Extract message links with explicit schemes and punctuation boundaries

tools/chatgui/src/message_links.cpp

@@ -0,0 +1,28 @@
+#include "message_links.h"
+#include <QRegularExpression>
+#include <QUrl>
+
+/* Отделить адрес от текста, сохраняя сбалансированные скобки внутри пути/query. */
+QVector<MessageLink> messageLinks(const QString& text) {
+    static const QRegularExpression pattern(QStringLiteral(
+        "utun://[A-Za-z0-9+/]+={0,2}|(?i:https?://)[^\\s\\p{Z}\\p{Cc}<>\"'`«»“”‘’]+"));
+    QVector<MessageLink> links;
+    auto matches = pattern.globalMatch(text);
+    while (matches.hasNext()) {
+        const auto match = matches.next();
+        QString url = match.captured();
+        if (!url.startsWith(QStringLiteral("utun://"))) {
+            while (!url.isEmpty()) {
+                const QChar last = url.back();
+                const int closing = QStringLiteral(")]}").indexOf(last);
+                const bool unmatched = closing >= 0 && url.count(last) > url.count(QStringLiteral("([{").at(closing));
+                if (!QStringLiteral(".,;:!?").contains(last) && !unmatched) break;
+                url.chop(1);
+            }
+            const QUrl parsed(url, QUrl::StrictMode);
+            if (!parsed.isValid() || parsed.host().isEmpty()) continue;
+        }
+        links.append({(int)match.capturedStart(), (int)url.length(), url});
+    }
+    return links;
+}