refactor(tests): simplify logging messages and enhance manual upload journey tests

surfsense_backend/tests/e2e/fakes/chat_llm.py

@@ -44,6 +44,10 @@ SLACK_CANARY_CHANNEL = "slack-e2e-canary"
 CLICKUP_CANARY_TOKEN = "SURFSENSE_E2E_CANARY_TOKEN_CLICKUP_001"
 CLICKUP_CANARY_TITLE = "E2E Canary ClickUp Task"
 CLICKUP_CANARY_TASK_ID = "fake-clickup-task-canary-001"
+MANUAL_UPLOAD_MD_CANARY_TOKEN = "E2E-MANUAL-UPLOAD-MD-CANARY-7f3a"
+MANUAL_UPLOAD_MD_CANARY_FILE = "canary.md"
+MANUAL_UPLOAD_PDF_CANARY_TOKEN = "E2E-MANUAL-UPLOAD-PDF-CANARY-9d2b"
+MANUAL_UPLOAD_PDF_CANARY_FILE = "canary.pdf"
 NO_RELEVANT_CONTENT_SENTINEL = "No relevant indexed content found."
 NO_RELEVANT_CONTENT_QUERY = "E2E_NO_RELEVANT_CONTENT_SMOKE"
 
@@ -208,6 +212,30 @@ class FakeChatLLM(BaseChatModel):
             latest_human,
             ("clickup", CLICKUP_CANARY_TITLE),
         )
+        wants_manual_upload = _contains_any(
+            latest_human,
+            (
+                "uploaded",
+                "manual upload",
+                MANUAL_UPLOAD_MD_CANARY_FILE,
+                MANUAL_UPLOAD_PDF_CANARY_FILE,
+                MANUAL_UPLOAD_MD_CANARY_TOKEN,
+                MANUAL_UPLOAD_PDF_CANARY_TOKEN,
+            ),
+        )
+        wants_manual_upload_pdf = wants_manual_upload and _contains_any(
+            latest_human,
+            ("pdf", MANUAL_UPLOAD_PDF_CANARY_FILE, MANUAL_UPLOAD_PDF_CANARY_TOKEN),
+        )
+        wants_manual_upload_md = wants_manual_upload and _contains_any(
+            latest_human,
+            (
+                "markdown",
+                ".md",
+                MANUAL_UPLOAD_MD_CANARY_FILE,
+                MANUAL_UPLOAD_MD_CANARY_TOKEN,
+            ),
+        )
         has_gmail_evidence = (
             GMAIL_CANARY_SUBJECT in prompt_text
             or GMAIL_CANARY_MESSAGE_ID in prompt_text
@@ -285,6 +313,14 @@ class FakeChatLLM(BaseChatModel):
             or CLICKUP_CANARY_TOKEN in prompt_text
             or CLICKUP_CANARY_TASK_ID in prompt_text
         )
+        has_manual_upload_md_evidence = (
+            MANUAL_UPLOAD_MD_CANARY_FILE in prompt_text
+            or MANUAL_UPLOAD_MD_CANARY_TOKEN in prompt_text
+        )
+        has_manual_upload_pdf_evidence = (
+            MANUAL_UPLOAD_PDF_CANARY_FILE in prompt_text
+            or MANUAL_UPLOAD_PDF_CANARY_TOKEN in prompt_text
+        )
 
         if wants_clickup and has_clickup_evidence:
             return f"ClickUp content found: {CLICKUP_CANARY_TOKEN}"
@@ -316,6 +352,10 @@ class FakeChatLLM(BaseChatModel):
             return f"Drive PDF content found: {COMPOSIO_DRIVE_PDF_CANARY_TOKEN}"
         if wants_drive and has_drive_evidence:
             return f"Drive content found: {DRIVE_CANARY_TOKEN}"
+        if wants_manual_upload_pdf and has_manual_upload_pdf_evidence:
+            return f"Manual upload PDF content found: {MANUAL_UPLOAD_PDF_CANARY_TOKEN}"
+        if wants_manual_upload_md and has_manual_upload_md_evidence:
+            return f"Manual upload MD content found: {MANUAL_UPLOAD_MD_CANARY_TOKEN}"
         if (
             has_notion_evidence
             and not has_confluence_evidence
@@ -468,6 +508,36 @@ class FakeChatLLM(BaseChatModel):
             and not has_slack_evidence
         ):
             return f"ClickUp content found: {CLICKUP_CANARY_TOKEN}"
+        if (
+            has_manual_upload_pdf_evidence
+            and not has_confluence_evidence
+            and not has_jira_evidence
+            and not has_linear_evidence
+            and not has_notion_evidence
+            and not has_calendar_evidence
+            and not has_gmail_evidence
+            and not has_drive_evidence
+            and not has_onedrive_evidence
+            and not has_dropbox_evidence
+            and not has_slack_evidence
+            and not has_clickup_evidence
+        ):
+            return f"Manual upload PDF content found: {MANUAL_UPLOAD_PDF_CANARY_TOKEN}"
+        if (
+            has_manual_upload_md_evidence
+            and not has_confluence_evidence
+            and not has_jira_evidence
+            and not has_linear_evidence
+            and not has_notion_evidence
+            and not has_calendar_evidence
+            and not has_gmail_evidence
+            and not has_drive_evidence
+            and not has_onedrive_evidence
+            and not has_dropbox_evidence
+            and not has_slack_evidence
+            and not has_clickup_evidence
+        ):
+            return f"Manual upload MD content found: {MANUAL_UPLOAD_MD_CANARY_TOKEN}"
         return NO_RELEVANT_CONTENT_SENTINEL
 
     def _tool_call_message_for(self, messages: list[BaseMessage]) -> AIMessage | None:

surfsense_backend/tests/e2e/run_backend.py

@@ -91,8 +91,7 @@ logging.basicConfig(
 )
 logger = logging.getLogger("surfsense.e2e.backend")
 logger.warning(
-    "*** SURFSENSE E2E BACKEND ENTRYPOINT — fake Composio + LLM + embeddings, "
+    "*** SURFSENSE E2E BACKEND ENTRYPOINT — fake Composio + LLM + embeddings ***"
-    "this MUST NOT be reachable in production. ***"
 )
 
 

surfsense_backend/tests/e2e/run_celery.py

@@ -78,8 +78,7 @@ logging.basicConfig(
 )
 logger = logging.getLogger("surfsense.e2e.celery")
 logger.warning(
-    "*** SURFSENSE E2E CELERY WORKER — fake Composio + LLM + embeddings, "
+    "*** SURFSENSE E2E CELERY WORKER — fake Composio + LLM + embeddings ***"
-    "this MUST NOT be reachable in production. ***"
 )
 
 

surfsense_web/tests/documents/file-upload/journey.spec.ts

@@ -1,6 +1,8 @@
 import path from "node:path";
 import type { APIRequestContext, Page } from "@playwright/test";
-import { expect, test } from "../../fixtures";
+import { expect, manualUploadWithChatTest as test } from "../../fixtures";
+import type { ChatThreadRow } from "../../fixtures/chat-thread.fixture";
+import { streamChatToCompletion } from "../../helpers/api/chat";
 import { getEditorContent, listDocuments } from "../../helpers/api/documents";
 import { CANARY_TOKENS } from "../../helpers/canary";
 import { waitForDocumentByTitle } from "../../helpers/waits/indexing";
@@ -9,6 +11,7 @@ type UploadFixture = {
 	path: string;
 	name: string;
 	canary: string;
+	chatQuery: string;
 };
 
 type SearchSpace = {
@@ -19,12 +22,14 @@ const MD_FILE: UploadFixture = {
 	path: path.join(__dirname, "fixtures", "canary.md"),
 	name: "canary.md",
 	canary: CANARY_TOKENS.manualUploadMdCanary,
+	chatQuery: "What is in my uploaded canary.md markdown file?",
 };
 
 const PDF_FILE: UploadFixture = {
 	path: path.join(__dirname, "fixtures", "canary.pdf"),
 	name: "canary.pdf",
 	canary: CANARY_TOKENS.manualUploadPdfCanary,
+	chatQuery: "What is in my uploaded canary.pdf file?",
 };
 
 async function uploadAndAssert({
@@ -32,12 +37,14 @@ async function uploadAndAssert({
 	request,
 	apiToken,
 	searchSpace,
+	chatThread,
 	file,
 }: {
 	page: Page;
 	request: APIRequestContext;
 	apiToken: string;
 	searchSpace: SearchSpace;
+	chatThread: ChatThreadRow;
 	file: UploadFixture;
 }) {
 	await page.goto(`/dashboard/${searchSpace.id}/new-chat`, {
@@ -68,6 +75,16 @@ async function uploadAndAssert({
 	const editor = await getEditorContent(request, apiToken, searchSpace.id, uploaded.id);
 	expect(editor.source_markdown).toContain(file.canary);
 	expect(editor.chunk_count).toBeGreaterThan(0);
+
+	const chat = await streamChatToCompletion(request, apiToken, {
+		searchSpaceId: searchSpace.id,
+		threadId: chatThread.id,
+		query: file.chatQuery,
+	});
+	expect(
+		chat.assistantText,
+		`chat agent should surface manual upload canary after indexing; got: ${chat.assistantText.slice(0, 200)}`
+	).toContain(file.canary);
 }
 
 test.describe("Manual file upload journey", () => {
@@ -76,10 +93,18 @@ test.describe("Manual file upload journey", () => {
 		request,
 		apiToken,
 		searchSpace,
+		chatThread,
 	}) => {
 		test.setTimeout(180_000);
 
-		await uploadAndAssert({ page, request, apiToken, searchSpace, file: MD_FILE });
+		await uploadAndAssert({
+			page,
+			request,
+			apiToken,
+			searchSpace,
+			chatThread,
+			file: MD_FILE,
+		});
 	});
 
 	test("user uploads a PDF (DOCUMENT branch via real Docling)", async ({
@@ -87,9 +112,17 @@ test.describe("Manual file upload journey", () => {
 		request,
 		apiToken,
 		searchSpace,
+		chatThread,
 	}) => {
 		test.setTimeout(240_000); // Docling cold-start can take 30-60s on first invocation.
 
-		await uploadAndAssert({ page, request, apiToken, searchSpace, file: PDF_FILE });
+		await uploadAndAssert({
+			page,
+			request,
+			apiToken,
+			searchSpace,
+			chatThread,
+			file: PDF_FILE,
+		});
 	});
 });

surfsense_web/tests/fixtures/index.ts

@@ -36,6 +36,7 @@
  *             └─ clickupWithChatTest — chatThread
  *         └─ slackFixtures — slackConnector
  *             └─ slackWithChatTest — chatThread
+ *         └─ manualUploadWithChatTest — chatThread (no connector; uses base search space)
  *
  * To add a new connector (Gmail, Slack, manual upload, etc.):
  *   1. Add a fixture file under `fixtures/connectors/<name>.fixture.ts`.
@@ -144,3 +145,11 @@ export const clickupWithChatTest = clickupFixtures.extend<ChatThreadFixtures>(ch
 export const slackTest = slackFixtures;
 /** `test` for Slack specs that also need a chat thread. */
 export const slackWithChatTest = slackFixtures.extend<ChatThreadFixtures>(chatThreadFixtures);
+/**
+ * `test` for manual upload specs that also need a chat thread.
+ *
+ * Manual upload has no connector fixture — the user uploads files directly via
+ * the Documents-sidebar UI — so this composes chat onto the bare search-space.
+ */
+export const manualUploadWithChatTest =
+	searchSpaceFixtures.extend<ChatThreadFixtures>(chatThreadFixtures);