diff options
Diffstat (limited to 'src/fscanner.l')
| -rw-r--r-- | src/fscanner.l | 1052 |
1 files changed, 1052 insertions, 0 deletions
diff --git a/src/fscanner.l b/src/fscanner.l new file mode 100644 index 0000000..a0e6959 --- /dev/null +++ b/src/fscanner.l @@ -0,0 +1,1052 @@ +%{ +/*=========================================================================== + Copyright (c) 1998-2000, The Santa Cruz Operation + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + + *Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimer. + + *Redistributions in binary form must reproduce the above copyright notice, + this list of conditions and the following disclaimer in the documentation + and/or other materials provided with the distribution. + + *Neither name of The Santa Cruz Operation nor the names of its contributors + may be used to endorse or promote products derived from this software + without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS + IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, + THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE + LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + INTERRUPTION) + HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH + DAMAGE. + =========================================================================*/ + +/* cscope - interactive C symbol cross-reference + * + * C symbol scanner + */ +#include "global.h" +#include "alloc.h" +#include "scanner.h" +#include "lookup.h" + +#include <assert.h> + +/* the line counting has been moved from character reading for speed */ +/* comments are discarded */ + +#ifndef FLEX_SCANNER +# error Sorry, this scanner needs flex. It is not usable with AT&T Lex. +#endif + +#define IFLEVELINC 5 /* #if nesting level size increment */ + +static char const rcsid[] = "$Id: fscanner.l,v 1.12 2006/09/30 20:29:14 broeker Exp $"; + +int first; /* buffer index for first char of symbol */ +int last; /* buffer index for last char of symbol */ +int lineno; /* symbol line number */ +int myylineno = 1; + +/* HBB 20001007: new variables, emulating yytext in a way that allows + * the yymore() simulation, my_yymore(), to be used even in the presence of + * yyless(). */ +size_t my_yyleng = 0; +char *my_yytext = NULL; + +static BOOL arraydimension; /* inside array dimension declaration */ +static BOOL bplisting; /* breakpoint listing */ +static int braces; /* unmatched left brace count */ +static BOOL classdef; /* c++ class definition */ +static BOOL elseelif; /* #else or #elif found */ +static BOOL esudef; /* enum/struct/union global definition */ +static BOOL external; /* external definition */ +static int externalbraces; /* external definition outer brace count */ +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 */ +static int parens; /* unmatched left parenthesis count */ +static BOOL ppdefine; /* preprocessor define statement */ +static BOOL pseudoelif; /* pseudo-#elif */ +static BOOL oldtype; /* next identifier is an old type */ +static BOOL rules; /* lex/yacc rules */ +static BOOL sdl; /* sdl file */ +static BOOL structfield; /* structure field declaration */ +static int tagdef; /* class/enum/struct/union tag definition */ +static BOOL template; /* function template */ +static int templateparens; /* function template outer parentheses count */ +static int typedefbraces = -1; /* initial typedef brace count */ +static int token; /* token found */ +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$]* +number \.?[0-9][.0-9a-fA-FlLuUxX]* +comment "/*"([^*]*("*"+[^/])?)*"*/"|"//"[^\n]*\n +ws [ \t\r\v\f] +wsnl [ \t\r\v\f\n]|{comment} + +/* flex options: stack of start conditions, and don't use yywrap() */ +%option stack +%option noyywrap + +%start SDL +%a 4000 +%o 7000 + +/* exclusive start conditions. not available in AT&T lex -> use flex! */ +%x IN_PREPROC WAS_ENDIF WAS_IDENTIFIER WAS_ESU IN_DQUOTE IN_SQUOTE COMMENT + +%% + +%\{ { /* lex/yacc C declarations/definitions */ + global = YES; + goto more; + /* NOTREACHED */ + } +%\} { + global = NO; + goto more; + /* NOTREACHED */ + } +^%% { /* lex/yacc rules delimiter */ + braces = 0; + if (rules == NO) { + /* this %% starts the section containing the rules */ + rules = YES; + + /* Copy yytext to private buffer, to be able to add further + * content following it: */ + my_yymore(); + + /* simulate a yylex() or yyparse() definition */ + (void) strcat(my_yytext, " /* "); + first = strlen(my_yytext); + if (lex == YES) { + (void) strcat(my_yytext, "yylex"); + } else { + /* yacc: yyparse implicitly calls yylex */ + char *s = " yylex()"; + char *cp = s + strlen(s); + while (--cp >= s) { + unput(*cp); + } + (void) strcat(my_yytext, "yyparse"); + } + last = strlen(my_yytext); + (void) strcat(my_yytext, " */"); + my_yyleng = strlen(my_yytext); + return(FCNDEF); + } else { + /* were in the rules section, now comes the closing one */ + rules = NO; + global = YES; + last = first; + my_yymore(); + return(FCNEND); + /* NOTREACHED */ + } + } + +<SDL>STATE[ \t]+({identifier}|\*) { /* sdl state, treat as function def */ + braces = 1; + fcndef = YES; + token = FCNDEF; + goto findident; + /* NOTREACHED */ + } +<SDL>ENDSTATE[ \t] { /* end of an sdl state, treat as end of a function */ + goto endstate; + /* NOTREACHED */ + } + +\{ { /* count unmatched left braces for fcn def detection */ + ++braces; + + /* mark an untagged enum/struct/union so its beginning + can be found */ + if (tagdef) { + if (braces == 1) { + esudef = YES; + } + token = tagdef; + tagdef = '\0'; + last = first; + my_yymore(); + return(token); + } + goto more; + /* NOTREACHED */ + } + +\#{ws}* { /* start a preprocessor line */ + if (rules == NO) /* don't consider CPP for lex/yacc rules */ + BEGIN(IN_PREPROC); + yyleng = 1; /* get rid of the blanks, if any */ + goto more; + /* NOTREACHED */ + } +<IN_PREPROC>endif([^a-zA-Z0-9_$\n].*)? { /* #endif */ + /* delay treatment of #endif depending on whether an + * #if comes right after it, or not */ + /* HBB 20010619: new pattern allows trailing garbage + * after the #endif */ + BEGIN(WAS_ENDIF); + goto more; + /* NOTREACHED */ + } +<WAS_ENDIF>\n{wsnl}*#{ws}*if(ndef|def)?{ws}+ { + /* attempt to correct erroneous brace count caused by: + * + * #if ... + * ... { + * #endif + * #if ... + * ... { + * #endif + */ + /* the current #if must not have an #else or #elif */ + if (elseelif == YES) { + goto endif; + /* NOTREACHED */ + } + pseudoelif = YES; + BEGIN(INITIAL); + yyless(1); /* rescan all but the line ending */ + yy_set_bol(1); + goto eol; + /* NOTREACHED */ + } +<WAS_ENDIF>\n{wsnl}* { /* an #endif with no #if right after it */ + endif: + if (iflevel > 0) { + /* get the maximum brace count for this #if */ + if (braces < maxifbraces[--iflevel]) { + braces = maxifbraces[iflevel]; + } + } + BEGIN(INITIAL); + yyless(1); + yy_set_bol(1); + goto eol; + /* NOTREACHED */ + } + +<IN_PREPROC>ifndef{ws}+ | +<IN_PREPROC>ifdef{ws}+ | +<IN_PREPROC>if{ws}+ { /* #if directive */ + elseelif = NO; + if (pseudoelif == YES) { + pseudoelif = NO; + goto elif; + /* NOTREACHED */ + } + /* make sure there is room for the current brace count */ + if (iflevel == miflevel) { + miflevel += IFLEVELINC; + maxifbraces = myrealloc(maxifbraces, miflevel * sizeof(int)); + preifbraces = myrealloc(preifbraces, miflevel * sizeof(int)); + } + /* push the current brace count */ + preifbraces[iflevel] = braces; + maxifbraces[iflevel++] = 0; + BEGIN(INITIAL); + goto more; + /* NOTREACHED */ + } +<IN_PREPROC>else({ws}.*)? { /* #else --- eat up whole line */ + elseelif = YES; + if (iflevel > 0) { + + /* save the maximum brace count for this #if */ + if (braces > maxifbraces[iflevel - 1]) { + maxifbraces[iflevel - 1] = braces; + } + /* restore the brace count to before the #if */ + braces = preifbraces[iflevel - 1]; + } + BEGIN(INITIAL); + goto more; + /* NOTREACHED */ + } +<IN_PREPROC>elif{ws}+ { /* #elif */ + /* elseelif = YES; --- HBB I doubt this is correct */ + elif: + if (iflevel > 0) { + + /* save the maximum brace count for this #if */ + if (braces > maxifbraces[iflevel - 1]) { + maxifbraces[iflevel - 1] = braces; + } + /* restore the brace count to before the #if */ + braces = preifbraces[iflevel - 1]; + } + BEGIN(INITIAL); + goto more; + /* NOTREACHED */ + } + +<IN_PREPROC>include{ws}*\"[^"\n]+\" | +<IN_PREPROC>include{ws}*<[^>\n]+> { /* #include file */ + char *s; + char remember = yytext[yyleng-1]; + + my_yymore(); + s = strpbrk(my_yytext, "\"<"); + my_yytext[my_yyleng-1] = '\0'; + incfile(s + 1, s); + my_yytext[my_yyleng-1] = remember; + first = s - my_yytext; + last = my_yyleng - 1; + if (compress == YES) { + my_yytext[0] = '\2'; /* compress the keyword */ + } + BEGIN(INITIAL); + return(INCLUDE); + /* NOTREACHED */ + } + +\} { + /* could be the last enum member initializer */ + if (braces == initializerbraces) { + initializerbraces = -1; + initializer = NO; + } + if (--braces <= 0) { + endstate: + braces = 0; + classdef = NO; + } + if (braces == 0 || (braces == 1 && classdef == YES)) { + + /* if the end of an enum/struct/union definition */ + if (esudef == YES) { + esudef = NO; + } + /* if the end of the function */ + else if (fcndef == YES) { + fcndef = NO; + last = first; + my_yymore(); + return(FCNEND); + } + } + goto more; + /* NOTREACHED */ + } + +\( { /* count unmatched left parentheses for function templates */ + ++parens; + goto more; + /* NOTREACHED */ + } +\) { + if (--parens <= 0) { + parens = 0; + } + /* if the end of a function template */ + if (parens == templateparens) { + templateparens = -1; + template = NO; + } + goto more; + /* NOTREACHED */ + } += { /* if a global definition initializer */ + if (global == YES && ppdefine == NO && my_yytext[0] != '#') { + initializerbraces = braces; + initializer = YES; + } + goto more; + /* NOTREACHED */ + } +: { /* a if global structure field */ + if (global == YES && ppdefine == NO && my_yytext[0] != '#') { + structfield = YES; + } + goto more; + /* NOTREACHED */ + } +\, { + if (braces == initializerbraces) { + initializerbraces = -1; + initializer = NO; + } + structfield = NO; + goto more; + /* NOTREACHED */ + } +; { /* if the enum/struct/union was not a definition */ + if (braces == 0) { + esudef = NO; + } + /* if the end of a typedef */ + if (braces == typedefbraces) { + typedefbraces = -1; + } + /* if the end of a external definition */ + if (braces == externalbraces) { + externalbraces = -1; + external = NO; + } + structfield = NO; + initializer = NO; + goto more; + /* NOTREACHED */ + } +<IN_PREPROC>define{ws}+{identifier} { + + /* preprocessor macro or constant definition */ + ppdefine = YES; + token = DEFINE; + if (compress == YES) { + my_yytext[0] = '\1'; /* compress the keyword */ + } + findident: + /* search backwards through yytext[] to find the identifier */ + /* NOTE: this had better be left to flex, by use of + * yet another starting condition */ + my_yymore(); + first = my_yyleng - 1; + while (my_yytext[first] != ' ' && my_yytext[first] != '\t') { + --first; + } + ++first; + last = my_yyleng; + BEGIN(INITIAL); + goto definition; + /* NOTREACHED */ + } +<IN_PREPROC>.|\n | +<IN_PREPROC>{identifier} { /* unknown preprocessor line */ + BEGIN(INITIAL); + goto more; + /* NOTREACHED */ + } + +class{wsnl}+{identifier}({wsnl}|{identifier}|[():])*\{ { /* class definition */ + classdef = YES; + tagdef = 'c'; + yyless(5); /* eat up 'class', and re-scan */ + yy_set_bol(0); + goto more; + /* NOTREACHED */ + } + +("enum"|"struct"|"union") { + ident_start = first; + BEGIN(WAS_ESU); + goto more; + } +<WAS_ESU>{ +({wsnl}+{identifier}){wsnl}*\{ { /* e/s/u definition */ + tagdef = my_yytext[ident_start]; + BEGIN(WAS_IDENTIFIER); + goto ident; + } +{wsnl}*\{ { /* e/s/u definition without a tag */ + tagdef = my_yytext[ident_start]; + BEGIN(INITIAL); + if (braces == 0) { + esudef = YES; + } + last = first; + yyless(0); /* re-scan all this as normal text */ + tagdef = '\0'; + goto more; + } +({wsnl}+{identifier})?{wsnl}* | +.|\n { /* e/s/u usage */ + BEGIN(WAS_IDENTIFIER); + goto ident; + } +} + +if{wsnl}*\( { /* ignore 'if' */ + yyless(2); + yy_set_bol(0); + goto more; +} + +{identifier} { /* identifier found: do nothing, yet. (!) */ + BEGIN(WAS_IDENTIFIER); + ident_start = first; + goto more; + /* NOTREACHED */ + } + +<WAS_IDENTIFIER>{ +{ws}*\(({wsnl}|{identifier}|{number}|[*&[\]=,.])*\)([()]|{wsnl})*[:a-zA-Z_#{] { + /* a function definition */ + /* note: "#define a (b) {" and "#if defined(a)\n#" + * are not fcn definitions! */ + /* warning: "if (...)" must not overflow yytext, + * so the content of function argument definitions + * is restricted, in particular parentheses are + * not allowed */ + /* FIXME HBB 20001003: the above 'not allowed' may well be the + * reason for the parsing bug concerning function pointer usage, + * I suspect. --- I think my new special-case rule for 'if' + * could be helpful in removing that limitation */ + if ((braces == 0 && ppdefine == NO && my_yytext[0] != '#' && rules == NO) || + (braces == 1 && classdef == YES)) { + fcndef = YES; + token = FCNDEF; + goto fcn; + /* NOTREACHED */ + } + goto fcncal; + /* NOTREACHED */ + } +{ws}*\(([*&[\]=,.]|{identifier}|{number}|{wsnl})* { /* function call */ + fcncal: if (fcndef == YES || ppdefine == YES || rules == YES) { + token = FCNCALL; + goto fcn; + /* NOTREACHED */ + } + if (template == NO) { + templateparens = parens; + template = YES; + } + goto ident; + /* NOTREACHED */ + } +("*"|{wsnl})+{identifier} { /* typedef name or modifier use */ + goto ident; + /* NOTREACHED */ + } +.|\n { /* general identifer usage */ + char *s; + + if (global == YES && ppdefine == NO && my_yytext[0] != '#' && + external == NO && initializer == NO && + arraydimension == NO && structfield == NO && + template == NO && fcndef == NO) { + if (esudef == YES) { + /* if enum/struct/union */ + token = MEMBERDEF; + } else { + token = GLOBALDEF; + } + } else { + ident: + token = IDENT; + } + fcn: + if (YYSTATE == WAS_IDENTIFIER) { + /* Position back to the actual identifier: */ + last = first; + first = ident_start; + yyless(0); + /* HBB 20001008: if the anti-backup-pattern above matched, + * and the matched context ended with a \n, then the scanner + * believes it's at the start of a new line. But the yyless() + * should feeds that \n back into the input, so that's + * wrong. --> force 'beginning-of-line' status off. */ + yy_set_bol(0); + BEGIN(INITIAL); + } else { + my_yymore(); + last = my_yyleng; + } + definition: + + /* if a long line */ + if (yyleng > STMTMAX) { + int c; + + /* skip to the end of the line */ + warning("line too long"); + while ((c = skipcomment_input()) > LEXEOF) { + if (c == '\n') { + unput(c); + break; + } + } + } + /* truncate a long symbol */ + if (yyleng > PATLEN) { + warning("symbol too long"); + my_yyleng = first + PATLEN; + my_yytext[my_yyleng] = '\0'; + } + + /* if found word was a keyword: */ + if ((s = lookup(my_yytext + first)) != NULL) { + first = my_yyleng; + + /* if the start of a typedef */ + if (s == typedeftext) { + typedefbraces = braces; + oldtype = YES; + } + /* if an enum/struct/union */ + /* (needed for "typedef struct tag name;" so + tag isn't marked as the typedef name) */ + else if (s == enumtext || s == structtext || s == uniontext) { + /* do nothing */ + } else if (s == externtext) { + /* if an external definition */ + externalbraces = braces; + external = YES; + } else if (templateparens == parens && template == YES) { + /* keyword doesn't start a function + * template */ + templateparens = -1; + template = NO; + } else { + /* identifier after typedef was a + * keyword */ + oldtype = NO; + } + } else { + /* not a keyword --> found an identifier */ + /* last = yyleng; */ + + /* if a class/enum/struct/union tag definition */ + /* FIXME HBB 20001001: why reject "class"? */ + if (tagdef && strnotequal(my_yytext + first, "class")) { + token = tagdef; + tagdef = '\0'; + if (braces == 0) { + esudef = YES; + } + } else if (braces == typedefbraces && oldtype == NO && + arraydimension == NO) { + /* if a typedef name */ + token = TYPEDEF; + } else { + oldtype = NO; + } + /* my_yymore(); */ + return(token); + /* NOTREACHED */ + } + } +} + +\[ { /* array dimension (don't worry or about subscripts) */ + arraydimension = YES; + goto more; + /* NOTREACHED */ + } +\] { + arraydimension = NO; + goto more; + /* NOTREACHED */ + } +\\\n { /* preprocessor statement is continued on next line */ + /* save the '\\' to the output file, but not the '\n': */ + yyleng = 1; + my_yymore(); + goto eol; + /* NOTREACHED */ + } +\n { /* end of the line */ + if (ppdefine == YES) { /* end of a #define */ + ppdefine = NO; + yyless(yyleng - 1); + last = first; + my_yymore(); + return(DEFINEEND); + } + /* skip the first 8 columns of a breakpoint listing line */ + /* and skip the file path in the page header */ + if (bplisting == YES) { + int c, i; + + /* 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; + case '.': /* header line */ + case '/': + /* skip to the end of the line */ + while ((c = skipcomment_input()) > LEXEOF) { + if (c == '\n') { + unput(c); + break; + } + } + break; + case '\n': /* empty line */ + unput('\n'); + break; + } + } + eol: + ++myylineno; + first = 0; + last = 0; + if (symbols > 0) { + /* no my_yymore(): \n doesn't need to be in my_yytext */ + return(NEWLINE); + } + /* line ended --> flush my_yytext */ + if (my_yytext) + *my_yytext = '\0'; + my_yyleng = 0; + lineno = myylineno; + } + +\' { /* character constant */ + if (sdl == NO) + BEGIN(IN_SQUOTE); + goto more; + /* NOTREACHED */ + } +<IN_SQUOTE>\' { + BEGIN(INITIAL); + goto more; + /* NOTREACHED */ + } +\" { /* string constant */ + BEGIN(IN_DQUOTE); + goto more; + /* NOTREACHED */ + } +<IN_DQUOTE>\" { + BEGIN(INITIAL); + goto more; + /* NOTREACHED */ + } +<IN_DQUOTE,IN_SQUOTE>{ +\n { /* syntax error: unexpected EOL */ + BEGIN(INITIAL); + goto eol; + /* NOTREACHED */ + } +\\. | +. { + goto more; + /* NOTREACHED */ + } +\\\n { /* line continuation inside a string! */ + myylineno++; + goto more; + /* NOTREACHED */ + } +} + +^{ws}+ { /* don't save leading white space */ + } + +{ws}+\n { /* eat whitespace at end of line */ + unput('\n'); + } + +[\t\r\v\f]+ { /* eat non-blank whitespace sequences, replace + * by single blank */ + unput(' '); + } + +{ws}{2,} { /* compress sequential whitespace here, not in putcrossref() */ + unput(' '); + } + +"/*" yy_push_state(COMMENT); +<COMMENT>{ +[^*\n]* | +"*"+[^*/\n]* ; /* do nothing */ +[^*\n]*\n | +"*"+[^*/\n]*\n { + if (ppdefine == NO) { + goto eol; + } else { + ++myylineno; + } + /* NOTREACHED */ + } +"*"+"/" { + /* replace the comment by a single blank */ + unput(' '); + yy_pop_state(); + } +} + +"//".*\n? { + /* C++-style one-line comment */ + goto eol; + /* NOTREACHED */ + } + +{number} | /* number */ +<SDL>STATE[ \t]+ | /* ... and other syntax error catchers... */ +. { /* punctuation and operators */ + more: + my_yymore(); + first = my_yyleng; + } + +%% + +void +initscanner(char *srcfile) +{ + char *s; + + if (maxifbraces == NULL) { + maxifbraces = mymalloc(miflevel * sizeof(int)); + preifbraces = mymalloc(miflevel * sizeof(int)); + } + first = 0; /* buffer index for first char of symbol */ + last = 0; /* buffer index for last char of symbol */ + lineno = 1; /* symbol line number */ + myylineno = 1; /* input line number */ + arraydimension = NO; /* inside array dimension declaration */ + bplisting = NO; /* breakpoint listing */ + braces = 0; /* unmatched left brace count */ + classdef = NO; /* c++ class definition */ + elseelif = NO; /* #else or #elif found */ + esudef = NO; /* enum/struct/union global definition */ + external = NO; /* external definition */ + externalbraces = -1; /* external definition outer brace count */ + fcndef = NO; /* function definition */ + global = YES; /* file global scope (outside functions) */ + iflevel = 0; /* #if nesting level */ + initializer = NO; /* data initializer */ + initializerbraces = -1; /* data initializer outer brace count */ + lex = NO; /* lex file */ + parens = 0; /* unmatched left parenthesis count */ + ppdefine = NO; /* preprocessor define statement */ + pseudoelif = NO; /* pseudo-#elif */ + oldtype = NO; /* next identifier is an old type */ + rules = NO; /* lex/yacc rules */ + sdl = NO; /* sdl file */ + structfield = NO; /* structure field declaration */ + tagdef = '\0'; /* class/enum/struct/union tag definition */ + template = NO; /* function template */ + templateparens = -1; /* function template outer parentheses count */ + typedefbraces = -1; /* initial typedef braces count */ + ident_start = 0; /* start of previously found identifier */ + + if (my_yytext) + *my_yytext = '\0'; + my_yyleng = 0; + + BEGIN(INITIAL); + + /* if this is not a C file */ + if ((s = strrchr(srcfile, '.')) != NULL) { + switch (*++s) { /* this switch saves time on C files */ + case 'b': + if (strcmp(s, "bp") == 0) { /* breakpoint listing */ + bplisting = YES; + } + break; + case 'l': + if (strcmp(s, "l") == 0) { /* lex */ + lex = YES; + global = NO; + } + break; + case 's': + if (strcmp(s, "sd") == 0) { /* sdl */ + sdl = YES; + BEGIN(SDL); + } + break; + case 'y': + if (strcmp(s, "y") == 0) { /* yacc */ + global = NO; + } + break; + } + } +} + +#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 +my_yymore(void) +{ + static size_t yytext_size = 0; + + /* my_yytext is an ever-growing buffer. It will not ever + * 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); + } + + strncpy (my_yytext + my_yyleng, yytext, yyleng+1); + my_yyleng += yyleng; +} |