html to string

1import { decode } from 'html-entities';
2
3function htmlToString(returnText: string) {
4  //-- remove BR tags and replace them with line break
5  returnText = returnText.replace(/<br>/gi, "\n");
6  returnText = returnText.replace(/<br\s\/>/gi, "\n");
7  returnText = returnText.replace(/<br\/>/gi, "\n");
8
9  // //-- remove P and A tags but preserve what's inside of them
10  returnText = returnText.replace(/<p.*>/gi, "\n");
11  returnText = returnText.replace(/<a.*href="(.*?)".*>(.*?)<\/a>/gi, " $2 ($1)");
12
13  // //-- remove all inside SCRIPT and STYLE tags
14  returnText = returnText.replace(/<script.*>[\w\W]{1,}(.*?)[\w\W]{1,}<\/script>/gi, "");
15  returnText = returnText.replace(/<style.*>[\w\W]{1,}(.*?)[\w\W]{1,}<\/style>/gi, "");
16
17  // //-- remove all else
18  returnText = returnText.replace(/<(?:.|\s)*?>/g, "");
19
20  // //-- get rid of more than 2 multiple line breaks:
21  returnText = returnText.replace(/(?:(?:\r\n|\r|\n)\s*){2,}/gim, "\n\n");
22
23  //-- get rid of more than 2 spaces:
24  returnText = returnText.replace(/ +(?= )/g, '');
25
26  returnText = returnText.replace(/<[^>]+>/g, '').replace(/&lt;/g, "<").replace(/&gt;/g, ">")
27  returnText = decode(returnText);
28  
29  return returnText;
30}
31