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 /* Prevent compilation of static input() function in generated scanner
  23    code.  This function is never actually used, and GCC 4.3 will emit
  24    an error for that. */
  25 #define YY_NO_INPUT
  26
  27 int filter_snarfage = 0;
  28 int print = 1;
  29
  30 enum t_state {
  31     SKIP,
  32     MULTILINE,
  33     MULTILINE_COOKIE,
  34     COOKIE
  35 };
  36
  37 enum t_state state = SKIP;
  38 int cookie_was_last = 0;
  39
  40 #define OUT_RAW(type,text) if (print) printf ("(%s . \"%s\")\n", #type, text)
  41
  42 #define OUT_T(type)        OUT_RAW (type, yytext)
  43 #define OUT_S              if (print) printf ("%s\n", yytext)
  44 #define OUT(type)          if (print) printf ("%s\n", #type)
  45
  46 #define IS_COOKIE cookie_was_last = 1
  47 #define IS_NOT_COOKIE cookie_was_last = 0
  48
  49 %}
  50
  51 %%
  52
  53 \/\*(\n|[^*]|\*[^/])*\*\/                  { OUT_T (comment); }
  54
  55 ({SPACE}*(\\\n)*{SPACE}*)+                 ;
  56
  57 ({SPACE}*\n*{SPACE}*)+                     { OUT(eol); }
  58
  59 #.*\n                                      { OUT(hash); IS_NOT_COOKIE; }
  60
  61 {LETTER}({LETTER}|{DIGIT})*                { OUT_T (id); IS_NOT_COOKIE; }
  62
  63 0[xX]{HEXDIGIT}+{INTQUAL}?                 { OUT_RAW (int_hex, yytext + 2); IS_NOT_COOKIE; }
  64 0{OCTDIGIT}+{INTQUAL}?                     { OUT_RAW (int_oct, yytext + 1); IS_NOT_COOKIE; }
  65 {DIGIT}+{INTQUAL}?                         { OUT_T (int_dec); IS_NOT_COOKIE; }
  66
  67 L?\'(\\.|[^\\\'])+\'                       { OUT_T (char); IS_NOT_COOKIE; }
  68
  69 {DIGIT}+{EXPONENT}{FLOQUAL}?               { OUT_T (flo_dec); IS_NOT_COOKIE; }
  70 {DIGIT}*"."{DIGIT}+({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
  71 {DIGIT}+"."{DIGIT}*({EXPONENT})?{FLOQUAL}? { OUT_T (flo_dec); IS_NOT_COOKIE; }
  72
  73 L?\"(\\.|[^\\\"])*\"                       { OUT_S; IS_NOT_COOKIE; }
  74
  75 "..."                                      { OUT (ellipsis); IS_NOT_COOKIE; }
  76
  77 ">>="                                      { OUT (shift_right_assign); IS_NOT_COOKIE; }
  78 "<<="                                      { OUT (shift_left_assign); IS_NOT_COOKIE; }
  79 "+="                                       { OUT (add_assign); IS_NOT_COOKIE; }
  80 "-="                                       { OUT (sub_assign); IS_NOT_COOKIE; }
  81 "*="                                       { OUT (mul-assign); IS_NOT_COOKIE; }
  82 "/="                                       { OUT (div_assign); IS_NOT_COOKIE; }
  83 "%="                                       { OUT (mod_assign); IS_NOT_COOKIE; }
  84 "&="                                       { OUT (logand_assign); IS_NOT_COOKIE; }
  85 "^="                                       { OUT (logxor_assign); IS_NOT_COOKIE; }
  86 "|="                                       { OUT (logior_assign); IS_NOT_COOKIE; }
  87 ">>"                                       { OUT (right_shift); IS_NOT_COOKIE; }
  88 "<<"                                       { OUT (left_shift); IS_NOT_COOKIE; }
  89 "++"                                       { OUT (inc); IS_NOT_COOKIE; }
  90 "--"                                       { OUT (dec); IS_NOT_COOKIE; }
  91 "->"                                       { OUT (ptr); IS_NOT_COOKIE; }
  92 "&&"                                       { OUT (and); IS_NOT_COOKIE; }
  93 "||"                                       { OUT (or); IS_NOT_COOKIE; }
  94 "<="                                       { OUT (le); IS_NOT_COOKIE; }
  95 ">="                                       { OUT (ge); IS_NOT_COOKIE; }
  96 "=="                                       { OUT (eq); IS_NOT_COOKIE; }
  97 "!="                                       { OUT (ne); IS_NOT_COOKIE; }
  98 ";"                                        { OUT (semicolon); IS_NOT_COOKIE; }
  99
 100 ("{"|"<%")                                 {
 101   OUT (brace_open);
 102   if (filter_snarfage && cookie_was_last && state == COOKIE)
 103     state = MULTILINE;
 104   IS_NOT_COOKIE; }
 105
 106 ("}"|"%>")                                 {
 107   OUT (brace_close);
 108   if (filter_snarfage && cookie_was_last && state == MULTILINE_COOKIE) {
 109     state = SKIP;
 110     print = 0;
 111   }
 112   IS_NOT_COOKIE; }
 113
 114 ","                                        { OUT (comma); IS_NOT_COOKIE; }
 115 ":"                                        { OUT (colon); IS_NOT_COOKIE; }
 116 "="                                        { OUT (assign); IS_NOT_COOKIE; }
 117 "("                                        { OUT (paren_open); IS_NOT_COOKIE; }
 118 ")"                                        { OUT (paren_close); IS_NOT_COOKIE; }
 119 ("["|"<:")                                 { OUT (bracket_open); IS_NOT_COOKIE; }
 120 ("]"|":>")                                 { OUT (bracket_close); IS_NOT_COOKIE; }
 121 "."                                        { OUT (dot); IS_NOT_COOKIE; }
 122 "&"                                        { OUT (amp); IS_NOT_COOKIE; }
 123 "!"                                        { OUT (bang); IS_NOT_COOKIE; }
 124 "~"                                        { OUT (tilde); IS_NOT_COOKIE; }
 125 "-"                                        { OUT (minus); IS_NOT_COOKIE; }
 126 "+"                                        { OUT (plus); IS_NOT_COOKIE; }
 127 "*"                                        { OUT (star); IS_NOT_COOKIE; }
 128 "/"                                        { OUT (slash); IS_NOT_COOKIE; }
 129 "%"                                        { OUT (percent); IS_NOT_COOKIE; }
 130 "<"                                        { OUT (lt); IS_NOT_COOKIE; }
 131 ">"                                        { OUT (gt); IS_NOT_COOKIE; }
 132
 133 \^{WS}*\^                                  {
 134   if (filter_snarfage)
 135     switch (state) {
 136     case SKIP:
 137       state = COOKIE;
 138       print = 1;
 139       OUT (snarf_cookie);
 140       break;
 141     case MULTILINE:
 142     case MULTILINE_COOKIE:
 143       state = MULTILINE_COOKIE;
 144       OUT (snarf_cookie);
 145       break;
 146     case COOKIE:
 147       state = SKIP;
 148       OUT (snarf_cookie);
 149       print = 0;
 150       break;
 151     default:
 152       /* whoops */
 153       abort ();
 154       break;
 155     }
 156   else
 157     OUT (snarf_cookie);
 158
 159   IS_COOKIE; }
 160
 161 "^"                                        { OUT (caret); IS_NOT_COOKIE; }
 162 "|"                                        { OUT (pipe); IS_NOT_COOKIE; }
 163 "?"                                        { OUT (question); IS_NOT_COOKIE; }
 164
 165 .                                          { fprintf (stderr, "*%s", yytext);  fflush (stderr); IS_NOT_COOKIE; }
 166
 167 %%
 168
 169 int
 170 main (int argc, char *argv[])
 171 {
 172     if (argc > 1 && !strcmp (argv[1], "--filter-snarfage")) {
 173         filter_snarfage = 1;
 174         print = 0;
 175     }
 176
 177     yylex ();
 178
 179     return EXIT_SUCCESS;
 180 }
 181
 182 /*
 183   Local Variables:
 184   c-file-style: "gnu"
 185   End:
 186 */