From c6d7cef491d1202c3d535ca24d92e746bee50d99 Mon Sep 17 00:00:00 2001
From: RMidhunSuresh <rmidhunsuresh@gmail.com>
Date: Tue, 11 May 2021 22:21:37 +0530
Subject: [PATCH] Refactor regex to improve readability

- Split regex into components
- Add informative comments

Signed-off-by: RMidhunSuresh <rmidhunsuresh@gmail.com>
---
 .../session/room/timeline/linkify/regex.js    | 30 ++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)

diff --git a/src/domain/session/room/timeline/linkify/regex.js b/src/domain/session/room/timeline/linkify/regex.js
index fe8c94eb..7530d301 100644
--- a/src/domain/session/room/timeline/linkify/regex.js
+++ b/src/domain/session/room/timeline/linkify/regex.js
@@ -1 +1,29 @@
-export const regex = /(?:https|http|ftp):\/\/[a-zA-Z0-9:.\[\]#-]+(?:[\/#][^\s]*[^\s.,?!]|[^\s\u{80}-\u{10ffff}.,?!])/gui
+const scheme = "(?:https|http|ftp):\\/\\/";
+const host = "[a-zA-Z0-9:.\\[\\]-]";
+
+/*
+A URL containing path (/) or fragment (#) component
+is allowed to end with any character which is not 
+space nor punctuation. The ending character may be 
+non-ASCII.
+*/
+const end = "[^\\s.,?!]";
+const additional = `[\\/#][^\\s]*${end}`;
+
+/*
+Similarly, a URL not containing path or fragment must
+also end with a character that is not space nor punctuation.
+However the ending character must also be ASCII.
+*/
+const nonASCII = "\\u{80}-\\u{10ffff}";
+const endASCII = `[^\\s${nonASCII}.,?!]`;
+
+/*
+URL must not contain non-ascii characters in host but may contain
+them in path or fragment components.
+https://matrix.org/<smiley> - valid
+https://matrix.org<smiley> - invalid
+*/
+const urlRegex = `${scheme}${host}+(?:${additional}|${endASCII})`;
+
+export const regex = new RegExp(urlRegex, "gui");