@acedatacloud/skills 2026.714.3 → 2026.715.0

package/skills/xiaohongshu/scripts/vendor/xhs/comment.py

@@ -1,283 +0,0 @@
-"""评论操作，对应 Go xiaohongshu/comment_feed.go。"""
-
-from __future__ import annotations
-
-import logging
-
-from .cdp import Page
-from .feed_detail import _check_end_container, _check_page_accessible, _get_comment_count
-from .human import sleep_random
-from .selectors import (
-    COMMENT_INPUT_FIELD,
-    COMMENT_INPUT_TRIGGER,
-    COMMENT_SUBMIT_BUTTON,
-    PARENT_COMMENT,
-    REPLY_BUTTON,
-)
-from .urls import make_feed_detail_url
-
-logger = logging.getLogger(__name__)
-
-
-def post_comment(
-    page: Page,
-    feed_id: str,
-    xsec_token: str,
-    content: str,
-    xsec_source: str = "pc_feed",
-) -> None:
-    """发表评论到 Feed。
-
-    Args:
-        page: CDP 页面对象。
-        feed_id: Feed ID。
-        xsec_token: xsec_token。
-        content: 评论内容。
-
-    Raises:
-        RuntimeError: 评论失败。
-    """
-    url = make_feed_detail_url(feed_id, xsec_token, xsec_source)
-    logger.info("打开 feed 详情页: %s", url)
-
-    page.navigate(url)
-    page.wait_for_load()
-    page.wait_dom_stable()
-    sleep_random(800, 1500)
-
-    _check_page_accessible(page)
-    initial_matches = _comment_content_matches(page, content)
-
-    # 点击评论输入触发区域
-    if not page.has_element(COMMENT_INPUT_TRIGGER):
-        raise RuntimeError("未找到评论输入框，该帖子可能不支持评论或网页端不可访问")
-
-    page.click_element(COMMENT_INPUT_TRIGGER)
-    sleep_random(400, 800)
-
-    # 输入评论内容（CDP 逐字输入）
-    page.wait_for_element(COMMENT_INPUT_FIELD, timeout=5)
-    page.input_content_editable(COMMENT_INPUT_FIELD, content)
-    sleep_random(600, 1200)
-
-    # 点击提交
-    if not page.has_element(COMMENT_SUBMIT_BUTTON):
-        raise RuntimeError("评论提交按钮不可用")
-    page.click_element(COMMENT_SUBMIT_BUTTON)
-    sleep_random(800, 1500)
-
-    if not _comment_editor_empty(page):
-        raise RuntimeError("评论未确认提交成功")
-    if not _wait_for_comment_content(page, content, initial_matches):
-        raise RuntimeError("评论未在页面中出现")
-
-    logger.info("评论发送成功: feed=%s", feed_id)
-
-
-def reply_comment(
-    page: Page,
-    feed_id: str,
-    xsec_token: str,
-    content: str,
-    comment_id: str = "",
-    user_id: str = "",
-    xsec_source: str = "pc_feed",
-) -> None:
-    """回复指定评论。
-
-    通过 comment_id 或 user_id 定位评论，然后回复。
-
-    Args:
-        page: CDP 页面对象。
-        feed_id: Feed ID。
-        xsec_token: xsec_token。
-        content: 回复内容。
-        comment_id: 评论 ID（优先使用）。
-        user_id: 用户 ID（备选）。
-
-    Raises:
-        RuntimeError: 回复失败。
-    """
-    if not comment_id and not user_id:
-        raise ValueError("comment_id 和 user_id 至少提供一个")
-
-    url = make_feed_detail_url(feed_id, xsec_token, xsec_source)
-    logger.info("打开 feed 详情页进行回复: %s", url)
-
-    page.navigate(url)
-    page.wait_for_load()
-    page.wait_dom_stable()
-    sleep_random(800, 1500)
-
-    _check_page_accessible(page)
-    sleep_random(1500, 2500)
-    initial_matches = _comment_content_matches(page, content)
-
-    # 查找目标评论
-    target_selector = _find_and_scroll_to_comment(page, comment_id, user_id)
-    if not target_selector:
-        raise RuntimeError(f"未找到评论 (commentID: {comment_id}, userID: {user_id})")
-
-    sleep_random(800, 1500)
-
-    # 点击回复按钮
-    reply_selector = f"{target_selector} {REPLY_BUTTON}"
-    page.click_element(reply_selector)
-    sleep_random(800, 1500)
-
-    # 输入回复内容（CDP 逐字输入）
-    page.wait_for_element(COMMENT_INPUT_FIELD, timeout=5)
-    page.input_content_editable(COMMENT_INPUT_FIELD, content)
-    sleep_random(600, 1200)
-
-    # 点击提交
-    if not page.has_element(COMMENT_SUBMIT_BUTTON):
-        raise RuntimeError("回复提交按钮不可用")
-    page.click_element(COMMENT_SUBMIT_BUTTON)
-    sleep_random(1500, 2500)
-
-    if not _comment_editor_empty(page):
-        raise RuntimeError("回复未确认提交成功")
-    if not _wait_for_comment_content(page, content, initial_matches):
-        raise RuntimeError("回复未在页面中出现")
-
-    logger.info("回复评论成功")
-
-
-def _find_and_scroll_to_comment(
-    page: Page,
-    comment_id: str,
-    user_id: str,
-    max_attempts: int = 100,
-) -> str | None:
-    """查找并滚动到目标评论。"""
-    logger.info("开始查找评论 - commentID: %s, userID: %s", comment_id, user_id)
-
-    # 先滚动到评论区
-    page.scroll_element_into_view(".comments-container")
-    sleep_random(800, 1500)
-
-    last_count = 0
-    stagnant = 0
-
-    for attempt in range(max_attempts):
-        if _locate_target_comment(page, comment_id, user_id, attempt):
-            return f"#comment-{comment_id}" if comment_id else '[data-acedata-reply-target="true"]'
-
-        # 检查是否到底
-        if _check_end_container(page):
-            logger.info("已到达评论底部，未找到目标评论")
-            break
-
-        # 停滞检测
-        current_count = _get_comment_count(page)
-        if current_count != last_count:
-            last_count = current_count
-            stagnant = 0
-        else:
-            stagnant += 1
-        if stagnant >= 10:
-            logger.info("评论数量停滞超过10次")
-            break
-
-        # 滚动到最后一条评论
-        if current_count > 0:
-            page.scroll_nth_element_into_view(PARENT_COMMENT, current_count - 1)
-            sleep_random(200, 500)
-
-        # 继续滚动
-        page.evaluate("window.scrollBy(0, window.innerHeight * 0.8)")
-        sleep_random(400, 800)
-
-        if _locate_target_comment(page, comment_id, user_id, attempt):
-            return f"#comment-{comment_id}" if comment_id else '[data-acedata-reply-target="true"]'
-
-        sleep_random(600, 1200)
-
-    return None
-
-
-def _locate_target_comment(page: Page, comment_id: str, user_id: str, attempt: int) -> bool:
-    if comment_id:
-        selector = f"#comment-{comment_id}"
-        if page.has_element(selector):
-            logger.info("通过 commentID 找到评论 (尝试 %d 次)", attempt + 1)
-            page.scroll_element_into_view(selector)
-            return True
-
-    if user_id:
-        found = page.evaluate(
-            f"""
-            (() => {{
-                const userId = {_js_str(user_id)};
-                const els = document.querySelectorAll(
-                    '.parent-comment, .comment-item, .comment'
-                );
-                for (const el of els) {{
-                    if ([...el.querySelectorAll('[data-user-id]')]
-                        .some(node => node.getAttribute('data-user-id') === userId)) {{
-                        document.querySelectorAll('[data-acedata-reply-target]')
-                            .forEach(node => node.removeAttribute('data-acedata-reply-target'));
-                        el.setAttribute('data-acedata-reply-target', 'true');
-                        el.scrollIntoView({{behavior: 'smooth', block: 'center'}});
-                        return true;
-                    }}
-                }}
-                return false;
-            }})()
-            """
-        )
-        if found:
-            logger.info("通过 userID 找到评论 (尝试 %d 次)", attempt + 1)
-            return True
-    return False
-
-
-def _js_str(s: str) -> str:
-    """将 Python 字符串转为 JS 字面量（含引号）。"""
-    import json
-
-    return json.dumps(s)
-
-
-def _comment_editor_empty(page: Page) -> bool:
-    return bool(
-        page.evaluate(
-            f"""
-            (() => {{
-                const editor = document.querySelector({_js_str(COMMENT_INPUT_FIELD)});
-                if (!editor) return true;
-                return !((editor.value || editor.textContent || '').trim());
-            }})()
-            """
-        )
-    )
-
-
-def _comment_content_matches(page: Page, content: str) -> int:
-    return int(
-        page.evaluate(
-            f"""
-            (() => {{
-                const expected = {_js_str(content)};
-                return [...document.querySelectorAll(
-                    '.parent-comment .content, .comment-item .content, .comment .content'
-                )].filter(node => (node.textContent || '').trim() === expected).length;
-            }})()
-            """
-        )
-        or 0
-    )
-
-
-def _wait_for_comment_content(
-    page: Page, content: str, initial_matches: int, timeout: float = 10
-) -> bool:
-    import time
-
-    deadline = time.monotonic() + timeout
-    while time.monotonic() < deadline:
-        if _comment_content_matches(page, content) > initial_matches:
-            return True
-        time.sleep(0.3)
-    return False

package/skills/xiaohongshu/scripts/vendor/xhs/errors.py

@@ -1,92 +0,0 @@
-"""小红书自动化异常体系。"""
-
-
-class XHSError(Exception):
-    """小红书自动化基础异常。"""
-
-
-class NoFeedsError(XHSError):
-    """没有捕获到 feeds 数据。"""
-
-    def __init__(self) -> None:
-        super().__init__("没有捕获到 feeds 数据")
-
-
-class NoFeedDetailError(XHSError):
-    """没有捕获到 feed 详情数据。"""
-
-    def __init__(self) -> None:
-        super().__init__("没有捕获到 feed 详情数据")
-
-
-class NotLoggedInError(XHSError):
-    """未登录。"""
-
-    def __init__(self) -> None:
-        super().__init__("未登录，请先扫码登录")
-
-
-class PageNotAccessibleError(XHSError):
-    """页面不可访问。"""
-
-    def __init__(self, reason: str) -> None:
-        self.reason = reason
-        super().__init__(f"笔记不可访问: {reason}")
-
-
-class UploadTimeoutError(XHSError):
-    """上传超时。"""
-
-
-class PublishError(XHSError):
-    """发布失败。"""
-
-
-class AccountRiskControlError(PublishError):
-    """账号被风控，无法发布。
-
-    XHS 后端业务错误码（HTTP 200 + code≠0）：
-      - -9136: 因违反社区规范禁止发笔记
-    """
-
-    def __init__(self, code: int, msg: str) -> None:
-        self.code = code
-        self.msg = msg
-        super().__init__(f"账号被风控（code={code}）：{msg}")
-
-
-class TitleTooLongError(PublishError):
-    """标题超过长度限制。"""
-
-    def __init__(self, current: str, maximum: str) -> None:
-        self.current = current
-        self.maximum = maximum
-        super().__init__(f"当前输入长度为{current}，最大长度为{maximum}")
-
-
-class ContentTooLongError(PublishError):
-    """正文超过长度限制。"""
-
-    def __init__(self, current: str, maximum: str) -> None:
-        self.current = current
-        self.maximum = maximum
-        super().__init__(f"当前输入长度为{current}，最大长度为{maximum}")
-
-
-class RateLimitError(XHSError):
-    """请求频率过高，验证码获取失败。"""
-
-    def __init__(self) -> None:
-        super().__init__("请求太频繁，验证码获取失败，请重启浏览器后重试")
-
-
-class CDPError(XHSError):
-    """CDP 通信异常。"""
-
-
-class ElementNotFoundError(XHSError):
-    """页面元素未找到。"""
-
-    def __init__(self, selector: str) -> None:
-        self.selector = selector
-        super().__init__(f"未找到元素: {selector}")