office-text-extractor

similar libraries

there are other great libraries that do the same job and have inspired this project, such as:

• any-text
• officeparser
• textract

however, office-text-extractor has the following differences:

• parses file based on its mime type, not its file extension.
• does not spawn a child process to use a tool installed on the device.
• reads and returns text from the file if it contains plain text.

libraries used

this package uses some amazing existing libraries that perform better than the ones that originally existed in this module, and are therefore used instead:

• pdf-parse, for parsing pdf files
• xlsx, for parsing xlsx files
• mammoth, for parsing docx files

a big thank you to the contributors of these projects!

installation

node

from version 2.0.0 onwards, this package is pure esm. please read this article for a guide on how to ensure your project can import this library.

to use office-text-extractor in an Node project, install it using npm/pnpm/yarn:

1> npm install office-text-extractor
2> pnpm add office-text-extractor
3> yarn add office-text-extractor

browser

the library currently cannot be used in the browser due to its usage of the node:buffer library. pull requests that can replace node:buffer with a different library are welcome!

usage

an example of using the library to extract text is as follows:

1import { readFile } from 'node:fs/promises'
2import { getTextExtractor } from 'office-text-extractor'
3
4// this function returns a new instance of the `TextExtractor` class, with the default
5// extraction methods (docx, pptx, xlsx, pdf) registered.
6const extractor = getTextExtractor()
7
8// extract text from a url, because that's a neat first example :p
9const url = 'https://raw.githubusercontent.com/gamemaker1/office-text-extractor/rewrite/test/fixtures/docs/pptx.pptx'
10const text = await extractor.extractText({ input: url, type: 'url' })
11
12// you can extract text from a file too, like so:
13const path = 'stuff/boring.pdf'
14const text = await extractor.extractText({ input: path, type: 'file' })
15
16// if you have a buffer with the file in it, you can pass that too:
17const buffer = await readFile(path)
18const text = await extractor.extractText({ input: buffer, type: 'buffer' })
19
20console.log(text)

the following is an example of how to create and use your own text extraction method:

1import { type Buffer } from 'node:buffer'
2import { TextExtractor, type TextExtractionMethod } from 'office-text-extractor'
3
4/**
5 * Extracts text from images.
6 */
7class ImageExtractor implements TextExtractionMethod {
8  /**
9   * The mime types of the file that the extractor accepts.
10   */
11  mimes = ['image/png', 'image/jpeg']
12
13  /**
14   * Extracts text from the image file passed by the user.
15   */
16  apply = async (input: Buffer): Promise<string> {
17    const text = await processImage(input)
18    return text
19  }
20}
21
22// create a new extractor and register our extraction method
23const extractor = new TextExtractor()
24extractor.addMethod(new ImageExtractor())
25
26// then use it like you would normally
27const text = await extractor.extractText({ input: '...', type: '...' }
28console.log(text)

license

this project is licensed under the ISC license. please see license.md for more details.