libguile/c-tokenize.lex

   1 %option noyywrap
   2 %option nounput
   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
  22 int yylex(void);
  23
  24 int yyget_lineno (void);
  25 FILE *yyget_in (void);
  26 FILE *yyget_out (void);
  27 int yyget_leng (void);
  28 char *yyget_text (void);
  29 void yyset_lineno (int line_number);
  30 void yyset_in (FILE * in_str);
  31 void yyset_out (FILE * out_str);
  32 int yyget_debug (void);
  33 void yyset_debug (int  bdebug);
  34 int yylex_destroy (void);
  35
  36 int filter_snarfage = 0;
  37 int print = 1;
  38
  39 enum t_state {
  40     SKIP,
  41     MULTILINE,
  42     MULTILINE_COOKIE,
  43     COOKIE
  44 };
  45
  46 enum t_state state = SKIP;
  47 int cookie_was_last = 0;
  48
  49 #define OUT_RAW(type,text) if (print) printf ("(%s . \"%s\")\n", #type, text)
  50
  51 #define OUT_T(type)        OUT_RAW (type, yytext)
  52 #define OUT_S              if (print) printf ("%s\n", yytext)
  53 #define OUT(type)          if (print) printf ("%s\n", #type)
  54
  55 #define IS_COOKIE cookie_was_last = 1
  56 #define IS_NOT_COOKIE cookie_was_last = 0
  57
  58 %}
  59
  60 %%
  61
  62 \/\*(\n|[^*]|\*[^/])*\*\/                  { OUT_T (comment); }
  63
  64 ({SPACE}*(\\\n)*{SPACE}*)+                 ;
  65
  66 ({SPACE}*\n*{SPACE}*)+                     { OUT(eol); }
  67
  68 #.*\n                                      { OUT(hash); IS_NOT_COOKIE; }
  69
  70 {LETTER}({LETTER}|{DIGIT})*                { OUT_T (id); IS_NOT_COOKIE; }
  71
  72 0[xX]{HEXDIGIT}+{INTQUAL}?                 { OUT_RAW (int_hex, yytext + 2); IS_NOT_COOKIE; }
  73 0{OCTDIGIT}+{INTQUAL}?                     { OUT_RAW (int_oct, yytext + 1); IS_NOT_COOKIE; }
  74 {DIGIT}+{INTQUAL}?                         { OUT_T (int_dec); IS_NOT_COOKIE; }
  75
  76 L?\'(\\.|[^\\\'])+\'                       { OUT_T (char); IS_NOT_COOKIE; }
  77
  78 {DIGIT}+{EXPONENT}{FLOQUAL}?               { OUT_T (flo_dec); IS_NOT_COOKIE; }
  79 {DIGIT}*"."{DIGIT}+({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
  80 {DIGIT}+"."{DIGIT}*({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
  81
  82 L?\"(\\.|[^\\\"])*\"                       { OUT_S; IS_NOT_COOKIE; }
  83
  84 "..."                                      { OUT (ellipsis); IS_NOT_COOKIE; }
  85
  86 ">>="                                      { OUT (shift_right_assign); IS_NOT_COOKIE; }
  87 "<<="                                      { OUT (shift_left_assign); IS_NOT_COOKIE; }
  88 "+="                                       { OUT (add_assign); IS_NOT_COOKIE; }
  89 "-="                                       { OUT (sub_assign); IS_NOT_COOKIE; }
  90 "*="                                       { OUT (mul-assign); IS_NOT_COOKIE; }
  91 "/="                                       { OUT (div_assign); IS_NOT_COOKIE; }
  92 "%="                                       { OUT (mod_assign); IS_NOT_COOKIE; }
  93 "&="                                       { OUT (logand_assign); IS_NOT_COOKIE; }
  94 "^="                                       { OUT (logxor_assign); IS_NOT_COOKIE; }
  95 "|="                                       { OUT (logior_assign); IS_NOT_COOKIE; }
  96 ">>"                                       { OUT (right_shift); IS_NOT_COOKIE; }
  97 "<<"                                       { OUT (left_shift); IS_NOT_COOKIE; }
  98 "++"                                       { OUT (inc); IS_NOT_COOKIE; }
  99 "--"                                       { OUT (dec); IS_NOT_COOKIE; }
 100 "->"                                       { OUT (ptr); IS_NOT_COOKIE; }
 101 "&&"                                       { OUT (and); IS_NOT_COOKIE; }
 102 "||"                                       { OUT (or); IS_NOT_COOKIE; }
 103 "<="                                       { OUT (le); IS_NOT_COOKIE; }
 104 ">="                                       { OUT (ge); IS_NOT_COOKIE; }
 105 "=="                                       { OUT (eq); IS_NOT_COOKIE; }
 106 "!="                                       { OUT (ne); IS_NOT_COOKIE; }
 107 ";"                                        { OUT (semicolon); IS_NOT_COOKIE; }
 108
 109 ("{"|"<%")                                 {
 110   OUT (brace_open);
 111   if (filter_snarfage && cookie_was_last && state == COOKIE)
 112     state = MULTILINE;
 113   IS_NOT_COOKIE; }
 114
 115 ("}"|"%>")                                 {
 116   OUT (brace_close);
 117   if (filter_snarfage && cookie_was_last && state == MULTILINE_COOKIE) {
 118     state = SKIP;
 119     print = 0;
 120   }
 121   IS_NOT_COOKIE; }
 122
 123 ","                                        { OUT (comma); IS_NOT_COOKIE; }
 124 ":"                                        { OUT (colon); IS_NOT_COOKIE; }
 125 "="                                        { OUT (assign); IS_NOT_COOKIE; }
 126 "("                                        { OUT (paren_open); IS_NOT_COOKIE; }
 127 ")"                                        { OUT (paren_close); IS_NOT_COOKIE; }
 128 ("["|"<:")                                 { OUT (bracket_open); IS_NOT_COOKIE; }
 129 ("]"|":>")                                 { OUT (bracket_close); IS_NOT_COOKIE; }
 130 "."                                        { OUT (dot); IS_NOT_COOKIE; }
 131 "&"                                        { OUT (amp); IS_NOT_COOKIE; }
 132 "!"                                        { OUT (bang); IS_NOT_COOKIE; }
 133 "~"                                        { OUT (tilde); IS_NOT_COOKIE; }
 134 "-"                                        { OUT (minus); IS_NOT_COOKIE; }
 135 "+"                                        { OUT (plus); IS_NOT_COOKIE; }
 136 "*"                                        { OUT (star); IS_NOT_COOKIE; }
 137 "/"                                        { OUT (slash); IS_NOT_COOKIE; }
 138 "%"                                        { OUT (percent); IS_NOT_COOKIE; }
 139 "<"                                        { OUT (lt); IS_NOT_COOKIE; }
 140 ">"                                        { OUT (gt); IS_NOT_COOKIE; }
 141
 142 \^{WS}*\^                                  {
 143   if (filter_snarfage)
 144     switch (state) {
 145     case SKIP:
 146       state = COOKIE;
 147       print = 1;
 148       OUT (snarf_cookie);
 149       break;
 150     case MULTILINE:
 151     case MULTILINE_COOKIE:
 152       state = MULTILINE_COOKIE;
 153       OUT (snarf_cookie);
 154       break;
 155     case COOKIE:
 156       state = SKIP;
 157       OUT (snarf_cookie);
 158       print = 0;
 159       break;
 160     default:
 161       /* whoops */
 162       abort ();
 163       break;
 164     }
 165   else
 166     OUT (snarf_cookie);
 167
 168   IS_COOKIE; }
 169
 170 "^"                                        { OUT (caret); IS_NOT_COOKIE; }
 171 "|"                                        { OUT (pipe); IS_NOT_COOKIE; }
 172 "?"                                        { OUT (question); IS_NOT_COOKIE; }
 173
 174 .                                          { fprintf (stderr, "*%s", yytext);  fflush (stderr); IS_NOT_COOKIE; }
 175
 176 %%
 177
 178 int
 179 main (int argc, char *argv[])
 180 {
 181     if (argc > 1 && !strcmp (argv[1], "--filter-snarfage")) {
 182         filter_snarfage = 1;
 183         print = 0;
 184     }
 185
 186     yylex ();
 187
 188     return EXIT_SUCCESS;
 189 }
 190
 191 /*
 192   Local Variables:
 193   c-file-style: "gnu"
 194   End:
 195 */