hyparquet - npm Package Compare versions

Comparing version 0.2.3 to 0.2.4

package.json

		{
		"name": "hyparquet",
		"version": "0.2.3",
		"version": "0.2.4",
		"description": "parquet file parser for javascript",
		@@ -30,4 +30,4 @@ "keywords": [
		"devDependencies": {
		"@types/node": "20.11.8",
		"@typescript-eslint/eslint-plugin": "6.19.1",
		"@types/node": "20.11.16",
		"@typescript-eslint/eslint-plugin": "6.20.0",
		"@vitest/coverage-v8": "1.2.2",
		@@ -34,0 +34,0 @@ "eslint": "8.56.0",

src/column.js

		@@ -48,15 +48,15 @@ import { CompressionCodec, ConvertedType, Encoding, PageType } from './constants.js'
		// decompress bytes
		/** @type {Uint8Array \| undefined} */
		let page
		const uncompressed_page_size = Number(header.uncompressed_page_size)
		const { codec } = columnMetadata
		if (codec === CompressionCodec.GZIP) {
		throw new Error('parquet gzip compression not supported')
		} else if (codec === CompressionCodec.SNAPPY) {
		if (codec === CompressionCodec.SNAPPY) {
		page = new Uint8Array(uncompressed_page_size)
		snappyUncompress(compressedBytes, page)
		} else if (codec === CompressionCodec.LZO) {
		throw new Error('parquet lzo compression not supported')
		} else {
		const compressor = Object.entries(CompressionCodec).find(([, value]) => value === codec)
		throw new Error(`parquet unsupported compression codec: ${codec} ${compressor?.[0]}`)
		}
		if (!page \|\| page.length !== uncompressed_page_size) {
		throw new Error('parquet decompressed page size does not match header')
		if (page?.length !== uncompressed_page_size) {
		throw new Error(`parquet decompressed page length ${page?.length} does not match header ${uncompressed_page_size}`)
		}
		@@ -63,0 +63,0 @@

src/metadata.js

		@@ -10,2 +10,9 @@ import { schemaTree } from './schema.js'
		*
		* You must provide the byteLength of the buffer, typically from a HEAD request.
		*
		* In theory, you could use suffix-range requests to fetch the end of the file,
		* and save a round trip. But in practice, this doesn't work because chrome
		* deems suffix-range requests as a not-safe-listed header, and will require
		* a pre-flight. So the byteLength is required.
		*
		* To make this efficient, we initially request the last 512kb of the file,
		@@ -16,3 +23,3 @@ * which is likely to contain the metadata. If the metadata length exceeds the
		* This ensures that we either make one 512kb initial request for the metadata,
		* or two requests for exactly the metadata size.
		* or a second request for up to the metadata size.
		*
		@@ -27,3 +34,3 @@ * @typedef {import("./types.d.ts").AsyncBuffer} AsyncBuffer
		// fetch last bytes (footer) of the file
		const footerOffset = asyncBuffer.byteLength - initialFetchSize
		const footerOffset = Math.max(0, asyncBuffer.byteLength - initialFetchSize)
		const footerBuffer = await asyncBuffer.slice(footerOffset)
		@@ -37,3 +44,3 @@ // check if metadata size fits inside the initial fetch
		const metadataBuffer = await asyncBuffer.slice(metadataOffset, footerOffset)
		// combine the buffers
		// combine initial fetch with the new slice
		const combinedBuffer = new ArrayBuffer(metadataLength + 8)
		@@ -40,0 +47,0 @@ const combinedView = new Uint8Array(combinedBuffer)

New alerts