New lexical class SCE_C_STRINGRAW for C++0x raw strings. Feature #3054629.

3 files changed, 46 insertions, 2 deletions

diff --git a/include/SciLexer.h b/include/SciLexer.h
index 87dbfe510..118e3927e 100644
--- a/include/SciLexer.h
+++ b/include/SciLexer.h
@@ -151,6 +151,7 @@
 #define SCE_C_COMMENTDOCKEYWORD 17
 #define SCE_C_COMMENTDOCKEYWORDERROR 18
 #define SCE_C_GLOBALCLASS 19
+#define SCE_C_STRINGRAW 20
 #define SCE_D_DEFAULT 0
 #define SCE_D_COMMENT 1
 #define SCE_D_COMMENTLINE 2
diff --git a/include/Scintilla.iface b/include/Scintilla.iface
index 2f28862a5..c78a0e6e5 100644
--- a/include/Scintilla.iface
+++ b/include/Scintilla.iface
@@ -2397,6 +2397,7 @@ val SCE_C_WORD2=16
 val SCE_C_COMMENTDOCKEYWORD=17
 val SCE_C_COMMENTDOCKEYWORDERROR=18
 val SCE_C_GLOBALCLASS=19
+val SCE_C_STRINGRAW=20
 # Lexical states for SCLEX_D
 lex D=SCLEX_D SCE_D_
 val SCE_D_DEFAULT=0
diff --git a/lexers/LexCPP.cxx b/lexers/LexCPP.cxx
index cb021b244..5fc769f51 100644
--- a/lexers/LexCPP.cxx
+++ b/lexers/LexCPP.cxx
@@ -33,6 +33,7 @@
 #include "CharacterSet.h"
 #include "LexerModule.h"
 #include "OptionSet.h"
+#include "SparseState.h"
 
 #ifdef SCI_NAMESPACE
 using namespace Scintilla;
@@ -317,6 +318,7 @@ class LexerCPP : public ILexer {
 	std::map<std::string, std::string> preprocessorDefinitionsStart;
 	OptionsCPP options;
 	OptionSetCPP osCPP;
+	SparseState<std::string> rawStringTerminators;
 public:
 	LexerCPP(bool caseSensitive_) :
 		caseSensitive(caseSensitive_),
@@ -503,6 +505,8 @@ void SCI_METHOD LexerCPP::Lex(unsigned int startPos, int length, int initStyle,
 	}
 
 	const int maskActivity = 0x3F;
+	std::string rawStringTerminator = rawStringTerminators.ValueAt(lineCurrent-1);
+	bool changedRawStringState = rawStringTerminators.Delete(lineCurrent);
 
 	int activitySet = preproc.IsInactive() ? 0x40 : 0;
 
@@ -535,6 +539,10 @@ void SCI_METHOD LexerCPP::Lex(unsigned int startPos, int length, int initStyle,
 		if (sc.atLineEnd) {
 			lineCurrent++;
 			vlls.Add(lineCurrent, preproc);
+			if (rawStringTerminator != "") {
+				rawStringTerminators.Set(lineCurrent-1, rawStringTerminator);
+				changedRawStringState = true;
+			}
 		}
 
 		// Handle line continuation generically.
@@ -578,6 +586,20 @@ void SCI_METHOD LexerCPP::Lex(unsigned int startPos, int length, int initStyle,
 					} else if (keywords4.InList(s)) {
 						sc.ChangeState(SCE_C_GLOBALCLASS|activitySet);
 					}
+					if (sc.ch == '\"') {
+						// Should check all s in "L|u|U|u8" "R"
+						const bool raw = sc.chPrev == 'R';
+						size_t lenS = strlen(s);
+						if (raw)
+							s[lenS--] = '\0';
+						bool valid = 
+							(lenS == 0) ||
+							((lenS == 1) && ((s[0] == 'L') || (s[0] == 'u') || (s[0] == 'U'))) ||
+							((lenS == 2) && (s[0] == 'u') && (s[1] == '8'));
+						if (valid) {
+							sc.ChangeState((raw ? SCE_C_STRINGRAW : SCE_C_STRING)|activitySet);
+						}
+					}
 					sc.SetState(SCE_C_DEFAULT|activitySet);
 				}
 				break;
@@ -662,6 +684,14 @@ void SCI_METHOD LexerCPP::Lex(unsigned int startPos, int length, int initStyle,
 					sc.ForwardSetState(SCE_C_DEFAULT|activitySet);
 				}
 				break;
+			case SCE_C_STRINGRAW:
+				if (sc.Match(rawStringTerminator.c_str())) {
+					for (size_t termPos=rawStringTerminator.size(); termPos; termPos--)
+						sc.Forward();
+					sc.SetState(SCE_C_DEFAULT|activitySet);
+					rawStringTerminator = "";
+				}
+				break;
 			case SCE_C_CHARACTER:
 				if (sc.atLineEnd) {
 					sc.ChangeState(SCE_C_STRINGEOL|activitySet);
@@ -753,7 +783,19 @@ void SCI_METHOD LexerCPP::Lex(unsigned int startPos, int length, int initStyle,
 				       || !FollowsPostfixOperator(sc, styler))) {
 				sc.SetState(SCE_C_REGEX|activitySet);	// JavaScript's RegEx
 			} else if (sc.ch == '\"') {
-				sc.SetState(SCE_C_STRING|activitySet);
+				if (sc.chPrev == 'R') {
+					sc.SetState(SCE_C_STRINGRAW|activitySet);
+					rawStringTerminator = ")";
+					for (int termPos = sc.currentPos + 1;;termPos++) {
+						char chTerminator = styler.SafeGetCharAt(termPos, '(');
+						if (chTerminator == '(')
+							break;
+						rawStringTerminator += chTerminator;
+					}
+					rawStringTerminator += '\"';
+				} else {
+					sc.SetState(SCE_C_STRING|activitySet);
+				}
 				isIncludePreprocessor = false;	// ensure that '>' won't end the string
 			} else if (isIncludePreprocessor && sc.ch == '<') {
 				sc.SetState(SCE_C_STRING|activitySet);
@@ -848,7 +890,7 @@ void SCI_METHOD LexerCPP::Lex(unsigned int startPos, int length, int initStyle,
 		}
 		continuationLine = false;
 	}
-	if (definitionsChanged)
+	if (definitionsChanged || changedRawStringState)
 		styler.ChangeLexerState(startPos, startPos + length);
 	sc.Complete();
 }