From 3066e3b33b2e91cb14a1ff03a5dfd784919ae3ba Mon Sep 17 00:00:00 2001
From: Ib Green <ib@foursquare.com>
Date: Sun, 28 Jan 2024 17:34:26 -0500
Subject: [PATCH] feat(parquet): restore ParquetWasm loader

---
 .../src/lib/get-schema-from-fgb-header.ts     |  1 +
 modules/geotiff/test/tiff.spec.ts             |  2 +-
 modules/parquet/package.json                  |  6 +-
 modules/parquet/src/index.ts                  | 11 ++--
 modules/parquet/src/{ => lib}/constants.ts    |  6 +-
 .../src/lib/wasm/encode-parquet-wasm.ts       | 66 +++++++++++++------
 .../parquet/src/lib/wasm/load-wasm-browser.ts | 19 ------
 .../parquet/src/lib/wasm/load-wasm-node.ts    |  9 ---
 modules/parquet/src/lib/wasm/load-wasm.ts     | 19 +++++-
 .../src/lib/wasm/parse-parquet-wasm.ts        | 44 +++++--------
 modules/parquet/src/parquet-wasm-loader.ts    | 12 ++--
 modules/parquet/src/parquet-wasm-writer.ts    | 20 ++++--
 .../parquet/src/parquetjs/parser/decoders.ts  |  2 +-
 .../src/parquetjs/parser/parquet-reader.ts    |  2 +-
 modules/parquet/test/index.ts                 |  3 +-
 ...s.disabled => parquet-wasm-loader.spec.ts} | 23 ++-----
 modules/zip/src/parse-zip/zip-composition.ts  |  5 ++
 package.json                                  |  6 +-
 yarn.lock                                     |  8 +--
 19 files changed, 137 insertions(+), 127 deletions(-)
 rename modules/parquet/src/{ => lib}/constants.ts (59%)
 delete mode 100644 modules/parquet/src/lib/wasm/load-wasm-browser.ts
 delete mode 100644 modules/parquet/src/lib/wasm/load-wasm-node.ts
 rename modules/parquet/test/{parquet-wasm-loader.spec.ts.disabled => parquet-wasm-loader.spec.ts} (82%)

diff --git a/modules/flatgeobuf/src/lib/get-schema-from-fgb-header.ts b/modules/flatgeobuf/src/lib/get-schema-from-fgb-header.ts
index 17203a9fe6..1df9390f30 100644
--- a/modules/flatgeobuf/src/lib/get-schema-from-fgb-header.ts
+++ b/modules/flatgeobuf/src/lib/get-schema-from-fgb-header.ts
@@ -86,6 +86,7 @@ enum fgbColumnType {
 }
 
 /** Convert FGB types to arrow like types */
+// eslint-disable-next-line complexity
 function getTypeFromFGBType(fgbType: fgbColumnType /* fgb.ColumnMeta['type'] */): DataType {
   switch (fgbType) {
     case fgbColumnType.Byte:
diff --git a/modules/geotiff/test/tiff.spec.ts b/modules/geotiff/test/tiff.spec.ts
index 8d708c53bf..69be96baf9 100644
--- a/modules/geotiff/test/tiff.spec.ts
+++ b/modules/geotiff/test/tiff.spec.ts
@@ -60,7 +60,7 @@ test('Correct OME-XML.', async (t) => {
   const tiff = await fromFile(TIFF_URL);
   const {metadata} = await loadGeoTiff(tiff);
   const {Name, Pixels} = metadata;
-  t.equal(Name, 'multi-channel.ome.tif', "Name should be 'multi-channel.ome.tif'.");
+  t.equal(Name, 'multi-channel.ome.tif', 'Name should be "multi-channel.ome.tif".');
   // @ts-ignore
   t.equal(Pixels.SizeC, 3, 'Should have three channels.');
   // @ts-ignore
diff --git a/modules/parquet/package.json b/modules/parquet/package.json
index 62405a84bb..21123827af 100644
--- a/modules/parquet/package.json
+++ b/modules/parquet/package.json
@@ -53,9 +53,7 @@
     "util": false,
     "events": false,
     "./src/polyfills/buffer/buffer-polyfill.node.ts": "./src/polyfills/buffer/buffer-polyfill.browser.ts",
-    "./dist/polyfills/buffer/buffer-polyfill.node.js": "./dist/polyfills/buffer/buffer-polyfill.browser.js",
-    "./src/lib/wasm/load-wasm-node.ts": "./src/lib/wasm/load-wasm-browser.ts",
-    "./dist/lib/wasm/load-wasm-node.js": "./dist/lib/wasm/load-wasm-browser.js"
+    "./dist/polyfills/buffer/buffer-polyfill.node.js": "./dist/polyfills/buffer/buffer-polyfill.browser.js"
   },
   "comments": [
     "base64-js and ieee754 are used by buffer polyfill"
@@ -76,7 +74,7 @@
     "lz4js": "^0.2.0",
     "node-int64": "^0.4.0",
     "object-stream": "0.0.1",
-    "parquet-wasm": "^0.3.1",
+    "parquet-wasm": "^0.6.0-beta.1",
     "snappyjs": "^0.6.0",
     "thrift": "^0.19.0",
     "util": "^0.12.5",
diff --git a/modules/parquet/src/index.ts b/modules/parquet/src/index.ts
index bd7292cc4f..90983fa5fe 100644
--- a/modules/parquet/src/index.ts
+++ b/modules/parquet/src/index.ts
@@ -15,12 +15,13 @@ export {
   ParquetColumnarLoader
 } from './parquet-loader';
 
-// import type {ParquetWasmLoaderOptions} from './lib/wasm/parse-parquet-wasm';
-// import {parseParquetWasm} from './lib/wasm/parse-parquet-wasm';
-// import {ParquetWasmLoader as ParquetWasmWorkerLoader} from './parquet-wasm-loader';
-
 export {ParquetWriter as _ParquetWriter} from './parquet-writer';
-// export {ParquetWasmWriter} from './parquet-wasm-writer';
+
+// EXPERIMENTAL - expose Parquet WASM loaders/writer
+
+export type {ParquetWasmLoaderOptions} from './parquet-wasm-loader';
+export {ParquetWasmLoader, ParquetWasmWorkerLoader} from './parquet-wasm-loader';
+export {ParquetWasmWriter} from './parquet-wasm-writer';
 
 // EXPERIMENTAL - expose the internal parquetjs API
 
diff --git a/modules/parquet/src/constants.ts b/modules/parquet/src/lib/constants.ts
similarity index 59%
rename from modules/parquet/src/constants.ts
rename to modules/parquet/src/lib/constants.ts
index d93a3ca991..b9861233c7 100644
--- a/modules/parquet/src/constants.ts
+++ b/modules/parquet/src/lib/constants.ts
@@ -2,7 +2,11 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) vis.gl contributors
 
-// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)
+// __VERSION__ is injected by babel-plugin-version-inline
+// @ts-ignore TS2304: Cannot find name '__VERSION__'.
+export const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';
+export const PARQUET_WASM_URL = 'https://unpkg.com/parquet-wasm@0.6.0-beta.1/esm/arrow1_bg.wasm';
+
 /**
  * Parquet File Magic String
  */
diff --git a/modules/parquet/src/lib/wasm/encode-parquet-wasm.ts b/modules/parquet/src/lib/wasm/encode-parquet-wasm.ts
index 7203db757f..28eaca3838 100644
--- a/modules/parquet/src/lib/wasm/encode-parquet-wasm.ts
+++ b/modules/parquet/src/lib/wasm/encode-parquet-wasm.ts
@@ -2,39 +2,67 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) vis.gl contributors
 
-import type {WriterOptions} from '@loaders.gl/loader-utils';
 import type {ArrowTable} from '@loaders.gl/arrow';
 
 import * as arrow from 'apache-arrow';
 import {loadWasm} from './load-wasm';
 
-export type ParquetWriterOptions = WriterOptions & {
-  parquet?: {
-    wasmUrl?: string;
-  };
-};
+import type {ParquetWriterOptions} from '../../parquet-wasm-writer';
 
 /**
  * Encode Arrow arrow.Table to Parquet buffer
  */
 export async function encode(
   table: ArrowTable,
-  options?: ParquetWriterOptions
+  options: ParquetWriterOptions
 ): Promise<ArrayBuffer> {
-  const wasmUrl = options?.parquet?.wasmUrl;
+  const wasmUrl = options.parquet?.wasmUrl!;
   const wasm = await loadWasm(wasmUrl);
 
+  // Serialize the table to the IPC format.
   const arrowTable: arrow.Table = table.data;
+  const ipcStream = arrow.tableToIPC(arrowTable);
 
-  // Serialize a table to the IPC format.
-  const writer = arrow.RecordBatchStreamWriter.writeAll(arrowTable);
-  const arrowIPCBytes = writer.toUint8Array(true);
-
-  // TODO: provide options for how to write table.
-  const writerProperties = new wasm.WriterPropertiesBuilder().build();
-  const parquetBytes = wasm.writeParquet(arrowIPCBytes, writerProperties);
-  return parquetBytes.buffer.slice(
-    parquetBytes.byteOffset,
-    parquetBytes.byteLength + parquetBytes.byteOffset
-  );
+  // Pass the IPC stream to the Parquet writer.
+  const wasmTable = wasm.Table.fromIPCStream(ipcStream);
+  const wasmProperties = new wasm.WriterPropertiesBuilder().build();
+  try {
+    const parquetBytes = wasm.writeParquet(wasmTable, wasmProperties);
+    // const parquetBytes = wasm.writeParquet(wasmTable, wasmProperties);
+    return parquetBytes.buffer.slice(
+      parquetBytes.byteOffset,
+      parquetBytes.byteLength + parquetBytes.byteOffset
+    );
+  } finally {
+    // wasmTable.free();
+    // wasmProperties.free();
+  }
 }
+
+// type WriteOptions = {
+//   compression?: number;
+//   dictionaryEnabled?: boolean;
+//   encoding?: number;
+//   maxRowGroupSize?: number;
+//   maxStatisticsSize?: number;
+//   statisticsEnabled?: boolean;
+//   writeBatchSize?: number;
+//   dataPageSizeLimit?: number;
+//   dictionaryPageSizeLimit?: number;
+// };
+
+// columnCompression: Record<string, number>;
+// columnDictionaryEnabled: Record<string, boolean>;
+// columnEncoding: Record<string, number>;
+// columnMaxStatisticsSize
+// compression:Record<string, number>;
+// setCreatedBy
+// setDataPageSizeLimit
+// setDictionaryEnabled
+// setDictionaryPageSizeLimit
+// setEncoding
+// setMaxRowGroupSize
+// setMaxStatisticsSize
+// setStatisticsEnabled
+// setWriteBatchSize
+// setWriterVersion
diff --git a/modules/parquet/src/lib/wasm/load-wasm-browser.ts b/modules/parquet/src/lib/wasm/load-wasm-browser.ts
deleted file mode 100644
index 13f35a1d7c..0000000000
--- a/modules/parquet/src/lib/wasm/load-wasm-browser.ts
+++ /dev/null
@@ -1,19 +0,0 @@
-// loaders.gl
-// SPDX-License-Identifier: MIT
-// Copyright (c) vis.gl contributors
-
-import * as wasmEsm from 'parquet-wasm/esm2/arrow1';
-
-let cached: typeof wasmEsm | null = null;
-
-export async function loadWasm(wasmUrl?: string) {
-  if (cached !== null) {
-    return cached;
-  }
-
-  // For ESM bundles, need to await the default export, which loads the WASM
-  await wasmEsm.default(wasmUrl);
-  cached = wasmEsm;
-
-  return wasmEsm;
-}
diff --git a/modules/parquet/src/lib/wasm/load-wasm-node.ts b/modules/parquet/src/lib/wasm/load-wasm-node.ts
deleted file mode 100644
index 5ee44eebb6..0000000000
--- a/modules/parquet/src/lib/wasm/load-wasm-node.ts
+++ /dev/null
@@ -1,9 +0,0 @@
-// loaders.gl
-// SPDX-License-Identifier: MIT
-// Copyright (c) vis.gl contributors
-
-import * as wasmNode from 'parquet-wasm/node/arrow1';
-
-export async function loadWasm(wasmUrl?: string) {
-  return wasmNode;
-}
diff --git a/modules/parquet/src/lib/wasm/load-wasm.ts b/modules/parquet/src/lib/wasm/load-wasm.ts
index 6c89cf2b1f..0c7c2d447a 100644
--- a/modules/parquet/src/lib/wasm/load-wasm.ts
+++ b/modules/parquet/src/lib/wasm/load-wasm.ts
@@ -2,4 +2,21 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) vis.gl contributors
 
-export {loadWasm} from './load-wasm-node';
+// eslint-disable-next-line import/default
+import initWasm from 'parquet-wasm';
+import * as parquetWasm from 'parquet-wasm';
+import {PARQUET_WASM_URL} from '../constants';
+
+let initializePromise: any;
+
+export async function loadWasm(wasmUrl: string = PARQUET_WASM_URL) {
+  if (!initializePromise && typeof initWasm === 'function') {
+    if (!wasmUrl) {
+      throw new Error('ParquetLoader: No wasmUrl provided');
+    }
+    // @ts-ignore
+    initializePromise = initWasm(wasmUrl);
+  }
+  await initializePromise;
+  return parquetWasm;
+}
diff --git a/modules/parquet/src/lib/wasm/parse-parquet-wasm.ts b/modules/parquet/src/lib/wasm/parse-parquet-wasm.ts
index b86e77eb3b..3ae930fd28 100644
--- a/modules/parquet/src/lib/wasm/parse-parquet-wasm.ts
+++ b/modules/parquet/src/lib/wasm/parse-parquet-wasm.ts
@@ -3,43 +3,31 @@
 // Copyright (c) vis.gl contributors
 
 // eslint-disable
-import type {LoaderOptions} from '@loaders.gl/loader-utils';
 import type {ArrowTable} from '@loaders.gl/arrow';
 import {serializeArrowSchema} from '@loaders.gl/arrow';
-import * as arrow from 'apache-arrow';
+import type {ParquetWasmLoaderOptions} from '../../parquet-wasm-loader';
 import {loadWasm} from './load-wasm';
-
-export type ParquetWasmLoaderOptions = LoaderOptions & {
-  parquet?: {
-    type?: 'arrow-table';
-    wasmUrl?: string;
-  };
-};
+import * as arrow from 'apache-arrow';
 
 export async function parseParquetWasm(
   arrayBuffer: ArrayBuffer,
-  options?: ParquetWasmLoaderOptions
+  options: ParquetWasmLoaderOptions
 ): Promise<ArrowTable> {
+  const arr = new Uint8Array(arrayBuffer);
+
   const wasmUrl = options?.parquet?.wasmUrl;
   const wasm = await loadWasm(wasmUrl);
+  const wasmTable = wasm.readParquet(arr);
+  try {
+    const ipcStream = wasmTable.intoIPCStream();
+    const arrowTable = arrow.tableFromIPC(ipcStream);
 
-  const arr = new Uint8Array(arrayBuffer);
-  const arrowIPCUint8Arr = wasm.readParquet(arr);
-  const arrowIPCBuffer = arrowIPCUint8Arr.buffer.slice(
-    arrowIPCUint8Arr.byteOffset,
-    arrowIPCUint8Arr.byteLength + arrowIPCUint8Arr.byteOffset
-  );
-
-  const reader = arrow.RecordBatchStreamReader.from(arrowIPCBuffer);
-  const recordBatches: arrow.RecordBatch[] = [];
-  for (const recordBatch of reader) {
-    recordBatches.push(recordBatch);
+    return {
+      shape: 'arrow-table',
+      schema: serializeArrowSchema(arrowTable.schema),
+      data: arrowTable
+    };
+  } finally {
+    // wasmTable.free();
   }
-  const arrowTable = new arrow.Table(recordBatches);
-
-  return {
-    shape: 'arrow-table',
-    schema: serializeArrowSchema(arrowTable.schema),
-    data: arrowTable
-  };
 }
diff --git a/modules/parquet/src/parquet-wasm-loader.ts b/modules/parquet/src/parquet-wasm-loader.ts
index 2fda7c68fb..8f6bc82fe9 100644
--- a/modules/parquet/src/parquet-wasm-loader.ts
+++ b/modules/parquet/src/parquet-wasm-loader.ts
@@ -6,10 +6,7 @@ import type {Loader, LoaderWithParser, LoaderOptions} from '@loaders.gl/loader-u
 import type {ArrowTable} from '@loaders.gl/arrow';
 
 import {parseParquetWasm} from './lib/wasm/parse-parquet-wasm';
-
-// __VERSION__ is injected by babel-plugin-version-inline
-// @ts-ignore TS2304: Cannot find name '__VERSION__'.
-const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';
+import {VERSION, PARQUET_WASM_URL} from './lib/constants';
 
 /** Parquet WASM loader options */
 export type ParquetWasmLoaderOptions = LoaderOptions & {
@@ -34,7 +31,7 @@ export const ParquetWasmWorkerLoader: Loader<ArrowTable, never, ParquetWasmLoade
   options: {
     parquet: {
       type: 'arrow-table',
-      wasmUrl: 'https://unpkg.com/parquet-wasm@0.3.1/esm2/arrow1_bg.wasm'
+      wasmUrl: PARQUET_WASM_URL
     }
   }
 };
@@ -42,5 +39,8 @@ export const ParquetWasmWorkerLoader: Loader<ArrowTable, never, ParquetWasmLoade
 /** Parquet WASM table loader */
 export const ParquetWasmLoader: LoaderWithParser<ArrowTable, never, ParquetWasmLoaderOptions> = {
   ...ParquetWasmWorkerLoader,
-  parse: parseParquetWasm
+  parse(arrayBuffer: ArrayBuffer, options?: ParquetWasmLoaderOptions) {
+    options = {parquet: {...ParquetWasmLoader.options.parquet, ...options?.parquet}, ...options};
+    return parseParquetWasm(arrayBuffer, options);
+  }
 };
diff --git a/modules/parquet/src/parquet-wasm-writer.ts b/modules/parquet/src/parquet-wasm-writer.ts
index 673fb318c9..36c225ffa1 100644
--- a/modules/parquet/src/parquet-wasm-writer.ts
+++ b/modules/parquet/src/parquet-wasm-writer.ts
@@ -4,11 +4,16 @@
 
 import type {WriterWithEncoder} from '@loaders.gl/loader-utils';
 import type {ArrowTable} from '@loaders.gl/arrow';
-import {encode, ParquetWriterOptions} from './lib/wasm/encode-parquet-wasm';
+import {encode} from './lib/wasm/encode-parquet-wasm';
+import type {WriterOptions} from '@loaders.gl/loader-utils';
 
-// __VERSION__ is injected by babel-plugin-version-inline
-// @ts-ignore TS2304: Cannot find name '__VERSION__'.
-const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';
+import {VERSION, PARQUET_WASM_URL} from './lib/constants';
+
+export type ParquetWriterOptions = WriterOptions & {
+  parquet?: {
+    wasmUrl?: string;
+  };
+};
 
 /** Parquet WASM writer */
 export const ParquetWasmWriter: WriterWithEncoder<ArrowTable, never, ParquetWriterOptions> = {
@@ -21,8 +26,11 @@ export const ParquetWasmWriter: WriterWithEncoder<ArrowTable, never, ParquetWrit
   binary: true,
   options: {
     parquet: {
-      wasmUrl: 'https://unpkg.com/parquet-wasm@0.3.1/esm2/arrow1_bg.wasm'
+      wasmUrl: PARQUET_WASM_URL
     }
   },
-  encode
+  encode(arrowTable: ArrowTable, options?: ParquetWriterOptions) {
+    options = {parquet: {...ParquetWasmWriter.options.parquet, ...options?.parquet}, ...options};
+    return encode(arrowTable, options);
+  }
 };
diff --git a/modules/parquet/src/parquetjs/parser/decoders.ts b/modules/parquet/src/parquetjs/parser/decoders.ts
index 34ec642b89..1bc580892d 100644
--- a/modules/parquet/src/parquetjs/parser/decoders.ts
+++ b/modules/parquet/src/parquetjs/parser/decoders.ts
@@ -19,7 +19,7 @@ import {
   Type
 } from '../parquet-thrift';
 import {decompress} from '../compression';
-import {PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING} from '../../constants';
+import {PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING} from '../../lib/constants';
 import {decodePageHeader, getThriftEnum, getBitWidth} from '../utils/read-utils';
 
 /**
diff --git a/modules/parquet/src/parquetjs/parser/parquet-reader.ts b/modules/parquet/src/parquetjs/parser/parquet-reader.ts
index 558bf4b681..9be78db03d 100644
--- a/modules/parquet/src/parquetjs/parser/parquet-reader.ts
+++ b/modules/parquet/src/parquetjs/parser/parquet-reader.ts
@@ -5,7 +5,7 @@ import {ParquetSchema} from '../schema/schema';
 import {decodeSchema} from './decoders';
 import {materializeRows} from '../schema/shred';
 
-import {PARQUET_MAGIC, PARQUET_MAGIC_ENCRYPTED} from '../../constants';
+import {PARQUET_MAGIC, PARQUET_MAGIC_ENCRYPTED} from '../../lib/constants';
 import {ColumnChunk, CompressionCodec, FileMetaData, RowGroup, Type} from '../parquet-thrift';
 import {
   ParquetRowGroup,
diff --git a/modules/parquet/test/index.ts b/modules/parquet/test/index.ts
index d4602cf5cc..d4db1a4cb7 100644
--- a/modules/parquet/test/index.ts
+++ b/modules/parquet/test/index.ts
@@ -18,4 +18,5 @@ import './parquet-writer.spec';
 import './geoparquet-loader.spec';
 
 import './parquet-columnar-loader.spec';
-// import './parquet-wasm-loader.spec';
+
+import './parquet-wasm-loader.spec';
diff --git a/modules/parquet/test/parquet-wasm-loader.spec.ts.disabled b/modules/parquet/test/parquet-wasm-loader.spec.ts
similarity index 82%
rename from modules/parquet/test/parquet-wasm-loader.spec.ts.disabled
rename to modules/parquet/test/parquet-wasm-loader.spec.ts
index 02a4e043a0..3945d1a7d3 100644
--- a/modules/parquet/test/parquet-wasm-loader.spec.ts.disabled
+++ b/modules/parquet/test/parquet-wasm-loader.spec.ts
@@ -8,7 +8,6 @@ import * as arrow from 'apache-arrow';
 import {WASM_SUPPORTED_FILES} from './data/files';
 
 const PARQUET_DIR = '@loaders.gl/parquet/test/data';
-const WASM_URL = 'node_modules/parquet-wasm/esm2/arrow1_bg.wasm';
 
 setLoaderOptions({
   _workerType: 'test'
@@ -23,11 +22,7 @@ test('ParquetLoader#loader objects', (t) => {
 
 test('Load Parquet file', async (t) => {
   const url = `${PARQUET_DIR}/geoparquet/example.parquet`;
-  const table = await load(url, ParquetWasmLoader, {
-    parquet: {
-      wasmUrl: WASM_URL
-    }
-  });
+  const table = await load(url, ParquetWasmLoader, {});
   const arrowTable = table.data;
   t.equal(arrowTable.numRows, 5);
   t.deepEqual(table.schema?.fields.map((f) => f.name), [
@@ -41,17 +36,13 @@ test('Load Parquet file', async (t) => {
   t.end();
 });
 
-test('ParquetWasmLoader#load', async (t) => {
+test.only('ParquetWasmLoader#load', async (t) => {
   t.comment('SUPPORTED FILES');
   for (const {title, path} of WASM_SUPPORTED_FILES) {
     const url = `${PARQUET_DIR}/apache/${path}`;
-    const table = await load(url, ParquetWasmLoader, {
-      parquet: {
-        wasmUrl: WASM_URL
-      }
-    });
+    const table = await load(url, ParquetWasmLoader);
     const arrowTable = table.data;
-    t.ok(arrowTable, `GOOD(${title})`);
+    t.ok(arrowTable instanceof arrow.Table, `GOOD(${title})`);
   }
 
   t.end();
@@ -62,15 +53,9 @@ test('ParquetWasmWriter#writer/loader round trip', async (t) => {
 
   const parquetBuffer = await encode(table, ParquetWasmWriter, {
     worker: false,
-    parquet: {
-      wasmUrl: WASM_URL
-    }
   });
   const newTable = await load(parquetBuffer, ParquetWasmLoader, {
     worker: false,
-    parquet: {
-      wasmUrl: WASM_URL
-    }
   });
 
   t.deepEqual(table.data.schema, newTable.data.schema);
diff --git a/modules/zip/src/parse-zip/zip-composition.ts b/modules/zip/src/parse-zip/zip-composition.ts
index a78e92ed85..f9fa8079f7 100644
--- a/modules/zip/src/parse-zip/zip-composition.ts
+++ b/modules/zip/src/parse-zip/zip-composition.ts
@@ -200,12 +200,17 @@ export function getFileIterator(
  * @param subfolder relative path from the root folder.
  * @returns list of paths
  */
+<<<<<<< HEAD
 export async function getAllFiles(
   basePath: string,
   subfolder: string = '',
   fsPassed?: NodeFilesystem
 ): Promise<string[]> {
   const fs = fsPassed ? fsPassed : new NodeFilesystem({});
+=======
+export async function getAllFiles(basePath: string, subfolder: string = ''): Promise<string[]> {
+  fs ||= new NodeFilesystem({});
+>>>>>>> 7e5932d46 (feat(parquet): restore ParquetWasm loader)
   const files = await fs.readdir(pathJoin(basePath, subfolder));
 
   const arrayOfFiles: string[] = [];
diff --git a/package.json b/package.json
index 8155eca42a..6b7726b1e2 100644
--- a/package.json
+++ b/package.json
@@ -56,16 +56,18 @@
   "dependencies": {},
   "resolutions_notes": [
     "Ensure we use recent typescript",
-    "Note: prettier 3 required for JSON import assertions"
+    "Note: prettier 3 required for JSON import assertions",
+    "apache-arrow version should be aligned with parquet-wasm version"
   ],
   "resolutions": {
     "@typescript-eslint/eslint-plugin": "^6.0.0",
     "@typescript-eslint/parser": "^6.0.0",
+    "apache-arrow": "^15.0.0",
     "prettier": "3.0.3",
     "typescript": "^5.3.0"
   },
   "volta": {
-    "node": "18.18.2",
+    "node": "18.19.0",
     "yarn": "1.22.19"
   }
 }
diff --git a/yarn.lock b/yarn.lock
index e15586f265..77692bcf59 100644
--- a/yarn.lock
+++ b/yarn.lock
@@ -9139,10 +9139,10 @@ parent-module@^1.0.0:
   dependencies:
     callsites "^3.0.0"
 
-parquet-wasm@^0.3.1:
-  version "0.3.1"
-  resolved "https://registry.yarnpkg.com/parquet-wasm/-/parquet-wasm-0.3.1.tgz#4c7306cac00e1bfacb00ee9e395a9ef8a4c8623f"
-  integrity sha512-8OMYwh7N+IcQBeKTuDgpvJR7xh/YKPgnQcs29CQW0Q4ziwroRu7N6tQmmsjiCaQ/W7enBNppquj6VAe2fA1yYQ==
+parquet-wasm@^0.6.0-beta.1:
+  version "0.6.0-beta.1"
+  resolved "https://registry.yarnpkg.com/parquet-wasm/-/parquet-wasm-0.6.0-beta.1.tgz#f86245a69802aa44d947182237f84fabc394654f"
+  integrity sha512-uA4w7XEssSub8+1zBmlnd9KlIKIfK43Syw2GrWK4pi0YG7N1ohx5T5PQbeNAqBjGZMFeC6bGw5FdGVAyKu9r1Q==
 
 parse-data-uri@^0.2.0:
   version "0.2.0"