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

%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>

int yylex(void);
  
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); }

#                                          { 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
ac13d9d2	1	%option noyywrap
e80bea70	2	%option nounput
ac13d9d2 ML	3	%pointer
	4
	5	EOL \n
	6	SPACE [ \t\v\f]
	7	WS [ \t\v\n\f]
	8	DIGIT [0-9]
	9	LETTER [a-zA-Z_]
	10	OCTDIGIT [0-7]
	11	HEXDIGIT [a-fA-F0-9]
	12	EXPONENT [Ee][+-]?{DIGIT}+
	13	FLOQUAL (f\|F\|l\|L)
	14	INTQUAL (l\|L\|ll\|LL\|lL\|Ll\|u\|U)
	15
	16	%{
	17
	18	#include <stdio.h>
	19	#include <stdlib.h>
	20	#include <string.h>
	21
e80bea70 RB	22	int yylex(void);
e80bea70 RB	23
ac13d9d2 ML	24	int filter_snarfage = 0;
	25	int print = 1;
	26
	27	enum t_state {
	28	SKIP,
	29	MULTILINE,
	30	MULTILINE_COOKIE,
62024674	31	COOKIE
ac13d9d2 ML	32	};
	33
	34	enum t_state state = SKIP;
	35	int cookie_was_last = 0;
	36
	37	#define OUT_RAW(type,text) if (print) printf ("(%s . \"%s\")\n", #type, text)
	38
	39	#define OUT_T(type) OUT_RAW (type, yytext)
	40	#define OUT_S if (print) printf ("%s\n", yytext)
	41	#define OUT(type) if (print) printf ("%s\n", #type)
	42
	43	#define IS_COOKIE cookie_was_last = 1
	44	#define IS_NOT_COOKIE cookie_was_last = 0
	45
	46	%}
	47
	48	%%
	49
	50	\/\(\n\|[^]\|\[^/])\*\/ { OUT_T (comment); }
	51
	52	({SPACE}(\\\n){SPACE}*)+ ;
	53
	54	({SPACE}\n{SPACE}*)+ { OUT(eol); }
	55
	56	# { OUT(hash); IS_NOT_COOKIE; }
	57
	58	{LETTER}({LETTER}\|{DIGIT})* { OUT_T (id); IS_NOT_COOKIE; }
	59
	60	0[xX]{HEXDIGIT}+{INTQUAL}? { OUT_RAW (int_hex, yytext + 2); IS_NOT_COOKIE; }
	61	0{OCTDIGIT}+{INTQUAL}? { OUT_RAW (int_oct, yytext + 1); IS_NOT_COOKIE; }
	62	{DIGIT}+{INTQUAL}? { OUT_T (int_dec); IS_NOT_COOKIE; }
	63
	64	L?\'(\\.\|[^\\\'])+\' { OUT_T (char); IS_NOT_COOKIE; }
	65
	66	{DIGIT}+{EXPONENT}{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
	67	{DIGIT}*"."{DIGIT}+({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
	68	{DIGIT}+"."{DIGIT}*({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
	69
	70	L?\"(\\.\|[^\\\"])*\" { OUT_S; IS_NOT_COOKIE; }
	71
	72	"..." { OUT (ellipsis); IS_NOT_COOKIE; }
	73
	74	">>=" { OUT (shift_right_assign); IS_NOT_COOKIE; }
	75	"<<=" { OUT (shift_left_assign); IS_NOT_COOKIE; }
	76	"+=" { OUT (add_assign); IS_NOT_COOKIE; }
	77	"-=" { OUT (sub_assign); IS_NOT_COOKIE; }
	78	"*=" { OUT (mul-assign); IS_NOT_COOKIE; }
	79	"/=" { OUT (div_assign); IS_NOT_COOKIE; }
	80	"%=" { OUT (mod_assign); IS_NOT_COOKIE; }
	81	"&=" { OUT (logand_assign); IS_NOT_COOKIE; }
	82	"^=" { OUT (logxor_assign); IS_NOT_COOKIE; }
	83	"\|=" { OUT (logior_assign); IS_NOT_COOKIE; }
	84	">>" { OUT (right_shift); IS_NOT_COOKIE; }
	85	"<<" { OUT (left_shift); IS_NOT_COOKIE; }
	86	"++" { OUT (inc); IS_NOT_COOKIE; }
	87	"--" { OUT (dec); IS_NOT_COOKIE; }
	88	"->" { OUT (ptr); IS_NOT_COOKIE; }
	89	"&&" { OUT (and); IS_NOT_COOKIE; }
	90	"\|\|" { OUT (or); IS_NOT_COOKIE; }
	91	"<=" { OUT (le); IS_NOT_COOKIE; }
	92	">=" { OUT (ge); IS_NOT_COOKIE; }
	93	"==" { OUT (eq); IS_NOT_COOKIE; }
	94	"!=" { OUT (ne); IS_NOT_COOKIE; }
	95	";" { OUT (semicolon); IS_NOT_COOKIE; }
96
97	("{"\|"<%") {
98	OUT (brace_open);
99	if (filter_snarfage && cookie_was_last && state == COOKIE)
100	state = MULTILINE;
101	IS_NOT_COOKIE; }
102
103	("}"\|"%>") {
104	OUT (brace_close);
105	if (filter_snarfage && cookie_was_last && state == MULTILINE_COOKIE) {
106	state = SKIP;
107	print = 0;
108	}
109	IS_NOT_COOKIE; }
110
111	"," { OUT (comma); IS_NOT_COOKIE; }
112	":" { OUT (colon); IS_NOT_COOKIE; }
113	"=" { OUT (assign); IS_NOT_COOKIE; }
114	"(" { OUT (paren_open); IS_NOT_COOKIE; }
115	")" { OUT (paren_close); IS_NOT_COOKIE; }
116	("["\|"<:") { OUT (bracket_open); IS_NOT_COOKIE; }
117	("]"\|":>") { OUT (bracket_close); IS_NOT_COOKIE; }
118	"." { OUT (dot); IS_NOT_COOKIE; }
119	"&" { OUT (amp); IS_NOT_COOKIE; }
120	"!" { OUT (bang); IS_NOT_COOKIE; }
121	"~" { OUT (tilde); IS_NOT_COOKIE; }
122	"-" { OUT (minus); IS_NOT_COOKIE; }
123	"+" { OUT (plus); IS_NOT_COOKIE; }
124	"*" { OUT (star); IS_NOT_COOKIE; }
125	"/" { OUT (slash); IS_NOT_COOKIE; }
126	"%" { OUT (percent); IS_NOT_COOKIE; }
127	"<" { OUT (lt); IS_NOT_COOKIE; }
128	">" { OUT (gt); IS_NOT_COOKIE; }
129
130	\^{WS}*\^ {
131	if (filter_snarfage)
132	switch (state) {
133	case SKIP:
134	state = COOKIE;
135	print = 1;
136	OUT (snarf_cookie);
137	break;
138	case MULTILINE:
139	case MULTILINE_COOKIE:
140	state = MULTILINE_COOKIE;
141	OUT (snarf_cookie);
142	break;
143	case COOKIE:
144	state = SKIP;
145	OUT (snarf_cookie);
146	print = 0;
147	break;
148	default:
149	/* whoops */
150	abort ();
151	break;
152	}
153	else
154	OUT (snarf_cookie);
155
156	IS_COOKIE; }
157
158	"^" { OUT (caret); IS_NOT_COOKIE; }
159	"\|" { OUT (pipe); IS_NOT_COOKIE; }
160	"?" { OUT (question); IS_NOT_COOKIE; }
161
162	. { fprintf (stderr, "*%s", yytext); fflush (stderr); IS_NOT_COOKIE; }
163
164	%%
165
166	int
167	main (int argc, char *argv[])
168	{
169	if (argc > 1 && !strcmp (argv[1], "--filter-snarfage")) {
170	filter_snarfage = 1;
171	print = 0;
172	}
173
174	yylex ();
175
176	return EXIT_SUCCESS;
177	}
178
179	/*
180	Local Variables:
181	c-file-style: "gnu"
182	End:
183	*/