Zstandard

README.md

@@ -26,9 +26,9 @@ Parquet compression types supported with `hyparquet-compressors`:
  - [X] Snappy
  - [x] GZip
  - [ ] LZO
- - [ ] Brotli
+ - [X] Brotli
  - [X] LZ4
- - [ ] ZSTD
+ - [X] ZSTD
  - [X] LZ4_RAW
 
 # References

package.json

@@ -27,6 +27,7 @@
   },
   "dependencies": {
     "brotli": "1.3.3",
+    "fzstd": "0.1.1",
     "hysnappy": "0.3.1",
     "pako": "2.1.0"
   },

src/index.js

@@ -1,5 +1,6 @@
 // @ts-ignore
 import BROTLI from 'brotli/decompress'
+import { decompress as ZSTD } from 'fzstd'
 import { snappyUncompressor } from 'hysnappy'
 import pako from 'pako'
 import { LZ4, LZ4_RAW } from './lz4.js'
@@ -11,7 +12,7 @@ export const compressors = {
   SNAPPY: snappyUncompressor(),
   GZIP: input => pako.ungzip(input),
   BROTLI,
-  ZSTD: () => new Uint8Array(), // TODO
+  ZSTD: input => ZSTD(input),
   LZ4,
   LZ4_RAW,
 }

test/zstd.test.js

@@ -0,0 +1,17 @@
+import fs from 'fs'
+import { parquetRead, toJson } from 'hyparquet'
+import { describe, expect, it } from 'vitest'
+import { compressors } from '../src/index.js'
+
+describe('zstd compressor', () => {
+  it('read zstd compressed parquet file wiki_1k', async () => {
+    const buffer = fs.readFileSync('test/files/wiki_1k.zstd.parquet')
+    const file = new Uint8Array(buffer).buffer
+    const expected = fs.readFileSync('test/files/wiki_1k.zstd.json').toString()
+
+    await parquetRead({ file, compressors, onComplete: data => {
+      expect(data.length).toBe(1024)
+      expect(toJson(data)).toEqual(JSON.parse(expected))
+    } })
+  })
+})