[bpt/guile.git] / libguile / c-tokenize.lex

%top{
/* Include <config.h> before anything else because Gnulib headers such
   as <stdio.h> rely on it.

   However, when cross-compiling, don't include <config.h> because it
   contains information about the host, not about the build.  */
#ifndef CROSS_COMPILING
# include <config.h>
#endif
}

%option noyywrap
%option nounput
%pointer

EOL             \n
SPACE           [ \t\v\f]
WS              [ \t\v\n\f]
DIGIT		[0-9]
LETTER		[a-zA-Z_]
OCTDIGIT	[0-7]
HEXDIGIT	[a-fA-F0-9]
EXPONENT	[Ee][+-]?{DIGIT}+
FLOQUAL		(f|F|l|L)
INTQUAL		(l|L|ll|LL|lL|Ll|u|U)

%{
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

/* Prevent compilation of static input() function in generated scanner
   code.  This function is never actually used, and GCC 4.3 will emit
   an error for that. */
#define YY_NO_INPUT

int filter_snarfage = 0;
int print = 1; 

enum t_state {
    SKIP,
    MULTILINE,
    MULTILINE_COOKIE,
    COOKIE
};

enum t_state state = SKIP;
int cookie_was_last = 0; 

#define OUT_RAW(type,text) if (print) printf ("(%s . \"%s\")\n", #type, text)
 
#define OUT_T(type)        OUT_RAW (type, yytext)
#define OUT_S              if (print) printf ("%s\n", yytext)
#define OUT(type)          if (print) printf ("%s\n", #type)

#define IS_COOKIE cookie_was_last = 1
#define IS_NOT_COOKIE cookie_was_last = 0
 
%}

%%

\/\*(\n|[^*]|\*[^/])*\*\/                  { OUT_T (comment); }

({SPACE}*(\\\n)*{SPACE}*)+                 ;

({SPACE}*\n*{SPACE}*)+                     { OUT(eol); }

#.*\n                                      { OUT(hash); IS_NOT_COOKIE; }

{LETTER}({LETTER}|{DIGIT})*                { OUT_T (id); IS_NOT_COOKIE; }

0[xX]{HEXDIGIT}+{INTQUAL}?                 { OUT_RAW (int_hex, yytext + 2); IS_NOT_COOKIE; }
0{OCTDIGIT}+{INTQUAL}?                     { OUT_RAW (int_oct, yytext + 1); IS_NOT_COOKIE; }
{DIGIT}+{INTQUAL}?                         { OUT_T (int_dec); IS_NOT_COOKIE; }

L?\'(\\.|[^\\\'])+\'                       { OUT_T (char); IS_NOT_COOKIE; }

{DIGIT}+{EXPONENT}{FLOQUAL}?               { OUT_T (flo_dec); IS_NOT_COOKIE; }
{DIGIT}*"."{DIGIT}+({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
{DIGIT}+"."{DIGIT}*({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }

L?\"(\\.|[^\\\"])*\"                       { OUT_S; IS_NOT_COOKIE; }

"..."                                      { OUT (ellipsis); IS_NOT_COOKIE; }

">>="                                      { OUT (shift_right_assign); IS_NOT_COOKIE; }
"<<="                                      { OUT (shift_left_assign); IS_NOT_COOKIE; }
"+="                                       { OUT (add_assign); IS_NOT_COOKIE; }
"-="                                       { OUT (sub_assign); IS_NOT_COOKIE; }
"*="                                       { OUT (mul-assign); IS_NOT_COOKIE; }
"/="                                       { OUT (div_assign); IS_NOT_COOKIE; }
"%="                                       { OUT (mod_assign); IS_NOT_COOKIE; }
"&="                                       { OUT (logand_assign); IS_NOT_COOKIE; }
"^="                                       { OUT (logxor_assign); IS_NOT_COOKIE; }
"|="                                       { OUT (logior_assign); IS_NOT_COOKIE; }
">>"                                       { OUT (right_shift); IS_NOT_COOKIE; }
"<<"                                       { OUT (left_shift); IS_NOT_COOKIE; }
"++"                                       { OUT (inc); IS_NOT_COOKIE; }
"--"                                       { OUT (dec); IS_NOT_COOKIE; }
"->"                                       { OUT (ptr); IS_NOT_COOKIE; }
"&&"                                       { OUT (and); IS_NOT_COOKIE; }
"||"                                       { OUT (or); IS_NOT_COOKIE; }
"<="                                       { OUT (le); IS_NOT_COOKIE; }
">="                                       { OUT (ge); IS_NOT_COOKIE; }
"=="                                       { OUT (eq); IS_NOT_COOKIE; }
"!="                                       { OUT (ne); IS_NOT_COOKIE; }
";"                                        { OUT (semicolon); IS_NOT_COOKIE; }

("{"|"<%")                                 {
  OUT (brace_open);
  if (filter_snarfage && cookie_was_last && state == COOKIE)
    state = MULTILINE;
  IS_NOT_COOKIE; }

("}"|"%>")                                 {
  OUT (brace_close);
  if (filter_snarfage && cookie_was_last && state == MULTILINE_COOKIE) {
    state = SKIP;
    print = 0;
  }
  IS_NOT_COOKIE; }

","                                        { OUT (comma); IS_NOT_COOKIE; }
":"                                        { OUT (colon); IS_NOT_COOKIE; }
"="                                        { OUT (assign); IS_NOT_COOKIE; }
"("                                        { OUT (paren_open); IS_NOT_COOKIE; }
")"                                        { OUT (paren_close); IS_NOT_COOKIE; }
("["|"<:")                                 { OUT (bracket_open); IS_NOT_COOKIE; }
("]"|":>")                                 { OUT (bracket_close); IS_NOT_COOKIE; }
"."                                        { OUT (dot); IS_NOT_COOKIE; }
"&"                                        { OUT (amp); IS_NOT_COOKIE; }
"!"                                        { OUT (bang); IS_NOT_COOKIE; }
"~"                                        { OUT (tilde); IS_NOT_COOKIE; }
"-"                                        { OUT (minus); IS_NOT_COOKIE; }
"+"                                        { OUT (plus); IS_NOT_COOKIE; }
"*"                                        { OUT (star); IS_NOT_COOKIE; }
"/"                                        { OUT (slash); IS_NOT_COOKIE; }
"%"                                        { OUT (percent); IS_NOT_COOKIE; }
"<"                                        { OUT (lt); IS_NOT_COOKIE; }
">"                                        { OUT (gt); IS_NOT_COOKIE; }

\^{WS}*\^                                  {
  if (filter_snarfage)
    switch (state) {
    case SKIP:
      state = COOKIE;
      print = 1;
      OUT (snarf_cookie);
      break;
    case MULTILINE:
    case MULTILINE_COOKIE:
      state = MULTILINE_COOKIE;
      OUT (snarf_cookie);
      break;
    case COOKIE:
      state = SKIP;
      OUT (snarf_cookie);
      print = 0;
      break;
    default:
      /* whoops */
      abort ();
      break;
    }
  else
    OUT (snarf_cookie);
  
  IS_COOKIE; }

"^"                                        { OUT (caret); IS_NOT_COOKIE; }
"|"                                        { OUT (pipe); IS_NOT_COOKIE; }
"?"                                        { OUT (question); IS_NOT_COOKIE; }

.                                          { fprintf (stderr, "*%s", yytext);  fflush (stderr); IS_NOT_COOKIE; }

%%

int
main (int argc, char *argv[])
{
    if (argc > 1 && !strcmp (argv[1], "--filter-snarfage")) {
        filter_snarfage = 1;
        print = 0;
    }

    yylex ();

    return EXIT_SUCCESS;
}

/*
  Local Variables:
  c-file-style: "gnu"
  End:
*/
Commit	Line	Data
e1bb79fd LC	1	%top{
e1bb79fd LC	2	/* Include <config.h> before anything else because Gnulib headers such
8cb0d6d7 LC	3	as <stdio.h> rely on it.
	4
	5	However, when cross-compiling, don't include <config.h> because it
	6	contains information about the host, not about the build. */
	7	#ifndef CROSS_COMPILING
	8	# include <config.h>
	9	#endif
e1bb79fd LC	10	}
e1bb79fd LC	11
ac13d9d2	12	%option noyywrap
e80bea70	13	%option nounput
ac13d9d2 ML	14	%pointer
	15
	16	EOL \n
	17	SPACE [ \t\v\f]
	18	WS [ \t\v\n\f]
	19	DIGIT [0-9]
	20	LETTER [a-zA-Z_]
	21	OCTDIGIT [0-7]
	22	HEXDIGIT [a-fA-F0-9]
	23	EXPONENT [Ee][+-]?{DIGIT}+
	24	FLOQUAL (f\|F\|l\|L)
	25	INTQUAL (l\|L\|ll\|LL\|lL\|Ll\|u\|U)
	26
	27	%{
ac13d9d2 ML	28	#include <stdio.h>
	29	#include <stdlib.h>
	30	#include <string.h>
6d4e1f62 NJ	31
	32	/* Prevent compilation of static input() function in generated scanner
	33	code. This function is never actually used, and GCC 4.3 will emit
	34	an error for that. */
	35	#define YY_NO_INPUT
	36
ac13d9d2 ML	37	int filter_snarfage = 0;
	38	int print = 1;
	39
	40	enum t_state {
	41	SKIP,
	42	MULTILINE,
	43	MULTILINE_COOKIE,
62024674	44	COOKIE
ac13d9d2 ML	45	};
	46
	47	enum t_state state = SKIP;
	48	int cookie_was_last = 0;
	49
	50	#define OUT_RAW(type,text) if (print) printf ("(%s . \"%s\")\n", #type, text)
	51
	52	#define OUT_T(type) OUT_RAW (type, yytext)
	53	#define OUT_S if (print) printf ("%s\n", yytext)
	54	#define OUT(type) if (print) printf ("%s\n", #type)
	55
	56	#define IS_COOKIE cookie_was_last = 1
	57	#define IS_NOT_COOKIE cookie_was_last = 0
	58
	59	%}
	60
	61	%%
	62
	63	\/\(\n\|[^]\|\[^/])\*\/ { OUT_T (comment); }
	64
	65	({SPACE}(\\\n){SPACE}*)+ ;
	66
	67	({SPACE}\n{SPACE}*)+ { OUT(eol); }
	68
6d7c6cf4	69	#.*\n { OUT(hash); IS_NOT_COOKIE; }
ac13d9d2 ML	70
	71	{LETTER}({LETTER}\|{DIGIT})* { OUT_T (id); IS_NOT_COOKIE; }
	72
	73	0[xX]{HEXDIGIT}+{INTQUAL}? { OUT_RAW (int_hex, yytext + 2); IS_NOT_COOKIE; }
	74	0{OCTDIGIT}+{INTQUAL}? { OUT_RAW (int_oct, yytext + 1); IS_NOT_COOKIE; }
	75	{DIGIT}+{INTQUAL}? { OUT_T (int_dec); IS_NOT_COOKIE; }
	76
	77	L?\'(\\.\|[^\\\'])+\' { OUT_T (char); IS_NOT_COOKIE; }
	78
	79	{DIGIT}+{EXPONENT}{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
	80	{DIGIT}*"."{DIGIT}+({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
	81	{DIGIT}+"."{DIGIT}*({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
	82
	83	L?\"(\\.\|[^\\\"])*\" { OUT_S; IS_NOT_COOKIE; }
	84
	85	"..." { OUT (ellipsis); IS_NOT_COOKIE; }
	86
	87	">>=" { OUT (shift_right_assign); IS_NOT_COOKIE; }
	88	"<<=" { OUT (shift_left_assign); IS_NOT_COOKIE; }
	89	"+=" { OUT (add_assign); IS_NOT_COOKIE; }
	90	"-=" { OUT (sub_assign); IS_NOT_COOKIE; }
	91	"*=" { OUT (mul-assign); IS_NOT_COOKIE; }
	92	"/=" { OUT (div_assign); IS_NOT_COOKIE; }
	93	"%=" { OUT (mod_assign); IS_NOT_COOKIE; }
	94	"&=" { OUT (logand_assign); IS_NOT_COOKIE; }
	95	"^=" { OUT (logxor_assign); IS_NOT_COOKIE; }
	96	"\|=" { OUT (logior_assign); IS_NOT_COOKIE; }
	97	">>" { OUT (right_shift); IS_NOT_COOKIE; }
	98	"<<" { OUT (left_shift); IS_NOT_COOKIE; }
	99	"++" { OUT (inc); IS_NOT_COOKIE; }
	100	"--" { OUT (dec); IS_NOT_COOKIE; }
	101	"->" { OUT (ptr); IS_NOT_COOKIE; }
	102	"&&" { OUT (and); IS_NOT_COOKIE; }
	103	"\|\|" { OUT (or); IS_NOT_COOKIE; }
	104	"<=" { OUT (le); IS_NOT_COOKIE; }
	105	">=" { OUT (ge); IS_NOT_COOKIE; }
	106	"==" { OUT (eq); IS_NOT_COOKIE; }
	107	"!=" { OUT (ne); IS_NOT_COOKIE; }
	108	";" { OUT (semicolon); IS_NOT_COOKIE; }
	109
	110	("{"\|"<%") {
	111	OUT (brace_open);
	112	if (filter_snarfage && cookie_was_last && state == COOKIE)
	113	state = MULTILINE;
	114	IS_NOT_COOKIE; }
	115
	116	("}"\|"%>") {
	117	OUT (brace_close);
	118	if (filter_snarfage && cookie_was_last && state == MULTILINE_COOKIE) {
	119	state = SKIP;
	120	print = 0;
	121	}
	122	IS_NOT_COOKIE; }
	123
	124	"," { OUT (comma); IS_NOT_COOKIE; }
	125	":" { OUT (colon); IS_NOT_COOKIE; }
	126	"=" { OUT (assign); IS_NOT_COOKIE; }
	127	"(" { OUT (paren_open); IS_NOT_COOKIE; }
	128	")" { OUT (paren_close); IS_NOT_COOKIE; }
	129	("["\|"<:") { OUT (bracket_open); IS_NOT_COOKIE; }
	130	("]"\|":>") { OUT (bracket_close); IS_NOT_COOKIE; }
	131	"." { OUT (dot); IS_NOT_COOKIE; }
	132	"&" { OUT (amp); IS_NOT_COOKIE; }
	133	"!" { OUT (bang); IS_NOT_COOKIE; }
134	"~" { OUT (tilde); IS_NOT_COOKIE; }
135	"-" { OUT (minus); IS_NOT_COOKIE; }
136	"+" { OUT (plus); IS_NOT_COOKIE; }
137	"*" { OUT (star); IS_NOT_COOKIE; }
138	"/" { OUT (slash); IS_NOT_COOKIE; }
139	"%" { OUT (percent); IS_NOT_COOKIE; }
140	"<" { OUT (lt); IS_NOT_COOKIE; }
141	">" { OUT (gt); IS_NOT_COOKIE; }
142
143	\^{WS}*\^ {
144	if (filter_snarfage)
145	switch (state) {
146	case SKIP:
147	state = COOKIE;
148	print = 1;
149	OUT (snarf_cookie);
150	break;
151	case MULTILINE:
152	case MULTILINE_COOKIE:
153	state = MULTILINE_COOKIE;
154	OUT (snarf_cookie);
155	break;
156	case COOKIE:
157	state = SKIP;
158	OUT (snarf_cookie);
159	print = 0;
160	break;
161	default:
162	/* whoops */
163	abort ();
164	break;
165	}
166	else
167	OUT (snarf_cookie);
168
169	IS_COOKIE; }
170
171	"^" { OUT (caret); IS_NOT_COOKIE; }
172	"\|" { OUT (pipe); IS_NOT_COOKIE; }
173	"?" { OUT (question); IS_NOT_COOKIE; }
174
175	. { fprintf (stderr, "*%s", yytext); fflush (stderr); IS_NOT_COOKIE; }
176
177	%%
178
179	int
180	main (int argc, char *argv[])
181	{
182	if (argc > 1 && !strcmp (argv[1], "--filter-snarfage")) {
183	filter_snarfage = 1;
184	print = 0;
185	}
186
187	yylex ();
188
189	return EXIT_SUCCESS;
190	}
191
192	/*
193	Local Variables:
194	c-file-style: "gnu"
195	End:
196	*/