fix(KeyValueStore): big buffers should not crash (#1734)

Closes #1732 
Closes #1710 

Simple fix really, tell shapeshift inputs can be buffers too (I'll also
see if we can add a generic "just check that its an object" in
shapeshift for times like these)

package.json

@@ -43,6 +43,7 @@
         "postci:build": "node ./scripts/typescript_fixes.mjs",
         "test": "jest --silent",
         "test:e2e": "node test/e2e/run.mjs",
+        "test:full": "cross-env CRAWLEE_DIFFICULT_TESTS=1 jest --silent",
         "coverage": "jest --coverage",
         "publish:next": "lerna publish from-package --contents dist --dist-tag next --force-publish",
         "release:next": "npm run build && npm run publish:next",
@@ -80,6 +81,7 @@
         "basic-auth-parser": "^0.0.2",
         "body-parser": "^1.20.0",
         "commitlint": "^17.1.2",
+        "cross-env": "^7.0.3",
         "deep-equal": "^2.0.5",
         "eslint": "^8.24.0",
         "express": "^4.18.1",

packages/memory-storage/src/resource-clients/key-value-store.ts

@@ -211,7 +211,7 @@ export class KeyValueStoreClient extends BaseClient {
     async setRecord(record: storage.KeyValueStoreRecord): Promise<void> {
         s.object({
             key: s.string.lengthGreaterThan(0),
-            value: s.union(s.null, s.string, s.number, s.object({}).passthrough),
+            value: s.union(s.null, s.string, s.number, s.instance(Buffer), s.object({}).passthrough),
             contentType: s.string.lengthGreaterThan(0).optional,
         }).parse(record);
 

packages/memory-storage/test/no-crash-on-big-buffers.test.ts

@@ -0,0 +1,45 @@
+// https://github.com/apify/crawlee/issues/1732
+// https://github.com/apify/crawlee/issues/1710
+
+import { rm } from 'node:fs/promises';
+import { resolve } from 'node:path';
+import { MemoryStorage } from '@crawlee/memory-storage';
+import type { KeyValueStoreClient, KeyValueStoreInfo } from '@crawlee/types';
+
+describe('MemoryStorage should not crash when saving a big buffer', () => {
+    const tmpLocation = resolve(__dirname, './tmp/no-buffer-crash');
+    const storage = new MemoryStorage({
+        localDataDirectory: tmpLocation,
+        persistStorage: false,
+    });
+
+    let kvs: KeyValueStoreInfo;
+    let store: KeyValueStoreClient;
+
+    beforeAll(async () => {
+        kvs = await storage.keyValueStores().getOrCreate();
+        store = storage.keyValueStore(kvs.id);
+    });
+
+    afterAll(async () => {
+        await rm(tmpLocation, { force: true, recursive: true });
+    });
+
+    test('should not crash when saving a big buffer', async () => {
+        let zip: Buffer;
+
+        if (process.env.CRAWLEE_DIFFICULT_TESTS) {
+            const numbers = Array.from(([...Array(18_100_000).keys()]).map((i) => i * 3_000_000));
+
+            zip = Buffer.from([...numbers]);
+        } else {
+            zip = Buffer.from([...Array(100_000)].map((i) => i * 8));
+        }
+
+        try {
+            await store.setRecord({ key: 'owo.zip', value: zip });
+        } catch (err) {
+            expect(err).not.toBeDefined();
+        }
+    });
+});