover_url,
                        duration=duration,
                        status=status,
                        publish_time=note.get('time', ''),
                        play_count=note.get('view_count', 0),
                        like_count=note.get('likes', 0),
                        comment_count=note.get('comments_count', 0),
                        share_count=note.get('shared_count', 0),
                        collect_count=note.get('collected_count', 0),
                    ))
                return parsed

            import json
            if captured_data:
                print(f"[{self.platform_name}] 成功捕获到 API 数据", flush=True)
                data = captured_data.get('data', {})
                notes = data.get('notes', [])
                print(f"[{self.platform_name}] notes 数量: {len(notes)}", flush=True)
                
                # 从 tags 获取总数
                tags = data.get('tags', [])
                for tag in tags:
                    if tag.get('id') == 'special.note_time_desc':
                        total = tag.get('notes_count', 0)
                        break
                
                works.extend(parse_notes(notes))

                # 分页抓取剩余页面：不依赖 data.page（有些情况下会误报 -1），直到拿不到新数据为止
                max_pages = 30
                page_num = 1  # 已经拿了 page=0
                seen_note_ids = set([w.work_id for w in works])
                has_more = True

                while has_more and page_num < max_pages:
                    try:
                        next_resp = await self.page.evaluate(
                            """async (p) => {
                                const res = await fetch(`https://edith.xiaohongshu.com/web_api/sns/v5/creator/note/user/posted?tab=0&page=${p}`, {
                                    method: 'GET',
                                    credentials: 'include',
                                    headers: { 'Accept': 'application/json' }
                                });
                                return await res.json();
                            }""",
                            page_num
                        )
                    except Exception as fetch_err:
                        print(f"[{self.platform_name}] 分页请求异常 page={page_num}: {fetch_err}", flush=True)
                        break

                    if not next_resp:
                        break

                    if not (next_resp.get('success') or next_resp.get('code') == 0):
                        break

                    next_data = next_resp.get('data', {})
                    next_notes = next_data.get('notes', []) or []

                    if not next_notes:
                        has_more = False
                        break

                    parsed_next = parse_notes(next_notes)
                    new_items = [w for w in parsed_next if w.work_id and w.work_id not in seen_note_ids]
                    if not new_items:
                        # 没有新数据，停止
                        has_more = False
                        break

                    for w in new_items:
                        seen_note_ids.add(w.work_id)
                    works.extend(new_items)

                    # 更新总数（若第一页未拿到）
                    if not total and next_data.get('tags'):
                        for tag in next_data.get('tags', []):
                            if tag.get('id') == 'special.note_time_desc':
                                total = tag.get('notes_count', 0)
                                break

                    page_num += 1
                
                # 分页完毕，has_more 表示是否还存在更多（以最后一页标记为准）
                if not has_more:
                    print(f"[{self.platform_name}] 已抓取所有分页，共 {len(works)} 条", flush=True)
            else:
                print(f"[{self.platform_name}] 未能捕获到 API 数据", flush=True)
            
        except Exception as e:
            import traceback
            print(f"[{self.platform_name}] 发生异常: {e}", flush=True)
            traceback.print_exc()
            return WorksResult(
                success=False,
                platform=self.platform_name,
                error=str(e)
            )
        finally:
            # 确保关闭浏览器
            await self.close_browser()
        
        return WorksResult(
            success=True,
            platform=self.platform_name,
            works=works,
            total=total or len(works),
            has_more=has_more
        )
    
    async def get_comments(self, cookies: str, work_id: str, cursor: str = "") -> CommentsResult:
        """获取小红书作品评论 - 通过创作者后台评论管理页面"""
        print(f"\n{'='*60}")
        print(f"[{self.platform_name}] 获取作品评论")
        print(f"[{self.platform_name}] work_id={work_id}, cursor={cursor}")
        print(f"{'='*60}")
        
        comments: List[CommentItem] = []
        total = 0
        has_more = False
        next_cursor = ""
        captured_data = {}
        
        try:
            await self.init_browser()
            cookie_list = self.parse_cookies(cookies)
            await self.set_cookies(cookie_list)
            
            if not self.page:
                raise Exception("Page not initialized")
            
            # 设置 API 响应监听器
            async def handle_response(response):
                nonlocal captured_data
                url = response.url
                # 监听评论相关 API - 创作者后台和普通页面的 API
                if '/comment/' in url and ('page' in url or 'list' in url):
                    try:
                        json_data = await response.json()
                        print(f"[{self.platform_name}] 捕获到评论 API: {url[:100]}...", flush=True)
                        if json_data.get('success') or json_data.get('code') == 0:
                            data = json_data.get('data', {})
                            comment_list = data.get('comments') or data.get('list') or []
                            if comment_list:
                                captured_data = json_data
                                print(f"[{self.platform_name}] 评论 API 响应成功，comments={len(comment_list)}", flush=True)
                            else:
                                print(f"[{self.platform_name}] 评论 API 响应成功但无评论", flush=True)
                    except Exception as e:
                        print(f"[{self.platform_name}] 解析评论响应失败: {e}", flush=True)
            
            self.page.on('response', handle_response)
            print(f"[{self.platform_name}] 已注册评论 API 响应监听器", flush=True)
            
            # 访问创作者后台评论管理页面
            comment_url = "https://creator.xiaohongshu.com/creator/comment"
            print(f"[{self.platform_name}] 访问评论管理页面: {comment_url}", flush=True)
            await self.page.goto(comment_url, wait_until="domcontentloaded", timeout=30000)
            await asyncio.sleep(5)
            
            # 检查是否被重定向到登录页
            current_url = self.page.url
            print(f"[{self.platform_name}] 当前页面 URL: {current_url}", flush=True)
            if "login" in current_url:
                raise Exception("Cookie 已过期，请重新登录")
            
            # 等待评论加载
            if not captured_data:
                print(f"[{self.platform_name}] 等待评论 API 响应...", flush=True)
                # 尝试滚动页面触发评论加载
                await self.page.evaluate('window.scrollBy(0, 500)')
                await asyncio.sleep(3)
            
            if not captured_data:
                # 再等待一会，可能评论 API 加载较慢
                print(f"[{self.platform_name}] 继续等待评论加载...", flush=True)
                await asyncio.sleep(5)
            
            # 移除监听器
            self.page.remove_listener('response', handle_response)
            
            # 解析评论数据
            if captured_data:
                data = captured_data.get('data', {})
                comment_list = data.get('comments') or data.get('list') or []
                has_more = data.get('has_more', False)
                next_cursor = data.get('cursor', '')
                
                print(f"[{self.platform_name}] 解析评论: has_more={has_more}, comments={len(comment_list)}", flush=True)
                
                for comment in comment_list:
                    cid = comment.get('id', '')
                    if not cid:
                        continue
                    
                    user_info = comment.get('user_info', {})
                    
                    # 解析子评论
                    replies = []
                    sub_comments = comment.get('sub_comments', []) or []
                    for sub in sub_comments:
                        sub_user = sub.get('user_info', {})
                        replies.append(CommentItem(
                            comment_id=sub.get('id', ''),
                            work_id=work_id,
                            content=sub.get('content', ''),
                            author_id=sub_user.get('user_id', ''),
                            author_name=sub_user.get('nickname', ''),
                            author_avatar=sub_user.get('image', ''),
                            like_count=sub.get('like_count', 0),
                            create_time=sub.get('create_time', ''),
                        ))
                    
                    comments.append(CommentItem(
                        comment_id=cid,
                        work_id=work_id,
                        content=comment.get('content', ''),
                        author_id=user_info.get('user_id', ''),
                        author_name=user_info.get('nickname', ''),
                        author_avatar=user_info.get('image', ''),
                        like_count=comment.get('like_count', 0),
                        reply_count=comment.get('sub_comment_count', 0),
                        create_time=comment.get('create_time', ''),
                        replies=replies,
                    ))
                
                total = len(comments)
                print(f"[{self.platform_name}] 解析到 {total} 条评论", flush=True)
            else:
                print(f"[{self.platform_name}] 未捕获到评论 API 响应", flush=True)
            
        except Exception as e:
            import traceback
            traceback.print_exc()
            return CommentsResult(
                success=False,
                platform=self.platform_name,
                work_id=work_id,
                error=str(e)
            )
        finally:
            await self.close_browser()
        
        result = CommentsResult(
            success=True,
            platform=self.platform_name,
            work_id=work_id,
            comments=comments,
            total=total,
            has_more=has_more
        )
        result.__dict__['cursor'] = next_cursor
        return result
    
    async def get_all_comments(self, cookies: str) -> dict:
        """获取所有作品的评论 - 通过评论管理页面"""
        print(f"\n{'='*60}")
        print(f"[{self.platform_name}] 获取所有作品评论")
        print(f"{'='*60}")
        
        all_work_comments = []
        captured_comments = []
        captured_notes = {}  # note_id -> note_info
        
        try:
            await self.init_browser()
            cookie_list = self.parse_cookies(cookies)
            await self.set_cookies(cookie_list)
            
            if not self.page:
                raise Exception("Page not initialized")
            
            # 设置 API 响应监听器
            async def handle_response(response):
                nonlocal captured_comments, captured_notes
                url = response.url
                try:
                    # 监听评论列表 API - 多种格式
                    if '/comment/' in url and ('page' in url or 'list' in url):
                        json_data = await response.json()
                        print(f"[{self.platform_name}] 捕获到评论 API: {url[:100]}...", flush=True)
                        
                        if json_data.get('success') or json_data.get('code') == 0:
                            data = json_data.get('data', {})
                            comments = data.get('comments', []) or data.get('list', [])
                            
                            # 从 URL 中提取 note_id
                            import re
                            note_id_match = re.search(r'note_id=([^&]+)', url)
                            note_id = note_id_match.group(1) if note_id_match else ''
                            
                            if comments:
                                for comment in comments:
                                    # 添加 note_id 到评论中
                                    if note_id and 'note_id' not in comment:
                                        comment['note_id'] = note_id
                                    captured_comments.append(comment)
                                
                                print(f"[{self.platform_name}] 捕获到 {len(comments)} 条评论 (note_id={note_id})，总计: {len(captured_comments)}", flush=True)
                    
                    # 监听笔记列表 API
                    if '/note/' in url and ('list' in url or 'posted' in url or 'manager' in url):
                        json_data = await response.json()
                        if json_data.get('success') or json_data.get('code') == 0:
                            data = json_data.get('data', {})
                            notes = data.get('notes', []) or data.get('list', [])
                            print(f"[{self.platform_name}] 捕获到笔记列表 API: {len(notes)} 个笔记", flush=True)
                            for note in notes:
                                note_id = note.get('note_id', '') or note.get('id', '')
                                if note_id:
                                    cover_url = ''
                                    cover = note.get('cover', {})
                                    if isinstance(cover, dict):
                                        cover_url = cover.get('url', '') or cover.get('url_default', '')
                                    elif isinstance(cover, str):
                                        cover_url = cover
                                    
                                    captured_notes[note_id] = {
                                        'title': note.get('title', '') or note.get('display_title', ''),
                                        'cover': cover_url,
                                    }
                except Exception as e:
                    print(f"[{self.platform_name}] 解析响应失败: {e}", flush=True)
            
            self.page.on('response', handle_response)
            print(f"[{self.platform_name}] 已注册 API 响应监听器", flush=True)
            
            # 访问评论管理页面
            print(f"[{self.platform_name}] 访问评论管理页面...", flush=True)
            await self.page.goto("https://creator.xiaohongshu.com/creator/comment", wait_until="domcontentloaded", timeout=30000)
            await asyncio.sleep(5)
            
            # 检查登录状态
            current_url = self.page.url
            if "login" in current_url:
                raise Exception("Cookie 已过期，请重新登录")
            
            print(f"[{self.platform_name}] 页面加载完成，当前捕获: {len(captured_comments)} 条评论, {len(captured_notes)} 个笔记", flush=True)
            
            # 滚动加载更多评论
            for i in range(5):
                await self.page.evaluate('window.scrollBy(0, 500)')
                await asyncio.sleep(1)
            
            await asyncio.sleep(3)
            
            # 移除监听器
            self.page.remove_listener('response', handle_response)
            
            print(f"[{self.platform_name}] 最终捕获: {len(captured_comments)} 条评论, {len(captured_notes)} 个笔记", flush=True)
            
            # 按作品分组评论
            work_comments_map = {}  # note_id -> work_comments
            for comment in captured_comments:
                # 获取笔记信息
                note_info = comment.get('note_info', {}) or comment.get('note', {})
                note_id = comment.get('note_id', '') or note_info.get('note_id', '') or note_info.get('id', '')
                
                if not note_id:
                    continue
                
                if note_id not in work_comments_map:
                    saved_note = captured_notes.get(note_id, {})
                    cover_url = ''
                    cover = note_info.get('cover', {})
                    if isinstance(cover, dict):
                        cover_url = cover.get('url', '') or cover.get('url_default', '')
                    elif isinstance(cover, str):
                        cover_url = cover
                    if not cover_url:
                        cover_url = saved_note.get('cover', '')
                    
                    work_comments_map[note_id] = {
                        'work_id': note_id,
                        'title': note_info.get('title', '') or note_info.get('display_title', '') or saved_note.get('title', ''),
                        'cover_url': cover_url,
                        'comments': []
                    }
                
                cid = comment.get('id', '') or comment.get('comment_id', '')
                if not cid:
                    continue
                
                user_info = comment.get('user_info', {}) or comment.get('user', {})
                
                work_comments_map[note_id]['comments'].append({
                    'comment_id': cid,
                    'author_id': user_info.get('user_id', '') or user_info.get('id', ''),
                    'author_name': user_info.get('nickname', '') or user_info.get('name', ''),
                    'author_avatar': user_info.get('image', '') or user_info.get('avatar', ''),
                    'content': comment.get('content', ''),
                    'like_count': comment.get('like_count', 0),
                    'create_time': comment.get('create_time', ''),
                })
            
            all_work_comments = list(work_comments_map.values())
            total_comments = sum(len(w['comments']) for w in all_work_comments)
            print(f"[{self.platform_name}] 获取到 {len(all_work_comments)} 个作品的 {total_comments} 条评论", flush=True)
            
        except Exception as e:
            import traceback
            traceback.print_exc()
            return {
                'success': False,
                'platform': self.platform_name,
                'error': str(e),
                'work_comments': []
            }
        finally:
            await self.close_browser()
        
        return {
            'success': True,
            'platform': self.platform_name,
            'work_comments': all_work_comments,
            'total': len(all_work_comments)
        }