From a041d64a113c864bfe4c8776e670090d9ee0d75f Mon Sep 17 00:00:00 2001
From: cxl <cxl@f0d560ea-af0d-0410-9eb7-867de7ffcac7>
Date: Thu, 11 Sep 2008 20:46:49 +0000
Subject: [PATCH] New heurestics singificantly improves C++ parser in theide

git-svn-id: svn://ultimatepp.org/upp/trunk@427 f0d560ea-af0d-0410-9eb7-867de7ffcac7
---
 uppsrc/CppBase/CppBase.h  | 15 +++++++++++++--
 uppsrc/CppBase/Parser.cpp | 17 +++++++++++++++--
 uppsrc/CppBase/Pre.cpp    |  2 ++
 uppsrc/CppBase/cpplex.cpp | 33 ++++++++++++++++++++++++++++++---
 uppsrc/CtrlLib/init       |  2 +-
 5 files changed, 61 insertions(+), 8 deletions(-)

diff --git a/uppsrc/CppBase/CppBase.h b/uppsrc/CppBase/CppBase.h
index aa2f9f3ac..3d23c5d17 100644
--- a/uppsrc/CppBase/CppBase.h
+++ b/uppsrc/CppBase/CppBase.h
@@ -107,21 +107,25 @@ class Lex {
 		int    code;
 		String text;
 		double number;
+		bool   grounding;
 	};
 
 	bool statsCollected;
 	LexSymbolStat symbolStat;
 	BiVector<Term> term;
-
+	int            body;
+	bool           grounding;
 
 	bool Char(int c)                 { if(*ptr == c) { ptr++; return true; } else return false; }
-	void AddCode(int code)           { Term& tm = term.AddTail(); tm.code = code; tm.ptr = pos; }
+	void AddCode(int code)           { Term& tm = term.AddTail(); tm.code = code; tm.ptr = pos; tm.grounding = grounding; }
 	void AssOp(int noass, int ass)   { AddCode(Char('=') ? ass : noass); }
 	void Next();
 	bool Prepare(int pos);
 	int  GetCharacter();
 
 public:
+	struct Grounding {};
+
 	int         Code(int pos = 0);
 	bool        IsId(int pos = 0);
 	String      Id(int pos = 0);
@@ -141,6 +145,13 @@ public:
 	int         Id(const String& s)         { return id.FindAdd(s) + 256; }
 
 	int         GetBracesLevel() const      { return braceslevel; }
+	void        ClearBracesLevel()          { braceslevel = 0; }
+	
+	void        BeginBody()                 { body++; }
+	void        EndBody()                   { body--; }
+	void        ClearBody()                 { body = 0; }
+	bool        IsBody() const              { return body; }
+	void        SkipToGrounding();
 
 	const char *Pos(int pos = 0);
 	int         operator[](int pos)         { return Code(pos); }
diff --git a/uppsrc/CppBase/Parser.cpp b/uppsrc/CppBase/Parser.cpp
index d7b64a83d..4221426f0 100644
--- a/uppsrc/CppBase/Parser.cpp
+++ b/uppsrc/CppBase/Parser.cpp
@@ -869,6 +869,7 @@ void Parser::Statement()
 bool Parser::EatBody()
 {
 	if(lex != '{') return false;
+	lex.BeginBody();
 	maxScopeDepth = currentScopeDepth = dobody ? 0 : 1;
 	if(dobody) {
 		inbody = true;
@@ -888,6 +889,7 @@ bool Parser::EatBody()
 			maxScopeDepth = max(level, maxScopeDepth);
 		}
 	}
+	lex.EndBody();
 	return true;
 }
 
@@ -1398,9 +1400,20 @@ void Parser::Do(Stream& in, const Vector<String>& ignore, CppBase& _base, const
 		try {
 			Do();
 		}
+		catch(Lex::Grounding) {
+			lex.ClearBracesLevel();
+			lex.ClearBody();
+		}
 		catch(Error) {
-			Resume(lex.GetBracesLevel());
-			Key(';');
+			if(lex.IsBody()) {
+				Resume(lex.GetBracesLevel());
+				Key(';');
+			}
+			else {
+				++lex;
+				lex.SkipToGrounding();
+				lex.ClearBracesLevel();
+			}
 		}
 }
 
diff --git a/uppsrc/CppBase/Pre.cpp b/uppsrc/CppBase/Pre.cpp
index 0d6efe265..41ce66780 100644
--- a/uppsrc/CppBase/Pre.cpp
+++ b/uppsrc/CppBase/Pre.cpp
@@ -73,6 +73,8 @@ SrcFile PreProcess(Stream& in)
 			SLPos(res);
 		}
 		const char *rm = ln;
+		if(IsAlNum(*rm))
+			res.text << '\1';
 		while(*rm == ' ' || *rm == '\t') rm++;
 		if(*rm == '\0')
 			res.blankLinesRemoved++;
diff --git a/uppsrc/CppBase/cpplex.cpp b/uppsrc/CppBase/cpplex.cpp
index ad0ce0e47..e8286d3ce 100644
--- a/uppsrc/CppBase/cpplex.cpp
+++ b/uppsrc/CppBase/cpplex.cpp
@@ -86,7 +86,7 @@ Lex::Lex()
 	for(int i = 0; cppk[i]; i++)
 		id.Add(cppk[i]);
 	endkey = id.GetCount();
-	braceslevel = ignore_low = ignore_high = 0;
+	braceslevel = ignore_low = ignore_high = body = 0;
 }
 
 void Lex::Init(const char *s, const Vector<String>& ig)
@@ -104,13 +104,12 @@ void Lex::StartStatCollection()
 	statsCollected = true;
 }
 
-const LexSymbolStat & Lex::FinishStatCollection()
+const LexSymbolStat& Lex::FinishStatCollection()
 {
 	statsCollected = false;
 	return symbolStat;
 }
 
-
 int Lex::GetCharacter()
 {
 	if(*ptr == '\0') return t_eof;
@@ -151,7 +150,10 @@ int Lex::GetCharacter()
 
 void Lex::Next()
 {
+	grounding = false;
 	while((byte)*ptr <= ' ') {
+		if(*ptr == '\1')
+			grounding = true;
 		if(*ptr == '\0') return;
 		ptr++;
 	}
@@ -369,6 +371,8 @@ void Lex::Get(int n)
 {
 	while(n--) {
 		if(term.GetCount()) {
+			if(body && term.Head().grounding)
+				throw Grounding();
 			int chr = term.Head().code;
 			if(statsCollected)
 				symbolStat.IncStat(chr);
@@ -381,6 +385,29 @@ void Lex::Get(int n)
 		}
 		if(term.GetCount() == 0)
 			Next();
+		if(term.GetCount() == 0)
+			break;
+	}
+}
+
+void Lex::SkipToGrounding()
+{
+	for(;;) {
+		if(term.GetCount() == 0)
+			Next();
+		if(term.GetCount() == 0)
+			break;
+		int chr = term.Head().code;
+		if(chr == t_eof)
+			return;
+		if(term.Head().grounding)
+			return;
+		if(chr == '{')
+			braceslevel++;
+		else
+		if(chr == '}')
+			braceslevel--;
+		term.DropHead();
 	}
 }
 
diff --git a/uppsrc/CtrlLib/init b/uppsrc/CtrlLib/init
index eb04e0781..58fc8e6f3 100644
--- a/uppsrc/CtrlLib/init
+++ b/uppsrc/CtrlLib/init
@@ -3,7 +3,7 @@
 #include "CtrlCore/init"
 #include "RichText/init"
 #include "PdfDraw/init"
-#define BLITZ_INDEX__ FF3059E544238824EE029458E30F1FEC9
+#define BLITZ_INDEX__ F9FBD74A341BB7E4C8A3BC288CE65AF08
 #include "CtrlLib.icpp"
 #undef BLITZ_INDEX__
 #endif