Detecting URLs in text is great fun.
When should characters that are allowed in URLs not be considered part of it? Cases I've dealt with today: dot/period, question mark, closing parenthesis. Examples…
This is a boring URL: https://example.com/path
Text <https://example.com/path>
Text (https://example.com/path)
Text (https://example.com/path).
This is a URL containing parentheses: https://example.com/(path)
Text <https://example.com/(path)>
Text (https://example.com/(path))
Text (https://example.com/(path)).
This is a URL containing unmatched parentheses: https://example.com/(path))
Text <https://example.com/(path))>
Text (https://example.com/(path)))
Text (https://example.com/(path))).
This is a URL ending in a period: https://example.com/path.
Text <https://example.com/path.>
Text (https://example.com/path.)
Text (https://example.com/path.).
This is a URL ending in a question mark: https://example.com/path?
Text <https://example.com/path?>
Text (https://example.com/path?)
Text (https://example.com/path?).
Looks like the Mastodon web UI could do better, too. Especially when URLs are wrapped in angle brackets there's no reason to end them early.
@cketti It's actually doing better than many other things I've seen though... 😕