@sap/cds-dk/lib/init/indent.js

Command line client and development toolkit for the SAP Cloud Application Programming Model

const indentRegEx = new RegExp(/^(\s*)/)
const leadingSpacesRegEx = new RegExp(/^( +)/)
const leadingTabsRegEx = new RegExp(/^(\t+)/)

/**
 * Greatest Common Divisor Euclidean algorithm
 * @param {number[]} xs - list of number to find the gcd of.
 * @returns {number} - greatest common divisor of all numbers in xs.
 */
function gcd (xs) {
    const euclid = (a, b) => b === 0 ? a : euclid(b, a % b)
    const [a, b, ...tail] = xs
    return tail.length
        ? gcd([euclid(a, b)].concat(tail))
        : euclid(a, b ?? 0)  // coalesce in case xs consists of exactly one element
}

/**
 * Detects indent used for a file.
 * The algorithm inspects the indentation of every line to find whether tabs or spaces are more prevalent.
 * Once that has been determined, it derives the spacing by calculating the gcd of all relevant lines.
 * I.e. if there are lines indented with 2, 4, or 8 spaces, it will find '  ' as indent.
 * If lines are indented using 1 or 2 tabs, '\t' will be returned.
 * TODO: filter out outliers, i.e. if the user used three spaces in one line, should ' ' space be used for spacing? Or should that be ignored?
 * @param {string[]} file - lines of a file.
 * @returns {string | null} - the string used to space (e.g. 2 spaces, 4 spaces, 1 tab, etc.) or null if the file was empty.
 */
function detectIndent (file) {
    if (file.length === 0) return null
    if (file.length === 1) return indentRegEx.exec(file[0])[0]

    const indents = file
        .map(line => indentRegEx.exec(line)[0])
        .filter(Boolean)
        .reduce((bins, indent) => {
            const type = leadingSpacesRegEx.test(indent)
                ? 'spaces'
                : leadingTabsRegEx.test(indent)
                ? 'tabs'
                : undefined
            if (type) {
                const length = indent.split('').length
                const bin = bins[type]
                bin.total++
                bin.indents[length] = (bin.indents[length] ?? 0) + 1
            }
            return bins
        }, { tabs: { indents: {}, total: 0, character: '\t' }, spaces: { indents: {}, total: 0, character: ' ' } })

    const preferred = indents.tabs.total > indents.spaces.total ? indents.tabs : indents.spaces
    const width = gcd(Object.keys(preferred.indents)) ?? 2
    return preferred.character.repeat(width)
}

module.exports = {
    detectIndent
}