lib/Haanga/Compiler/Lexer.lex

   1 <?php
   2 /*
   3   +---------------------------------------------------------------------------------+
   4   | Copyright (c) 2010 César Rodas and Menéame Comunicacions S.L.                   |
   5   +---------------------------------------------------------------------------------+
   6   | Redistribution and use in source and binary forms, with or without              |
   7   | modification, are permitted provided that the following conditions are met:     |
   8   | 1. Redistributions of source code must retain the above copyright               |
   9   |    notice, this list of conditions and the following disclaimer.                |
  10   |                                                                                 |
  11   | 2. Redistributions in binary form must reproduce the above copyright            |
  12   |    notice, this list of conditions and the following disclaimer in the          |
  13   |    documentation and/or other materials provided with the distribution.         |
  14   |                                                                                 |
  15   | 3. All advertising materials mentioning features or use of this software        |
  16   |    must display the following acknowledgement:                                  |
  17   |    This product includes software developed by César D. Rodas.                  |
  18   |                                                                                 |
  19   | 4. Neither the name of the César D. Rodas nor the                               |
  20   |    names of its contributors may be used to endorse or promote products         |
  21   |    derived from this software without specific prior written permission.        |
  22   |                                                                                 |
  23   | THIS SOFTWARE IS PROVIDED BY CÉSAR D. RODAS ''AS IS'' AND ANY                   |
  24   | EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED       |
  25   | WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE          |
  26   | DISCLAIMED. IN NO EVENT SHALL CÉSAR D. RODAS BE LIABLE FOR ANY                  |
  27   | DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES      |
  28   | (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;    |
  29   | LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND     |
  30   | ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT      |
  31   | (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS   |
  32   | SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE                     |
  33   +---------------------------------------------------------------------------------+
  34   | Authors: César Rodas <crodas@php.net>                                           |
  35   +---------------------------------------------------------------------------------+
  36 */
  37
  38 class HG_Parser Extends Haanga_Compiler_Parser
  39 {
  40     /* subclass to made easier references to constants */
  41 }
  42
  43 class Haanga_Compiler_Lexer
  44 {
  45     private $data;
  46     private $N;
  47     public $token;
  48     public $value;
  49     private $line;
  50     private $state = 1;
  51
  52     function __construct($data, $compiler)
  53     {
  54         $this->data     = $data;
  55         $this->compiler = $compiler;
  56         $this->N        = 0;
  57         $this->line     = 1;
  58     }
  59
  60     static function init($template, $compiler)
  61     {
  62         $lexer  = new Haanga_Compiler_Lexer($template, $compiler);
  63         $parser = new Haanga_Compiler_Parser($lexer);
  64
  65         $parser->compiler = $compiler;
  66
  67         try {
  68             for($i=0; ; $i++) {
  69                 if  (!$lexer->yylex()) {
  70                     break;
  71                 }
  72                 $parser->doParse($lexer->token, $lexer->value);
  73             }
  74         } catch (Exception $e) {
  75             throw new Haanga_Compiler_Exception($e->getMessage(). ' on line '.$lexer->getLine());
  76         }
  77         $parser->doParse(0, 0);
  78         return (array)$parser->body;
  79     }
  80
  81     function getLine()
  82     {
  83         return $this->line;
  84     }
  85
  86     public $custom_tags=array();
  87
  88     function is_custom_tag()
  89     {
  90         static $tag=NULL;
  91         if (!$tag) {
  92             $tag = Haanga_Extension::getInstance('Tag');
  93         }
  94         $value = $tag->isValid($this->value);
  95         $this->token = $value ? $value : HG_Parser::T_ALPHA;
  96     }
  97
  98 /*!lex2php
  99 %input          $this->data
 100 %counter        $this->N
 101 %token          $this->token
 102 %value          $this->value
 103 %line           $this->line
 104 alpha           = /([a-zA-Z_][a-zA-Z_0-9]*)/
 105 number          = /[0-9]/
 106 numerals        = /([0-9])+/
 107 whitespace      = /[ \r\t\n]+/
 108 html            = /([^{]+(.[^%{#])?)+/
 109 comment         = /([^\#]+\#\})+/
 110 custom_tag_end  = /end([a-zA-Z][a-zA-Z0-9]*)/
 111 token_end       = /[^a-zA-Z0-9_\.]/
 112 single_string   = /[^'\\]+/
 113 double_string   = /[^"\\]+/
 114 */
 115 /*!lex2php
 116 %statename IN_HTML
 117 "{%" {
 118     $this->token = HG_Parser::T_OPEN_TAG;
 119     $this->yypushstate(self::IN_CODE);
 120 }
 121
 122 "{#" {
 123     $this->token = HG_Parser::T_COMMENT_OPEN;
 124     $this->yypushstate(self::IN_COMMENT);
 125 }
 126
 127
 128 "{{" {
 129     $this->token = HG_Parser::T_PRINT_OPEN;
 130     $this->yypushstate(self::IN_PRINT);
 131 }
 132
 133 html {
 134     $this->token = HG_Parser::T_HTML;
 135 }
 136
 137 */
 138 /*!lex2php
 139 %statename IN_CODE
 140 "%}" {
 141     $this->token = HG_Parser::T_CLOSE_TAG;
 142     $this->yypopstate();
 143 }
 144
 145 "->" {
 146     $this->token = HG_Parser::T_OBJ;
 147 }
 148
 149
 150 "." {
 151     $this->token = HG_Parser::T_DOT;
 152 }
 153
 154 "buffer" token_end {
 155     $this->token = HG_Parser::T_BUFFER;
 156 }
 157
 158
 159 "for" token_end {
 160     $this->token = HG_Parser::T_FOR;
 161 }
 162
 163 "empty" token_end {
 164     $this->token = HG_Parser::T_EMPTY;
 165 }
 166
 167 "load" token_end {
 168     $this->token = HG_Parser::T_LOAD;
 169 }
 170
 171 "block" token_end {
 172     $this->token = HG_Parser::T_BLOCK;
 173 }
 174
 175 "&&" {
 176     $this->token = HG_Parser::T_AND;
 177 }
 178
 179 "AND" token_end {
 180     $this->token = HG_Parser::T_AND;
 181 }
 182
 183 "||" {
 184     $this->token = HG_Parser::T_OR;
 185 }
 186
 187 "OR" token_end {
 188     $this->token = HG_Parser::T_OR;
 189 }
 190
 191 "==" {
 192     $this->token = HG_Parser::T_EQ;
 193 }
 194
 195 "!=" {
 196     $this->token = HG_Parser::T_NE;
 197 }
 198
 199 ">=" {
 200     $this->token = HG_Parser::T_GE;
 201 }
 202
 203 "not" token_end {
 204     $this->token = HG_Parser::T_NOT;
 205 }
 206
 207 "!" token_end {
 208     $this->token = HG_Parser::T_NOT;
 209 }
 210
 211
 212 "[" {
 213     $this->token = HG_Parser::T_BRACKETS_OPEN;
 214 }
 215
 216 "]" {
 217     $this->token = HG_Parser::T_BRACKETS_CLOSE;
 218 }
 219
 220 ">" {
 221     $this->token = HG_Parser::T_GT;
 222 }
 223
 224 "<" {
 225     $this->token = HG_Parser::T_LT;
 226 }
 227 "=<" {
 228     $this->token = HG_Parser::T_LE;
 229 }
 230
 231 "|" {
 232     $this->token = HG_Parser::T_PIPE;
 233 }
 234
 235 ":" {
 236     $this->token = HG_Parser::T_COLON;
 237 }
 238
 239 "filter" token_end {
 240     $this->token = HG_Parser::T_FILTER;
 241 }
 242
 243 "regroup" token_end {
 244     $this->token = HG_Parser::T_REGROUP;
 245 }
 246
 247 "endfilter" token_end {
 248     $this->token = HG_Parser::T_END_FILTER;
 249 }
 250
 251 "autoescape" token_end {
 252     $this->token = HG_Parser::T_AUTOESCAPE;
 253 }
 254
 255 "spacefull" token_end {
 256     $this->token = HG_Parser::T_SPACEFULL;
 257 }
 258
 259
 260 "endautoescape" token_end {
 261     $this->token = HG_Parser::T_END_AUTOESCAPE;
 262 }
 263
 264
 265 "endblock" token_end {
 266     $this->token = HG_Parser::T_END_BLOCK;
 267 }
 268
 269 "ifchanged" token_end {
 270     $this->token = HG_Parser::T_IFCHANGED;
 271 }
 272
 273 "ifequal" token_end {
 274     $this->token = HG_Parser::T_IFEQUAL;
 275 }
 276
 277 "endifequal" token_end {
 278     $this->token = HG_Parser::T_END_IFEQUAL;
 279 }
 280
 281 "ifnotequal" token_end {
 282     $this->token = HG_Parser::T_IFNOTEQUAL;
 283 }
 284
 285 "endifnotequal" token_end {
 286     $this->token = HG_Parser::T_END_IFNOTEQUAL;
 287 }
 288
 289
 290 "else" token_end {
 291     $this->token = HG_Parser::T_ELSE;
 292 }
 293
 294 "endifchanged" token_end {
 295     $this->token = HG_Parser::T_ENDIFCHANGED;
 296 }
 297
 298
 299 "in" token_end {
 300     $this->token = HG_Parser::T_IN;
 301 }
 302
 303 "endfor" token_end {
 304     $this->token = HG_Parser::T_CLOSEFOR;
 305 }
 306
 307 "with" token_end {
 308     $this->token = HG_Parser::T_WITH;
 309 }
 310
 311 "endwith" token_end {
 312     $this->token = HG_Parser::T_ENDWITH;
 313 }
 314
 315 "as" {
 316     $this->token = HG_Parser::T_AS;
 317 }
 318
 319 "on" {
 320     $this->token = HG_Parser::T_ON;
 321 }
 322
 323 "off" {
 324     $this->token = HG_Parser::T_OFF;
 325 }
 326
 327 "by" {
 328     $this->token = HG_Parser::T_BY;
 329 }
 330
 331 "if" token_end {
 332     $this->token = HG_Parser::T_IF;
 333 }
 334
 335 "else" token_end {
 336     $this->token = HG_Parser::T_ELSE;
 337 }
 338
 339 "endif" token_end {
 340     $this->token = HG_Parser::T_ENDIF;
 341 }
 342
 343 "_("  {
 344     $this->token = HG_Parser::T_INTL;
 345 }
 346
 347
 348 "(" {
 349     $this->token = HG_Parser::T_LPARENT;
 350 }
 351
 352 ")" {
 353     $this->token = HG_Parser::T_RPARENT;
 354 }
 355
 356 "%" {
 357     $this->token = HG_Parser::T_MOD;
 358 }
 359
 360 "," {
 361     $this->token = HG_Parser::T_COMMA;
 362 }
 363
 364 "+" {
 365     $this->token = HG_Parser::T_PLUS;
 366 }
 367 "-" {
 368     $this->token = HG_Parser::T_MINUS;
 369 }
 370 "*" {
 371     $this->token = HG_Parser::T_TIMES;
 372 }
 373
 374 "/" {
 375     $this->token = HG_Parser::T_DIV;
 376 }
 377
 378 "'" {
 379     $this->token = HG_Parser::T_STRING_SINGLE_INIT;
 380     $this->yypushstate(self::IN_STRING_SINGLE);
 381 }
 382
 383 "\"" {
 384     $this->token = HG_Parser::T_STRING_DOUBLE_INIT;
 385     $this->yypushstate(self::IN_STRING_DOUBLE);
 386 }
 387
 388 custom_tag_end {
 389     $this->token = HG_Parser::T_CUSTOM_END;
 390 }
 391
 392 "extends" token_end {
 393     $this->token = HG_Parser::T_EXTENDS;
 394 }
 395
 396 "include" token_end {
 397     $this->token = HG_Parser::T_INCLUDE;
 398 }
 399
 400 numerals {
 401     $this->token = HG_Parser::T_NUMERIC;
 402 }
 403
 404 numerals "."  numerals {
 405     $this->token = HG_Parser::T_NUMERIC;
 406 }
 407
 408 alpha  {
 409     $this->is_custom_tag();
 410 }
 411
 412 whitespace {
 413     return FALSE;
 414 }
 415
 416 */
 417 /*!lex2php
 418 %statename IN_PRINT
 419 "}}" {
 420     $this->token = HG_Parser::T_PRINT_CLOSE;
 421     $this->yypopstate();
 422 }
 423
 424 "|" {
 425     $this->token = HG_Parser::T_PIPE;
 426 }
 427
 428 ":" {
 429     $this->token = HG_Parser::T_COLON;
 430 }
 431
 432 "->" {
 433     $this->token = HG_Parser::T_OBJ;
 434 }
 435
 436
 437 "." {
 438     $this->token = HG_Parser::T_DOT;
 439 }
 440
 441 "[" {
 442     $this->token = HG_Parser::T_BRACKETS_OPEN;
 443 }
 444
 445 "]" {
 446     $this->token = HG_Parser::T_BRACKETS_CLOSE;
 447 }
 448
 449 numerals {
 450     $this->token = HG_Parser::T_NUMERIC;
 451 }
 452
 453 numerals "."  numerals {
 454     $this->token = HG_Parser::T_NUMERIC;
 455 }
 456
 457 "'" {
 458     $this->token = HG_Parser::T_STRING_SINGLE_INIT;
 459     $this->yypushstate(self::IN_STRING_SINGLE);
 460 }
 461
 462 "\"" {
 463     $this->token = HG_Parser::T_STRING_DOUBLE_INIT;
 464     $this->yypushstate(self::IN_STRING_DOUBLE);
 465 }
 466
 467 alpha {
 468     $this->token = HG_Parser::T_ALPHA;
 469 }
 470
 471 whitespace {
 472     return FALSE;
 473 }
 474 */
 475
 476 /*!lex2php
 477 %statename IN_STRING_DOUBLE
 478
 479 "\\" "\""  {
 480     $this->token = HG_Parser::T_STRING_CONTENT;
 481     $this->value = "\"";
 482     $this->N    += 1;
 483 }
 484
 485 "\'"  {
 486     $this->token = HG_Parser::T_STRING_CONTENT;
 487     $this->value = "'";
 488     $this->N    += 1;
 489 }
 490
 491
 492 "\"" {
 493     $this->token = HG_Parser::T_STRING_DOUBLE_END;
 494     $this->yypopstate();
 495 }
 496
 497 double_string {
 498     $this->token = HG_Parser::T_STRING_CONTENT;
 499 }
 500
 501 */
 502
 503 /*!lex2php
 504 %statename IN_STRING_SINGLE
 505 "\'"  {
 506     $this->token = HG_Parser::T_STRING_CONTENT;
 507     $this->value = "'";
 508     $this->N    += 1;
 509 }
 510
 511 "\\" "\""  {
 512     $this->token = HG_Parser::T_STRING_CONTENT;
 513     $this->value = "\"";
 514     $this->N    += 1;
 515 }
 516
 517
 518 "'" {
 519     $this->token = HG_Parser::T_STRING_SINGLE_END;
 520     $this->yypopstate();
 521 }
 522
 523 single_string {
 524     $this->token = HG_Parser::T_STRING_CONTENT;
 525 }
 526
 527 */
 528
 529 /*!lex2php
 530 %statename IN_COMMENT
 531 comment {
 532     $this->token = HG_Parser::T_COMMENT;
 533     $this->yypopstate();
 534 }
 535 */
 536 }