C++: Fix crash with invalid raw string literal

While parsing a document Bind::asStringLiteral() Token::spell() was
called for a raw string literal token with a Token::literal nullptr.

This is due scanRawStringLiteral() not properly aborting for
invalid/incomplete code and that the code paths handling
multi-line-raw-strings were not limited to the highlighting case.

Address both cases.

Task-number: QTCREATORBUG-18941
Change-Id: I489d288ccbd7b59be396dada846613ff555436cf
Reviewed-by: Eike Ziller <eike.ziller@qt.io>
Reviewed-by: Orgad Shaneh <orgads@gmail.com>

3 files changed, 31 insertions, 6 deletions

diff --git a/src/libs/3rdparty/cplusplus/Lexer.cpp b/src/libs/3rdparty/cplusplus/Lexer.cpp
index e62c9412a4..079ae0ca6e 100644
--- a/src/libs/3rdparty/cplusplus/Lexer.cpp
+++ b/src/libs/3rdparty/cplusplus/Lexer.cpp
@@ -211,7 +211,7 @@ void Lexer::scan_helper(Token *tok)
         _state = 0;
         scanCppComment(originalKind);
         return;
-    } else if (isRawStringLiteral(s._tokenKind)) {
+    } else if (!control() && isRawStringLiteral(s._tokenKind)) {
         tok->f.kind = s._tokenKind;
         if (scanUntilRawStringLiteralEndSimple())
             _state = 0;
@@ -755,13 +755,17 @@ void Lexer::scanRawStringLiteral(Token *tok, unsigned char hint)
             yyinp();
         } else if (_yychar == ')') {
             yyinp();
-            if (delimLength == -1)
-                break;
+            if (delimLength == -1) {
+                tok->f.kind = T_ERROR;
+                return;
+            }
             closingDelimCandidate = _currentChar;
         } else {
             if (delimLength == -1) {
-                if (_yychar == '\\' || std::isspace(_yychar))
-                    break;
+                if (_yychar == '\\' || std::isspace(_yychar)) {
+                    tok->f.kind = T_ERROR;
+                    return;
+                }
                 yyinp();
             } else {
                 if (!closingDelimCandidate) {
@@ -804,7 +808,7 @@ void Lexer::scanRawStringLiteral(Token *tok, unsigned char hint)
     else
         tok->f.kind = T_RAW_STRING_LITERAL;
 
-    if (!closed)
+    if (!control() && !closed)
         s._tokenKind = tok->f.kind;
 }
 
diff --git a/tests/auto/cplusplus/lexer/tst_lexer.cpp b/tests/auto/cplusplus/lexer/tst_lexer.cpp
index a2c415d33c..1721efaedd 100644
--- a/tests/auto/cplusplus/lexer/tst_lexer.cpp
+++ b/tests/auto/cplusplus/lexer/tst_lexer.cpp
@@ -366,6 +366,14 @@ void tst_SimpleLexer::literals_data()
                             << T_RAW_STRING_LITERAL
                                ;
     QTest::newRow("raw-string-literals") << source << expectedTokenKindList;
+
+    source = "R\"\\" ;
+    expectedTokenKindList = TokenKindList() << T_ERROR;
+    QTest::newRow("invalid-raw-string-literals1") << source << expectedTokenKindList;
+
+    source = "R\")" ;
+    expectedTokenKindList = TokenKindList() << T_ERROR;
+    QTest::newRow("invalid-raw-string-literals2") << source << expectedTokenKindList;
 }
 
 void tst_SimpleLexer::preprocessor()
diff --git a/tests/auto/cplusplus/misc/tst_misc.cpp b/tests/auto/cplusplus/misc/tst_misc.cpp
index aff5a12ce0..808408df40 100644
--- a/tests/auto/cplusplus/misc/tst_misc.cpp
+++ b/tests/auto/cplusplus/misc/tst_misc.cpp
@@ -48,6 +48,8 @@ private slots:
     void astPathOnGeneratedTokens();
 
     void typeMatcher();
+
+    void doNotCrashForInvalidRawString();
 };
 
 void tst_Misc::diagnosticClient_error()
@@ -266,5 +268,16 @@ void tst_Misc::typeMatcher()
     }
 }
 
+void tst_Misc::doNotCrashForInvalidRawString()
+{
+    const QByteArray src("\n"
+                         "void f() { enum { Size = sizeof(R\"[^\\s]+([^]+)*\") }; }"
+                         "}\n"
+                         );
+    Document::Ptr doc = Document::create("crash");
+    doc->setUtf8Source(src);
+    doc->check();
+}
+
 QTEST_MAIN(tst_Misc)
 #include "tst_misc.moc"