You've already forked joplin
mirror of
https://github.com/laurent22/joplin.git
synced 2026-06-03 16:53:41 +02:00
46 lines
1.6 KiB
TypeScript
46 lines
1.6 KiB
TypeScript
import packToWriter, { FileApi } from './packToWriter';
|
|
|
|
describe('htmlpack/packToWriter', () => {
|
|
|
|
test('should write chunks as they arrive, not buffer them in memory', async () => {
|
|
// packToWriter replaced packToString to pack notes with very large files without
|
|
// loading them fully into memory. This was the OOM bug from issue #13903.
|
|
//
|
|
// The tests in index.test.ts use small files, so they cannot catch this bug.
|
|
// A bad version that keeps all chunks in memory and writes them at the end would
|
|
// still produce the correct output and pass those tests. But it would cause OOM
|
|
// again on a large file.
|
|
//
|
|
// This test checks the real behavior: streamFileDataUri sends chunks one by one,
|
|
// and after each chunk it checks that writeChunk was already called. If the code
|
|
// was keeping all chunks in memory, writeChunk would not run until the end, so
|
|
// this test would fail on the first chunk.
|
|
|
|
const writtenChunks: string[] = [];
|
|
let onChunkCalls = 0;
|
|
|
|
const fs: FileApi = {
|
|
exists: async () => true,
|
|
readFileText: async () => '',
|
|
readFileDataUri: async () => { throw new Error('readFileDataUri should not be called'); },
|
|
streamFileDataUri: async (_path, onChunk) => {
|
|
for (let i = 0; i < 5; i++) {
|
|
const writesBefore = writtenChunks.length;
|
|
onChunkCalls++;
|
|
await onChunk(`CHUNK_${i}`);
|
|
expect(writtenChunks.length).toBeGreaterThan(writesBefore);
|
|
expect(writtenChunks).toContain(`CHUNK_${i}`);
|
|
}
|
|
},
|
|
writeChunk: (chunk: string) => {
|
|
writtenChunks.push(chunk);
|
|
},
|
|
};
|
|
|
|
await packToWriter('/base', '<a href="big.bin">d</a>', fs);
|
|
|
|
expect(onChunkCalls).toBe(5);
|
|
});
|
|
|
|
});
|