feat!: use buffer npm module to add browser support

Author: gamemaker1

package.json

@@ -44,9 +44,10 @@
 		"test": "run-s test:compile test:integration",
 		"test:compile": "tsc --noEmit",
 		"test:quality": "xo source/ test/",
-		"test:integration": "ava"
+		"test:integration": "TSIMP_DIAG=ignore ava"
 	},
 	"dependencies": {
+		"buffer": "6.0.3",
 		"fflate": "0.8.2",
 		"file-type": "19.3.0",
 		"got": "14.4.1",

pnpm-lock.yaml

@@ -1,7 +1,7 @@
 // source/lib.ts
 // The source code for the library.
 
-import { Buffer } from 'node:buffer'
+import { Buffer } from 'buffer/index.js'
 import { fileTypeFromBuffer as getFileType } from 'file-type'
 import { readFile, fetchUrl } from './util.js'
 

source/lib.ts

@@ -1,7 +1,7 @@
 // source/parsers/docx.ts
 // The text extracter for DOCX files.
 
-import { type Buffer } from 'node:buffer'
+import { type Buffer } from 'buffer/'
 import { extractRawText as parseWordFile } from 'mammoth'
 
 import type { TextExtractionMethod } from '../lib.js'
@@ -22,6 +22,7 @@ export class DocExtractor implements TextExtractionMethod {
 	 */
 	apply = async (input: Buffer): Promise<string> => {
 		// Convert the DOCX to text and return the text.
+		// @ts-expect-error: see feross/buffer#353, the types are incomplete.
 		const parsedDocx = await parseWordFile({ buffer: input })
 		return parsedDocx.value
 	}

source/parsers/doc.ts

@@ -1,7 +1,7 @@
 // source/parsers/excel.ts
 // The text extracter for Excel files.
 
-import { type Buffer } from 'node:buffer'
+import { type Buffer } from 'buffer/'
 import Xlsx, { utils as sheetUtils } from 'xlsx'
 import { dump as convertToYaml } from 'js-yaml'
 

source/parsers/excel.ts

@@ -1,7 +1,7 @@
 // source/parsers/pdf.ts
 // The text extracter for PDF files.
 
-import { type Buffer } from 'node:buffer'
+import { type Buffer } from 'buffer/'
 // @ts-expect-error There are no types for this package.
 import parsePdf from 'pdf-parse/lib/pdf-parse.js'
 

source/parsers/pdf.ts

@@ -3,7 +3,7 @@
 
 /* eslint-disable @typescript-eslint/no-unsafe-assignment */
 
-import { type Buffer } from 'node:buffer'
+import { type Buffer } from 'buffer/'
 import { unzip } from 'fflate'
 import { parseStringPromise as xmlToJson } from 'xml2js'
 import encoding from 'text-encoding'

source/parsers/ppt.ts

@@ -1,11 +1,11 @@
 // source/util.ts
 // Utility functions to help with the handling of input.
 
-import { type Buffer } from 'node:buffer'
 import { readFile as read } from 'node:fs/promises'
 import { got as fetch } from 'got'
+import { type Buffer } from 'buffer/'
 
 export const readFile = async (filePath: string): Promise<Buffer> =>
-	read(filePath)
+	(await read(filePath)) as unknown as Buffer
 export const fetchUrl = async (url: string): Promise<Buffer> =>
-	fetch(url).buffer()
+	(await fetch(url).buffer()) as unknown as Buffer

source/util.ts

@@ -2,9 +2,9 @@
 // This file contains the integration test for the library.
 
 import { readFileSync } from 'node:fs'
-import { type Buffer } from 'node:buffer'
 import test from 'ava'
 
+import { type Buffer } from 'buffer/'
 import { getTextExtractor, type InputType } from '../../source/index.js'
 
 const extractor = getTextExtractor()