lib/tcpdf/tcpdf_parser.php

   1 <?php
   2 //============================================================+
   3 // File name   : tcpdf_parser.php
   4 // Version     : 1.0.001
   5 // Begin       : 2011-05-23
   6 // Last Update : 2012-05-03
   7 // Author      : Nicola Asuni - Tecnick.com LTD - Manor Coach House, Church Hill, Aldershot, Hants, GU12 4RQ, UK - www.tecnick.com - info@tecnick.com
   8 // License     : http://www.tecnick.com/pagefiles/tcpdf/LICENSE.TXT GNU-LGPLv3
   9 // -------------------------------------------------------------------
  10 // Copyright (C) 2011-2012  Nicola Asuni - Tecnick.com LTD
  11 //
  12 // This file is part of TCPDF software library.
  13 //
  14 // TCPDF is free software: you can redistribute it and/or modify it
  15 // under the terms of the GNU Lesser General Public License as
  16 // published by the Free Software Foundation, either version 3 of the
  17 // License, or (at your option) any later version.
  18 //
  19 // TCPDF is distributed in the hope that it will be useful, but
  20 // WITHOUT ANY WARRANTY; without even the implied warranty of
  21 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  22 // See the GNU Lesser General Public License for more details.
  23 //
  24 // You should have received a copy of the License
  25 // along with TCPDF. If not, see
  26 // <http://www.tecnick.com/pagefiles/tcpdf/LICENSE.TXT>.
  27 //
  28 // See LICENSE.TXT file for more information.
  29 // -------------------------------------------------------------------
  30 //
  31 // Description : This is a PHP class for parsing PDF documents.
  32 //
  33 //============================================================+
  34
  35 /**
  36  * @file
  37  * This is a PHP class for parsing PDF documents.<br>
  38  * @package com.tecnick.tcpdf
  39  * @author Nicola Asuni
  40  * @version 1.0.001
  41  */
  42
  43 // include class for decoding filters
  44 require_once(dirname(__FILE__).'/tcpdf_filters.php');
  45
  46 /**
  47  * @class TCPDF_PARSER
  48  * This is a PHP class for parsing PDF documents.<br>
  49  * @package com.tecnick.tcpdf
  50  * @brief This is a PHP class for parsing PDF documents..
  51  * @version 1.0.001
  52  * @author Nicola Asuni - info@tecnick.com
  53  */
  54 class TCPDF_PARSER {
  55
  56         /**
  57          * Raw content of the PDF document.
  58          * @private
  59          */
  60         private $pdfdata = '';
  61
  62         /**
  63          * XREF data.
  64          * @protected
  65          */
  66         protected $xref = array();
  67
  68         /**
  69          * Array of PDF objects.
  70          * @protected
  71          */
  72         protected $objects = array();
  73
  74         /**
  75          * Class object for decoding filters.
  76          * @private
  77          */
  78         private $FilterDecoders;
  79
  80 // -----------------------------------------------------------------------------
  81
  82         /**
  83          * Parse a PDF document an return an array of objects.
  84          * @param $data (string) PDF data to parse.
  85          * @public
  86          * @since 1.0.000 (2011-05-24)
  87          */
  88         public function __construct($data) {
  89                 if (empty($data)) {
  90                         $this->Error('Empty PDF data.');
  91                 }
  92                 $this->pdfdata = $data;
  93                 // get length
  94                 $pdflen = strlen($this->pdfdata);
  95                 // initialize class for decoding filters
  96                 $this->FilterDecoders = new TCPDF_FILTERS();
  97                 // get xref and trailer data
  98                 $this->xref = $this->getXrefData();
  99                 // parse all document objects
 100                 $this->objects = array();
 101                 foreach ($this->xref['xref'] as $obj => $offset) {
 102                         if (!isset($this->objects[$obj])) {
 103                                 $this->objects[$obj] = $this->getIndirectObject($obj, $offset, true);
 104                         }
 105                 }
 106                 // release some memory
 107                 unset($this->pdfdata);
 108                 $this->pdfdata = '';
 109         }
 110
 111         /**
 112          * Return an array of parsed PDF document objects.
 113          * @return (array) Array of parsed PDF document objects.
 114          * @public
 115          * @since 1.0.000 (2011-06-26)
 116          */
 117         public function getParsedData() {
 118                 return array($this->xref, $this->objects);
 119         }
 120
 121         /**
 122          * Get xref (cross-reference table) and trailer data from PDF document data.
 123          * @param $offset (int) xref offset (if know).
 124          * @param $xref (array) previous xref array (if any).
 125          * @return Array containing xref and trailer data.
 126          * @protected
 127          * @since 1.0.000 (2011-05-24)
 128          */
 129         protected function getXrefData($offset=0, $xref=array()) {
 130                 if ($offset == 0) {
 131                         // find last startxref
 132                         if (preg_match_all('/[\r\n]startxref[\s]*[\r\n]+([0-9]+)[\s]*[\r\n]+%%EOF/i', $this->pdfdata, $matches, PREG_SET_ORDER, $offset) == 0) {
 133                                 $this->Error('Unable to find startxref');
 134                         }
 135                         $matches = array_pop($matches);
 136                         $startxref = $matches[1];
 137                 } else {
 138                         // get the first xref at the specified offset
 139                         if (preg_match('/[\r\n]startxref[\s]*[\r\n]+([0-9]+)[\s]*[\r\n]+%%EOF/i', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $offset) == 0) {
 140                                 $this->Error('Unable to find startxref');
 141                         }
 142                         $startxref = $matches[1][0];
 143                 }
 144                 // check xref position
 145                 if (strpos($this->pdfdata, 'xref', $startxref) != $startxref) {
 146                         $this->Error('Unable to find xref');
 147                 }
 148                 // extract xref data (object indexes and offsets)
 149                 $xoffset = $startxref + 5;
 150                 // initialize object number
 151                 $obj_num = 0;
 152                 $offset = $xoffset;
 153                 while (preg_match('/^([0-9]+)[\s]([0-9]+)[\s]?([nf]?)/im', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $offset) > 0) {
 154                         $offset = (strlen($matches[0][0]) + $matches[0][1]);
 155                         if ($matches[3][0] == 'n') {
 156                                 // create unique object index: [object number]_[generation number]
 157                                 $index = $obj_num.'_'.intval($matches[2][0]);
 158                                 // check if object already exist
 159                                 if (!isset($xref['xref'][$index])) {
 160                                         // store object offset position
 161                                         $xref['xref'][$index] = intval($matches[1][0]);
 162                                 }
 163                                 ++$obj_num;
 164                                 $offset += 2;
 165                         } elseif ($matches[3][0] == 'f') {
 166                                 ++$obj_num;
 167                                 $offset += 2;
 168                         } else {
 169                                 // object number (index)
 170                                 $obj_num = intval($matches[1][0]);
 171                         }
 172                 }
 173                 // get trailer data
 174                 if (preg_match('/trailer[\s]*<<(.*)>>[\s]*[\r\n]+startxref[\s]*[\r\n]+/isU', $this->pdfdata, $matches, PREG_OFFSET_CAPTURE, $xoffset) > 0) {
 175                         $trailer_data = $matches[1][0];
 176                         if (!isset($xref['trailer'])) {
 177                                 // get only the last updated version
 178                                 $xref['trailer'] = array();
 179                                 // parse trailer_data
 180                                 if (preg_match('/Size[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
 181                                         $xref['trailer']['size'] = intval($matches[1]);
 182                                 }
 183                                 if (preg_match('/Root[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
 184                                         $xref['trailer']['root'] = intval($matches[1]).'_'.intval($matches[2]);
 185                                 }
 186                                 if (preg_match('/Encrypt[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
 187                                         $xref['trailer']['encrypt'] = intval($matches[1]).'_'.intval($matches[2]);
 188                                 }
 189                                 if (preg_match('/Info[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
 190                                         $xref['trailer']['info'] = intval($matches[1]).'_'.intval($matches[2]);
 191                                 }
 192                                 if (preg_match('/ID[\s]*[\[][\s]*[<]([^>]*)[>][\s]*[<]([^>]*)[>]/i', $trailer_data, $matches) > 0) {
 193                                         $xref['trailer']['id'] = array();
 194                                         $xref['trailer']['id'][0] = $matches[1];
 195                                         $xref['trailer']['id'][1] = $matches[2];
 196                                 }
 197                         }
 198                         if (preg_match('/Prev[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
 199                                 // get previous xref
 200                                 $xref = $this->getXrefData(intval($matches[1]), $xref);
 201                         }
 202                 } else {
 203                         $this->Error('Unable to find trailer');
 204                 }
 205                 return $xref;
 206         }
 207
 208         /**
 209          * Get object type, raw value and offset to next object
 210          * @param $offset (int) Object offset.
 211          * @return array containing object type, raw value and offset to next object
 212          * @protected
 213          * @since 1.0.000 (2011-06-20)
 214          */
 215         protected function getRawObject($offset=0) {
 216                 $objtype = ''; // object type to be returned
 217                 $objval = ''; // object value to be returned
 218                 // skip initial white space chars: \x00 null (NUL), \x09 horizontal tab (HT), \x0A line feed (LF), \x0C form feed (FF), \x0D carriage return (CR), \x20 space (SP)
 219                 $offset += strspn($this->pdfdata, "\x00\x09\x0a\x0c\x0d\x20", $offset);
 220                 // get first char
 221                 $char = $this->pdfdata{$offset};
 222                 // get object type
 223                 switch ($char) {
 224                         case '%': { // \x25 PERCENT SIGN
 225                                 // skip comment and search for next token
 226                                 $next = strcspn($this->pdfdata, "\r\n", $offset);
 227                                 if ($next > 0) {
 228                                         $offset += $next;
 229                                         return $this->getRawObject($this->pdfdata, $offset);
 230                                 }
 231                                 break;
 232                         }
 233                         case '/': { // \x2F SOLIDUS
 234                                 // name object
 235                                 $objtype = $char;
 236                                 ++$offset;
 237                                 if (preg_match('/^([^\x00\x09\x0a\x0c\x0d\x20\s\x28\x29\x3c\x3e\x5b\x5d\x7b\x7d\x2f\x25]+)/', substr($this->pdfdata, $offset, 256), $matches) == 1) {
 238                                         $objval = $matches[1]; // unescaped value
 239                                         $offset += strlen($objval);
 240                                 }
 241                                 break;
 242                         }
 243                         case '(':   // \x28 LEFT PARENTHESIS
 244                         case ')': { // \x29 RIGHT PARENTHESIS
 245                                 // literal string object
 246                                 $objtype = $char;
 247                                 ++$offset;
 248                                 $strpos = $offset;
 249                                 if ($char == '(') {
 250                                         $open_bracket = 1;
 251                                         while ($open_bracket > 0) {
 252                                                 if (!isset($this->pdfdata{$strpos})) {
 253                                                         break;
 254                                                 }
 255                                                 $ch = $this->pdfdata{$strpos};
 256                                                 switch ($ch) {
 257                                                         case '\\': { // REVERSE SOLIDUS (5Ch) (Backslash)
 258                                                                 // skip next character
 259                                                                 ++$strpos;
 260                                                                 break;
 261                                                         }
 262                                                         case '(': { // LEFT PARENHESIS (28h)
 263                                                                 ++$open_bracket;
 264                                                                 break;
 265                                                         }
 266                                                         case ')': { // RIGHT PARENTHESIS (29h)
 267                                                                 --$open_bracket;
 268                                                                 break;
 269                                                         }
 270                                                 }
 271                                                 ++$strpos;
 272                                         }
 273                                         $objval = substr($this->pdfdata, $offset, ($strpos - $offset - 1));
 274                                         $offset = $strpos;
 275                                 }
 276                                 break;
 277                         }
 278                         case '[':   // \x5B LEFT SQUARE BRACKET
 279                         case ']': { // \x5D RIGHT SQUARE BRACKET
 280                                 // array object
 281                                 $objtype = $char;
 282                                 ++$offset;
 283                                 if ($char == '[') {
 284                                         // get array content
 285                                         $objval = array();
 286                                         do {
 287                                                 // get element
 288                                                 $element = $this->getRawObject($offset);
 289                                                 $offset = $element[2];
 290                                                 $objval[] = $element;
 291                                         } while ($element[0] != ']');
 292                                         // remove closing delimiter
 293                                         array_pop($objval);
 294                                 }
 295                                 break;
 296                         }
 297                         case '<':   // \x3C LESS-THAN SIGN
 298                         case '>': { // \x3E GREATER-THAN SIGN
 299                                 if (isset($this->pdfdata{($offset + 1)}) AND ($this->pdfdata{($offset + 1)} == $char)) {
 300                                         // dictionary object
 301                                         $objtype = $char.$char;
 302                                         $offset += 2;
 303                                         if ($char == '<') {
 304                                                 // get array content
 305                                                 $objval = array();
 306                                                 do {
 307                                                         // get element
 308                                                         $element = $this->getRawObject($offset);
 309                                                         $offset = $element[2];
 310                                                         $objval[] = $element;
 311                                                 } while ($element[0] != '>>');
 312                                                 // remove closing delimiter
 313                                                 array_pop($objval);
 314                                         }
 315                                 } else {
 316                                         // hexadecimal string object
 317                                         $objtype = $char;
 318                                         ++$offset;
 319                                         if (($char == '<') AND (preg_match('/^([0-9A-Fa-f]+)[>]/iU', substr($this->pdfdata, $offset), $matches) == 1)) {
 320                                                 $objval = $matches[1];
 321                                                 $offset += strlen($matches[0]);
 322                                         }
 323                                 }
 324                                 break;
 325                         }
 326                         default: {
 327                                 if (substr($this->pdfdata, $offset, 6) == 'endobj') {
 328                                         // indirect object
 329                                         $objtype = 'endobj';
 330                                         $offset += 6;
 331                                 } elseif (substr($this->pdfdata, $offset, 4) == 'null') {
 332                                         // null object
 333                                         $objtype = 'null';
 334                                         $offset += 4;
 335                                         $objval = 'null';
 336                                 } elseif (substr($this->pdfdata, $offset, 4) == 'true') {
 337                                         // boolean true object
 338                                         $objtype = 'boolean';
 339                                         $offset += 4;
 340                                         $objval = 'true';
 341                                 } elseif (substr($this->pdfdata, $offset, 5) == 'false') {
 342                                         // boolean false object
 343                                         $objtype = 'boolean';
 344                                         $offset += 5;
 345                                         $objval = 'false';
 346                                 } elseif (substr($this->pdfdata, $offset, 6) == 'stream') {
 347                                         // start stream object
 348                                         $objtype = 'stream';
 349                                         $offset += 6;
 350                                         if (preg_match('/^[\r\n]+(.*)[\r\n]*endstream/isU', substr($this->pdfdata, $offset), $matches) == 1) {
 351                                                 $objval = $matches[1];
 352                                                 $offset += strlen($matches[0]);
 353                                         }
 354                                 } elseif (substr($this->pdfdata, $offset, 9) == 'endstream') {
 355                                         // end stream object
 356                                         $objtype = 'endstream';
 357                                         $offset += 9;
 358                                 } elseif (preg_match('/^([0-9]+)[\s]+([0-9]+)[\s]+R/iU', substr($this->pdfdata, $offset, 33), $matches) == 1) {
 359                                         // indirect object reference
 360                                         $objtype = 'ojbref';
 361                                         $offset += strlen($matches[0]);
 362                                         $objval = intval($matches[1]).'_'.intval($matches[2]);
 363                                 } elseif (preg_match('/^([0-9]+)[\s]+([0-9]+)[\s]+obj/iU', substr($this->pdfdata, $offset, 33), $matches) == 1) {
 364                                         // object start
 365                                         $objtype = 'ojb';
 366                                         $objval = intval($matches[1]).'_'.intval($matches[2]);
 367                                         $offset += strlen ($matches[0]);
 368                                 } elseif (($numlen = strspn($this->pdfdata, '+-.0123456789', $offset)) > 0) {
 369                                         // numeric object
 370                                         $objtype = 'numeric';
 371                                         $objval = substr($this->pdfdata, $offset, $numlen);
 372                                         $offset += $numlen;
 373                                 }
 374                                 break;
 375                         }
 376                 }
 377                 return array($objtype, $objval, $offset);
 378         }
 379
 380         /**
 381          * Get content of indirect object.
 382          * @param $obj_ref (string) Object number and generation number separated by underscore character.
 383          * @param $offset (int) Object offset.
 384          * @param $decoding (boolean) If true decode streams.
 385          * @return array containing object data.
 386          * @protected
 387          * @since 1.0.000 (2011-05-24)
 388          */
 389         protected function getIndirectObject($obj_ref, $offset=0, $decoding=true) {
 390                 $obj = explode('_', $obj_ref);
 391                 if (($obj === false) OR (count($obj) != 2)) {
 392                         $this->Error('Invalid object reference: '.$obj);
 393                         return;
 394                 }
 395                 $objref = $obj[0].' '.$obj[1].' obj';
 396                 if (strpos($this->pdfdata, $objref, $offset) != $offset) {
 397                         // an indirect reference to an undefined object shall be considered a reference to the null object
 398                         return array('null', 'null', $offset);
 399                 }
 400                 // starting position of object content
 401                 $offset += strlen($objref);
 402                 // get array of object content
 403                 $objdata = array();
 404                 $i = 0; // object main index
 405                 do {
 406                         // get element
 407                         $element = $this->getRawObject($offset);
 408                         $offset = $element[2];
 409                         // decode stream using stream's dictionary information
 410                         if ($decoding AND ($element[0] == 'stream') AND (isset($objdata[($i - 1)][0])) AND ($objdata[($i - 1)][0] == '<<')) {
 411                                 $element[3] = $this->decodeStream($objdata[($i - 1)][1], substr($element[1], 1));
 412                         }
 413                         $objdata[$i] = $element;
 414                         ++$i;
 415                 } while ($element[0] != 'endobj');
 416                 // remove closing delimiter
 417                 array_pop($objdata);
 418                 // return raw object content
 419                 return $objdata;
 420         }
 421
 422         /**
 423          * Get the content of object, resolving indect object reference if necessary.
 424          * @param $obj (string) Object value.
 425          * @return array containing object data.
 426          * @protected
 427          * @since 1.0.000 (2011-06-26)
 428          */
 429         protected function getObjectVal($obj) {
 430                 if ($obj[0] == 'objref') {
 431                         // reference to indirect object
 432                         if (isset($this->objects[$obj[1]])) {
 433                                 // this object has been already parsed
 434                                 return $this->objects[$obj[1]];
 435                         } elseif (isset($this->xref[$obj[1]])) {
 436                                 // parse new object
 437                                 $this->objects[$obj[1]] = $this->getIndirectObject($obj[1], $this->xref[$obj[1]], false);
 438                                 return $this->objects[$obj[1]];
 439                         }
 440                 }
 441                 return $obj;
 442         }
 443
 444         /**
 445          * Decode the specified stream.
 446          * @param $sdic (array) Stream's dictionary array.
 447          * @param $stream (string) Stream to decode.
 448          * @return array containing decoded stream data and remaining filters.
 449          * @protected
 450          * @since 1.0.000 (2011-06-22)
 451          */
 452         protected function decodeStream($sdic, $stream) {
 453                 // get stream lenght and filters
 454                 $slength = strlen($stream);
 455                 $filters = array();
 456                 foreach ($sdic as $k => $v) {
 457                         if ($v[0] == '/') {
 458                                 if (($v[1] == 'Length') AND (isset($sdic[($k + 1)])) AND ($sdic[($k + 1)][0] == 'numeric')) {
 459                                         // get declared stream lenght
 460                                         $declength = intval($sdic[($k + 1)][1]);
 461                                         if ($declength < $slength) {
 462                                                 $stream = substr($stream, 0, $declength);
 463                                                 $slength = $declength;
 464                                         }
 465                                 } elseif (($v[1] == 'Filter') AND (isset($sdic[($k + 1)]))) {
 466                                         // resolve indirect object
 467                                         $objval = $this->getObjectVal($sdic[($k + 1)]);
 468                                         if ($objval[0] == '/') {
 469                                                 // single filter
 470                                                 $filters[] = $objval[1];
 471                                         } elseif ($objval[0] == '[') {
 472                                                 // array of filters
 473                                                 foreach ($objval[1] as $flt) {
 474                                                         if ($flt[0] == '/') {
 475                                                                 $filters[] = $flt[1];
 476                                                         }
 477                                                 }
 478                                         }
 479                                 }
 480                         }
 481                 }
 482                 // decode the stream
 483                 $remaining_filters = array();
 484                 foreach ($filters as $filter) {
 485                         if (in_array($filter, $this->FilterDecoders->getAvailableFilters())) {
 486                                 $stream = $this->FilterDecoders->decodeFilter($filter, $stream);
 487                         } else {
 488                                 // add missing filter to array
 489                                 $remaining_filters[] = $filter;
 490                         }
 491                 }
 492                 return array($stream, $remaining_filters);
 493         }
 494
 495         /**
 496          * This method is automatically called in case of fatal error; it simply outputs the message and halts the execution.
 497          * @param $msg (string) The error message
 498          * @public
 499          * @since 1.0.000 (2011-05-23)
 500          */
 501         public function Error($msg) {
 502                 // exit program and print error
 503                 die('<strong>TCPDF_PARSER ERROR: </strong>'.$msg);
 504         }
 505
 506 } // END OF TCPDF_PARSER CLASS
 507
 508 //============================================================+
 509 // END OF FILE
 510 //============================================================+