@mikezimm/fps-core-v7/lib/logic/Strings/drillParse/getDigitSequence.js

Library of reusable core interfaces, types and constants migrated from fps-library-v2

/**
 * https://github.com/mikezimm/drilldown7/issues/499
 *   Extracts the first occurrence of a digit sequence of a specified length from a given string,
 *   optionally allowing a trailing suffix of letters.
 *
 * @param input - The input string to search.
 * @param digitCount - The exact number of digits to look for (e.g., 5 for ZIP code).
 * @param trailingCharacters - Optional. Number of trailing characters to allow (default = 0).
 * @param onlyCAPS - Optional. If true, only allows trailing characters if all are uppercase (default = false).
 *
 * @returns The matched digit sequence (with or without suffix depending on parameters), or an empty string if no match is found.
 *
 * ✅ Examples:
 * getDigitSequence("Zip: 48226", 5)                     → "48226"
 * getDigitSequence("CostCenter-12345A", 5)              → "" (suffix not allowed)
 * getDigitSequence("Dept: 12345A", 5, 1)                → "12345A"
 * getDigitSequence("Region: 12345ab", 5, 2)             → "12345ab"
 * getDigitSequence("Region: 12345ab", 5, 2, true)       → "" (suffix not all caps)
 * getDigitSequence("Region: 12345AB", 5, 2, true)       → "12345AB"
 * getDigitSequence("ID: 1234A", 4, 2)                   → "1234A"
 *
 * ❗ Notes:
 * - Uses negative lookbehind (`(?<!\\d)`) to ensure the match is not part of a longer number.
 * - Uses negative lookahead (`(?!\\w)`) to prevent the match from being part of a larger word.
 * - Underscores `_` and dashes `-` are considered non-word characters, so they count as boundaries.
 * - `|` and `/` are also boundary-friendly in this regex, as they are not word characters.
 */
export function getDigitSequence(input, digitCount, trailingCharacters = 0, onlyCAPS = true) {
    if (!input || digitCount < 1)
        return '';
    const digitGroup = `\\d{${digitCount}}`;
    // Construct the trailing character group if needed
    let trailingGroup = '';
    if (trailingCharacters > 0) {
        const letterPattern = onlyCAPS ? '[A-Z]' : '[A-Za-z]';
        trailingGroup = `${letterPattern}{1,${trailingCharacters}}`;
    }
    // Compose the full pattern in readable blocks
    const patternString = [
        `(?<!\\d)`,
        `(${digitGroup})`,
        trailingGroup,
        `(?!\\w)` // Not followed by a word character
    ].join('');
    const regex = new RegExp(patternString);
    const match = input.match(regex);
    return match ? match[0] : '';
}
/**
 * https://github.com/mikezimm/drilldown7/issues/499
 *  Use this option to get the number sequence based on extra characters but only include the number part
 * @param input
 * @param digitCount
 * @param trailingCharacters
 * @param onlyCAPS
 * @returns
 */
export function getDigitSequenceTrimmed(input, digitCount, trailingCharacters = 0, onlyCAPS = false) {
    const fullMatch = getDigitSequence(input, digitCount, trailingCharacters, onlyCAPS);
    if (!fullMatch)
        return '';
    // Always remove trailing non-digit characters from the match
    return fullMatch.replace(/[^0-9]+$/, '');
}
//# sourceMappingURL=getDigitSequence.js.map