PlainNormalizer.ts

// this is a Node module. require is a must to work across different envs
const URL = require("url-parse");

import { normalizeTextSpacings } from "../utils";

export interface SealRemovalOptions {
  sealUrl: string;
}

export const normalizePlainPart = (
  text: string,
  sealRemovalOptions?: SealRemovalOptions
): string => {
  text = cleanupHiddenCharacters(text);

  text = patchOutlookSafelinksWrappers(text);

  if (sealRemovalOptions) {
    text = removeSeal(text, sealRemovalOptions.sealUrl);
  }
  text = normalizeTextSpacings(text);

  return text.trim();
};

const patchOutlookSafelinksWrappers = (text: string) => {
  const links = text.match(
    /<https:.+?(safelinks\.protection\.outlook\.com).+?>/g
  );

  if (links) {
    links.forEach((link) => {
      const url = new URL(link.slice(1, link.length - 1), true);
      const originalUrl = url.query["url"];
      text = text.replace(link, `<${originalUrl}>`);
    });
  }

  return text;
};

/**
 * Function removes seal from the plain text.
 * @param plain
 * @param sealUrl
 */
const removeSeal = (plain: string, sealUrl: string): string => {
  // For cases [<image-alt>]<<seal-url>>
  const sealRegex = `\\[.+?]\\s*<${sealUrl}>`;

  // For cases <<seal-url>>[<image-alt>]
  const sealRegexReversed = `<${sealUrl}>\\s*\\[.+?]`;

  return plain
    .replace(new RegExp(sealRegex), "")
    .replace(new RegExp(sealRegexReversed), "");
};

export const cleanupHiddenCharacters = (s: string): string => {
  const removeSymbols = new RegExp(/[\u200B]+/g);
  return s.replace(removeSymbols, "");
};