Using regexp for words

[mikdiet]

I'm looking for an opportunity for switching from yaspeller to cspell, and I want to reuse my custom dictionary.

Yaspeller allowed me to define words using simplified regexp, e.g.:

"(un|)knownword(s|)" # will match knownword, knownwords, unknownword, unknownwords

Is it a way to define the same in cspell?

P.s. I of course could define 4 words instead of 1 regexp in the example above, but my language defines lots of prefixes/suffixes/endings, real example in https://github.com/rusrails/rusrails/blob/master/.yaspeller.json#L15

[Jason3S]

@mikdiet,

Great question:

Please see: #4942 (comment)

I had not realized that yaspeller supported explicit substitutions.

It is also possible to write your own word list expansion function:

cspell.confiig.mjs

const wordlist = ['(un|)knownword(s|)'];

function expandWordlist(list) {
// expansion code goes here
}

const config = {
  words: expandWordlist(wordlist),
};

export default config;

[Jason3S]

Here is a working example. It loads the words from words.txt in the same directory as cspell.config.mjs and expands them returning a custom dictionary called custom-words.

cspell.config.mjs

import { promises as fs } from 'fs';

function expand(pattern, options = { begin: '(', end: ')', sep: '|' }, start = 0) {
    const len = pattern.length;
    const parts = [];
    function push(word) {
        if (Array.isArray(word)) {
            parts.push(...word);
        } else {
            parts.push(word);
        }
    }
    let i = start;
    let curWord = '';
    while (i < len) {
        const ch = pattern[i++];
        if (ch === options.end) {
            break;
        }
        if (ch === options.begin) {
            const nested = expand(pattern, options, i);
            i = nested.idx;
            curWord = nested.parts.flatMap((p) => (Array.isArray(curWord) ? curWord.map((w) => w + p) : [curWord + p]));
            continue;
        }
        if (ch === options.sep) {
            push(curWord);
            curWord = '';
            continue;
        }
        curWord = Array.isArray(curWord) ? curWord.map((w) => w + ch) : curWord + ch;
    }
    push(curWord);
    return { parts, idx: i };
}

function expandWord(pattern, options = { begin: '(', end: ')', sep: '|' }) {
    return expand(pattern, options).parts;
}

function expandWords(wordList) {
    const words = wordList
        .split(/\n/g)
        .map((w) => w.trim())
        .filter((w) => !!w && !w.startsWith('#'))
        .flatMap((w) => expandWord(w));
    return words;
}

const wordsFile = new URL('words.txt', import.meta.url);

const wordList = await fs.readFile(wordsFile, 'utf8');

const config = {
    id: 'example-config',
    dictionaryDefinitions: [
        {
            name: 'custom-words',
            words: expandWords(wordList),
        },
    ],
    dictionaries: ['custom-words'],
};

export default config;

[mikdiet]

It's so helpful, I thought of preliminary CI step with similar approach, but custom script/dictionary is much better!