perf: use lazy load provider for IDB and SQLITE (#3351)

This commit is contained in:
Peng Xiao
2023-07-26 00:56:48 +08:00
committed by GitHub
parent e3f66d7e22
commit 20ee9d485d
25 changed files with 481 additions and 758 deletions
@@ -3,18 +3,18 @@
*/
import 'fake-indexeddb/auto';
import { setTimeout } from 'node:timers/promises';
import { __unstableSchemas, AffineSchemas } from '@blocksuite/blocks/models';
import { assertExists } from '@blocksuite/global/utils';
import type { Page } from '@blocksuite/store';
import { uuidv4, Workspace } from '@blocksuite/store';
import { openDB } from 'idb';
import { afterEach, beforeEach, describe, expect, test, vi } from 'vitest';
import { IndexeddbPersistence } from 'y-indexeddb';
import { applyUpdate, Doc, encodeStateAsUpdate } from 'yjs';
import type { WorkspacePersist } from '../index';
import {
CleanupWhenConnectingError,
createIndexedDBProvider,
dbVersion,
DEFAULT_DB_NAME,
@@ -43,7 +43,7 @@ function initEmptyPage(page: Page) {
async function getUpdates(id: string): Promise<Uint8Array[]> {
const db = await openDB(rootDBName, dbVersion);
const store = await db
const store = db
.transaction('workspace', 'readonly')
.objectStore('workspace');
const data = (await store.get(id)) as WorkspacePersist | undefined;
@@ -74,7 +74,10 @@ describe('indexeddb provider', () => {
test('connect', async () => {
const provider = createIndexedDBProvider(workspace.doc);
provider.connect();
await provider.whenSynced;
// todo: has a better way to know when data is synced
await setTimeout(200);
const db = await openDB(rootDBName, dbVersion);
{
const store = db
@@ -96,9 +99,9 @@ describe('indexeddb provider', () => {
const frameId = page.addBlock('affine:note', {}, pageBlockId);
page.addBlock('affine:paragraph', {}, frameId);
}
await new Promise(resolve => setTimeout(resolve, 1000));
await setTimeout(200);
{
const store = await db
const store = db
.transaction('workspace', 'readonly')
.objectStore('workspace');
const data = (await store.get(id)) as WorkspacePersist | undefined;
@@ -130,22 +133,11 @@ describe('indexeddb provider', () => {
}
});
test('disconnect suddenly', async () => {
const provider = createIndexedDBProvider(workspace.doc, rootDBName);
const fn = vi.fn();
provider.connect();
provider.disconnect();
expect(fn).toBeCalledTimes(0);
await provider.whenSynced.catch(fn);
expect(fn).toBeCalledTimes(1);
});
test('connect and disconnect', async () => {
const provider = createIndexedDBProvider(workspace.doc, rootDBName);
provider.connect();
expect(provider.connected).toBe(true);
const p1 = provider.whenSynced;
await p1;
await setTimeout(200);
const snapshot = encodeStateAsUpdate(workspace.doc);
provider.disconnect();
expect(provider.connected).toBe(false);
@@ -164,8 +156,7 @@ describe('indexeddb provider', () => {
expect(provider.connected).toBe(false);
provider.connect();
expect(provider.connected).toBe(true);
const p2 = provider.whenSynced;
await p2;
await setTimeout(200);
{
const updates = await getUpdates(workspace.id);
expect(updates).not.toEqual([]);
@@ -173,13 +164,12 @@ describe('indexeddb provider', () => {
expect(provider.connected).toBe(true);
provider.disconnect();
expect(provider.connected).toBe(false);
expect(p1).not.toBe(p2);
});
test('cleanup', async () => {
const provider = createIndexedDBProvider(workspace.doc);
provider.connect();
await provider.whenSynced;
await setTimeout(200);
const db = await openDB(rootDBName, dbVersion);
{
@@ -190,8 +180,8 @@ describe('indexeddb provider', () => {
expect(keys).contain(workspace.id);
}
provider.disconnect();
await provider.cleanup();
provider.disconnect();
{
const store = db
@@ -202,17 +192,6 @@ describe('indexeddb provider', () => {
}
});
test('cleanup when connecting', async () => {
const provider = createIndexedDBProvider(workspace.doc);
provider.connect();
await expect(() => provider.cleanup()).rejects.toThrowError(
CleanupWhenConnectingError
);
await provider.whenSynced;
provider.disconnect();
await provider.cleanup();
});
test('merge', async () => {
setMergeCount(5);
const provider = createIndexedDBProvider(workspace.doc, rootDBName);
@@ -226,7 +205,7 @@ describe('indexeddb provider', () => {
page.addBlock('affine:paragraph', {}, frameId);
}
}
await provider.whenSynced;
await setTimeout(200);
{
const updates = await getUpdates(id);
expect(updates.length).lessThanOrEqual(5);
@@ -242,14 +221,12 @@ describe('indexeddb provider', () => {
{
const provider = createIndexedDBProvider(doc, rootDBName);
provider.connect();
await provider.whenSynced;
provider.disconnect();
}
{
const newDoc = new Workspace.Y.Doc();
const provider = createIndexedDBProvider(newDoc, rootDBName);
provider.connect();
await provider.whenSynced;
provider.disconnect();
newDoc.getMap('map').forEach((value, key) => {
expect(value).toBe(parseInt(key));
@@ -257,42 +234,6 @@ describe('indexeddb provider', () => {
}
});
test('migration', async () => {
{
const yDoc = new Doc();
yDoc.getMap().set('foo', 'bar');
const persistence = new IndexeddbPersistence('test', yDoc);
await persistence.whenSynced;
await persistence.destroy();
}
{
const yDoc = new Doc({
guid: 'test',
});
const provider = createIndexedDBProvider(yDoc);
provider.connect();
await provider.whenSynced;
await new Promise(resolve => setTimeout(resolve, 0));
expect(yDoc.getMap().get('foo')).toBe('bar');
}
localStorage.clear();
{
indexedDB.databases = vi.fn(async () => {
throw new Error('not supported');
});
await expect(indexedDB.databases).rejects.toThrow('not supported');
const yDoc = new Doc({
guid: 'test',
});
expect(indexedDB.databases).toBeCalledTimes(1);
const provider = createIndexedDBProvider(yDoc);
provider.connect();
await provider.whenSynced;
expect(indexedDB.databases).toBeCalledTimes(2);
expect(yDoc.getMap().get('foo')).toBe('bar');
}
});
test('beforeunload', async () => {
const oldAddEventListener = window.addEventListener;
window.addEventListener = vi.fn((event: string, fn, options) => {
@@ -311,7 +252,8 @@ describe('indexeddb provider', () => {
const map = doc.getMap('map');
map.set('1', 1);
provider.connect();
await provider.whenSynced;
await setTimeout(200);
expect(window.addEventListener).toBeCalledTimes(1);
expect(window.removeEventListener).toBeCalledTimes(1);
@@ -396,7 +338,7 @@ describe('subDoc', () => {
map.set('2', 'test');
const provider = createIndexedDBProvider(doc);
provider.connect();
await provider.whenSynced;
await setTimeout(200);
provider.disconnect();
json1 = doc.toJSON();
}
@@ -406,7 +348,7 @@ describe('subDoc', () => {
});
const provider = createIndexedDBProvider(doc);
provider.connect();
await provider.whenSynced;
await setTimeout(200);
const map = doc.getMap();
const subDoc = map.get('1') as Doc;
subDoc.load();
@@ -431,7 +373,7 @@ describe('subDoc', () => {
});
await page1.waitForLoaded();
const { paragraphBlockId: paragraphBlockIdPage2 } = initEmptyPage(page1);
await new Promise(resolve => setTimeout(resolve, 1000));
await setTimeout(200);
provider.disconnect();
{
const newWorkspace = new Workspace({
@@ -441,15 +383,17 @@ describe('subDoc', () => {
newWorkspace.register(AffineSchemas).register(__unstableSchemas);
const provider = createIndexedDBProvider(newWorkspace.doc, rootDBName);
provider.connect();
await provider.whenSynced;
await setTimeout(200);
const page0 = newWorkspace.getPage('page0') as Page;
await page0.waitForLoaded();
await setTimeout(200);
{
const block = page0.getBlockById(paragraphBlockIdPage1);
assertExists(block);
}
const page1 = newWorkspace.getPage('page1') as Page;
await page1.waitForLoaded();
await setTimeout(200);
{
const block = page1.getBlockById(paragraphBlockIdPage2);
assertExists(block);
@@ -465,7 +409,7 @@ describe('utils', () => {
initEmptyPage(page);
const provider = createIndexedDBProvider(workspace.doc, rootDBName);
provider.connect();
await provider.whenSynced;
await setTimeout(200);
provider.disconnect();
const update = (await downloadBinary(
workspace.id,
@@ -478,16 +422,12 @@ describe('utils', () => {
});
newWorkspace.register(AffineSchemas).register(__unstableSchemas);
applyUpdate(newWorkspace.doc, update);
await new Promise<void>(resolve =>
setTimeout(() => {
expect(workspace.doc.toJSON()['meta']).toEqual(
newWorkspace.doc.toJSON()['meta']
);
expect(Object.keys(workspace.doc.toJSON()['spaces'])).toEqual(
Object.keys(newWorkspace.doc.toJSON()['spaces'])
);
resolve();
}, 0)
await setTimeout();
expect(workspace.doc.toJSON()['meta']).toEqual(
newWorkspace.doc.toJSON()['meta']
);
expect(Object.keys(workspace.doc.toJSON()['spaces'])).toEqual(
Object.keys(newWorkspace.doc.toJSON()['spaces'])
);
});
+2 -262
View File
@@ -1,26 +1,17 @@
import { openDB } from 'idb';
import {
applyUpdate,
diffUpdate,
Doc,
encodeStateAsUpdate,
encodeStateVector,
UndoManager,
} from 'yjs';
import type {
BlockSuiteBinaryDB,
IndexedDBProvider,
WorkspaceMilestone,
} from './shared';
import type { BlockSuiteBinaryDB, WorkspaceMilestone } from './shared';
import { dbVersion, DEFAULT_DB_NAME, upgradeDB } from './shared';
import { tryMigrate } from './utils';
const indexeddbOrigin = 'indexeddb-provider-origin';
const snapshotOrigin = 'snapshot-origin';
let mergeCount = 500;
/**
* @internal
*/
@@ -40,10 +31,6 @@ export const writeOperation = async (op: Promise<unknown>) => {
});
};
export function setMergeCount(count: number) {
mergeCount = count;
}
export function revertUpdate(
doc: Doc,
snapshotUpdate: Uint8Array,
@@ -142,257 +129,10 @@ export const getMilestones = async (
return milestone.milestone;
};
type SubDocsEvent = {
added: Set<Doc>;
removed: Set<Doc>;
loaded: Set<Doc>;
};
/**
* We use `doc.guid` as the unique key, please make sure it not changes.
*/
export const createIndexedDBProvider = (
doc: Doc,
dbName: string = DEFAULT_DB_NAME,
/**
* In the future, migrate will be removed and there will be a separate function
*/
migrate = true
): IndexedDBProvider => {
let resolve: () => void;
let reject: (reason?: unknown) => void;
let early = true;
let connected = false;
const dbPromise = openDB<BlockSuiteBinaryDB>(dbName, dbVersion, {
upgrade: upgradeDB,
});
const updateHandlerMap = new WeakMap<
Doc,
(update: Uint8Array, origin: unknown) => void
>();
const destroyHandlerMap = new WeakMap<Doc, () => void>();
const subDocsHandlerMap = new WeakMap<Doc, (event: SubDocsEvent) => void>();
const createOrGetHandleUpdate = (id: string, doc: Doc) => {
if (updateHandlerMap.has(doc)) {
// eslint-disable-next-line @typescript-eslint/no-non-null-assertion
return updateHandlerMap.get(doc)!;
}
const fn = async function handleUpdate(
update: Uint8Array,
origin: unknown
) {
const db = await dbPromise;
if (!connected) {
return;
}
if (origin === indexeddbOrigin) {
return;
}
const store = db
.transaction('workspace', 'readwrite')
.objectStore('workspace');
let data = await store.get(id);
if (!data) {
data = {
id,
updates: [],
};
}
data.updates.push({
timestamp: Date.now(),
update,
});
if (data.updates.length > mergeCount) {
const updates = data.updates.map(({ update }) => update);
const doc = new Doc();
doc.transact(() => {
updates.forEach(update => {
applyUpdate(doc, update, indexeddbOrigin);
});
}, indexeddbOrigin);
const update = encodeStateAsUpdate(doc);
data = {
id,
updates: [
{
timestamp: Date.now(),
update,
},
],
};
await writeOperation(store.put(data));
} else {
await writeOperation(store.put(data));
}
};
updateHandlerMap.set(doc, fn);
return fn;
};
/* deepscan-disable UNUSED_PARAM */
const createOrGetHandleDestroy = (_: string, doc: Doc) => {
if (destroyHandlerMap.has(doc)) {
// eslint-disable-next-line @typescript-eslint/no-non-null-assertion
return destroyHandlerMap.get(doc)!;
}
const fn = async function handleDestroy() {
unTrackDoc(doc.guid, doc);
};
destroyHandlerMap.set(doc, fn);
return fn;
};
/* deepscan-disable UNUSED_PARAM */
const createOrGetHandleSubDocs = (_: string, doc: Doc) => {
if (subDocsHandlerMap.has(doc)) {
// eslint-disable-next-line @typescript-eslint/no-non-null-assertion
return subDocsHandlerMap.get(doc)!;
}
const fn = async function handleSubDocs(event: SubDocsEvent) {
event.removed.forEach(doc => {
unTrackDoc(doc.guid, doc);
});
event.loaded.forEach(doc => {
trackDoc(doc.guid, doc);
});
};
subDocsHandlerMap.set(doc, fn);
return fn;
};
function trackDoc(id: string, doc: Doc) {
doc.on('update', createOrGetHandleUpdate(id, doc));
doc.on('destroy', createOrGetHandleDestroy(id, doc));
doc.on('subdocs', createOrGetHandleSubDocs(id, doc));
doc.subdocs.forEach(doc => {
trackDoc(doc.guid, doc);
});
}
function unTrackDoc(id: string, doc: Doc) {
doc.subdocs.forEach(doc => {
unTrackDoc(doc.guid, doc);
});
doc.off('update', createOrGetHandleUpdate(id, doc));
doc.off('destroy', createOrGetHandleDestroy(id, doc));
doc.off('subdocs', createOrGetHandleSubDocs(id, doc));
}
async function saveDocOperation(id: string, doc: Doc) {
const db = await dbPromise;
const store = db
.transaction('workspace', 'readwrite')
.objectStore('workspace');
const data = await store.get(id);
if (!connected) {
return;
}
if (!data) {
await writeOperation(
db.put('workspace', {
id,
updates: [
{
timestamp: Date.now(),
update: encodeStateAsUpdate(doc),
},
],
})
);
} else {
const updates = data.updates.map(({ update }) => update);
const fakeDoc = new Doc();
fakeDoc.transact(() => {
updates.forEach(update => {
applyUpdate(fakeDoc, update, indexeddbOrigin);
});
}, indexeddbOrigin);
const newUpdate = diffUpdate(
encodeStateAsUpdate(doc),
encodeStateAsUpdate(fakeDoc)
);
await writeOperation(
store.put({
...data,
updates: [
...data.updates,
{
timestamp: Date.now(),
update: newUpdate,
},
],
})
);
doc.transact(() => {
updates.forEach(update => {
applyUpdate(doc, update, indexeddbOrigin);
});
}, indexeddbOrigin);
}
}
const apis = {
connect: async () => {
if (connected) return;
apis.whenSynced = new Promise<void>((_resolve, _reject) => {
early = true;
resolve = _resolve;
reject = _reject;
});
connected = true;
trackDoc(doc.guid, doc);
// only the runs `await` below, otherwise the logic is incorrect
const db = await dbPromise;
if (migrate) {
// Tips:
// this is only backward compatible with the yjs official version of y-indexeddb
await tryMigrate(db, doc.guid, dbName);
}
if (!connected) {
return;
}
// recursively save all docs into indexeddb
const docs: [string, Doc][] = [];
docs.push([doc.guid, doc]);
while (docs.length > 0) {
const [id, doc] = docs.pop() as [string, Doc];
await saveDocOperation(id, doc);
doc.subdocs.forEach(doc => {
docs.push([doc.guid, doc]);
});
}
early = false;
resolve();
},
disconnect() {
connected = false;
if (early) {
reject(new EarlyDisconnectError());
}
unTrackDoc(doc.guid, doc);
},
async cleanup() {
if (connected) {
throw new CleanupWhenConnectingError();
}
await (await dbPromise).delete('workspace', doc.guid);
},
whenSynced: Promise.resolve(),
get connected() {
return connected;
},
};
return apis;
};
export * from './provider';
export * from './shared';
export * from './utils';
+133
View File
@@ -0,0 +1,133 @@
import {
createLazyProvider,
type DatasourceDocAdapter,
writeOperation,
} from '@affine/y-provider';
import { openDB } from 'idb';
import type { Doc } from 'yjs';
import { diffUpdate, mergeUpdates } from 'yjs';
import {
type BlockSuiteBinaryDB,
dbVersion,
DEFAULT_DB_NAME,
type IndexedDBProvider,
type UpdateMessage,
upgradeDB,
} from './shared';
let mergeCount = 500;
export function setMergeCount(count: number) {
mergeCount = count;
}
const createDatasource = ({
dbName,
mergeCount,
}: {
dbName: string;
mergeCount?: number;
}) => {
const dbPromise = openDB<BlockSuiteBinaryDB>(dbName, dbVersion, {
upgrade: upgradeDB,
});
const adapter = {
queryDocState: async (guid, options) => {
try {
const db = await dbPromise;
const store = db
.transaction('workspace', 'readonly')
.objectStore('workspace');
const data = await store.get(guid);
if (!data) {
return false;
}
const { updates } = data;
const update = mergeUpdates(updates.map(({ update }) => update));
const diff = options?.stateVector
? diffUpdate(update, options?.stateVector)
: update;
return diff;
} catch (err: any) {
if (!err.message?.includes('The database connection is closing.')) {
throw err;
}
return false;
}
},
sendDocUpdate: async (guid, update) => {
try {
const db = await dbPromise;
const store = db
.transaction('workspace', 'readwrite')
.objectStore('workspace');
// TODO: maybe we do not need to get data every time
const { updates } = (await store.get(guid)) ?? { updates: [] };
let rows: UpdateMessage[] = [
...updates,
{ timestamp: Date.now(), update },
];
if (mergeCount && rows.length >= mergeCount) {
const merged = mergeUpdates(rows.map(({ update }) => update));
rows = [{ timestamp: Date.now(), update: merged }];
}
await writeOperation(
store.put({
id: guid,
updates: rows,
})
);
} catch (err: any) {
if (!err.message?.includes('The database connection is closing.')) {
throw err;
}
}
},
} satisfies DatasourceDocAdapter;
return {
...adapter,
disconnect: () => {
dbPromise.then(db => db.close()).catch(console.error);
},
cleanup: async () => {
const db = await dbPromise;
await db.clear('workspace');
},
};
};
export const createIndexedDBProvider = (
doc: Doc,
dbName: string = DEFAULT_DB_NAME
): IndexedDBProvider => {
let datasource: ReturnType<typeof createDatasource> | null = null;
let provider: ReturnType<typeof createLazyProvider> | null = null;
return {
connect: () => {
datasource = createDatasource({ dbName, mergeCount });
provider = createLazyProvider(doc, datasource);
provider.connect();
},
disconnect: () => {
datasource?.disconnect();
provider?.disconnect();
datasource = null;
provider = null;
},
cleanup: async () => {
await datasource?.cleanup();
},
get connected() {
return provider?.connected || false;
},
};
};
-1
View File
@@ -12,7 +12,6 @@ export interface IndexedDBProvider {
connect: () => void;
disconnect: () => void;
cleanup: () => Promise<void>;
whenSynced: Promise<void>;
readonly connected: boolean;
}