Exclude lonely question marks from the linkify regex

Many URLs we see in the KDE rooms end with a question mark, without a
space. The linkify regex for plain URLs incorrectly considered them as
part of the link, which usually breaks them when opened in a web
browser. Now the regex excludes these, unless they are accompanied by
another character (so links like kde.org/realurl?is=true will still
work.)

(cherry picked from commit 1b7f482d0b)
This commit is contained in:
Joshua Goins
2024-03-18 15:46:22 -04:00
parent f33780e996
commit 498f6d9e64

View File

@@ -37,7 +37,7 @@ static const QRegularExpression strikethrough{QStringLiteral("<del>(.*?)</del>")
static const QRegularExpression mxcImage{QStringLiteral(R"AAA(<img(.*?)src="mxc:\/\/(.*?)\/(.*?)"(.*?)>)AAA")};
static const QRegularExpression plainUrl(
QStringLiteral(
R"(<a.*?<\/a>(*SKIP)(*F)|\b((www\.(?!\.)(?!(\w|\.|-)+@)|(https?|ftp):(//)?\w|(magnet|matrix):)(&(?![lg]t;)|[^&\s<>'"])+(&(?![lg]t;)|[^&!,.\s<>'"\]):])))"),
R"(<a.*?<\/a>(*SKIP)(*F)|\b((www\.(?!\.)(?!(\w|\.|-)+@)|(https?|ftp):(//)?\w|(magnet|matrix):)(&(?![lg]t;)|[^&\s<>'"])+(&(?![lg]t;)|[^?&!,.\s<>'"\]):])))"),
QRegularExpression::CaseInsensitiveOption | QRegularExpression::UseUnicodePropertiesOption);
static const QRegularExpression
url(QStringLiteral(R"(\b((www\.(?!\.)(?!(\w|\.|-)+@)|https?:(//)?\w)(&(?![lg]t;)|[^&\s<>'"])+(&(?![lg]t;)|[^&!,.\s<>'"\]):])))"),