[Cscope-cvs] CVS: cscope/src fscanner.l,1.17,1.18

[Hans-Bernhard B. <br...@us...>]

Update of /cvsroot/cscope/cscope/src
In directory sfp-cvs-1.v30.ch3.sourceforge.com:/tmp/cvs-serv30004

Modified Files:
	fscanner.l 
Log Message:
Remove hand-coded comment-skipping function in favour of letting flex handle it.

Index: fscanner.l
===================================================================
RCS file: /cvsroot/cscope/cscope/src/fscanner.l,v
retrieving revision 1.17
retrieving revision 1.18
diff -C2 -r1.17 -r1.18
*** fscanner.l	2 Aug 2012 21:48:08 -0000	1.17
--- fscanner.l	20 Nov 2014 21:18:51 -0000	1.18
***************
*** 76,84 ****
  static	BOOL	fcndef;			/* function definition */
  static	BOOL	global;			/* file global scope (outside functions) */
! static	int	iflevel;		/* #if nesting level */
  static	BOOL	initializer;		/* data initializer */
  static	int	initializerbraces;	/* data initializer outer brace count */
  static	BOOL	lex;			/* lex file */
! static	int	miflevel = IFLEVELINC;	/* maximum #if nesting level */
  static	int	*maxifbraces;		/* maximum brace count within #if */
  static	int	*preifbraces;		/* brace count before #if */
--- 76,84 ----
  static	BOOL	fcndef;			/* function definition */
  static	BOOL	global;			/* file global scope (outside functions) */
! static	size_t	iflevel;		/* #if nesting level */
  static	BOOL	initializer;		/* data initializer */
  static	int	initializerbraces;	/* data initializer outer brace count */
  static	BOOL	lex;			/* lex file */
! static	size_t	miflevel = IFLEVELINC;	/* maximum #if nesting level */
  static	int	*maxifbraces;		/* maximum brace count within #if */
  static	int	*preifbraces;		/* brace count before #if */
***************
*** 97,131 ****
  static	int 	ident_start;		/* begin of preceding identifier */
  
- /* If this is defined to 1, use flex rules rather than the input
-  * function to discard comments. The scanner gains quite a bit of
-  * speed this way, because of a large reduction of the number of I/O
-  * system/library calls.  The original skipcomment_input() called
-  * getc() so often that the call overhead of shared libraries
-  * vs. static linking, alone, already caused a sizeable performance
-  * hit (up to 40% gross gain on a cscope -cub of its own source
-  * dir). */
- #define COMMENTS_BY_FLEX 1
- 
- #if !COMMENTS_BY_FLEX
- static	int	skipcomment_input(void);
- static	int	comment(void);
- static	int	insidestring_input(int);
- #endif
- 
  static	void	my_yymore(void);
  
- #if COMMENTS_BY_FLEX
- # define skipcomment_input input
- #else
- 
- # define YY_INPUT(buf,result,max_size)				\
- {								\
- 	int c = skipcomment_input ();				\
- 	result = (c == EOF) ? YY_NULL : (buf[0] = c, 1);	\
- }
- 
- #endif /* !COMMENTS_BY_FLEX*/
- 
- 
  %}
  identifier	[a-zA-Z_$][a-zA-Z_0-9$]*
--- 97,102 ----
***************
*** 293,298 ****
  			if (iflevel == miflevel) {
  				miflevel += IFLEVELINC;
! 				maxifbraces = myrealloc(maxifbraces, miflevel * sizeof(int));
! 				preifbraces = myrealloc(preifbraces, miflevel * sizeof(int));
  			}
  			/* push the current brace count */
--- 264,269 ----
  			if (iflevel == miflevel) {
  				miflevel += IFLEVELINC;
! 				maxifbraces = myrealloc(maxifbraces, miflevel * sizeof(*maxifbraces));
! 				preifbraces = myrealloc(preifbraces, miflevel * sizeof(*preifbraces));
  			}
  			/* push the current brace count */
***************
*** 618,622 ****
  				/* skip to the end of the line */
  				warning("line too long");
! 				while ((c = skipcomment_input()) > LEXEOF) { 
  					if (c == '\n') {
  						unput(c);
--- 589,593 ----
  				/* skip to the end of the line */
  				warning("line too long");
! 				while ((c = input()) > LEXEOF) { 
  					if (c == '\n') {
  						unput(c);
***************
*** 717,724 ****
  
  				/* FIXME HBB 20001007: should call input() instead */
! 				switch (skipcomment_input()) {	/* tab and EOF just fall through */
  				case ' ':	/* breakpoint number line */
  				case '[':
! 					for (i = 1; i < 8 && skipcomment_input() > LEXEOF; ++i)
  						;
  					break;
--- 688,695 ----
  
  				/* FIXME HBB 20001007: should call input() instead */
! 				switch (input()) {	/* tab and EOF just fall through */
  				case ' ':	/* breakpoint number line */
  				case '[':
! 					for (i = 1; i < 8 && input() > LEXEOF; ++i)
  						;
  					break;
***************
*** 726,730 ****
  				case '/':
  					/* skip to the end of the line */
! 					while ((c = skipcomment_input()) > LEXEOF) {
  						if (c == '\n') {
  							unput(c);
--- 697,701 ----
  				case '/':
  					/* skip to the end of the line */
! 					while ((c = input()) > LEXEOF) {
  						if (c == '\n') {
  							unput(c);
***************
*** 850,855 ****
  	
  	if (maxifbraces == NULL) {
! 		maxifbraces = mymalloc(miflevel * sizeof(int));
! 		preifbraces = mymalloc(miflevel * sizeof(int));
  	}
  	first = 0;		/* buffer index for first char of symbol */
--- 821,826 ----
  	
  	if (maxifbraces == NULL) {
! 		maxifbraces = mymalloc(miflevel * sizeof(*maxifbraces));
! 		preifbraces = mymalloc(miflevel * sizeof(*preifbraces));
  	}
  	first = 0;		/* buffer index for first char of symbol */
***************
*** 919,1052 ****
  }
  
- #if !COMMENTS_BY_FLEX
- 
- /* A micro-scanner that serves as the input() function of the
-  * scanner. It throws away any comments in the input, correctly
-  * avoiding doing this inside string/character constants, and knows
-  * about backslash sequences. Now that the main scanner doesn't use
-  * yymore() any longer, this could be replaced by lex rules. Left for
-  * trying later. */
- 
- /* Status variable: If this is non-NUL, it's the character that
- * terminates a string we're currently in. */
- static int string_terminator = '\0';
- 
- /* Helper routine: treat 'c' as a character found inside a
-  * string. Check if this character might be the end of that
-  * string. Backslashes have to be taken care of, for the sake of
-  * "quotes like \"these\" found inside a string". */
- static int
- insidestring_input(int c)
- {
- 	static BOOL was_backslash = NO;
- 	
- 	if ((c == '\\') && (was_backslash == NO)) {
- 		/* escape character found --> treat next char specially */
- 		/* FIXME HBB 20001003: need treatment of backslash in the main
- 		 * scanner, too. It'll get false line counts in case of "\\'",
- 		 * otherwise --- they can occur as part of a lex pattern */
- 		was_backslash = YES;
- 		return c;
- 	}
- 
- 	if (((c == '\t') && (lex == YES))
- 	    /* Note: "\\\n" is removed even inside strings! */
- 	    || ((c == '\n') && (was_backslash == NO))
- 		|| (c == EOF)
- 		|| ((c == string_terminator) && (was_backslash == NO))
- 	   ) {
- 		/* Line ended, or end-of-string was found. That is a syntax
- 		 * error.  To recover, stop treatment as a string constant: */
- 		string_terminator = '\0';
- 	} else if (!isprint((unsigned char)c)) {
- 		/* mask unprintable characters */
- 		c = ' ';
- 	}
- 	
- 	was_backslash = NO;
- 	return c;
- }
- 
- /* Helper function: skip over input until end of comment is found (or
-  * we find that it wasn't really comment, in the first place): */
- static int
- comment(void)
- {
- 	int	c, lastc;
- 
- 	/* Coming here, we've just read in the opening '/' of a
- 	 * comment. */
- 	do {
- 		if ((c = getc(yyin)) == '*') {	/* C comment */
- 			lastc = '\0';
- 			while ((c = getc(yyin)) != EOF
- 				   /* fewer '/'s --> test them first! */
- 			       && (c != '/' || lastc != '*')
- 			      ) { 
- 				if (c == '\n') {
- 					/* keep the line number count */
- 					/* FIXME HBB 20001008: this is not synchronized
- 					 * properly with myylineno changes by the main
- 					 * scanner. A strong point in favour of moving
- 					 * this to lex-code that is, IMHO */
- 					++myylineno;
- 				}
- 				lastc = c;
- 			}
- 			/* return a blank for Reiser cpp token concatenation */
- 			/* FIXME HBB 20001008: what on earth is 'Reiser cpp'? ANSI
- 			 * C defines cpp to explicitly replace any comment by a
- 			 * blank. Pre-ANSI cpp's behaved differently, but do we
- 			 * really want that? If at all, it should only ever be a
- 			 * non-default option (like gcc's "-traditional-cpp")
- 			 * */
- 			if ((c = getc(yyin)) == '_' || isalnum(c)) {
- 				(void) ungetc(c, yyin);
- 				c = ' ';
- 				break;
- 			}
- 		} else if (c == '/') {		/* C++ comment */
- 			while ((c = getc(yyin)) != EOF && c != '\n') {
- 				; /* do nothing else */
- 			}
- 			break;
- 		} else {					/* not a comment */
- 			(void) ungetc(c, yyin);
- 			c = '/';
- 			break;
- 			/* NOTREACHED */
- 		}
- 		
- 	/* there may be an immediately following comment */
- 	} while (c == '/');
- 	return(c);
- }
- 
- /* The core of the actual input() function to be used by (f)lex. The
-  * calling scheme between this and the actual input() redefinition is
-  * a bit different for lex and flex. See the #ifdef FLEX_SCANNER part
-  * in the head section. */
- static int
- skipcomment_input(void)
- {
- 	int	c;
- 
- 	c = getc (yyin);
- 	if (string_terminator != '\0') {
- 		/* don't look for comments inside strings! */
- 		return insidestring_input(c);
- 	} else if (c == '/') {
- 		/* swallow everything until end of comment, if this is one */
- 		return comment (); 
- 	} else if (c == '"' || c == '\'') {
- 		/* a string is beginning here, so switch input method */
- 		string_terminator = c;
- 	}
- 	
- 	return c;
- }
- 
- #endif /* !COMMENTS_BY_FLEX */
- 
  #define MY_YY_ALLOCSTEP 1000
  static void
--- 890,893 ----
***************
*** 1058,1063 ****
  	 * shrink, nor will it be freed at end of program, for now */
  	while (my_yyleng + yyleng + 1 >= yytext_size) {
! 		my_yytext = myrealloc(my_yytext, 
! 				      		  yytext_size += MY_YY_ALLOCSTEP);
  	}
  	
--- 899,903 ----
  	 * shrink, nor will it be freed at end of program, for now */
  	while (my_yyleng + yyleng + 1 >= yytext_size) {
! 		my_yytext = myrealloc(my_yytext, yytext_size += MY_YY_ALLOCSTEP);
  	}