Simplify linkifyMessage

Parsing html as xml has inherent problems, most notable there are many matrix clients that don't escape ampersands in urls of mx-replies, etc. (See issue #18) This also removes the replacement of <mx-reply> as it isn't strictly needed. Also the QRegExp is replaced with the Qt5 QRegularExpression for perfomance and because it supports lookahead and lookbehind. I'm pretty sure that the original code also replaced href="" with href=\"\", which was probably wrong, but I'm not to sure about that. Fixes #18

Simplify linkifyMessage
23eef9e1 · Nicolas Werner · e1457d5c · 23eef9e1 · 23eef9e1
Commit 23eef9e1 authored 5 years ago by Nicolas Werner
--- a/src/Config.h
+++ b/src/Config.h
 #pragma once

-#include <QRegExp>
+#include <QRegularExpression>
 #include <QString>

 // Non-theme app configuration. Layouts, fonts spacing etc.
@@ -51,8 +51,11 @@ constexpr auto LABEL_BIG_SIZE_RATIO    = 2;

 namespace strings {
 const QString url_html = "<a href=\"\\1\">\\1</a>";
-const QRegExp url_regex(
-  "((www\\.(?!\\.)|[a-z][a-z0-9+.-]*://)[^\\s<>'\"]+[^!,\\.\\s<>'\"\\]\\)\\:])");
+const QRegularExpression url_regex(
+  // match an URL, that is not quoted, i.e.
+  //      vvvvvv match quote via negative lookahead/lookbehind                       vvvvv
+  //            vvvv atomic match url -> fail if there is a " before or after      vv
+  R"regex((?<!")(?>((www\.(?!\.)|[a-z][a-z0-9+.-]*://)[^\s<>'"]+[^!,\.\s<>'"\]\)\:]))(?!"))regex");
 }

 // Window geometry.

--- a/src/Utils.cpp
+++ b/src/Utils.cpp
@@ -291,65 +291,9 @@ utils::linkifyMessage(const QString &body)
 {
        // Convert to valid XML.
        auto doc = QString("<html>%1</html>").arg(body);
+        doc.replace(conf::strings::url_regex, conf::strings::url_html);

-        doc.replace("<mx-reply>", "");
-        doc.replace("</mx-reply>", "");
-        doc.replace("<br>", "<br></br>");
-
-        QXmlStreamReader xml{doc};
-
-        QString textString;
-        while (!xml.atEnd() && !xml.hasError()) {
-                auto t = xml.readNext();
-
-                switch (t) {
-                case QXmlStreamReader::Characters: {
-                        auto text = xml.text().toString();
-                        text.replace(conf::strings::url_regex, conf::strings::url_html);
-
-                        textString += text;
-                        break;
-                }
-                case QXmlStreamReader::StartDocument:
-                case QXmlStreamReader::EndDocument:
-                        break;
-                case QXmlStreamReader::StartElement: {
-                        if (xml.name() == "html")
-                                break;
-
-                        textString += QString("<%1").arg(xml.name().toString());
-
-                        const auto attrs = xml.attributes();
-                        for (const auto &e : attrs)
-                                textString += QString(" %1=\"%2\"")
-                                                .arg(e.name().toString())
-                                                .arg(e.value().toString());
-
-                        textString += ">";
-
-                        break;
-                }
-                case QXmlStreamReader::EndElement: {
-                        if (xml.name() == "html")
-                                break;
-
-                        textString += QString("</%1>").arg(xml.name().toString());
-                        break;
-                }
-                default: {
-                        break;
-                }
-                }
-        }
-
-        if (xml.hasError()) {
-                qWarning() << "error while parsing xml" << xml.errorString() << doc;
-                doc.replace("<html>", "");
-                doc.replace("</html>", "");
-                return doc;
-        }
-
-        return textString;
+        return doc;
 }

 QString