1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of PHPOffice Common |
4
|
|
|
* |
5
|
|
|
* PHPOffice Common is free software distributed under the terms of the GNU Lesser |
6
|
|
|
* General Public License version 3 as published by the Free Software Foundation. |
7
|
|
|
* |
8
|
|
|
* For the full copyright and license information, please read the LICENSE |
9
|
|
|
* file that was distributed with this source code. For the full list of |
10
|
|
|
* contributors, visit https://github.com/PHPOffice/Common/contributors. |
11
|
|
|
* |
12
|
|
|
* @link https://github.com/PHPOffice/Common |
13
|
|
|
* @copyright 2009-2016 PHPOffice Common contributors |
14
|
|
|
* @license http://www.gnu.org/licenses/lgpl.txt LGPL version 3 |
15
|
|
|
*/ |
16
|
|
|
|
17
|
|
|
namespace PhpOffice\Common\Microsoft; |
18
|
|
|
|
19
|
|
|
if (!defined('IDENTIFIER_OLE')) { |
20
|
|
|
define('IDENTIFIER_OLE', pack('CCCCCCCC', 0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1)); |
21
|
|
|
} |
22
|
|
|
|
23
|
|
|
class OLERead |
24
|
|
|
{ |
25
|
|
|
private $data = ''; |
26
|
|
|
|
27
|
|
|
// OLE identifier |
28
|
|
|
const IDENTIFIER_OLE = IDENTIFIER_OLE; |
29
|
|
|
|
30
|
|
|
// Size of a sector = 512 bytes |
31
|
|
|
const BIG_BLOCK_SIZE = 0x200; |
32
|
|
|
|
33
|
|
|
// Size of a short sector = 64 bytes |
34
|
|
|
const SMALL_BLOCK_SIZE = 0x40; |
35
|
|
|
|
36
|
|
|
// Size of a directory entry always = 128 bytes |
37
|
|
|
const PROPERTY_STORAGE_BLOCK_SIZE = 0x80; |
38
|
|
|
|
39
|
|
|
// Minimum size of a standard stream = 4096 bytes, streams smaller than this are stored as short streams |
40
|
|
|
const SMALL_BLOCK_THRESHOLD = 0x1000; |
41
|
|
|
|
42
|
|
|
// header offsets |
43
|
|
|
const NUM_BIG_BLOCK_DEPOT_BLOCKS_POS = 0x2c; |
44
|
|
|
const ROOT_START_BLOCK_POS = 0x30; |
45
|
|
|
const SMALL_BLOCK_DEPOT_BLOCK_POS = 0x3c; |
46
|
|
|
const EXTENSION_BLOCK_POS = 0x44; |
47
|
|
|
const NUM_EXTENSION_BLOCK_POS = 0x48; |
48
|
|
|
const BIG_BLOCK_DEPOT_BLOCKS_POS = 0x4c; |
49
|
|
|
|
50
|
|
|
// property storage offsets (directory offsets) |
51
|
|
|
const SIZE_OF_NAME_POS = 0x40; |
52
|
|
|
const TYPE_POS = 0x42; |
53
|
|
|
const START_BLOCK_POS = 0x74; |
54
|
|
|
const SIZE_POS = 0x78; |
55
|
|
|
|
56
|
|
|
public $summaryInformation = null; |
57
|
|
|
public $docSummaryInfos = null; |
58
|
|
|
public $powerpointDocument = null; |
59
|
|
|
public $currentUser = null; |
60
|
|
|
public $pictures = null; |
61
|
|
|
public $rootEntry = null; |
62
|
|
|
public $props = array(); |
63
|
|
|
public $smallBlockChain = null; |
64
|
|
|
public $bigBlockChain = null; |
65
|
|
|
public $entry = null; |
66
|
|
|
|
67
|
|
|
/** |
68
|
|
|
* Read the file |
69
|
|
|
* |
70
|
|
|
* @param $sFileName string Filename |
71
|
|
|
* @throws \Exception |
72
|
|
|
*/ |
73
|
|
|
public function read($sFileName) |
74
|
|
|
{ |
75
|
|
|
// Check if file exists and is readable |
76
|
|
|
if (!is_readable($sFileName)) { |
77
|
|
|
throw new \Exception("Could not open " . $sFileName . " for reading! File does not exist, or it is not readable."); |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
// Get the file identifier |
81
|
|
|
// Don't bother reading the whole file until we know it's a valid OLE file |
82
|
|
|
$this->data = file_get_contents($sFileName, false, null, 0, 8); |
83
|
|
|
|
84
|
|
|
// Check OLE identifier |
85
|
|
|
if ($this->data != self::IDENTIFIER_OLE) { |
86
|
|
|
throw new \Exception('The filename ' . $sFileName . ' is not recognised as an OLE file'); |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
// Get the file data |
90
|
|
|
$this->data = file_get_contents($sFileName); |
91
|
|
|
|
92
|
|
|
// Total number of sectors used for the SAT |
93
|
|
|
$numBigBlkDepotBlks = self::getInt4d($this->data, self::NUM_BIG_BLOCK_DEPOT_BLOCKS_POS); |
94
|
|
|
|
95
|
|
|
// SecID of the first sector of the directory stream |
96
|
|
|
$rootStartBlock = self::getInt4d($this->data, self::ROOT_START_BLOCK_POS); |
97
|
|
|
|
98
|
|
|
// SecID of the first sector of the SSAT (or -2 if not extant) |
99
|
|
|
$sbdStartBlock = self::getInt4d($this->data, self::SMALL_BLOCK_DEPOT_BLOCK_POS); |
100
|
|
|
|
101
|
|
|
// SecID of the first sector of the MSAT (or -2 if no additional sectors are used) |
102
|
|
|
$extensionBlock = self::getInt4d($this->data, self::EXTENSION_BLOCK_POS); |
103
|
|
|
|
104
|
|
|
// Total number of sectors used by MSAT |
105
|
|
|
$numExtensionBlocks = self::getInt4d($this->data, self::NUM_EXTENSION_BLOCK_POS); |
106
|
|
|
|
107
|
|
|
$bigBlockDepotBlocks = array(); |
108
|
|
|
$pos = self::BIG_BLOCK_DEPOT_BLOCKS_POS; |
109
|
|
|
|
110
|
|
|
$bbdBlocks = $numBigBlkDepotBlks; |
111
|
|
|
|
112
|
|
|
if ($numExtensionBlocks != 0) { |
113
|
|
|
$bbdBlocks = (self::BIG_BLOCK_SIZE - self::BIG_BLOCK_DEPOT_BLOCKS_POS)/4; |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
for ($i = 0; $i < $bbdBlocks; ++$i) { |
117
|
|
|
$bigBlockDepotBlocks[$i] = self::getInt4d($this->data, $pos); |
118
|
|
|
$pos += 4; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
for ($j = 0; $j < $numExtensionBlocks; ++$j) { |
122
|
|
|
$pos = ($extensionBlock + 1) * self::BIG_BLOCK_SIZE; |
123
|
|
|
$blocksToRead = min($numBigBlkDepotBlks - $bbdBlocks, self::BIG_BLOCK_SIZE / 4 - 1); |
124
|
|
|
|
125
|
|
|
for ($i = $bbdBlocks; $i < $bbdBlocks + $blocksToRead; ++$i) { |
126
|
|
|
$bigBlockDepotBlocks[$i] = self::getInt4d($this->data, $pos); |
127
|
|
|
$pos += 4; |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
$bbdBlocks += $blocksToRead; |
131
|
|
|
if ($bbdBlocks < $numBigBlkDepotBlks) { |
132
|
|
|
$extensionBlock = self::getInt4d($this->data, $pos); |
133
|
|
|
} |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
$this->bigBlockChain = ''; |
137
|
|
|
$bbs = self::BIG_BLOCK_SIZE / 4; |
138
|
|
|
for ($i = 0; $i < $numBigBlkDepotBlks; ++$i) { |
139
|
|
|
$pos = ($bigBlockDepotBlocks[$i] + 1) * self::BIG_BLOCK_SIZE; |
140
|
|
|
|
141
|
|
|
$this->bigBlockChain .= substr($this->data, $pos, 4*$bbs); |
142
|
|
|
$pos += 4*$bbs; |
143
|
|
|
} |
144
|
|
|
|
145
|
|
|
$sbdBlock = $sbdStartBlock; |
146
|
|
|
$this->smallBlockChain = ''; |
147
|
|
|
while ($sbdBlock != -2) { |
148
|
|
|
$pos = ($sbdBlock + 1) * self::BIG_BLOCK_SIZE; |
149
|
|
|
|
150
|
|
|
$this->smallBlockChain .= substr($this->data, $pos, 4*$bbs); |
151
|
|
|
$pos += 4*$bbs; |
152
|
|
|
|
153
|
|
|
$sbdBlock = self::getInt4d($this->bigBlockChain, $sbdBlock*4); |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
// read the directory stream |
157
|
|
|
$block = $rootStartBlock; |
158
|
|
|
$this->entry = $this->readData($block); |
159
|
|
|
|
160
|
|
|
$this->readPropertySets(); |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Extract binary stream data |
165
|
|
|
* |
166
|
|
|
* @return string |
167
|
|
|
*/ |
168
|
|
|
public function getStream($stream) |
169
|
|
|
{ |
170
|
|
|
if ($stream === null) { |
171
|
|
|
return null; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
$streamData = ''; |
175
|
|
|
|
176
|
|
|
if ($this->props[$stream]['size'] < self::SMALL_BLOCK_THRESHOLD) { |
177
|
|
|
$rootdata = $this->readData($this->props[$this->rootEntry]['startBlock']); |
178
|
|
|
|
179
|
|
|
$block = $this->props[$stream]['startBlock']; |
180
|
|
|
|
181
|
|
|
while ($block != -2) { |
182
|
|
|
$pos = $block * self::SMALL_BLOCK_SIZE; |
183
|
|
|
$streamData .= substr($rootdata, $pos, self::SMALL_BLOCK_SIZE); |
184
|
|
|
|
185
|
|
|
$block = self::getInt4d($this->smallBlockChain, $block*4); |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
return $streamData; |
189
|
|
|
} |
190
|
|
|
|
191
|
|
|
$numBlocks = $this->props[$stream]['size'] / self::BIG_BLOCK_SIZE; |
192
|
|
|
if ($this->props[$stream]['size'] % self::BIG_BLOCK_SIZE != 0) { |
193
|
|
|
++$numBlocks; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
if ($numBlocks == 0) { |
197
|
|
|
return ''; |
198
|
|
|
} |
199
|
|
|
|
200
|
|
|
$block = $this->props[$stream]['startBlock']; |
201
|
|
|
|
202
|
|
|
while ($block != -2) { |
203
|
|
|
$pos = ($block + 1) * self::BIG_BLOCK_SIZE; |
204
|
|
|
$streamData .= substr($this->data, $pos, self::BIG_BLOCK_SIZE); |
205
|
|
|
$block = self::getInt4d($this->bigBlockChain, $block*4); |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
return $streamData; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* Read a standard stream (by joining sectors using information from SAT) |
213
|
|
|
* |
214
|
|
|
* @param int $blID Sector ID where the stream starts |
215
|
|
|
* @return string Data for standard stream |
216
|
|
|
*/ |
217
|
|
|
private function readData($blID) |
218
|
|
|
{ |
219
|
|
|
$block = $blID; |
220
|
|
|
$data = ''; |
221
|
|
|
|
222
|
|
|
while ($block != -2) { |
223
|
|
|
$pos = ($block + 1) * self::BIG_BLOCK_SIZE; |
224
|
|
|
$data .= substr($this->data, $pos, self::BIG_BLOCK_SIZE); |
225
|
|
|
$block = self::getInt4d($this->bigBlockChain, $block*4); |
226
|
|
|
} |
227
|
|
|
return $data; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
/** |
231
|
|
|
* Read entries in the directory stream. |
232
|
|
|
*/ |
233
|
|
|
private function readPropertySets() |
234
|
|
|
{ |
235
|
|
|
$offset = 0; |
236
|
|
|
|
237
|
|
|
// loop through entires, each entry is 128 bytes |
238
|
|
|
$entryLen = strlen($this->entry); |
239
|
|
|
while ($offset < $entryLen) { |
240
|
|
|
// entry data (128 bytes) |
241
|
|
|
$data = substr($this->entry, $offset, self::PROPERTY_STORAGE_BLOCK_SIZE); |
242
|
|
|
|
243
|
|
|
// size in bytes of name |
244
|
|
|
$nameSize = ord($data[self::SIZE_OF_NAME_POS]) | (ord($data[self::SIZE_OF_NAME_POS+1]) << 8); |
245
|
|
|
|
246
|
|
|
// type of entry |
247
|
|
|
$type = ord($data[self::TYPE_POS]); |
248
|
|
|
|
249
|
|
|
// sectorID of first sector or short sector, if this entry refers to a stream (the case with workbook) |
250
|
|
|
// sectorID of first sector of the short-stream container stream, if this entry is root entry |
251
|
|
|
$startBlock = self::getInt4d($data, self::START_BLOCK_POS); |
252
|
|
|
|
253
|
|
|
$size = self::getInt4d($data, self::SIZE_POS); |
254
|
|
|
|
255
|
|
|
$name = str_replace("\x00", "", substr($data, 0, $nameSize)); |
256
|
|
|
if ($size > 0) { |
257
|
|
|
$this->props[] = array ( |
258
|
|
|
'name' => $name, |
259
|
|
|
'type' => $type, |
260
|
|
|
'startBlock' => $startBlock, |
261
|
|
|
'size' => $size); |
262
|
|
|
|
263
|
|
|
// tmp helper to simplify checks |
264
|
|
|
$upName = strtoupper($name); |
265
|
|
|
|
266
|
|
|
switch ($upName) { |
267
|
|
|
case 'ROOT ENTRY': |
268
|
|
|
case 'R': |
269
|
|
|
$this->rootEntry = count($this->props) - 1; |
270
|
|
|
break; |
271
|
|
|
case chr(1).'COMPOBJ': |
272
|
|
|
break; |
273
|
|
|
case chr(1).'OLE': |
274
|
|
|
break; |
275
|
|
|
case chr(5).'SUMMARYINFORMATION': |
276
|
|
|
$this->summaryInformation = count($this->props) - 1; |
277
|
|
|
break; |
278
|
|
|
case chr(5).'DOCUMENTSUMMARYINFORMATION': |
279
|
|
|
$this->docSummaryInfos = count($this->props) - 1; |
280
|
|
|
break; |
281
|
|
|
case 'CURRENT USER': |
282
|
|
|
$this->currentUser = count($this->props) - 1; |
283
|
|
|
break; |
284
|
|
|
case 'PICTURES': |
285
|
|
|
$this->pictures = count($this->props) - 1; |
286
|
|
|
break; |
287
|
|
|
case 'POWERPOINT DOCUMENT': |
288
|
|
|
$this->powerpointDocument = count($this->props) - 1; |
289
|
|
|
break; |
290
|
|
|
default: |
291
|
|
|
throw new \Exception('OLE Block Not defined: $upName : '.$upName. ' - $name : "'.$name.'"'); |
292
|
|
|
} |
293
|
|
|
} |
294
|
|
|
|
295
|
|
|
$offset += self::PROPERTY_STORAGE_BLOCK_SIZE; |
296
|
|
|
} |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
/** |
300
|
|
|
* Read 4 bytes of data at specified position |
301
|
|
|
* |
302
|
|
|
* @param string $data |
303
|
|
|
* @param int $pos |
304
|
|
|
* @return int |
305
|
|
|
*/ |
306
|
|
|
private static function getInt4d($data, $pos) |
307
|
|
|
{ |
308
|
|
|
// FIX: represent numbers correctly on 64-bit system |
309
|
|
|
// http://sourceforge.net/tracker/index.php?func=detail&aid=1487372&group_id=99160&atid=623334 |
310
|
|
|
// Hacked by Andreas Rehm 2006 to ensure correct result of the <<24 block on 32 and 64bit systems |
311
|
|
|
$or24 = ord($data[$pos + 3]); |
312
|
|
|
if ($or24 >= 128) { |
313
|
|
|
// negative number |
314
|
|
|
$ord24 = -abs((256 - $or24) << 24); |
315
|
|
|
} else { |
316
|
|
|
$ord24 = ($or24 & 127) << 24; |
317
|
|
|
} |
318
|
|
|
return ord($data[$pos]) | (ord($data[$pos + 1]) << 8) | (ord($data[$pos + 2]) << 16) | $ord24; |
319
|
|
|
} |
320
|
|
|
} |
321
|
|
|
|