klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-08 05:39:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove the 1000 files limit when syncing from Desktop, Obsidian clients (#605)

Browse Source 
### Major
- Push 1000 files at a time from the Desktop client for indexing
- Push 1000 files at a time from the Obsidian client for indexing
- Test 1000 file upload limit to index/update API endpoint

### Minor
- Show relevant error message in desktop app, e.g when can't connect to server
- Pass indexed filenames in API response for client validation
- Collect files to index in single dict to simplify index/update controller

Resolves #573
This commit is contained in:
Debanjum
2024-01-16 17:59:26 +05:30
committed by GitHub
parent c24389cff5 6ded4c1d75
commit 51c59d0059
5 changed files with 112 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files

85
src/interface/desktop/main.js
View File

@@ -114,8 +114,8 @@ function processDirectory(filesToPush, folder) {
const files = fs.readdirSync(folder.path, { withFileTypes: true, recursive: true }); const files = fs.readdirSync(folder.path, { withFileTypes: true, recursive: true });
for (const file of files) { for (const file of files) {
console.log(file);
if (file.isFile() && validFileTypes.includes(file.name.split('.').pop())) { if (file.isFile() && validFileTypes.includes(file.name.split('.').pop())) {
console.log(`Add ${file.name} in ${folder.path} for indexing`);
filesToPush.push(path.join(folder.path, file.name)); filesToPush.push(path.join(folder.path, file.name));
} }
@@ -151,7 +151,7 @@ function pushDataToKhoj (regenerate = false) {
} }
const lastSync = store.get('lastSync') || []; const lastSync = store.get('lastSync') || [];
const formData = new FormData(); const filesDataToPush = [];
for (const file of filesToPush) { for (const file of filesToPush) {
const stats = fs.statSync(file); const stats = fs.statSync(file);
if (!regenerate) { if (!regenerate) {
@@ -167,7 +167,7 @@ function pushDataToKhoj (regenerate = false) {
let mimeType = filenameToMimeType(file) + (encoding === "utf8" ? "; charset=UTF-8" : ""); let mimeType = filenameToMimeType(file) + (encoding === "utf8" ? "; charset=UTF-8" : "");
let fileContent = Buffer.from(fs.readFileSync(file, { encoding: encoding }), encoding); let fileContent = Buffer.from(fs.readFileSync(file, { encoding: encoding }), encoding);
let fileObj = new Blob([fileContent], { type: mimeType }); let fileObj = new Blob([fileContent], { type: mimeType });
formData.append('files', fileObj, file); filesDataToPush.push({blob: fileObj, path: file});
state[file] = { state[file] = {
success: true, success: true,
} }
@@ -184,52 +184,51 @@ function pushDataToKhoj (regenerate = false) {
for (const syncedFile of lastSync) { for (const syncedFile of lastSync) {
if (!filesToPush.includes(syncedFile.path)) { if (!filesToPush.includes(syncedFile.path)) {
fileObj = new Blob([""], { type: filenameToMimeType(syncedFile.path) }); fileObj = new Blob([""], { type: filenameToMimeType(syncedFile.path) });
formData.append('files', fileObj, syncedFile.path); filesDataToPush.push({blob: fileObj, path: syncedFile.path});
} }
} }
// Send collected files to Khoj server for indexing // Send collected files to Khoj server for indexing
if (!!formData?.entries()?.next().value) { const hostURL = store.get('hostURL') || KHOJ_URL;
const hostURL = store.get('hostURL') || KHOJ_URL; const headers = { 'Authorization': `Bearer ${store.get("khojToken")}` };
const headers = { let requests = [];
'Authorization': `Bearer ${store.get("khojToken")}`
}; // Request indexing files on server. With upto 1000 files in each request
axios.post(`${hostURL}/api/v1/index/update?force=${regenerate}&client=desktop`, formData, { headers }) for (let i = 0; i < filesDataToPush.length; i += 1000) {
.then(response => { const filesDataGroup = filesDataToPush.slice(i, i + 1000);
console.log(response.data); const formData = new FormData();
let lastSync = []; filesDataGroup.forEach(fileData => { formData.append('files', fileData.blob, fileData.path) });
for (const file of filesToPush) { let request = axios.post(`${hostURL}/api/v1/index/update?force=${regenerate}&client=desktop`, formData, { headers });
lastSync.push({ requests.push(request);
path: file, }
datetime: new Date().toISOString()
}); // Wait for requests batch to finish
} Promise
store.set('lastSync', lastSync); .all(requests)
}) .then(responses => {
.catch(error => { const lastSync = filesToPush
console.error(error); .filter(file => responses.find(response => response.data.includes(file)))
if (error.code == 'ECONNREFUSED') { .map(file => ({ path: file, datetime: new Date().toISOString() }));
const win = BrowserWindow.getAllWindows()[0]; store.set('lastSync', lastSync);
if (win) win.webContents.send('update-state', state); })
} else if (error.response.status == 429) { .catch(error => {
const win = BrowserWindow.getAllWindows()[0]; console.error(error);
if (win) win.webContents.send('needsSubscription', true); state["completed"] = false;
if (win) win.webContents.send('update-state', state); if (error?.response?.status === 429 && (win = BrowserWindow.getAllWindows()[0])) {
} state["error"] = `Looks like you're out of space to sync your files. <a href="https://app.khoj.dev/config">Upgrade your plan</a> to unlock more space.`;
state['completed'] = false } else if (error?.code === 'ECONNREFUSED') {
}) state["error"] = `Could not connect to Khoj server. Ensure you can connect to it at ${error.address}:${error.port}.`;
.finally(() => { } else {
// Syncing complete state["error"] = `Sync was unsuccessful at ${currentTime.toLocaleTimeString()}. Contact team@khoj.dev to report this issue.`;
syncing = false; }
const win = BrowserWindow.getAllWindows()[0]; })
if (win) win.webContents.send('update-state', state); .finally(() => {
});
} else {
// Syncing complete // Syncing complete
syncing = false; syncing = false;
const win = BrowserWindow.getAllWindows()[0]; if (win = BrowserWindow.getAllWindows()[0]) {
if (win) win.webContents.send('update-state', state); win.webContents.send('update-state', state);
} }
});
} }
pushDataToKhoj(); pushDataToKhoj();

2
src/interface/desktop/renderer.js
View File

@@ -158,7 +158,7 @@ window.updateStateAPI.onUpdateState((event, state) => {
nextSyncTime = new Date(); nextSyncTime = new Date();
nextSyncTime.setMinutes(Math.ceil((nextSyncTime.getMinutes() + 1) / 10) * 10); nextSyncTime.setMinutes(Math.ceil((nextSyncTime.getMinutes() + 1) / 10) * 10);
if (state.completed == false) { if (state.completed == false) {
syncStatusElement.innerHTML = `Sync was unsuccessful at ${currentTime.toLocaleTimeString()}. Contact team@khoj.dev to report this issue.`; if (state.error) syncStatusElement.innerHTML = state.error;
return; return;
} }
const options = { hour: '2-digit', minute: '2-digit' }; const options = { hour: '2-digit', minute: '2-digit' };

51
src/interface/obsidian/src/utils.ts
View File

@@ -31,40 +31,59 @@ function fileExtensionToMimeType (extension: string): string {
export async function updateContentIndex(vault: Vault, setting: KhojSetting, lastSyncedFiles: TFile[], regenerate: boolean = false): Promise<TFile[]> { export async function updateContentIndex(vault: Vault, setting: KhojSetting, lastSyncedFiles: TFile[], regenerate: boolean = false): Promise<TFile[]> {
// Get all markdown, pdf files in the vault // Get all markdown, pdf files in the vault
console.log(`Khoj: Updating Khoj content index...`) console.log(`Khoj: Updating Khoj content index...`)
const files = vault.getFiles().filter(file => file.extension === 'md' || file.extension === 'pdf'); const files = vault.getFiles().filter(file => file.extension === 'md' || file.extension === 'markdown' || file.extension === 'pdf');
const binaryFileTypes = ['pdf', 'png', 'jpg', 'jpeg'] const binaryFileTypes = ['pdf']
let countOfFilesToIndex = 0; let countOfFilesToIndex = 0;
let countOfFilesToDelete = 0; let countOfFilesToDelete = 0;
// Add all files to index as multipart form data // Add all files to index as multipart form data
const formData = new FormData(); const fileData = [];
for (const file of files) { for (const file of files) {
countOfFilesToIndex++; countOfFilesToIndex++;
const encoding = binaryFileTypes.includes(file.extension) ? "binary" : "utf8"; const encoding = binaryFileTypes.includes(file.extension) ? "binary" : "utf8";
const mimeType = fileExtensionToMimeType(file.extension) + (encoding === "utf8" ? "; charset=UTF-8" : ""); const mimeType = fileExtensionToMimeType(file.extension) + (encoding === "utf8" ? "; charset=UTF-8" : "");
const fileContent = encoding == 'binary' ? await vault.readBinary(file) : await vault.read(file); const fileContent = encoding == 'binary' ? await vault.readBinary(file) : await vault.read(file);
formData.append('files', new Blob([fileContent], { type: mimeType }), file.path); fileData.push({blob: new Blob([fileContent], { type: mimeType }), path: file.path});
} }
// Add any previously synced files to be deleted to multipart form data // Add any previously synced files to be deleted to multipart form data
for (const lastSyncedFile of lastSyncedFiles) { for (const lastSyncedFile of lastSyncedFiles) {
if (!files.includes(lastSyncedFile)) { if (!files.includes(lastSyncedFile)) {
countOfFilesToDelete++; countOfFilesToDelete++;
formData.append('files', new Blob([]), lastSyncedFile.path); fileData.push({blob: new Blob([]), path: lastSyncedFile.path});
} }
} }
// Call Khoj backend to update index with all markdown, pdf files // Iterate through all indexable files in vault, 1000 at a time
const response = await fetch(`${setting.khojUrl}/api/v1/index/update?force=${regenerate}&client=obsidian`, { let error_message = null;
method: 'POST', for (let i = 0; i < fileData.length; i += 1000) {
headers: { const filesGroup = fileData.slice(i, i + 1000);
'Authorization': `Bearer ${setting.khojApiKey}`, const formData = new FormData();
}, filesGroup.forEach(fileItem => { formData.append('files', fileItem.blob, fileItem.path) });
body: formData, // Call Khoj backend to update index with all markdown, pdf files
}); const response = await fetch(`${setting.khojUrl}/api/v1/index/update?force=${regenerate}&client=obsidian`, {
method: 'POST',
headers: {
'Authorization': `Bearer ${setting.khojApiKey}`,
},
body: formData,
});
if (!response.ok) { if (!response.ok) {
new Notice(`❗️Failed to update Khoj content index. Ensure Khoj server connected or raise issue on Khoj Discord/Github\nError: ${response.statusText}`); if (response.status === 429) {
error_message = `❗️Failed to sync your content with Khoj server. Requests were throttled. Upgrade your subscription or try again later.`;
break;
} else if (response.status === 404) {
error_message = `❗️Could not connect to Khoj server. Ensure you can connect to it.`;
break;
} else {
error_message = `❗️Failed to sync your content with Khoj server. Raise issue on Khoj Discord or Github\nError: ${response.statusText}`;
}
}
}
if (error_message) {
new Notice(error_message);
} else { } else {
console.log(`✅ Refreshed Khoj content index. Updated: ${countOfFilesToIndex} files, Deleted: ${countOfFilesToDelete} files.`); console.log(`✅ Refreshed Khoj content index. Updated: ${countOfFilesToIndex} files, Deleted: ${countOfFilesToDelete} files.`);
} }
@@ -92,7 +111,7 @@ export async function createNote(name: string, newLeaf = false): Promise<void> {
console.error('Khoj: Could not create note.\n' + (e as any).message); console.error('Khoj: Could not create note.\n' + (e as any).message);
throw e throw e
} }
} }
export async function createNoteAndCloseModal(query: string, modal: Modal, opt?: { newLeaf: boolean }): Promise<void> { export async function createNoteAndCloseModal(query: string, modal: Modal, opt?: { newLeaf: boolean }): Promise<void> {
try { try {

39
src/khoj/routers/indexer.py
View File

@@ -63,37 +63,23 @@ async def update(
), ),
): ):
user = request.user.object user = request.user.object
index_files: Dict[str, Dict[str, str]] = {"org": {}, "markdown": {}, "pdf": {}, "plaintext": {}}
try: try:
logger.info(f"📬 Updating content index via API call by {client} client") logger.info(f"📬 Updating content index via API call by {client} client")
org_files: Dict[str, str] = {}
markdown_files: Dict[str, str] = {}
pdf_files: Dict[str, bytes] = {}
plaintext_files: Dict[str, str] = {}
for file in files: for file in files:
file_type, encoding = get_file_type(file.content_type) file_type, encoding = get_file_type(file.content_type)
dict_to_update = None if file_type in index_files:
if file_type == "org": index_files[file_type][file.filename] = (
dict_to_update = org_files
elif file_type == "markdown":
dict_to_update = markdown_files
elif file_type == "pdf":
dict_to_update = pdf_files # type: ignore
elif file_type == "plaintext":
dict_to_update = plaintext_files
if dict_to_update is not None:
dict_to_update[file.filename] = (
file.file.read().decode("utf-8") if encoding == "utf-8" else file.file.read() # type: ignore file.file.read().decode("utf-8") if encoding == "utf-8" else file.file.read() # type: ignore
) )
else: else:
logger.warning(f"Skipped indexing unsupported file type sent by {client} client: {file.filename}") logger.warning(f"Skipped indexing unsupported file type sent by {client} client: {file.filename}")
indexer_input = IndexerInput( indexer_input = IndexerInput(
org=org_files, org=index_files["org"],
markdown=markdown_files, markdown=index_files["markdown"],
pdf=pdf_files, pdf=index_files["pdf"],
plaintext=plaintext_files, plaintext=index_files["plaintext"],
) )
if state.config == None: if state.config == None:
@@ -143,10 +129,10 @@ async def update(
return Response(content="Failed", status_code=500) return Response(content="Failed", status_code=500)
indexing_metadata = { indexing_metadata = {
"num_org": len(org_files), "num_org": len(index_files["org"]),
"num_markdown": len(markdown_files), "num_markdown": len(index_files["markdown"]),
"num_pdf": len(pdf_files), "num_pdf": len(index_files["pdf"]),
"num_plaintext": len(plaintext_files), "num_plaintext": len(index_files["plaintext"]),
} }
update_telemetry_state( update_telemetry_state(
@@ -162,7 +148,8 @@ async def update(
logger.info(f"📪 Content index updated via API call by {client} client") logger.info(f"📪 Content index updated via API call by {client} client")
return Response(content="OK", status_code=200) indexed_filenames = ",".join(file for ctype in index_files for file in index_files[ctype]) or ""
return Response(content=indexed_filenames, status_code=200)
def configure_search(search_models: SearchModels, search_config: Optional[SearchConfig]) -> Optional[SearchModels]: def configure_search(search_models: SearchModels, search_config: Optional[SearchConfig]) -> Optional[SearchModels]:

21
tests/test_client.py
View File

@@ -169,6 +169,7 @@ def test_index_update_normal_file_unsubscribed(client, api_user4: KhojApiUser):
assert response.status_code == 200 assert response.status_code == 200
# ----------------------------------------------------------------------------------------------------
@pytest.mark.django_db(transaction=True) @pytest.mark.django_db(transaction=True)
def test_index_update_big_files_no_billing(client): def test_index_update_big_files_no_billing(client):
# Arrange # Arrange
@@ -197,6 +198,26 @@ def test_index_update(client):
assert response.status_code == 200 assert response.status_code == 200
# ----------------------------------------------------------------------------------------------------
@pytest.mark.django_db(transaction=True)
def test_index_update_fails_if_more_than_1000_files(client, api_user4: KhojApiUser):
# Arrange
api_token = api_user4.token
state.billing_enabled = True
files = [("files", (f"path/to/filename{i}.org", f"Symphony No {i}", "text/org")) for i in range(1001)]
headers = {"Authorization": f"Bearer {api_token}"}
# Act
ok_response = client.post("/api/v1/index/update", files=files[:1000], headers=headers)
bad_response = client.post("/api/v1/index/update", files=files, headers=headers)
# Assert
assert ok_response.status_code == 200
assert bad_response.status_code == 400
assert bad_response.content.decode("utf-8") == '{"detail":"Too many files. Maximum number of files is 1000."}'
# ---------------------------------------------------------------------------------------------------- # ----------------------------------------------------------------------------------------------------
@pytest.mark.django_db(transaction=True) @pytest.mark.django_db(transaction=True)
def test_regenerate_with_valid_content_type(client): def test_regenerate_with_valid_content_type(client):