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

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