Handle Audiences new message list previews

app/modules/indexer/parser/nexus_audiences.py

@@ -24,6 +24,7 @@ class NexusAudiencesSiteUserInfo(NexusPhpSiteUserInfo):
         self._sys_mail_unread_page = None
         self.__next_mail_page = 1
         self.__seen_unread_message_links = set()
+        self.__message_list_previews = {}
 
     def _parse_message_unread(self, html_text):
         """
@@ -64,10 +65,8 @@ class NexusAudiencesSiteUserInfo(NexusPhpSiteUserInfo):
             if not StringUtils.is_valid_html_element(html):
                 return None
 
-            message_links = html.xpath(
+            message_links = self.__parse_table_unread_message_links(html)
-                '//tr[.//img[contains(concat(" ", normalize-space(@class), " "), " unreadpm ") '
+            message_links.extend(self.__parse_pm_item_unread_message_links(html))
-                'or @alt="Unread" or @title="未读"]]/td/a[contains(@href, "viewmessage")]/@href'
-            )
             new_message_links = self.__filter_new_message_links(message_links)
             if message_links and not new_message_links:
                 logger.warn(f"{self._site_name} 未读消息页只发现重复消息链接，停止后续翻页")
@@ -81,6 +80,30 @@ class NexusAudiencesSiteUserInfo(NexusPhpSiteUserInfo):
 
         return next_page
 
+    def _parse_message_content(self, html_text):
+        """
+        解析 Audiences 新版短消息详情页。
+        """
+        html = etree.HTML(html_text)
+        try:
+            if StringUtils.is_valid_html_element(html):
+                head = self.__extract_first_text(
+                    html,
+                    '//*[contains(concat(" ", normalize-space(@class), " "), " pm-hero__title ")]'
+                )
+                date = self.__extract_pm_view_meta(html, "日期")
+                content = self.__extract_first_text(
+                    html,
+                    '//*[contains(concat(" ", normalize-space(@class), " "), " pm-view__body ")]'
+                )
+                if not self.__is_empty_message_content(head, date, content):
+                    return head, date, content
+        finally:
+            if html is not None:
+                del html
+
+        return super()._parse_message_content(html_text)
+
     def _pase_unread_msgs(self):
         """
         解析 Audiences 未读消息，避免异常分页重复通知和空详情通知。
@@ -110,6 +133,7 @@ class NexusAudiencesSiteUserInfo(NexusPhpSiteUserInfo):
                     headers=self._mail_content_headers
                 )
             )
+            head, date, content = self.__fill_empty_message_content_from_list(msg_link, head, date, content)
             logger.debug(f"{self._site_name} 标题 {head} 时间 {date} 内容 {content}")
             if self.__is_empty_message_content(head, date, content):
                 logger.warn(f"{self._site_name} 信息链接 {msg_link} 解析结果为空，跳过消息通知")
@@ -137,6 +161,7 @@ class NexusAudiencesSiteUserInfo(NexusPhpSiteUserInfo):
         """
         self.__next_mail_page = 1
         self.__seen_unread_message_links.clear()
+        self.__message_list_previews.clear()
 
     def __filter_new_message_links(self, message_links: list) -> list:
         """
@@ -151,6 +176,118 @@ class NexusAudiencesSiteUserInfo(NexusPhpSiteUserInfo):
             new_message_links.append(message_link)
         return new_message_links
 
+    @staticmethod
+    def __parse_table_unread_message_links(html) -> list:
+        """
+        解析 Audiences 旧版表格消息列表中的未读消息链接。
+        """
+        return html.xpath(
+            '//tr[.//img[contains(concat(" ", normalize-space(@class), " "), " unreadpm ") '
+            'or @alt="Unread" or @title="未读"]]/td/a[contains(@href, "viewmessage")]/@href'
+        )
+
+    def __parse_pm_item_unread_message_links(self, html) -> list:
+        """
+        解析 Audiences 新版 pm-item 私信列表中的未读消息链接。
+        """
+        message_links = []
+        unread_rows = html.xpath(
+            '//*[contains(concat(" ", normalize-space(@class), " "), " pm-item-row ") '
+            'and contains(concat(" ", normalize-space(@class), " "), " is-unread ")]'
+        )
+        if not unread_rows:
+            unread_rows = html.xpath(
+                '//*[contains(concat(" ", normalize-space(@class), " "), " pm-item-row ") '
+                'and .//*[contains(concat(" ", normalize-space(@class), " "), " pm-item__status--unread ") '
+                'or @title="未读"]]'
+            )
+
+        for row in unread_rows:
+            row_links = row.xpath('.//a[contains(@href, "viewmessage")]/@href')
+            if not row_links:
+                continue
+            message_link = row_links[0].strip()
+            if not message_link:
+                continue
+            message_links.append(message_link)
+            self.__cache_pm_item_preview(message_link, row)
+        return message_links
+
+    def __cache_pm_item_preview(self, message_link: str, row):
+        """
+        缓存新版列表页预览，用于详情页结构变化时兜底生成站点消息。
+        """
+        head = self.__extract_pm_item_text(
+            row,
+            './/*[contains(concat(" ", normalize-space(@class), " "), " pm-item__subject ")]'
+        )
+        date = self.__extract_pm_item_text(
+            row,
+            './/*[contains(concat(" ", normalize-space(@class), " "), " pm-item__time ")]'
+        )
+        content = self.__extract_pm_item_text(
+            row,
+            './/*[contains(concat(" ", normalize-space(@class), " "), " pm-item__preview ")]'
+        )
+        self.__message_list_previews[urljoin(self._base_url, message_link)] = (head, date, content)
+
+    @staticmethod
+    def __extract_pm_item_text(row, xpath: str):
+        """
+        提取新版私信列表节点文本并规整空白字符。
+        """
+        nodes = row.xpath(xpath)
+        if not nodes:
+            return None
+        text = nodes[0].xpath("string(.)")
+        text = re.sub(r"\s+", " ", text.replace("\xa0", " ")).strip()
+        return text or None
+
+    @staticmethod
+    def __extract_first_text(html, xpath: str):
+        """
+        提取第一个匹配节点的规整文本。
+        """
+        nodes = html.xpath(xpath)
+        if not nodes:
+            return None
+        return NexusAudiencesSiteUserInfo.__normalize_text(nodes[0].xpath("string(.)"))
+
+    @staticmethod
+    def __extract_pm_view_meta(html, label: str):
+        """
+        按标签提取 Audiences 新版短消息详情页中的元信息。
+        """
+        values = html.xpath(
+            '//*[contains(concat(" ", normalize-space(@class), " "), " pm-view__meta ") '
+            f'and .//*[contains(concat(" ", normalize-space(@class), " "), " pm-view__label ") '
+            f'and normalize-space()="{label}"]]'
+            '//*[contains(concat(" ", normalize-space(@class), " "), " pm-view__value ")]'
+        )
+        if not values:
+            return None
+        return NexusAudiencesSiteUserInfo.__normalize_text(values[0].xpath("string(.)"))
+
+    @staticmethod
+    def __normalize_text(text: str):
+        """
+        规整 Audiences 新版消息页文本空白字符。
+        """
+        if not text:
+            return None
+        text = re.sub(r"\s+", " ", text.replace("\xa0", " ")).strip()
+        return text or None
+
+    def __fill_empty_message_content_from_list(self, msg_link: str, head, date, content):
+        """
+        使用列表页预览填补详情页解析不到的字段。
+        """
+        preview = self.__message_list_previews.get(urljoin(self._base_url, msg_link))
+        if not preview:
+            return head, date, content
+        preview_head, preview_date, preview_content = preview
+        return head or preview_head, date or preview_date, content or preview_content
+
     def __should_fetch_next_unread_page(self, new_message_links: list) -> bool:
         """
         判断是否还需要继续请求 Audiences 下一页未读消息列表。

tests/test_nexus_audiences_parser.py

@@ -269,6 +269,129 @@ def test_audiences_readpm_row_is_not_unread_message():
     assert msg_links == []
 
 
+def test_audiences_pm_item_unread_links_use_list_preview_when_detail_empty():
+    """
+    Audiences 新版 div 私信列表应能识别未读行，并在详情页不可解析时使用列表预览通知。
+    """
+    parser = NexusAudiencesSiteUserInfo(
+        site_name="Audiences",
+        url="https://audiences.me/",
+        site_cookie="",
+        apikey=None,
+        token=None,
+    )
+    parser.message_unread = 7
+    unread_rows = "".join(
+        f"""
+        <div class="pm-item-row is-unread">
+          <div class="pm-item">
+            <input class="pm-item__check" type="checkbox" name="messages[]" value="{4495900 + index}">
+            <span class="pm-item__status pm-item__status--unread" title="未读"></span>
+            <a class="pm-item__subject"
+               href="messages.php?action=viewmessage&amp;id={4495900 + index}">种子被删除</a>
+            <span class="pm-item__user"><i class="fas fa-user" aria-hidden="true"></i>系统</span>
+            <span class="pm-item__time">2026-06-22 22:32:11</span>
+          </div>
+          <div class="pm-item__preview">
+            你下载的种子'Wonder Wall S01E{index:02d} 2026 1080p WEB-DL H265 AAC-ADWeb'被管理员删除。
+          </div>
+        </div>
+        """
+        for index in range(1, 8)
+    )
+    list_html = f"""
+    <html>
+      <body>
+        <form action="messages.php" method="post">
+          <div class="pm-list">
+            {unread_rows}
+            <div class="pm-item-row">
+              <div class="pm-item">
+                <span class="pm-item__status" title="已读"></span>
+                <a class="pm-item__subject"
+                   href="messages.php?action=viewmessage&amp;id=4419171">已读消息</a>
+                <span class="pm-item__time">2026-06-07 14:27:45</span>
+              </div>
+            </div>
+          </div>
+        </form>
+      </body>
+    </html>
+    """
+    requested_urls = []
+
+    def fake_get_page_content(url, params=None, headers=None):
+        """
+        模拟新版列表页可读，但详情页结构暂不兼容导致解析为空。
+        """
+        requested_urls.append(url)
+        return "<html></html>" if "viewmessage" in url else list_html
+
+    parser._get_page_content = fake_get_page_content
+
+    parser._pase_unread_msgs()
+
+    detail_requests = [url for url in requested_urls if "viewmessage" in url]
+    assert len(detail_requests) == 7
+    assert len(parser.message_unread_contents) == 7
+    assert parser.message_unread_contents[0] == (
+        "种子被删除",
+        "2026-06-22 22:32:11",
+        "你下载的种子'Wonder Wall S01E01 2026 1080p WEB-DL H265 AAC-ADWeb'被管理员删除。",
+    )
+    assert "已读消息" not in [item[0] for item in parser.message_unread_contents]
+
+
+def test_audiences_pm_view_message_content_is_parsed():
+    """
+    Audiences 新版短消息详情页应解析 pm-view 中的标题、日期和正文。
+    """
+    parser = NexusAudiencesSiteUserInfo(
+        site_name="Audiences",
+        url="https://audiences.me/",
+        site_cookie="",
+        apikey=None,
+        token=None,
+    )
+    html_text = """
+    <html>
+      <body>
+        <td class="embedded">
+          <div class="pm-page">
+            <div class="pm-hero">
+              <div class="pm-hero__text">
+                <h1 class="pm-hero__title">种子被删除</h1>
+                <p class="pm-hero__sub">自 系统</p>
+              </div>
+            </div>
+            <article class="pm-view">
+              <header class="pm-view__head">
+                <div class="pm-view__meta">
+                  <span class="pm-view__label">自</span>
+                  <span class="pm-view__value">系统</span>
+                </div>
+                <div class="pm-view__meta">
+                  <span class="pm-view__label">日期</span>
+                  <span class="pm-view__value">2026-06-22 22:32:11 </span>
+                </div>
+              </header>
+              <div class="pm-view__body">
+                你下载的种子'Wonder Wall S01E20 2026 1080p WEB-DL H265 AAC-ADWeb'被管理员删除。原因：已完结剧集，清理单集。
+              </div>
+            </article>
+          </div>
+        </td>
+      </body>
+    </html>
+    """
+
+    head, date, content = parser._parse_message_content(html_text)
+
+    assert head == "种子被删除"
+    assert date == "2026-06-22 22:32:11"
+    assert content == "你下载的种子'Wonder Wall S01E20 2026 1080p WEB-DL H265 AAC-ADWeb'被管理员删除。原因：已完结剧集，清理单集。"
+
+
 def test_audiences_unread_mailbox_only_uses_user_box():
     """
     Audiences 只使用用户消息箱，首页不传 page，page=1 实际表示第二页。