From 66eebf614fb0be6028eee233884f8650899b6402 Mon Sep 17 00:00:00 2001 From: Anish Sarkar <104695310+AnishSarkar22@users.noreply.github.com> Date: Sat, 9 May 2026 05:04:00 +0530 Subject: [PATCH] test(e2e): cover connector PDF docling indexing in journeys --- .../connectors/composio/drive/journey.spec.ts | 84 ++++--- .../tests/connectors/dropbox/journey.spec.ts | 38 ++- .../connectors/google/drive/journey.spec.ts | 41 +++- .../tests/connectors/onedrive/journey.spec.ts | 226 ++++++++++-------- 4 files changed, 251 insertions(+), 138 deletions(-) diff --git a/surfsense_web/tests/connectors/composio/drive/journey.spec.ts b/surfsense_web/tests/connectors/composio/drive/journey.spec.ts index 62a074e4a..195c8b8e8 100644 --- a/surfsense_web/tests/connectors/composio/drive/journey.spec.ts +++ b/surfsense_web/tests/connectors/composio/drive/journey.spec.ts @@ -26,7 +26,7 @@ test.describe("Composio Drive journey", () => { composioDriveConnector, chatThread, }) => { - test.setTimeout(180_000); // worker cold-start + summarize + embed + chunk + test.setTimeout(240_000); // worker cold-start + Docling + summarize + embed + chunk await page.goto(`/dashboard/${searchSpace.id}/new-chat`, { waitUntil: "domcontentloaded", @@ -35,53 +35,63 @@ test.describe("Composio Drive journey", () => { const connectorDialog = page.getByRole("dialog", { name: "Manage Connectors" }); await expect(connectorDialog).toBeVisible(); + const selectedFiles = [ + { + id: FAKE_DRIVE_FILES.canary.id, + name: FAKE_DRIVE_FILES.canary.name, + mimeType: FAKE_DRIVE_FILES.canary.mimeType, + }, + { + id: FAKE_DRIVE_FILES.pdfComposio.id, + name: FAKE_DRIVE_FILES.pdfComposio.name, + mimeType: FAKE_DRIVE_FILES.pdfComposio.mimeType, + }, + ]; + const indexingOptions = { + max_files_per_folder: 10, + incremental_sync: false, + include_subfolders: false, + }; + await updateConnectorConfig(request, apiToken, composioDriveConnector.id, { ...composioDriveConnector.config, selected_folders: [], - selected_files: [ - { - id: FAKE_DRIVE_FILES.canary.id, - name: FAKE_DRIVE_FILES.canary.name, - mimeType: FAKE_DRIVE_FILES.canary.mimeType, - }, - ], - indexing_options: { - max_files_per_folder: 10, - incremental_sync: false, - include_subfolders: false, - }, + selected_files: selectedFiles, + indexing_options: indexingOptions, }); await triggerIndex(request, apiToken, composioDriveConnector.id, searchSpace.id, { - files: [ - { - id: FAKE_DRIVE_FILES.canary.id, - name: FAKE_DRIVE_FILES.canary.name, - mimeType: FAKE_DRIVE_FILES.canary.mimeType, - }, - ], - indexing_options: { - max_files_per_folder: 10, - incremental_sync: false, - include_subfolders: false, - }, + files: selectedFiles, + indexing_options: indexingOptions, }); await waitForIndexingComplete(request, apiToken, composioDriveConnector.id, searchSpace.id, { - timeoutMs: 150_000, + timeoutMs: 240_000, intervalMs: 1_500, - minDocuments: 1, + minDocuments: 2, }); await waitForDocumentByTitle(request, apiToken, searchSpace.id, FAKE_DRIVE_FILES.canary.name, { timeoutMs: 30_000, }); + await waitForDocumentByTitle( + request, + apiToken, + searchSpace.id, + FAKE_DRIVE_FILES.pdfComposio.name, + { timeoutMs: 60_000 } + ); const docs = await listDocuments(request, apiToken, searchSpace.id); const canaryDoc = docs.find((d) => d.title === FAKE_DRIVE_FILES.canary.name); + const pdfDoc = docs.find((d) => d.title === FAKE_DRIVE_FILES.pdfComposio.name); expect(canaryDoc, "canary document must exist after indexing").toBeDefined(); if (!canaryDoc) throw new Error("unreachable: canaryDoc asserted defined above"); + expect(canaryDoc.document_type).toBe("GOOGLE_DRIVE_FILE"); + expect(pdfDoc, "Composio Drive PDF document must exist after indexing").toBeDefined(); + if (!pdfDoc) throw new Error("unreachable: pdfDoc asserted defined above"); + expect(pdfDoc.document_type).toBe("GOOGLE_DRIVE_FILE"); // content holds the LLM summary; the raw file body lives in source_markdown. // editor-content is the same endpoint the UI hits when opening a document. @@ -91,8 +101,18 @@ test.describe("Composio Drive journey", () => { `canary token ${CANARY_TOKENS.driveCanaryFile} should appear in editor source_markdown; ` + `got first 200 chars: ${editor.source_markdown.slice(0, 200)}` ).toContain(CANARY_TOKENS.driveCanaryFile); + expect(editor.document_type).toBe("GOOGLE_DRIVE_FILE"); expect(editor.chunk_count).toBeGreaterThan(0); + const pdfEditor = await getEditorContent(request, apiToken, searchSpace.id, pdfDoc.id); + expect( + pdfEditor.source_markdown, + `PDF canary token ${CANARY_TOKENS.composioDrivePdfCanary} should appear in editor source_markdown; ` + + `got first 200 chars: ${pdfEditor.source_markdown.slice(0, 200)}` + ).toContain(CANARY_TOKENS.composioDrivePdfCanary); + expect(pdfEditor.document_type).toBe("GOOGLE_DRIVE_FILE"); + expect(pdfEditor.chunk_count).toBeGreaterThan(0); + const refreshedConnectors = await listConnectors(request, apiToken, searchSpace.id); const refreshed = refreshedConnectors.find((c) => c.id === composioDriveConnector.id); expect(refreshed?.last_indexed_at).not.toBeNull(); @@ -106,5 +126,15 @@ test.describe("Composio Drive journey", () => { chat.assistantText, `chat agent should surface canary token after indexing; got: ${chat.assistantText.slice(0, 200)}` ).toContain(CANARY_TOKENS.driveCanaryFile); + + const pdfChat = await streamChatToCompletion(request, apiToken, { + searchSpaceId: searchSpace.id, + threadId: chatThread.id, + query: "What is in my e2e-composio-canary.pdf Drive file?", + }); + expect( + pdfChat.assistantText, + `chat agent should surface Composio Drive PDF canary token after indexing; got: ${pdfChat.assistantText.slice(0, 200)}` + ).toContain(CANARY_TOKENS.composioDrivePdfCanary); }); }); diff --git a/surfsense_web/tests/connectors/dropbox/journey.spec.ts b/surfsense_web/tests/connectors/dropbox/journey.spec.ts index 4b33d4d5a..27616b2b9 100644 --- a/surfsense_web/tests/connectors/dropbox/journey.spec.ts +++ b/surfsense_web/tests/connectors/dropbox/journey.spec.ts @@ -22,7 +22,7 @@ test.describe("Native Dropbox journey", () => { nativeDropboxConnector, chatThread, }) => { - test.setTimeout(180_000); // worker cold-start + summarize + embed + chunk + test.setTimeout(240_000); // worker cold-start + Docling + summarize + embed + chunk expect(nativeDropboxConnector.connector_type).toBe("DROPBOX_CONNECTOR"); expect(nativeDropboxConnector.is_indexable).toBe(true); @@ -42,6 +42,11 @@ test.describe("Native Dropbox journey", () => { name: FAKE_DROPBOX_FILES.canary.name, mimeType: FAKE_DROPBOX_FILES.canary.mimeType, }, + { + id: FAKE_DROPBOX_FILES.pdf.id, + name: FAKE_DROPBOX_FILES.pdf.name, + mimeType: FAKE_DROPBOX_FILES.pdf.mimeType, + }, ]; // Keep the shared Drive-style body shape. Dropbox currently defaults // these internally because the indexer expects max_files/use_delta_sync. @@ -64,9 +69,9 @@ test.describe("Native Dropbox journey", () => { }); await waitForIndexingComplete(request, apiToken, nativeDropboxConnector.id, searchSpace.id, { - timeoutMs: 150_000, + timeoutMs: 240_000, intervalMs: 1_500, - minDocuments: 1, + minDocuments: 2, }); await waitForDocumentByTitle( @@ -76,13 +81,20 @@ test.describe("Native Dropbox journey", () => { FAKE_DROPBOX_FILES.canary.name, { timeoutMs: 30_000 } ); + await waitForDocumentByTitle(request, apiToken, searchSpace.id, FAKE_DROPBOX_FILES.pdf.name, { + timeoutMs: 60_000, + }); const docs = await listDocuments(request, apiToken, searchSpace.id); const canaryDoc = docs.find((d) => d.title === FAKE_DROPBOX_FILES.canary.name); + const pdfDoc = docs.find((d) => d.title === FAKE_DROPBOX_FILES.pdf.name); expect(canaryDoc, "Dropbox canary document must exist after indexing").toBeDefined(); if (!canaryDoc) throw new Error("unreachable: canaryDoc asserted defined above"); expect(canaryDoc.document_type).toBe("DROPBOX_FILE"); + expect(pdfDoc, "Dropbox PDF document must exist after indexing").toBeDefined(); + if (!pdfDoc) throw new Error("unreachable: pdfDoc asserted defined above"); + expect(pdfDoc.document_type).toBe("DROPBOX_FILE"); const editor = await getEditorContent(request, apiToken, searchSpace.id, canaryDoc.id); expect( @@ -93,6 +105,15 @@ test.describe("Native Dropbox journey", () => { expect(editor.document_type).toBe("DROPBOX_FILE"); expect(editor.chunk_count).toBeGreaterThan(0); + const pdfEditor = await getEditorContent(request, apiToken, searchSpace.id, pdfDoc.id); + expect( + pdfEditor.source_markdown, + `PDF canary token ${CANARY_TOKENS.dropboxPdfCanary} should appear in editor source_markdown; ` + + `got first 200 chars: ${pdfEditor.source_markdown.slice(0, 200)}` + ).toContain(CANARY_TOKENS.dropboxPdfCanary); + expect(pdfEditor.document_type).toBe("DROPBOX_FILE"); + expect(pdfEditor.chunk_count).toBeGreaterThan(0); + const refreshedConnectors = await listConnectors(request, apiToken, searchSpace.id); const refreshed = refreshedConnectors.find((c) => c.id === nativeDropboxConnector.id); expect(refreshed?.connector_type).toBe("DROPBOX_CONNECTOR"); @@ -109,5 +130,16 @@ test.describe("Native Dropbox journey", () => { "chat agent should surface Dropbox canary token after indexing; " + `got: ${chat.assistantText.slice(0, 200)}` ).toContain(CANARY_TOKENS.dropboxCanary); + + const pdfChat = await streamChatToCompletion(request, apiToken, { + searchSpaceId: searchSpace.id, + threadId: chatThread.id, + query: "What is in my e2e-dropbox-canary.pdf Dropbox file?", + }); + expect( + pdfChat.assistantText, + "chat agent should surface Dropbox PDF canary token after indexing; " + + `got: ${pdfChat.assistantText.slice(0, 200)}` + ).toContain(CANARY_TOKENS.dropboxPdfCanary); }); }); diff --git a/surfsense_web/tests/connectors/google/drive/journey.spec.ts b/surfsense_web/tests/connectors/google/drive/journey.spec.ts index 8044fae58..258939c36 100644 --- a/surfsense_web/tests/connectors/google/drive/journey.spec.ts +++ b/surfsense_web/tests/connectors/google/drive/journey.spec.ts @@ -26,7 +26,7 @@ test.describe("Native Google Drive journey", () => { nativeDriveConnector, chatThread, }) => { - test.setTimeout(180_000); // worker cold-start + summarize + embed + chunk + test.setTimeout(240_000); // worker cold-start + Docling + summarize + embed + chunk expect(nativeDriveConnector.connector_type).toBe("GOOGLE_DRIVE_CONNECTOR"); expect(nativeDriveConnector.is_indexable).toBe(true); @@ -46,6 +46,11 @@ test.describe("Native Google Drive journey", () => { name: FAKE_DRIVE_FILES.canary.name, mimeType: FAKE_DRIVE_FILES.canary.mimeType, }, + { + id: FAKE_DRIVE_FILES.pdfNative.id, + name: FAKE_DRIVE_FILES.pdfNative.name, + mimeType: FAKE_DRIVE_FILES.pdfNative.mimeType, + }, ]; const indexingOptions = { max_files_per_folder: 10, @@ -66,21 +71,32 @@ test.describe("Native Google Drive journey", () => { }); await waitForIndexingComplete(request, apiToken, nativeDriveConnector.id, searchSpace.id, { - timeoutMs: 150_000, + timeoutMs: 240_000, intervalMs: 1_500, - minDocuments: 1, + minDocuments: 2, }); await waitForDocumentByTitle(request, apiToken, searchSpace.id, FAKE_DRIVE_FILES.canary.name, { timeoutMs: 30_000, }); + await waitForDocumentByTitle( + request, + apiToken, + searchSpace.id, + FAKE_DRIVE_FILES.pdfNative.name, + { timeoutMs: 60_000 } + ); const docs = await listDocuments(request, apiToken, searchSpace.id); const canaryDoc = docs.find((d) => d.title === FAKE_DRIVE_FILES.canary.name); + const pdfDoc = docs.find((d) => d.title === FAKE_DRIVE_FILES.pdfNative.name); expect(canaryDoc, "native Drive canary document must exist after indexing").toBeDefined(); if (!canaryDoc) throw new Error("unreachable: canaryDoc asserted defined above"); expect(canaryDoc.document_type).toBe("GOOGLE_DRIVE_FILE"); + expect(pdfDoc, "native Drive PDF document must exist after indexing").toBeDefined(); + if (!pdfDoc) throw new Error("unreachable: pdfDoc asserted defined above"); + expect(pdfDoc.document_type).toBe("GOOGLE_DRIVE_FILE"); const editor = await getEditorContent(request, apiToken, searchSpace.id, canaryDoc.id); expect( @@ -91,6 +107,15 @@ test.describe("Native Google Drive journey", () => { expect(editor.document_type).toBe("GOOGLE_DRIVE_FILE"); expect(editor.chunk_count).toBeGreaterThan(0); + const pdfEditor = await getEditorContent(request, apiToken, searchSpace.id, pdfDoc.id); + expect( + pdfEditor.source_markdown, + `PDF canary token ${CANARY_TOKENS.drivePdfCanary} should appear in editor source_markdown; ` + + `got first 200 chars: ${pdfEditor.source_markdown.slice(0, 200)}` + ).toContain(CANARY_TOKENS.drivePdfCanary); + expect(pdfEditor.document_type).toBe("GOOGLE_DRIVE_FILE"); + expect(pdfEditor.chunk_count).toBeGreaterThan(0); + const refreshedConnectors = await listConnectors(request, apiToken, searchSpace.id); const refreshed = refreshedConnectors.find((c) => c.id === nativeDriveConnector.id); expect(refreshed?.connector_type).toBe("GOOGLE_DRIVE_CONNECTOR"); @@ -106,5 +131,15 @@ test.describe("Native Google Drive journey", () => { chat.assistantText, `chat agent should surface native Drive canary token after indexing; got: ${chat.assistantText.slice(0, 200)}` ).toContain(CANARY_TOKENS.driveCanaryFile); + + const pdfChat = await streamChatToCompletion(request, apiToken, { + searchSpaceId: searchSpace.id, + threadId: chatThread.id, + query: "What is in my e2e-canary.pdf native Drive file?", + }); + expect( + pdfChat.assistantText, + `chat agent should surface native Drive PDF canary token after indexing; got: ${pdfChat.assistantText.slice(0, 200)}` + ).toContain(CANARY_TOKENS.drivePdfCanary); }); }); diff --git a/surfsense_web/tests/connectors/onedrive/journey.spec.ts b/surfsense_web/tests/connectors/onedrive/journey.spec.ts index 81a9a52ff..5ec041415 100644 --- a/surfsense_web/tests/connectors/onedrive/journey.spec.ts +++ b/surfsense_web/tests/connectors/onedrive/journey.spec.ts @@ -1,17 +1,10 @@ import { expect, nativeOneDriveWithChatTest as test } from "../../fixtures"; import { streamChatToCompletion } from "../../helpers/api/chat"; -import { - listConnectors, - triggerIndex, - updateConnectorConfig, -} from "../../helpers/api/connectors"; +import { listConnectors, triggerIndex, updateConnectorConfig } from "../../helpers/api/connectors"; import { getEditorContent, listDocuments } from "../../helpers/api/documents"; import { CANARY_TOKENS, FAKE_ONEDRIVE_FILES } from "../../helpers/canary"; import { openConnectorPopup } from "../../helpers/ui/connector-popup"; -import { - waitForDocumentByTitle, - waitForIndexingComplete, -} from "../../helpers/waits/indexing"; +import { waitForDocumentByTitle, waitForIndexingComplete } from "../../helpers/waits/indexing"; /** * Proves the native OneDrive wiring from Microsoft OAuth fixture -> selection @@ -21,109 +14,132 @@ import { * persisted config and indexing contract the picker ultimately feeds. */ test.describe("Native OneDrive journey", () => { - test( - "user connects OneDrive, selects a file, indexes it, and chats with the canary token", - async ({ - page, + test("user connects OneDrive, selects a file, indexes it, and chats with the canary token", async ({ + page, + request, + apiToken, + searchSpace, + nativeOneDriveConnector, + chatThread, + }) => { + test.setTimeout(240_000); // worker cold-start + Docling + summarize + embed + chunk + + expect(nativeOneDriveConnector.connector_type).toBe("ONEDRIVE_CONNECTOR"); + expect(nativeOneDriveConnector.is_indexable).toBe(true); + expect(nativeOneDriveConnector.config._token_encrypted).toBe(true); + expect(nativeOneDriveConnector.config.composio_connected_account_id).toBeUndefined(); + + await page.goto(`/dashboard/${searchSpace.id}/new-chat`, { + waitUntil: "domcontentloaded", + }); + await openConnectorPopup(page); + const connectorDialog = page.getByRole("dialog", { name: "Manage Connectors" }); + await expect(connectorDialog).toBeVisible(); + + const selectedFiles = [ + { + id: FAKE_ONEDRIVE_FILES.canary.id, + name: FAKE_ONEDRIVE_FILES.canary.name, + mimeType: FAKE_ONEDRIVE_FILES.canary.mimeType, + }, + { + id: FAKE_ONEDRIVE_FILES.pdf.id, + name: FAKE_ONEDRIVE_FILES.pdf.name, + mimeType: FAKE_ONEDRIVE_FILES.pdf.mimeType, + }, + ]; + // Keep the shared Drive-style body shape. OneDrive currently defaults + // these internally because the indexer expects max_files/use_delta_sync. + const indexingOptions = { + max_files_per_folder: 10, + incremental_sync: false, + include_subfolders: false, + }; + + await updateConnectorConfig(request, apiToken, nativeOneDriveConnector.id, { + ...nativeOneDriveConnector.config, + selected_folders: [], + selected_files: selectedFiles, + indexing_options: indexingOptions, + }); + + await triggerIndex(request, apiToken, nativeOneDriveConnector.id, searchSpace.id, { + files: selectedFiles, + indexing_options: indexingOptions, + }); + + await waitForIndexingComplete(request, apiToken, nativeOneDriveConnector.id, searchSpace.id, { + timeoutMs: 240_000, + intervalMs: 1_500, + minDocuments: 2, + }); + + await waitForDocumentByTitle( request, apiToken, - searchSpace, - nativeOneDriveConnector, - chatThread, - }) => { - test.setTimeout(180_000); // worker cold-start + summarize + embed + chunk + searchSpace.id, + FAKE_ONEDRIVE_FILES.canary.name, + { timeoutMs: 30_000 } + ); + await waitForDocumentByTitle(request, apiToken, searchSpace.id, FAKE_ONEDRIVE_FILES.pdf.name, { + timeoutMs: 60_000, + }); - expect(nativeOneDriveConnector.connector_type).toBe("ONEDRIVE_CONNECTOR"); - expect(nativeOneDriveConnector.is_indexable).toBe(true); - expect(nativeOneDriveConnector.config._token_encrypted).toBe(true); - expect(nativeOneDriveConnector.config.composio_connected_account_id).toBeUndefined(); + const docs = await listDocuments(request, apiToken, searchSpace.id); + const canaryDoc = docs.find((d) => d.title === FAKE_ONEDRIVE_FILES.canary.name); + const pdfDoc = docs.find((d) => d.title === FAKE_ONEDRIVE_FILES.pdf.name); - await page.goto(`/dashboard/${searchSpace.id}/new-chat`, { - waitUntil: "domcontentloaded", - }); - await openConnectorPopup(page); - const connectorDialog = page.getByRole("dialog", { name: "Manage Connectors" }); - await expect(connectorDialog).toBeVisible(); + expect(canaryDoc, "OneDrive canary document must exist after indexing").toBeDefined(); + if (!canaryDoc) throw new Error("unreachable: canaryDoc asserted defined above"); + expect(canaryDoc.document_type).toBe("ONEDRIVE_FILE"); + expect(pdfDoc, "OneDrive PDF document must exist after indexing").toBeDefined(); + if (!pdfDoc) throw new Error("unreachable: pdfDoc asserted defined above"); + expect(pdfDoc.document_type).toBe("ONEDRIVE_FILE"); - const selectedFiles = [ - { - id: FAKE_ONEDRIVE_FILES.canary.id, - name: FAKE_ONEDRIVE_FILES.canary.name, - mimeType: FAKE_ONEDRIVE_FILES.canary.mimeType, - }, - ]; - // Keep the shared Drive-style body shape. OneDrive currently defaults - // these internally because the indexer expects max_files/use_delta_sync. - const indexingOptions = { - max_files_per_folder: 10, - incremental_sync: false, - include_subfolders: false, - }; + const editor = await getEditorContent(request, apiToken, searchSpace.id, canaryDoc.id); + expect( + editor.source_markdown, + `canary token ${CANARY_TOKENS.onedriveCanary} should appear in editor source_markdown; ` + + `got first 200 chars: ${editor.source_markdown.slice(0, 200)}` + ).toContain(CANARY_TOKENS.onedriveCanary); + expect(editor.document_type).toBe("ONEDRIVE_FILE"); + expect(editor.chunk_count).toBeGreaterThan(0); - await updateConnectorConfig(request, apiToken, nativeOneDriveConnector.id, { - ...nativeOneDriveConnector.config, - selected_folders: [], - selected_files: selectedFiles, - indexing_options: indexingOptions, - }); + const pdfEditor = await getEditorContent(request, apiToken, searchSpace.id, pdfDoc.id); + expect( + pdfEditor.source_markdown, + `PDF canary token ${CANARY_TOKENS.onedrivePdfCanary} should appear in editor source_markdown; ` + + `got first 200 chars: ${pdfEditor.source_markdown.slice(0, 200)}` + ).toContain(CANARY_TOKENS.onedrivePdfCanary); + expect(pdfEditor.document_type).toBe("ONEDRIVE_FILE"); + expect(pdfEditor.chunk_count).toBeGreaterThan(0); - await triggerIndex(request, apiToken, nativeOneDriveConnector.id, searchSpace.id, { - files: selectedFiles, - indexing_options: indexingOptions, - }); + const refreshedConnectors = await listConnectors(request, apiToken, searchSpace.id); + const refreshed = refreshedConnectors.find((c) => c.id === nativeOneDriveConnector.id); + expect(refreshed?.connector_type).toBe("ONEDRIVE_CONNECTOR"); + expect(refreshed?.is_indexable).toBe(true); + expect(refreshed?.last_indexed_at).not.toBeNull(); - await waitForIndexingComplete( - request, - apiToken, - nativeOneDriveConnector.id, - searchSpace.id, - { - timeoutMs: 150_000, - intervalMs: 1_500, - minDocuments: 1, - } - ); + const chat = await streamChatToCompletion(request, apiToken, { + searchSpaceId: searchSpace.id, + threadId: chatThread.id, + query: "What is in my e2e-onedrive-canary.txt OneDrive file?", + }); + expect( + chat.assistantText, + "chat agent should surface OneDrive canary token after indexing; " + + `got: ${chat.assistantText.slice(0, 200)}` + ).toContain(CANARY_TOKENS.onedriveCanary); - await waitForDocumentByTitle( - request, - apiToken, - searchSpace.id, - FAKE_ONEDRIVE_FILES.canary.name, - { timeoutMs: 30_000 } - ); - - const docs = await listDocuments(request, apiToken, searchSpace.id); - const canaryDoc = docs.find((d) => d.title === FAKE_ONEDRIVE_FILES.canary.name); - - expect(canaryDoc, "OneDrive canary document must exist after indexing").toBeDefined(); - if (!canaryDoc) throw new Error("unreachable: canaryDoc asserted defined above"); - expect(canaryDoc.document_type).toBe("ONEDRIVE_FILE"); - - const editor = await getEditorContent(request, apiToken, searchSpace.id, canaryDoc.id); - expect( - editor.source_markdown, - `canary token ${CANARY_TOKENS.onedriveCanary} should appear in editor source_markdown; ` + - `got first 200 chars: ${editor.source_markdown.slice(0, 200)}` - ).toContain(CANARY_TOKENS.onedriveCanary); - expect(editor.document_type).toBe("ONEDRIVE_FILE"); - expect(editor.chunk_count).toBeGreaterThan(0); - - const refreshedConnectors = await listConnectors(request, apiToken, searchSpace.id); - const refreshed = refreshedConnectors.find((c) => c.id === nativeOneDriveConnector.id); - expect(refreshed?.connector_type).toBe("ONEDRIVE_CONNECTOR"); - expect(refreshed?.is_indexable).toBe(true); - expect(refreshed?.last_indexed_at).not.toBeNull(); - - const chat = await streamChatToCompletion(request, apiToken, { - searchSpaceId: searchSpace.id, - threadId: chatThread.id, - query: "What is in my e2e-onedrive-canary.txt OneDrive file?", - }); - expect( - chat.assistantText, - "chat agent should surface OneDrive canary token after indexing; " + - `got: ${chat.assistantText.slice(0, 200)}` - ).toContain(CANARY_TOKENS.onedriveCanary); - } - ); + const pdfChat = await streamChatToCompletion(request, apiToken, { + searchSpaceId: searchSpace.id, + threadId: chatThread.id, + query: "What is in my e2e-onedrive-canary.pdf OneDrive file?", + }); + expect( + pdfChat.assistantText, + "chat agent should surface OneDrive PDF canary token after indexing; " + + `got: ${pdfChat.assistantText.slice(0, 200)}` + ).toContain(CANARY_TOKENS.onedrivePdfCanary); + }); });