lib/pear/OLE.php

   1 <?php
   2 /* vim: set expandtab tabstop=4 shiftwidth=4: */
   3 // +----------------------------------------------------------------------+
   4 // | PHP Version 4                                                        |
   5 // +----------------------------------------------------------------------+
   6 // | Copyright (c) 1997-2002 The PHP Group                                |
   7 // +----------------------------------------------------------------------+
   8 // | This source file is subject to version 2.02 of the PHP license,      |
   9 // | that is bundled with this package in the file LICENSE, and is        |
  10 // | available at through the world-wide-web at                           |
  11 // | http://www.php.net/license/2_02.txt.                                 |
  12 // | If you did not receive a copy of the PHP license and are unable to   |
  13 // | obtain it through the world-wide-web, please send a note to          |
  14 // | license@php.net so we can mail you a copy immediately.               |
  15 // +----------------------------------------------------------------------+
  16 // | Author: Xavier Noguer <xnoguer@php.net>                              |
  17 // | Based on OLE::Storage_Lite by Kawai, Takanori                        |
  18 // +----------------------------------------------------------------------+
  19 //
  20 // $Id$
  21
  22
  23 /**
  24 * Constants for OLE package
  25 */
  26 define('OLE_PPS_TYPE_ROOT',        5);
  27 define('OLE_PPS_TYPE_DIR',         1);
  28 define('OLE_PPS_TYPE_FILE',        2);
  29 define('OLE_DATA_SIZE_SMALL', 0x1000);
  30 define('OLE_LONG_INT_SIZE',        4);
  31 define('OLE_PPS_SIZE',          0x80);
  32
  33 require_once('PEAR.php');
  34 require_once 'OLE/PPS.php';
  35
  36 /**
  37 * OLE package base class.
  38 *
  39 * @author   Xavier Noguer <xnoguer@php.net>
  40 * @category Structures
  41 * @package  OLE
  42 */
  43 class OLE extends PEAR
  44 {
  45     /**
  46     * The file handle for reading an OLE container
  47     * @var resource
  48     */
  49     var $_file_handle;
  50
  51     /**
  52     * Array of PPS's found on the OLE container
  53     * @var array
  54     */
  55     var $_list;
  56
  57     /**
  58     * Creates a new OLE object
  59     * Remember to use ampersand when creating an OLE object ($my_ole =& new OLE();)
  60     * @access public
  61     */
  62     function OLE()
  63     {
  64         $this->_list = array();
  65     }
  66
  67     /**
  68     * Reads an OLE container from the contents of the file given.
  69     *
  70     * @acces public
  71     * @param string $file
  72     * @return mixed true on success, PEAR_Error on failure
  73     */
  74     function read($file)
  75     {
  76         /* consider storing offsets as constants */
  77         $big_block_size_offset = 30;
  78         $iBdbCnt_offset        = 44;
  79         $bd_start_offset       = 68;
  80
  81         $fh = @fopen($file, "r");
  82         if ($fh == false) {
  83             return $this->raiseError("Can't open file $file");
  84         }
  85         $this->_file_handle = $fh;
  86
  87         /* begin reading OLE attributes */
  88         fseek($fh, 0);
  89         $signature = fread($fh, 8);
  90         if ("\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1" != $signature) {
  91             return $this->raiseError("File doesn't seem to be an OLE container.");
  92         }
  93         fseek($fh, $big_block_size_offset);
  94         $packed_array = unpack("v", fread($fh, 2));
  95         $big_block_size   = pow(2, $packed_array['']);
  96
  97         $packed_array = unpack("v", fread($fh, 2));
  98         $small_block_size = pow(2, $packed_array['']);
  99         $i1stBdL = ($big_block_size - 0x4C) / OLE_LONG_INT_SIZE;
 100
 101         fseek($fh, $iBdbCnt_offset);
 102         $packed_array = unpack("V", fread($fh, 4));
 103         $iBdbCnt = $packed_array[''];
 104
 105         $packed_array = unpack("V", fread($fh, 4));
 106         $pps_wk_start = $packed_array[''];
 107
 108         fseek($fh, $bd_start_offset);
 109         $packed_array = unpack("V", fread($fh, 4));
 110         $bd_start = $packed_array[''];
 111         $packed_array = unpack("V", fread($fh, 4));
 112         $bd_count = $packed_array[''];
 113         $packed_array = unpack("V", fread($fh, 4));
 114         $iAll = $packed_array[''];  // this may be wrong
 115         /* create OLE_PPS objects from */
 116         $ret = $this->_readPpsWks($pps_wk_start, $big_block_size);
 117         if (PEAR::isError($ret)) {
 118             return $ret;
 119         }
 120         return true;
 121     }
 122
 123     /**
 124     * Destructor (using PEAR)
 125     * Just closes the file handle on the OLE file.
 126     *
 127     * @access private
 128     */
 129     function _OLE()
 130     {
 131         fclose($this->_file_handle);
 132     }
 133
 134     /**
 135     * Gets information about all PPS's on the OLE container from the PPS WK's
 136     * creates an OLE_PPS object for each one.
 137     *
 138     * @access private
 139     * @param integer $pps_wk_start   Position inside the OLE file where PPS WK's start
 140     * @param integer $big_block_size Size of big blobks in the OLE file
 141     * @return mixed true on success, PEAR_Error on failure
 142     */
 143     function _readPpsWks($pps_wk_start, $big_block_size)
 144     {
 145         $pointer = ($pps_wk_start + 1) * $big_block_size;
 146         while (1)
 147         {
 148             fseek($this->_file_handle, $pointer);
 149             $pps_wk = fread($this->_file_handle, OLE_PPS_SIZE);
 150             if (strlen($pps_wk) != OLE_PPS_SIZE) {
 151                 break; // Excel likes to add a trailing byte sometimes
 152                 //return $this->raiseError("PPS at $pointer seems too short: ".strlen($pps_wk));
 153             }
 154             $name_length = unpack("c", substr($pps_wk, 64, 2)); // FIXME (2 bytes??)
 155             $name_length = $name_length[''] - 2;
 156             $name = substr($pps_wk, 0, $name_length);
 157             $type = unpack("c", substr($pps_wk, 66, 1));
 158             if (($type[''] != OLE_PPS_TYPE_ROOT) and
 159                 ($type[''] != OLE_PPS_TYPE_DIR) and
 160                 ($type[''] != OLE_PPS_TYPE_FILE))
 161             {
 162                 return $this->raiseError("PPS at $pointer has unknown type: {$type['']}");
 163             }
 164             $prev = unpack("V", substr($pps_wk, 68, 4));
 165             $next = unpack("V", substr($pps_wk, 72, 4));
 166             $dir  = unpack("V", substr($pps_wk, 76, 4));
 167             // there is no magic number, it can take different values.
 168             //$magic = unpack("V", strrev(substr($pps_wk, 92, 4)));
 169             $time_1st = substr($pps_wk, 100, 8);
 170             $time_2nd = substr($pps_wk, 108, 8);
 171             $start_block = unpack("V", substr($pps_wk, 116, 4));
 172             $size = unpack("V", substr($pps_wk, 120, 4));
 173             // _data member will point to position in file!!
 174             // OLE_PPS object is created with an empty children array!!
 175             $this->_list[] = new OLE_PPS(null, '', $type[''], $prev[''], $next[''],
 176                                          $dir[''], OLE::OLE2LocalDate($time_1st),
 177                                          OLE::OLE2LocalDate($time_2nd),
 178                                          ($start_block[''] + 1) * $big_block_size, array());
 179             // give it a size
 180             $this->_list[count($this->_list) - 1]->Size = $size[''];
 181             // check if the PPS tree (starting from root) is complete
 182             if ($this->_ppsTreeComplete(0)) {
 183                 break;
 184             }
 185             $pointer += OLE_PPS_SIZE;
 186         }
 187     }
 188
 189     /**
 190     * It checks whether the PPS tree is complete (all PPS's read)
 191     * starting with the given PPS (not necessarily root)
 192     *
 193     * @access private
 194     * @param integer $index The index of the PPS from which we are checking
 195     * @return boolean Whether the PPS tree for the given PPS is complete
 196     */
 197     function _ppsTreeComplete($index)
 198     {
 199         if ($this->_list[$index]->NextPps != -1) {
 200             if (!isset($this->_list[$this->_list[$index]->NextPps])) {
 201                 return false;
 202             }
 203             else {
 204                 return $this->_ppsTreeComplete($this->_list[$index]->NextPps);
 205             }
 206         }
 207         if ($this->_list[$index]->DirPps != -1) {
 208             if (!isset($this->_list[$this->_list[$index]->DirPps])) {
 209                 return false;
 210             }
 211             else {
 212                 return $this->_ppsTreeComplete($this->_list[$index]->DirPps);
 213             }
 214         }
 215         return true;
 216     }
 217
 218     /**
 219     * Checks whether a PPS is a File PPS or not.
 220     * If there is no PPS for the index given, it will return false.
 221     *
 222     * @access public
 223     * @param integer $index The index for the PPS
 224     * @return bool true if it's a File PPS, false otherwise
 225     */
 226     function isFile($index)
 227     {
 228         if (isset($this->_list[$index])) {
 229             return ($this->_list[$index]->Type == OLE_PPS_TYPE_FILE);
 230         }
 231         return false;
 232     }
 233
 234     /**
 235     * Checks whether a PPS is a Root PPS or not.
 236     * If there is no PPS for the index given, it will return false.
 237     *
 238     * @access public
 239     * @param integer $index The index for the PPS.
 240     * @return bool true if it's a Root PPS, false otherwise
 241     */
 242     function isRoot($index)
 243     {
 244         if (isset($this->_list[$index])) {
 245             return ($this->_list[$index]->Type == OLE_PPS_TYPE_ROOT);
 246         }
 247         return false;
 248     }
 249
 250     /**
 251     * Gives the total number of PPS's found in the OLE container.
 252     *
 253     * @access public
 254     * @return integer The total number of PPS's found in the OLE container
 255     */
 256     function ppsTotal()
 257     {
 258         return count($this->_list);
 259     }
 260
 261     /**
 262     * Gets data from a PPS
 263     * If there is no PPS for the index given, it will return an empty string.
 264     *
 265     * @access public
 266     * @param integer $index    The index for the PPS
 267     * @param integer $position The position from which to start reading
 268     *                          (relative to the PPS)
 269     * @param integer $length   The amount of bytes to read (at most)
 270     * @return string The binary string containing the data requested
 271     */
 272     function getData($index, $position, $length)
 273     {
 274         // if position is not valid return empty string
 275         if (!isset($this->_list[$index]) or ($position >= $this->_list[$index]->Size) or ($position < 0)) {
 276             return '';
 277         }
 278         // Beware!!! _data member is actually a position
 279         fseek($this->_file_handle, $this->_list[$index]->_data + $position);
 280         return fread($this->_file_handle, $length);
 281     }
 282
 283     /**
 284     * Gets the data length from a PPS
 285     * If there is no PPS for the index given, it will return 0.
 286     *
 287     * @access public
 288     * @param integer $index    The index for the PPS
 289     * @return integer The amount of bytes in data the PPS has
 290     */
 291     function getDataLength($index)
 292     {
 293         if (isset($this->_list[$index])) {
 294             return $this->_list[$index]->Size;
 295         }
 296         return 0;
 297     }
 298
 299     /**
 300     * Utility function to transform ASCII text to Unicode
 301     *
 302     * @access public
 303     * @static
 304     * @param string $ascii The ASCII string to transform
 305     * @return string The string in Unicode
 306     */
 307     function Asc2Ucs($ascii)
 308     {
 309         $rawname = '';
 310         for ($i = 0; $i < strlen($ascii); $i++) {
 311             $rawname .= $ascii{$i}."\x00";
 312         }
 313         return $rawname;
 314     }
 315
 316     /**
 317     * Utility function
 318     * Returns a string for the OLE container with the date given
 319     *
 320     * @access public
 321     * @static
 322     * @param integer $date A timestamp
 323     * @return string The string for the OLE container
 324     */
 325     function LocalDate2OLE($date = null)
 326     {
 327         if (!isset($date)) {
 328             return "\x00\x00\x00\x00\x00\x00\x00\x00";
 329         }
 330
 331         // factor used for separating numbers into 4 bytes parts
 332         $factor = pow(2,32);
 333
 334         // days from 1-1-1601 until the beggining of UNIX era
 335         $days = 134774;
 336         // calculate seconds
 337         $big_date = $days*24*3600 + gmmktime(date("H",$date),date("i",$date),date("s",$date),
 338                                              date("m",$date),date("d",$date),date("Y",$date));
 339         // multiply just to make MS happy
 340         $big_date *= 10000000;
 341
 342         $high_part = floor($big_date/$factor);
 343         // lower 4 bytes
 344         $low_part = floor((($big_date/$factor) - $high_part)*$factor);
 345
 346         // Make HEX string
 347         $res = '';
 348
 349         for ($i=0; $i<4; $i++)
 350         {
 351             $hex = $low_part % 0x100;
 352             $res .= pack('c', $hex);
 353             $low_part /= 0x100;
 354         }
 355         for ($i=0; $i<4; $i++)
 356         {
 357             $hex = $high_part % 0x100;
 358             $res .= pack('c', $hex);
 359             $high_part /= 0x100;
 360         }
 361         return $res;
 362     }
 363
 364     /**
 365     * Returns a timestamp from an OLE container's date
 366     *
 367     * @access public
 368     * @static
 369     * @param integer $string A binary string with the encoded date
 370     * @return string The timestamp corresponding to the string
 371     */
 372     function OLE2LocalDate($string)
 373     {
 374         if (strlen($string) != 8) {
 375             return new PEAR_Error("Expecting 8 byte string");
 376         }
 377
 378         // factor used for separating numbers into 4 bytes parts
 379         $factor = pow(2,32);
 380         $high_part = 0;
 381         for ($i=0; $i<4; $i++)
 382         {
 383             $al = unpack('C', $string{(7 - $i)});
 384             $high_part += $al[''];
 385             if ($i < 3) {
 386                 $high_part *= 0x100;
 387             }
 388         }
 389         $low_part = 0;
 390         for ($i=4; $i<8; $i++)
 391         {
 392             $al = unpack('C', $string{(7 - $i)});
 393             $low_part += $al[''];
 394             if ($i < 7) {
 395                 $low_part *= 0x100;
 396             }
 397         }
 398         $big_date = ($high_part*$factor) + $low_part;
 399         // translate to seconds
 400         $big_date /= 10000000;
 401
 402         // days from 1-1-1601 until the beggining of UNIX era
 403         $days = 134774;
 404
 405         // translate to seconds from beggining of UNIX era
 406         $big_date -= $days*24*3600;
 407         return floor($big_date);
 408     }
 409 }
 410 ?>