fix(msteams): cap $top at 50 and paginate to fill list window (#70127)

gaurav0107 · gaurav0107 · commit 98d473152f1f · 2026-04-22T23:32:42.000+05:30
diff --git a/extensions/msteams/src/graph-messages.search.test.ts b/extensions/msteams/src/graph-messages.search.test.ts
@@ -115,7 +115,7 @@ describe("searchMessagesMSTeams", () => {
     expect(call.headers).toBeUndefined();
   });
 
-  it("requests a wider list window than the limit to allow local filtering", async () => {
+  it("requests Graph's max page size ($top=50) regardless of limit", async () => {
     mockState.fetchGraphJson.mockResolvedValue({ value: [] });
 
     await searchMessagesMSTeams({
@@ -126,23 +126,61 @@ describe("searchMessagesMSTeams", () => {
     });
 
     const calledPath = mockState.fetchGraphJson.mock.calls[0][0].path as string;
-    // list window = clamp(limit*10, 50, 200) = 50 for limit=5
+    // Graph caps $top at 50 on chat/channel message endpoints; wider windows
+    // must come from @odata.nextLink pagination, not a larger $top.
     expect(calledPath).toContain("$top=50");
+    expect(calledPath).not.toContain("$top=200");
   });
 
-  it("caps the list window at 200 even for the max limit", async () => {
-    mockState.fetchGraphJson.mockResolvedValue({ value: [] });
+  it("pages through @odata.nextLink until the list window is reached", async () => {
+    const page1Value = Array.from({ length: 50 }, (_, i) => ({
+      id: `msg-${i}`,
+      body: { content: `no match ${i}` },
+    }));
+    const page2Value = Array.from({ length: 50 }, (_, i) => ({
+      id: `msg-${50 + i}`,
+      body: { content: i === 5 ? "match needle here" : `no match ${50 + i}` },
+    }));
+    mockState.fetchGraphJson.mockResolvedValue({
+      value: page1Value,
+      "@odata.nextLink": "https://graph.microsoft.com/v1.0/chats/next-page",
+    });
+    mockState.fetchGraphAbsoluteUrl.mockResolvedValue({ value: page2Value });
+
+    const result = await searchMessagesMSTeams({
+      cfg: {} as OpenClawConfig,
+      to: CHAT_ID,
+      query: "needle",
+      limit: 50,
+    });
+
+    expect(mockState.fetchGraphAbsoluteUrl).toHaveBeenCalledTimes(1);
+    expect(mockState.fetchGraphAbsoluteUrl.mock.calls[0][0].url).toBe(
+      "https://graph.microsoft.com/v1.0/chats/next-page",
+    );
+    expect(result.messages).toHaveLength(1);
+    expect(result.messages[0].id).toBe("msg-55");
+  });
+
+  it("stops paginating once the list window is filled", async () => {
+    // With limit=5, list window = 50; page 1 already returns 50 items, so
+    // we should NOT follow @odata.nextLink.
+    mockState.fetchGraphJson.mockResolvedValue({
+      value: Array.from({ length: 50 }, (_, i) => ({
+        id: `msg-${i}`,
+        body: { content: `ping ${i}` },
+      })),
+      "@odata.nextLink": "https://graph.microsoft.com/v1.0/chats/next-page",
+    });
 
     await searchMessagesMSTeams({
       cfg: {} as OpenClawConfig,
       to: CHAT_ID,
-      query: "test",
-      limit: 100,
+      query: "ping",
+      limit: 5,
     });
 
-    const calledPath = mockState.fetchGraphJson.mock.calls[0][0].path as string;
-    // limit clamps to 50, list window = min(200, 50*10) = 200
-    expect(calledPath).toContain("$top=200");
+    expect(mockState.fetchGraphAbsoluteUrl).not.toHaveBeenCalled();
   });
 
   it("caps returned matches at the effective limit after local filtering", async () => {
diff --git a/extensions/msteams/src/graph-messages.ts b/extensions/msteams/src/graph-messages.ts
@@ -2,7 +2,6 @@ import type { OpenClawConfig } from "../runtime-api.js";
 import { createMSTeamsConversationStoreFs } from "./conversation-store-fs.js";
 import { stripHtmlFromTeamsMessage } from "./graph-thread.js";
 import {
-  type GraphResponse,
   deleteGraphRequest,
   escapeOData,
   fetchGraphAbsoluteUrl,
@@ -39,6 +38,11 @@ type GraphPinnedMessagesResponse = {
   "@odata.nextLink"?: string;
 };
 
+type GraphPagedMessagesResponse = {
+  value?: GraphMessage[];
+  "@odata.nextLink"?: string;
+};
+
 /**
  * Resolve the Graph API path prefix for a conversation.
  * If `to` contains "/" it's a `teamId/channelId` (channel path),
@@ -486,6 +490,8 @@ export type SearchMessagesMSTeamsResult = {
 
 const SEARCH_DEFAULT_LIMIT = 25;
 const SEARCH_MAX_LIMIT = 50;
+/** Graph caps `$top` at 50 on chat/channel message endpoints. */
+const SEARCH_GRAPH_PAGE_SIZE = 50;
 const SEARCH_LIST_WINDOW_MIN = 50;
 const SEARCH_LIST_WINDOW_MAX = 200;
 const SEARCH_LIST_WINDOW_MULTIPLIER = 10;
@@ -500,6 +506,10 @@ const SEARCH_LIST_WINDOW_MULTIPLIER = 10;
  * configuration at the cost of only searching the recent window; full-archive
  * search requires Delegated auth and is out of scope here.
  *
+ * The list window is retrieved in pages of 50 (Graph's `$top` max for these
+ * endpoints) via `@odata.nextLink` until the configured window size is reached
+ * or Graph runs out of messages.
+ *
  * Sender filtering still pushes down to Graph via `$filter` when provided,
  * since `$filter` is supported for app-only on this endpoint.
  */
@@ -526,25 +536,42 @@ export async function searchMessagesMSTeams(
 
   // Build query string manually (not URLSearchParams) to preserve literal $
   // in OData parameter names, consistent with other Graph calls in this module.
-  const parts = [`$top=${listWindow}`];
+  const parts = [`$top=${SEARCH_GRAPH_PAGE_SIZE}`];
   if (params.from) {
     parts.push(
       `$filter=${encodeURIComponent(`from/user/displayName eq '${escapeOData(params.from)}'`)}`,
     );
   }
 
   const path = `${basePath}/messages?${parts.join("&")}`;
-  const res = await fetchGraphJson<GraphResponse<GraphMessage>>({ token, path });
+  const maxPages = Math.ceil(listWindow / SEARCH_GRAPH_PAGE_SIZE);
+  const fetched: GraphMessage[] = [];
+  let page = await fetchGraphJson<GraphPagedMessagesResponse>({ token, path });
+  for (let i = 1; i <= maxPages; i++) {
+    for (const msg of page.value ?? []) {
+      fetched.push(msg);
+      if (fetched.length >= listWindow) {
+        break;
+      }
+    }
+    const nextLink = page["@odata.nextLink"];
+    if (!nextLink || fetched.length >= listWindow || i >= maxPages) {
+      break;
+    }
+    page = await fetchGraphAbsoluteUrl<GraphPagedMessagesResponse>({ token, url: nextLink });
+  }
 
+  // Note: Graph's `contentType` is "html" or "text". The production docs list
+  // "text" as the default, but real responses often omit the field. Treat any
+  // non-"text" body as HTML so queries always match rendered text rather than
+  // raw markup (e.g. "bold" must not match "<b>old"). This intentionally
+  // differs from `stripHtmlFromTeamsMessage`'s caller in graph-thread.ts,
+  // which is display-facing and prefers the documented default.
   const matches =
     needle.length === 0
-      ? (res.value ?? [])
-      : (res.value ?? []).filter((msg) => {
+      ? fetched
+      : fetched.filter((msg) => {
           const raw = msg.body?.content ?? "";
-          // Teams bodies default to HTML — strip tags so queries match the
-          // rendered text rather than raw markup (e.g. "bold" would otherwise
-          // match "<b>old"). Only skip stripping for explicit plain-text bodies;
-          // missing contentType falls through to the HTML-safe path.
           const text = msg.body?.contentType === "text" ? raw : stripHtmlFromTeamsMessage(raw);
           return text.toLowerCase().includes(needle);
         });
