test/cppscan6.rl

   1 /*
   2  * @LANG: indep
   3  *
   4  * const char *data = ts;
   5  * int len = te - ts;
   6  * cout << "<" << tok << "> ";
   7  * for ( int i = 0; i < len; i++ )
   8  *     cout << data[i];
   9  * cout << '\n';
  10  */
  11 ptr ts;
  12 ptr te;
  13 int act;
  14 int token;
  15 %%
  16 %%{
  17         machine scanner;
  18
  19         action comment {
  20                 token = 242;
  21                 prints "<";
  22                 printi token;
  23                 prints "> ";
  24                 print_token;
  25                 prints "\n";
  26         }
  27
  28
  29         main := |*
  30
  31         # Single and double literals.
  32         ( 'L'? "'" ( [^'\\\n] | '\\' any )* "'" )
  33                 => {
  34                         token = 193;
  35                         prints "<";
  36                         printi token;
  37                         prints "> ";
  38                         print_token;
  39                         prints "\n";
  40                 };
  41         ( 'L'? '"' ( [^"\\\n] | '\\' any )* '"' )
  42                 => {
  43                         token = 192;
  44                         prints "<";
  45                         printi token;
  46                         prints "> ";
  47                         print_token;
  48                         prints "\n";
  49                 };
  50
  51         # Identifiers
  52         ( [a-zA-Z_] [a-zA-Z0-9_]* )
  53                 =>{
  54                         token = 195;
  55                         prints "<";
  56                         printi token;
  57                         prints "> ";
  58                         print_token;
  59                         prints "\n";
  60                 };
  61
  62         # Floating literals.
  63         fract_const = digit* '.' digit+ | digit+ '.';
  64         exponent = [eE] [+\-]? digit+;
  65         float_suffix = [flFL];
  66
  67         ( fract_const exponent? float_suffix? |
  68                 digit+ exponent float_suffix? )
  69                 => {
  70                         token = 194;
  71                         prints "<";
  72                         printi token;
  73                         prints "> ";
  74                         print_token;
  75                         prints "\n";
  76                 };
  77
  78         # Integer decimal. Leading part buffered by float.
  79         ( ( '0' | [1-9] [0-9]* ) [ulUL]? )
  80                 => {
  81                         token = 218;
  82                         prints "<";
  83                         printi token;
  84                         prints "> ";
  85                         print_token;
  86                         prints "\n";
  87                 };
  88
  89         # Integer octal. Leading part buffered by float.
  90         ( '0' [0-9]+ [ulUL]? )
  91                 => {
  92                         token = 219;
  93                         prints "<";
  94                         printi token;
  95                         prints "> ";
  96                         print_token;
  97                         prints "\n";
  98                 };
  99
 100         # Integer hex. Leading 0 buffered by float.
 101         ( '0' ( 'x' [0-9a-fA-F]+ [ulUL]? ) )
 102                 => {
 103                         token = 220;
 104                         prints "<";
 105                         printi token;
 106                         prints "> ";
 107                         print_token;
 108                         prints "\n";
 109                 };
 110
 111         # Only buffer the second item, first buffered by symbol.
 112         '::' => {
 113                 token = 197;
 114                 prints "<";
 115                 printi token;
 116                 prints "> ";
 117                 print_token;
 118                 prints "\n";
 119         };
 120         '==' => {
 121                 token = 223;
 122                 prints "<";
 123                 printi token;
 124                 prints "> ";
 125                 print_token;
 126                 prints "\n";
 127         };
 128         '!=' => {
 129                 token = 224;
 130                 prints "<";
 131                 printi token;
 132                 prints "> ";
 133                 print_token;
 134                 prints "\n";
 135         };
 136         '&&' => {
 137                 token = 225;
 138                 prints "<";
 139                 printi token;
 140                 prints "> ";
 141                 print_token;
 142                 prints "\n";
 143         };
 144         '||' => {
 145                 token = 226;
 146                 prints "<";
 147                 printi token;
 148                 prints "> ";
 149                 print_token;
 150                 prints "\n";
 151         };
 152         '*=' => {
 153                 token = 227;
 154                 prints "<";
 155                 printi token;
 156                 prints "> ";
 157                 print_token;
 158                 prints "\n";
 159         };
 160         '/=' => {
 161                 token = 228;
 162                 prints "<";
 163                 printi token;
 164                 prints "> ";
 165                 print_token;
 166                 prints "\n";
 167         };
 168         '%=' => {
 169                 token = 229;
 170                 prints "<";
 171                 printi token;
 172                 prints "> ";
 173                 print_token;
 174                 prints "\n";
 175         };
 176         '+=' => {
 177                 token = 230;
 178                 prints "<";
 179                 printi token;
 180                 prints "> ";
 181                 print_token;
 182                 prints "\n";
 183         };
 184         '-=' => {
 185                 token = 231;
 186                 prints "<";
 187                 printi token;
 188                 prints "> ";
 189                 print_token;
 190                 prints "\n";
 191         };
 192         '&=' => {
 193                 token = 232;
 194                 prints "<";
 195                 printi token;
 196                 prints "> ";
 197                 print_token;
 198                 prints "\n";
 199         };
 200         '^=' => {
 201                 token = 233;
 202                 prints "<";
 203                 printi token;
 204                 prints "> ";
 205                 print_token;
 206                 prints "\n";
 207         };
 208         '|=' => {
 209                 token = 234;
 210                 prints "<";
 211                 printi token;
 212                 prints "> ";
 213                 print_token;
 214                 prints "\n";
 215         };
 216         '++' => {
 217                 token = 212;
 218                 prints "<";
 219                 printi token;
 220                 prints "> ";
 221                 print_token;
 222                 prints "\n";
 223         };
 224         '--' => {
 225                 token = 213;
 226                 prints "<";
 227                 printi token;
 228                 prints "> ";
 229                 print_token;
 230                 prints "\n";
 231         };
 232         '->' => {
 233                 token = 211;
 234                 prints "<";
 235                 printi token;
 236                 prints "> ";
 237                 print_token;
 238                 prints "\n";
 239         };
 240         '->*' => {
 241                 token = 214;
 242                 prints "<";
 243                 printi token;
 244                 prints "> ";
 245                 print_token;
 246                 prints "\n";
 247         };
 248         '.*' => {
 249                 token = 215;
 250                 prints "<";
 251                 printi token;
 252                 prints "> ";
 253                 print_token;
 254                 prints "\n";
 255         };
 256
 257         # Three char compounds, first item already buffered.
 258         '...' => {
 259                 token = 240;
 260                 prints "<";
 261                 printi token;
 262                 prints "> ";
 263                 print_token;
 264                 prints "\n";
 265         };
 266
 267         # Single char symbols.
 268         ( punct - [_"'] ) => {
 269                 token = first_token_char;
 270                 prints "<";
 271                 printi token;
 272                 prints "> ";
 273                 print_token;
 274                 prints "\n";
 275         };
 276
 277         # Comments and whitespace.
 278         '/!' ( any* $0 '!/' @1 ) => comment;
 279         '//' ( any* $0 '\n' @1 ) => comment;
 280         ( any - 33..126 )+ => {
 281                 token = 241;
 282                 prints "<";
 283                 printi token;
 284                 prints "> ";
 285                 print_token;
 286                 prints "\n";
 287         };
 288         *|;
 289 }%%
 290 /* _____INPUT_____
 291 "\"\\\"hi\" /!\n!/\n44 .44\n44. 44\n44 . 44\n44.44\n_hithere22"
 292 "'\\''\"\\n\\d'\\\"\"\nhi\n99\n.99\n99e-4\n->*\n||\n0x98\n0x\n//\n/! * !/"
 293 "'\n'\n"
 294 _____INPUT_____ */
 295 /* _____OUTPUT_____
 296 <192> "\"hi"
 297 <241>
 298 <242> /!
 299 !/
 300 <241>
 301
 302 <218> 44
 303 <241>
 304 <194> .44
 305 <241>
 306
 307 <194> 44.
 308 <241>
 309 <218> 44
 310 <241>
 311
 312 <218> 44
 313 <241>
 314 <46> .
 315 <241>
 316 <218> 44
 317 <241>
 318
 319 <194> 44.44
 320 <241>
 321
 322 <195> _hithere22
 323 ACCEPT
 324 <193> '\''
 325 <192> "\n\d'\""
 326 <241>
 327
 328 <195> hi
 329 <241>
 330
 331 <218> 99
 332 <241>
 333
 334 <194> .99
 335 <241>
 336
 337 <194> 99e-4
 338 <241>
 339
 340 <214> ->*
 341 <241>
 342
 343 <226> ||
 344 <241>
 345
 346 <220> 0x98
 347 <241>
 348
 349 <218> 0
 350 <195> x
 351 <241>
 352
 353 <242> //
 354
 355 <242> /! * !/
 356 ACCEPT
 357 FAIL
 358 _____OUTPUT_____ */