From 4ce03e9415bee694c2b0d8ab745077c6a29bb65e Mon Sep 17 00:00:00 2001
From: Xiao YiFang <yifang.xiao@gmail.com>
Date: Wed, 8 Jun 2022 21:13:07 +0800
Subject: [PATCH] html unescape

---
 htmlescape.cc  | 29 +++++++++++++++++++++++++++++
 htmlescape.hh  |  3 +++
 mdictparser.cc |  2 +-
 utils.hh       |  7 -------
 4 files changed, 33 insertions(+), 8 deletions(-)

diff --git a/htmlescape.cc b/htmlescape.cc
index 59086e8c..bb2c65e6 100644
--- a/htmlescape.cc
+++ b/htmlescape.cc
@@ -157,6 +157,35 @@ QString unescape( QString const & str, bool saveFormat )
   return str;
 }
 
+QString fromHtmlEscaped( QString const & str){
+  QString retVal = str;
+  QRegularExpression regExp("(?<lt>\\&lt\\;)|(?<gt>\\&gt\\;)|(?<amp>\\&amp\\;)|(?<quot>\\&quot\\;)", QRegularExpression::PatternOption::CaseInsensitiveOption);
+  auto match = regExp.match(str, 0);
+
+  while (match.hasMatch())
+  {
+    if (!match.captured("lt").isEmpty())
+    {
+      retVal.replace(match.capturedStart("lt"), match.capturedLength("lt"), "<");
+    }
+    else if (!match.captured("gt").isEmpty())
+    {
+      retVal.replace(match.capturedStart("gt"), match.capturedLength("gt"), ">");
+    }
+    else if (!match.captured("amp").isEmpty())
+    {
+      retVal.replace(match.capturedStart("amp"), match.capturedLength("amp"), "&");
+    }
+    else if (!match.captured("quot").isEmpty())
+    {
+      retVal.replace(match.capturedStart("quot"), match.capturedLength("quot"), "\"");
+    }
+    match = regExp.match(retVal, match.capturedStart() + 1);
+  }
+
+  return retVal;
+}
+
 string unescapeUtf8( const string &str, bool saveFormat )
 {
   return string( unescape( QString::fromUtf8( str.c_str(), str.size() ) ).toUtf8().data(), saveFormat );
diff --git a/htmlescape.hh b/htmlescape.hh
index f86e4136..39f3d161 100644
--- a/htmlescape.hh
+++ b/htmlescape.hh
@@ -4,6 +4,7 @@
 #ifndef __HTMLESCAPE_HH_INCLUDED__
 #define __HTMLESCAPE_HH_INCLUDED__
 
+#include <QString>
 #include <string>
 
 namespace Html {
@@ -24,6 +25,8 @@ string escapeForJavaScript( string const & );
 
 // Replace html entities
 QString unescape( QString const & str, bool saveFormat = false );
+
+QString fromHtmlEscaped( QString const & str);
 string unescapeUtf8( string const & str, bool saveFormat = false );
 
 }
diff --git a/mdictparser.cc b/mdictparser.cc
index f3124a46..a4ec7142 100644
--- a/mdictparser.cc
+++ b/mdictparser.cc
@@ -374,7 +374,7 @@ bool MdictParser::readHeader( QDataStream & in )
     {
 #if( QT_VERSION >= QT_VERSION_CHECK( 6, 0, 0 ) )
       styleSheets_[ lines[ i ].toInt() ] =
-        pair< QString, QString >( Html::unescape( lines[ i + 1 ] ),  Html::unescape( lines[ i + 2 ] ) );
+        pair< QString, QString >( Html::fromHtmlEscaped( lines[ i + 1 ] ),  Html::fromHtmlEscaped( lines[ i + 2 ] ) );
 #else
       styleSheets_[ lines[ i ].toInt() ] = pair< QString, QString >( lines[ i + 1 ], lines[ i + 2 ] );
 #endif
diff --git a/utils.hh b/utils.hh
index 6f4b0b1b..a12175af 100644
--- a/utils.hh
+++ b/utils.hh
@@ -78,13 +78,6 @@ inline QString rstripnull(const QString &str) {
   return "";
 }
 
-inline QString unescapeHtml(const QString &str) {
-  QTextDocument text;
-  text.setHtml(str);
-  return text.toPlainText();
-}
-
-
 inline bool isExternalLink(QUrl const &url) {
   return url.scheme() == "http" || url.scheme() == "https" || url.scheme() == "ftp" || url.scheme() == "mailto" ||
          url.scheme() == "file" || url.toString().startsWith( "//" );