Function splitIntoWords

ts/codediff/char-diffs.ts:110–141

(line: string)

 *     splitIntoWords_(line).join('') == line.
 */
export function splitIntoWords(line: string): string[] {
  const LC = 0,
    UC = 2,
    NUM = 3,
    WS = 4,
    SYM = 5;
  const charType = function (c: string) {
    if (/[a-z]/.exec(c)) return LC;
    if (/[A-Z]/.exec(c)) return UC;
    if (/[0-9]/.exec(c)) return NUM;
    if (/\s/.exec(c)) return WS;
    return SYM;
  };

  // Single words can be [A-Z][a-z]+, [A-Z]+, [a-z]+, [0-9]+ or \s+.
  const words = [];
  let lastType = -1;
  for (let i = 0; i < line.length; i++) {
    const c = line.charAt(i);
    const ct = charType(c);
    if (
      (ct == lastType && ct != SYM && ct != WS) ||
      (ct == LC && lastType == UC && words[words.length - 1].length == 1)
    ) {
      words[words.length - 1] += c;
    } else {
      words.push(c);
    }
    lastType = ct;
  }
  return words;
}

// codes are (span class, start, end) triples.
// This merges consecutive runs with the same class, which simplifies the HTML.