lexobjc.cxx.svn-base

Notepad++ is a generic source code editor (it tries to be anyway) and Notepad replacement written in

// Scintilla source code edit control
/** @file LexCPP.cxx
 ** Lexer for C++, C, Java, Javascript, Resource File and Objective-C
 **/
// Copyright 1998-2002 by Neil Hodgson <[e-mail]>
// The License.txt file describes the conditions under which this software may be distributed.
// Modified by Don <[e-mail]> 2004 to add lexer Object C

#include <stdlib.h>
#include <string.h>
#include <ctype.h>
#include <stdio.h>
#include <stdarg.h>

#include "Platform.h"

#include "PropSet.h"
#include "Accessor.h"
#include "StyleContext.h"
#include "KeyWords.h"
#include "Scintilla.h"
#include "SciLexer.h"

#define KEYWORD_BOXHEADER 1
#define KEYWORD_FOLDCONTRACTED 2

static bool IsOKBeforeRE(const int ch) {
	return (ch == '(') || (ch == '=') || (ch == ',');
}

static inline bool IsAWordChar(const int ch) {
	return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_');
}

static inline bool IsAWordStart(const int ch) {
	return (ch < 0x80) && (isalnum(ch) || ch == '_');
}

static inline bool IsADoxygenChar(const int ch) {
	return (islower(ch) || ch == '$' || ch == '@' ||
	        ch == '\\' || ch == '&' || ch == '<' ||
	        ch == '>' || ch == '#' || ch == '{' ||
	        ch == '}' || ch == '[' || ch == ']');
}

static inline bool IsStateComment(const int state) {
	return ((state == SCE_C_COMMENT) ||
	        (state == SCE_C_COMMENTLINE) ||
	        (state == SCE_C_COMMENTDOC) ||
	        (state == SCE_C_COMMENTDOCKEYWORD) ||
	        (state == SCE_C_COMMENTDOCKEYWORDERROR));
}

static inline bool IsStateString(const int state) {
	return ((state == SCE_C_STRING) || (state == SCE_C_VERBATIM));
}

static void ColouriseObjCDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
                            Accessor &styler, bool caseSensitive) {

	WordList &mainInstrsList = *keywordlists[0]; //Commun Instriction
	WordList &mainTypesList = *keywordlists[1]; //Commun Type
	WordList &DoxygenList = *keywordlists[2]; //Doxygen keyword
	WordList &objcDirectiveList = *keywordlists[3]; // objC Directive
	WordList &objcQualifierList = *keywordlists[4]; //objC Qualifier

	bool stylingWithinPreprocessor = styler.GetPropertyInt("styling.within.preprocessor") != 0;

	// Do not leak onto next line
	if (initStyle == SCE_C_STRINGEOL)
		initStyle = SCE_C_DEFAULT;

	int chPrevNonWhite = ' ';
	int visibleChars = 0;
	bool lastWordWasUUID = false;

	StyleContext sc(startPos, length, initStyle, styler);

	for (; sc.More(); sc.Forward()) 
	{
		if (sc.atLineStart && (sc.state == SCE_C_STRING)) 
		{
			// Prevent SCE_C_STRINGEOL from leaking back to previous line
			sc.SetState(SCE_C_STRING);
		}

		// Handle line continuation generically.
		if (sc.ch == '\\') 
		{
			if (sc.chNext == '\n' || sc.chNext == '\r') 
			{
				sc.Forward();
				if (sc.ch == '\r' && sc.chNext == '\n') 
				{
					sc.Forward();
				}
				continue;
			}
		}

		// Determine if the current state should terminate.
		switch (sc.state)
		{
			case SCE_C_OPERATOR :
			{
				sc.SetState(SCE_C_DEFAULT);
				break;
			} 
			case SCE_C_NUMBER :
			{
				if (!IsAWordChar(sc.ch)) 
					sc.SetState(SCE_C_DEFAULT);
				break;
			} 
			case SCE_C_IDENTIFIER : 
			{
				if (!IsAWordChar(sc.ch) || (sc.ch == '.')) 
				{
					char s[100];
					sc.GetCurrent(s, sizeof(s));
					if (s[0] == '@')
					{
						char *ps = s + 1;
						if (objcDirectiveList.InList(ps))
							sc.ChangeState(SCE_OBJC_DIRECTIVE);
					}
					else 
					{
						if (mainInstrsList.InList(s)) 
						{
							lastWordWasUUID = strcmp(s, "uuid") == 0;
							sc.ChangeState(SCE_C_WORD);
						} 
						else if (mainTypesList.InList(s)) 
						{
							sc.ChangeState(SCE_C_WORD2);
						}
						else if (objcQualifierList.InList(s)) 
						{
							sc.ChangeState(SCE_OBJC_QUALIFIER);
						}
					}
					sc.SetState(SCE_C_DEFAULT);
				}
				break;
			} 
			case SCE_C_PREPROCESSOR :
			{
				if (stylingWithinPreprocessor) 
				{
					if (IsASpace(sc.ch)) 
						sc.SetState(SCE_C_DEFAULT);
				} 
				else 
				{
					if ((sc.atLineEnd) || (sc.Match('/', '*')) || (sc.Match('/', '/'))) 
						sc.SetState(SCE_C_DEFAULT);
				}
				break;
			} 
			case SCE_C_COMMENT :
			{
				if (sc.Match('*', '/')) 
				{
					sc.Forward();
					sc.ForwardSetState(SCE_C_DEFAULT);
				}
				break;
			} 
			case SCE_C_COMMENTDOC :
			{
				if (sc.Match('*', '/')) 
				{
					sc.Forward();
					sc.ForwardSetState(SCE_C_DEFAULT);
				} 
				else if (sc.ch == '@' || sc.ch == '\\') 
				{
					sc.SetState(SCE_C_COMMENTDOCKEYWORD);
				}
				break;
			} 
			case SCE_C_COMMENTLINE :
			case SCE_C_COMMENTLINEDOC :
			{
				if (sc.atLineEnd) 
				{
					sc.SetState(SCE_C_DEFAULT);
					visibleChars = 0;
				}
				break;
			} 
			case SCE_C_COMMENTDOCKEYWORD :
			{
				if (sc.Match('*', '/')) 
				{
					sc.ChangeState(SCE_C_COMMENTDOCKEYWORDERROR);
					sc.Forward();
					sc.ForwardSetState(SCE_C_DEFAULT);
				} 
				else if (!IsADoxygenChar(sc.ch)) 
				{
					char s[100];
					if (caseSensitive) 
						sc.GetCurrent(s, sizeof(s));
					else 
						sc.GetCurrentLowered(s, sizeof(s));

					if (!isspace(sc.ch) || !DoxygenList.InList(s + 1)) 
						sc.ChangeState(SCE_C_COMMENTDOCKEYWORDERROR);

					sc.SetState(SCE_C_COMMENTDOC);
				}
				break;
			}
			case SCE_C_STRING :
			{
				if (sc.ch == '\\') 
				{
					if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\')
						sc.Forward();
				} 
				else if (sc.ch == '\"') 
				{
					sc.ForwardSetState(SCE_C_DEFAULT);
				} 
				else if (sc.atLineEnd) 
				{
					sc.ChangeState(SCE_C_STRINGEOL);
					sc.ForwardSetState(SCE_C_DEFAULT);
					visibleChars = 0;
				}
				break;
			} 
			case SCE_C_CHARACTER :
			{
				if (sc.atLineEnd) 
				{
					sc.ChangeState(SCE_C_STRINGEOL);
					sc.ForwardSetState(SCE_C_DEFAULT);
					visibleChars = 0;
				} 
				else if (sc.ch == '\\') 
				{
					if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') 
					{
						sc.Forward();
					}
				} 
				else if (sc.ch == '\'') 
				{
					sc.ForwardSetState(SCE_C_DEFAULT);
				}
				break;
			} 
			case SCE_C_REGEX :
			{
				if (sc.ch == '\r' || sc.ch == '\n' || sc.ch == '/') 
				{
					sc.ForwardSetState(SCE_C_DEFAULT);
				} 
				else if (sc.ch == '\\') 
				{
					// Gobble up the quoted character
					if (sc.chNext == '\\' || sc.chNext == '/') 
					{
						sc.Forward();
					}
				}
				break;
			} 
			case SCE_C_VERBATIM :
			{
				if (sc.ch == '\"') 
				{
					if (sc.chNext == '\"') 
						sc.Forward();
					else 
						sc.ForwardSetState(SCE_C_DEFAULT);
				}
				break;
			} 
			case SCE_C_UUID :
			{
				if (sc.ch == '\r' || sc.ch == '\n' || sc.ch == ')') 
					sc.SetState(SCE_C_DEFAULT);
				break;
			}
			default :
				break;
		}

		// Determine if a new state should be entered.
		if (sc.state == SCE_C_DEFAULT) 
		{
			if (sc.Match('@', '\"')) 
			{
				sc.SetState(SCE_C_VERBATIM);
				sc.Forward();
			}
			else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) 
			{
				if (lastWordWasUUID) 
				{
					sc.SetState(SCE_C_UUID);
					lastWordWasUUID = false;
				} 
				else 
				{
					sc.SetState(SCE_C_NUMBER);
				}
			} 
			else if (IsAWordStart(sc.ch) || (sc.ch == '@')) 
			{
				if (lastWordWasUUID) 
				{
					sc.SetState(SCE_C_UUID);
					lastWordWasUUID = false;
				} 
				else 
				{
					sc.SetState(SCE_C_IDENTIFIER);
				}
			} 
			else if (sc.Match('/', '*')) 
			{
				if (sc.Match("/**") || sc.Match("/*!")) 
					// Support of Qt/Doxygen doc. style
					sc.SetState(SCE_C_COMMENTDOC);
				else 
					sc.SetState(SCE_C_COMMENT);
				sc.Forward();	// Eat the * so it isn't used for the end of the comment
			} 
			else if (sc.Match('/', '/')) 
			{
				if (sc.Match("///") || sc.Match("//!"))	// Support of Qt/Doxygen doc. style
					sc.SetState(SCE_C_COMMENTLINEDOC);
				else
					sc.SetState(SCE_C_COMMENTLINE);
			} 
			else if (sc.ch == '/' && IsOKBeforeRE(chPrevNonWhite)) 
			{
				sc.SetState(SCE_C_REGEX);
			} 
			else if (sc.ch == '\"') 
			{
				sc.SetState(SCE_C_STRING);
			} 
			else if (sc.ch == '\'') 
			{
				sc.SetState(SCE_C_CHARACTER);
			} 
			else if (sc.ch == '#' && visibleChars == 0) 
			{
				// Preprocessor commands are alone on their line
				sc.SetState(SCE_C_PREPROCESSOR);
				// Skip whitespace between # and preprocessor word
				do {
					sc.Forward();
				} while ((sc.ch == ' ' || sc.ch == '\t') && sc.More());
				if (sc.atLineEnd) 
					sc.SetState(SCE_C_DEFAULT);
			}
			else if (isoperator(static_cast<char>(sc.ch)))
			{
				sc.SetState(SCE_C_OPERATOR);
			}
		}

		if (sc.atLineEnd) {
			// Reset states to begining of colourise so no surprises
			// if different sets of lines lexed.
			chPrevNonWhite = ' ';
			visibleChars = 0;
			lastWordWasUUID = false;
		}
		if (!IsASpace(sc.ch)) {
			chPrevNonWhite = sc.ch;
			visibleChars++;
		}