spech

Check your text for grammar and spelling error using multiple providers.

• Zero config
• Multiple libraries/services support

Usage

npx spech

By default, it finds **/*.md files and checks it using en-US language.

Use -l flag to specify another language:

npx spech -l ru

Providers

Hunspell

Hunspell is the most popular open-source spell checker which supports a great variety of languages. The checker uses WebAssembly binding for hunspell hunspell-asm.

GrammarBot

Free grammar checking API. With an API key, you can receive 250 requests/day (7500/mo) at no cost. Without an API key, requests are limited to 100 per day per IP address (3000/mo). The API supports only English (en-US and en-GB).

If you need more that 100 request per day, register a new key and pass it to GrammarBot config:

{
  name: 'grammarBot',
  apiKey: 'YOUKEY,
}

Yandex Speller

Free and very fast spell checker API for en, ru and uk languages. It provides free 10k requests/day or 10m characters/day.

Configuring

You can store configuration in two places:

• spech.config.js
• "spech" section of the package.json

spech.config.js

module.exports = {
  languages: ['en-us'],
  providers: [
    { name: 'hunspell' },
    { name: 'grammarBot', apiKey: 'YOUR_API_KEY' },
  ],
};

More details.

Dictionaries

You can place words/phrases which is marked as error to your dictionary file. Just create a file with .dic extension to the project root:

mydictionary.dic

# Simple word
browserify
# Regexp
/Component.tsx?/

More details.

API Usage

The most useful parts of the library are available through facade class SpellChecker.

Here is a simple example how it can be used:

const { Config, SpellChecker } = require('spech');

async function getMistakes() {
  const config = new Config({ ignoreCase: false });
  const checker = new SpellChecker(config);

  await checker.addDocumentsByMask(process.cwd(), 'docs/*.md');
  checker.addDictionaryPhrase('exceptionphrase');
  checker.addProviderByConfig({ name: 'hunspell' });

  const noMistakes = await checker.checkDocuments();
  if (noMistakes) {
    return [];
  }
  
  const corrections = checker.documents.map(doc => doc.corrections).flat();
  return corrections.map(correction => correction.fragment);
}