Align remote source processed row state

src/components/source-row.test.ts

@@ -49,6 +49,44 @@ describe("SourceRow", () => {
     expect(screen.getByText("Processed · 3 chunks")).toBeTruthy();
   });
 
+  it("labels unlocalized remote documents as processed without chunk counts", () => {
+    const onToggleIncluded = vi.fn();
+
+    render(
+      React.createElement(SourceRow, {
+        isArchiving: false,
+        isSelected: false,
+        onSelect: vi.fn(),
+        onToggleIncluded,
+        source: {
+          id: "knowhere-doc:default:doc_remote",
+          kind: "remote",
+          mimeType: "application/pdf",
+          title: "remote.pdf",
+          status: "ready",
+          chunkCount: 4,
+          documentId: "doc_remote",
+          excludedFromQuery: false,
+        },
+      }),
+    );
+
+    const checkbox = screen.getByRole("checkbox", {
+      name: "Use remote.pdf in answers",
+    });
+
+    expect(screen.getByText("Processed")).toBeTruthy();
+    expect(screen.queryByText("Processed · 4 chunks")).toBeNull();
+    expect(checkbox.getAttribute("aria-checked")).toBe("true");
+
+    fireEvent.click(checkbox);
+
+    expect(onToggleIncluded).toHaveBeenCalledWith(
+      "knowhere-doc:default:doc_remote",
+      false,
+    );
+  });
+
   it("shows source archive loading locally", () => {
     render(
       React.createElement(SourceRow, {

src/components/source-row.tsx

@@ -42,7 +42,6 @@ export function SourceRow({
   const isFailed = source.status === "failed";
   const canRetry = isFailed && source.originalFile !== undefined;
   const isLibrarySource = source.officialLibrary !== undefined;
-  const isRemoteSource = source.kind === "remote";
 
   const iconBg = fileIconTint(source.title);
 
@@ -63,7 +62,7 @@ export function SourceRow({
       >
         <Checkbox
           checked={!source.excludedFromQuery}
-          disabled={!isReady || !onToggleIncluded || isAdding || isRemoteSource}
+          disabled={!isReady || !onToggleIncluded || isAdding}
           onCheckedChange={(checked) =>
             onToggleIncluded?.(source.id, checked === true)
           }
@@ -193,11 +192,11 @@ export function SourceRow({
 
 function getReadySourceLabel(source: SourceView): string {
   if (source.officialLibrary !== undefined) return "Official Library";
-  if (source.kind === "remote") return "Remote";
   return "Processed";
 }
 
 function getReadySourceStatusText(source: SourceView): string {
+  if (source.kind === "remote") return getReadySourceLabel(source);
   if (typeof source.chunkCount !== "number") return getReadySourceLabel(source);
   return `${getReadySourceLabel(source)} · ${source.chunkCount} chunks`;
 }

src/domains/chat/index.test.ts

@@ -66,7 +66,7 @@ describe("answerQuestionWithRetrieval", () => {
         question: "What does the document say?",
         namespace: "notebook-workspace",
         sources,
-        excludedSourceIds: ["source_2"],
+        excludedSourceIds: ["source_2", "knowhere-doc:default:doc_remote"],
         retrieval,
         generateAnswer,
         messages: [],
@@ -79,13 +79,13 @@ describe("answerQuestionWithRetrieval", () => {
       topK: 8,
       useAgentic: true,
       dataType: 1,
-      excludeDocumentIds: ["doc_excluded"],
+      excludeDocumentIds: ["doc_excluded", "doc_remote"],
     });
     expect(generateAnswer).toHaveBeenCalledWith({
       question: "What does the document say?",
       messages: [],
       sources,
-      excludedSourceIds: ["source_2"],
+      excludedSourceIds: ["source_2", "knowhere-doc:default:doc_remote"],
       searchSources: expect.any(Function),
     });
     expect(answer).toEqual({

src/domains/chat/retrieval.ts

@@ -1,6 +1,7 @@
 import type { RetrievalQueryParams } from "@ontos-ai/knowhere-sdk"
 
 import type { Source } from "@/infrastructure/db/schema"
+import { decodeRemoteSourceId } from "@/domains/sources/remote-library"
 
 const RETRIEVAL_QUERY_CHAR_LIMIT = 600
 
@@ -25,10 +26,16 @@ export function excludeDocuments(
   excludedSourceIds: readonly string[],
 ): Pick<RetrievalQueryParams, "excludeDocumentIds"> {
   const excluded = new Set(excludedSourceIds)
-  const documentIds = sources
+  const localDocumentIds = sources
     .filter((source) => excluded.has(source.id))
     .map((source) => source.knowhereDocumentId)
     .filter((documentId): documentId is string => Boolean(documentId))
+  const remoteDocumentIds = excludedSourceIds
+    .map((sourceId) => decodeRemoteSourceId(sourceId)?.documentId)
+    .filter((documentId): documentId is string => Boolean(documentId))
+  const documentIds = Array.from(
+    new Set([...localDocumentIds, ...remoteDocumentIds]),
+  )
 
   return documentIds.length > 0 ? { excludeDocumentIds: documentIds } : {}
 }

src/domains/sources/remote-library.ts

@@ -280,7 +280,7 @@ function toRemoteSourceView(document: RemoteDocument): SourceView {
     mimeType: document.mimeType ?? "application/octet-stream",
     status: document.status,
     documentId: document.documentId,
-    excludedFromQuery: true,
+    excludedFromQuery: false,
   }
 }
 

src/domains/sources/route-service.test.ts

@@ -301,7 +301,7 @@ describe("source route service", () => {
         mimeType: "application/pdf",
         status: "ready",
         documentId: "doc_default",
-        excludedFromQuery: true,
+        excludedFromQuery: false,
       },
       {
         id: "knowhere-doc:notebook-workspace_1:doc_legacy",
@@ -311,7 +311,7 @@ describe("source route service", () => {
         mimeType: "application/octet-stream",
         status: "ready",
         documentId: "doc_legacy",
-        excludedFromQuery: true,
+        excludedFromQuery: false,
       },
     ]);
   });