C++: fix multi-byte character handling in input.

Temporary fix: if a single byte is found with the highest bit set, then
convert from utf8 to latin1. This can be removed when the lexer can
handle multi-byte characters.

Task-number: QTCREATORBUG-10141
Change-Id: I36a17aa18bd1b2378f12d0cecf4fd4957b38d8f2
Reviewed-by: Nikolai Kosjar <nikolai.kosjar@digia.com>
Reviewed-by: Orgad Shaneh <orgads@gmail.com>

1 files changed, 13 insertions, 0 deletions

diff --git a/src/plugins/cpptools/cpppreprocessor.cpp b/src/plugins/cpptools/cpppreprocessor.cpp
index 85ba300fc3..48cb7fc77c 100644
--- a/src/plugins/cpptools/cpppreprocessor.cpp
+++ b/src/plugins/cpptools/cpppreprocessor.cpp
@@ -351,6 +351,18 @@ void CppPreprocessor::stopSkippingBlocks(unsigned offset)
         m_currentDoc->stopSkippingBlocks(offset);
 }
 
+// This is a temporary fix to handle non-ascii characters. This can be removed when the lexer can
+// handle multi-byte characters.
+static QByteArray convertToLatin1(const QByteArray &contents)
+{
+    const char *p = contents.constData();
+    while (char ch = *p++)
+        if (ch & 0x80)
+            return QString::fromUtf8(contents).toLatin1();
+
+    return contents;
+}
+
 void CppPreprocessor::sourceNeeded(unsigned line, const QString &fileName, IncludeType type)
 {
     if (fileName.isEmpty())
@@ -368,6 +380,7 @@ void CppPreprocessor::sourceNeeded(unsigned line, const QString &fileName, Inclu
     unsigned editorRevision = 0;
     QByteArray contents;
     getFileContents(absoluteFileName, &contents, &editorRevision);
+    contents = convertToLatin1(contents);
     if (m_currentDoc) {
         if (contents.isEmpty() && !QFileInfo(absoluteFileName).isAbsolute()) {
             QString msg = QCoreApplication::translate(