fix(audiences): correctly filter unread messages by icon class and attributes

jxxghp · jxxghp · commit 6685bd0e0ecc · 2026-05-17T21:52:13.000+08:00
diff --git a/app/modules/indexer/parser/nexus_audiences.py b/app/modules/indexer/parser/nexus_audiences.py
@@ -42,6 +42,30 @@ def _parse_message_unread(self, html_text):
             if html is not None:
                 del html
 
+    def _parse_message_unread_links(self, html_text: str, msg_links: list):
+        """
+        解析 Audiences 未读消息链接。
+        """
+        html = etree.HTML(html_text)
+        try:
+            if not StringUtils.is_valid_html_element(html):
+                return None
+
+            message_links = html.xpath(
+                '//tr[.//img[contains(concat(" ", normalize-space(@class), " "), " unreadpm ") '
+                'or @alt="Unread" or @title="未读"]]/td/a[contains(@href, "viewmessage")]/@href'
+            )
+            msg_links.extend(message_links)
+            next_page = None
+            next_page_text = html.xpath('//a[contains(.//text(), "下一页") or contains(.//text(), "下一頁")]/@href')
+            if next_page_text:
+                next_page = next_page_text[-1].strip()
+        finally:
+            if html is not None:
+                del html
+
+        return next_page
+
     def _parse_user_traffic_info(self, html_text):
         """
         解析用户流量信息
diff --git a/tests/test_nexus_audiences_parser.py b/tests/test_nexus_audiences_parser.py
@@ -120,6 +120,26 @@ def test_audiences_table_unread_links_ignore_content_rows():
             <td class="rowfollow" nowrap=""><span title="2026-05-07 23:01:58">8天17时前</span></td>
             <td class="rowfollow"><input class="checkbox" type="checkbox" name="messages[]" value="4318000"></td>
           </tr>
+          <tr>
+            <td class="rowfollow" align="center">
+              <img class="readpm" src="pic/trans.gif" title="已读">
+            </td>
+            <td class="rowfollow" align="left">
+              <a href="messages.php?action=viewmessage&amp;id=4317999">无英文 alt 的已读消息</a>
+            </td>
+            <td class="rowfollow" align="left">系统</td>
+            <td class="rowfollow" nowrap=""><span title="2026-05-07 23:01:58">8天17时前</span></td>
+            <td class="rowfollow"><input class="checkbox" type="checkbox" name="messages[]" value="4317999"></td>
+          </tr>
+          <tr>
+            <td class="rowfollow" align="center"></td>
+            <td class="rowfollow" align="left">
+              <a href="messages.php?action=viewmessage&amp;id=4317998">无状态图标消息</a>
+            </td>
+            <td class="rowfollow" align="left">系统</td>
+            <td class="rowfollow" nowrap=""><span title="2026-05-07 23:01:58">8天17时前</span></td>
+            <td class="rowfollow"><input class="checkbox" type="checkbox" name="messages[]" value="4317998"></td>
+          </tr>
         </table>
       </body>
     </html>
@@ -130,3 +150,34 @@ def test_audiences_table_unread_links_ignore_content_rows():
 
     assert msg_links == ["messages.php?action=viewmessage&id=4318225"]
     assert next_page is None
+
+
+def test_audiences_readpm_row_is_not_unread_message():
+    parser = NexusAudiencesSiteUserInfo(
+        site_name="Audiences",
+        url="https://audiences.me/",
+        site_cookie="",
+        apikey=None,
+        token=None,
+    )
+    html_text = """
+    <html>
+      <body>
+        <table>
+          <tr>
+            <td class="rowfollow" align="center">
+              <img class="readpm" src="pic/trans.gif" alt="Read" title="已读">
+            </td>
+            <td class="rowfollow" align="left">
+              <a href="messages.php?action=viewmessage&amp;id=4318000">已读消息</a>
+            </td>
+          </tr>
+        </table>
+      </body>
+    </html>
+    """
+    msg_links = []
+
+    parser._parse_message_unread_links(html_text, msg_links)
+
+    assert msg_links == []
