apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-05-12 09:12:40 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

test(e2e): cover connector PDF docling indexing in journeys

Browse source
This commit is contained in:
Anish Sarkar 2026-05-09 05:04:00 +05:30
parent fbfde74cdc
commit 66eebf614f
4 changed files with 251 additions and 138 deletions
Download patch file Download diff file Expand all files Collapse all files

84
surfsense_web/tests/connectors/composio/drive/journey.spec.ts
View file

@ -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);
});
});

38
surfsense_web/tests/connectors/dropbox/journey.spec.ts
View file

@ -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);
});
});

41
surfsense_web/tests/connectors/google/drive/journey.spec.ts
View file

@ -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);
});
});

66
surfsense_web/tests/connectors/onedrive/journey.spec.ts
View file

@ -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
@ -21,9 +14,7 @@ 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 ({
test("user connects OneDrive, selects a file, indexes it, and chats with the canary token", async ({
page,
request,
apiToken,
@ -31,7 +22,7 @@ test.describe("Native OneDrive journey", () => {
nativeOneDriveConnector,
chatThread,
}) => {
test.setTimeout(180_000); // worker cold-start + summarize + embed + chunk
test.setTimeout(240_000); // worker cold-start + Docling + summarize + embed + chunk
expect(nativeOneDriveConnector.connector_type).toBe("ONEDRIVE_CONNECTOR");
expect(nativeOneDriveConnector.is_indexable).toBe(true);
@ -51,6 +42,11 @@ test.describe("Native OneDrive journey", () => {
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.
@ -72,17 +68,11 @@ test.describe("Native OneDrive journey", () => {
indexing_options: indexingOptions,
});
await waitForIndexingComplete(
request,
apiToken,
nativeOneDriveConnector.id,
searchSpace.id,
{
timeoutMs: 150_000,
await waitForIndexingComplete(request, apiToken, nativeOneDriveConnector.id, searchSpace.id, {
timeoutMs: 240_000,
intervalMs: 1_500,
minDocuments: 1,
}
);
minDocuments: 2,
});
await waitForDocumentByTitle(
request,
@ -91,13 +81,20 @@ test.describe("Native OneDrive journey", () => {
FAKE_ONEDRIVE_FILES.canary.name,
{ timeoutMs: 30_000 }
);
await waitForDocumentByTitle(request, apiToken, searchSpace.id, FAKE_ONEDRIVE_FILES.pdf.name, {
timeoutMs: 60_000,
});
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);
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 editor = await getEditorContent(request, apiToken, searchSpace.id, canaryDoc.id);
expect(
@ -108,6 +105,15 @@ test.describe("Native OneDrive journey", () => {
expect(editor.document_type).toBe("ONEDRIVE_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.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);
const refreshedConnectors = await listConnectors(request, apiToken, searchSpace.id);
const refreshed = refreshedConnectors.find((c) => c.id === nativeOneDriveConnector.id);
expect(refreshed?.connector_type).toBe("ONEDRIVE_CONNECTOR");
@ -124,6 +130,16 @@ test.describe("Native OneDrive journey", () => {
"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);
});
});