JavaScript 实现中英文空格 - Implementation

spacing.ts

/* Partial implementation from https://zhuanlan.zhihu.com/p/33612593 */
import _ from 'lodash';

/* 标点 */
const punctuationRegex = /\p{Punctuation}/u;
/* 空格 */
const spaceRegex = /\p{Separator}/u;
/* CJK 字符，中日韩 */
const cjkRegex = /\p{Script=Han}|\p{Script=Katakana}|\p{Script=Hiragana}|\p{Script=Hangul}/u;

const shouldSpace = (a: string, b: string): boolean => {
  if (cjkRegex.test(a)) {
    return !(punctuationRegex.test(b)
      || spaceRegex.test(b)
      || cjkRegex.test(b)
    );
  } else {
    return cjkRegex.test(b)
      && !punctuationRegex.test(a)
      && !spaceRegex.test(a);
  }
};

const join = (
  parts: string[],
  sepFunc: (a: string, b: string) => string,
): string => {
  return _.reduce<string, string>(parts, (r, p, i) => {
    const sep = i !== 0 ? sepFunc(p, parts[i - 1]) : "";
    return r + sep + p;
  }, "");
};

export const cjkspace = (
  string: TemplateStringsArray, ...subs: string[]
): string => {
  const parts = _.flatten(_.zip(string, subs.concat("")));
  const filtered = parts.filter((c) => c !== undefined && c !== "") as string[];
  return join(
    filtered,
    (a, b) => (shouldSpace(a, b) ? " " : ""),
  );
};