feat(wenxin): unwrap reference redirects and capture web page sources

Decode Wenxin reference redirect targets so citation URLs land on the
real source host, and broaden the reference/search field patterns to
pick up webPages-style payloads.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

apps/desktop-client/src/main/adapters/wenxin.test.ts

@@ -3,8 +3,10 @@ import { describe, expect, it } from "vitest";
 import { __wenxinTestUtils } from "./wenxin";
 
 const {
+  buildSourceItem,
   htmlTableToMarkdown,
   isObservationComplete,
+  normalizeUrl,
   parseWenxinHistoryResponse,
   parseWenxinSSEBody,
 } = __wenxinTestUtils;
@@ -163,6 +165,62 @@ describe("wenxin adapter helpers", () => {
     expect(history.citations[0]?.url).toBe("https://fallback.example.com/article");
   });
 
+  it("unwraps Wenxin reference redirect urls", () => {
+    expect(normalizeUrl("https://yiyan.baidu.com/eb/link?target=https%3A%2F%2Fwww.example.com%2Farticle%23ref")).toBe(
+      "https://www.example.com/article",
+    );
+
+    expect(buildSourceItem({
+      pageUrl: "https://yiyan.baidu.com/eb/link?url=https%3A%2F%2Fsource.example.com%2Fnews%3Fid%3D1",
+      title: "引用来源标题",
+      source: "来源站点",
+    })).toMatchObject({
+      url: "https://source.example.com/news?id=1",
+      title: "引用来源标题",
+      site_name: "来源站点",
+      host: "source.example.com",
+    });
+  });
+
+  it("extracts web page references from Wenxin reference list fields", () => {
+    const history = parseWenxinHistoryResponse({
+      code: 0,
+      msg: "success",
+      data: {
+        state: 1,
+        chats: [
+          {
+            id: "bot-1",
+            role: "robot",
+            stop: 1,
+            modelSign: "EB45T",
+            message: [
+              {
+                contentType: "text",
+                content: "这是答案",
+              },
+            ],
+            webPages: [
+              {
+                targetUrl: "https://yiyan.baidu.com/eb/link?target=https%3A%2F%2Fwww.reference-source.com%2Fa%23section",
+                title: "参考网页 A",
+                siteName: "reference-source.com",
+              },
+            ],
+          },
+        ],
+      },
+    }, "session-3");
+
+    expect(history.citations).toHaveLength(1);
+    expect(history.searchResults).toHaveLength(1);
+    expect(history.citations[0]).toMatchObject({
+      url: "https://www.reference-source.com/a",
+      title: "参考网页 A",
+      site_name: "reference-source.com",
+    });
+  });
+
   it("does not treat error-only observations as complete", () => {
     expect(isObservationComplete({
       sessionId: "session-1",