1
|
|
|
<?php |
2
|
|
|
/* |
3
|
|
|
* SPDX-License-Identifier: AGPL-3.0-only |
4
|
|
|
* SPDX-FileCopyrightText: Copyright 2007-2016 Zarafa Deutschland GmbH |
5
|
|
|
* SPDX-FileCopyrightText: Copyright 2020-2022 grommunio GmbH |
6
|
|
|
* |
7
|
|
|
* WBXMLDecoder decodes from Wap Binary XML |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
class WBXMLDecoder extends WBXMLDefs { |
11
|
|
|
private $in; |
12
|
|
|
private $inLog; |
|
|
|
|
13
|
|
|
private $tagcp = 0; |
14
|
|
|
private $ungetbuffer; |
15
|
|
|
private $log = false; |
16
|
|
|
private $logStack = []; |
17
|
|
|
private $inputBuffer = ""; |
18
|
|
|
private $isWBXML = true; |
19
|
|
|
private static $loopCounter = []; |
20
|
|
|
public const MAXLOOP = 5000; |
21
|
|
|
public const VERSION = 0x03; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* Counts the amount of times a code part has been executed. |
25
|
|
|
* When being executed too often, the code throws a WBMXLException. |
26
|
|
|
* |
27
|
|
|
* @param string $name |
28
|
|
|
* |
29
|
|
|
* @return bool |
30
|
|
|
* |
31
|
|
|
* @throws WBXMLException |
32
|
|
|
*/ |
33
|
|
|
public static function InWhile($name) { |
34
|
|
|
if (!isset(self::$loopCounter[$name])) { |
35
|
|
|
self::$loopCounter[$name] = 0; |
36
|
|
|
} |
37
|
|
|
else { |
38
|
|
|
++self::$loopCounter[$name]; |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
if (self::$loopCounter[$name] > self::MAXLOOP) { |
42
|
|
|
throw new WBXMLException(sprintf("Loop count in while too high, code '%s' exceeded max. amount of permitted loops", $name)); |
43
|
|
|
} |
44
|
|
|
|
45
|
|
|
return true; |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Resets the inWhile counter. |
50
|
|
|
* |
51
|
|
|
* @param string $name |
52
|
|
|
* |
53
|
|
|
* @return bool |
54
|
|
|
*/ |
55
|
|
|
public static function ResetInWhile($name) { |
56
|
|
|
if (isset(self::$loopCounter[$name])) { |
57
|
|
|
unset(self::$loopCounter[$name]); |
58
|
|
|
} |
59
|
|
|
|
60
|
|
|
return true; |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* WBXML Decode Constructor |
65
|
|
|
* We only handle ActiveSync WBXML, which is a subset of WBXML. |
66
|
|
|
* |
67
|
|
|
* @param stream $input the incoming data stream |
68
|
|
|
*/ |
69
|
|
|
public function __construct($input) { |
70
|
|
|
$this->log = SLog::IsWbxmlDebugEnabled(); |
71
|
|
|
|
72
|
|
|
$this->in = $input; |
73
|
|
|
|
74
|
|
|
$version = $this->getByte(); |
75
|
|
|
if ($version != self::VERSION) { |
76
|
|
|
$this->inputBuffer .= chr($version); |
77
|
|
|
$this->isWBXML = false; |
78
|
|
|
|
79
|
|
|
return; |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
$publicid = $this->getMBUInt(); |
83
|
|
|
if ($publicid !== 1) { |
84
|
|
|
throw new WBXMLException("Wrong publicid : " . $publicid); |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
$charsetid = $this->getMBUInt(); |
88
|
|
|
if ($charsetid !== 106) { |
89
|
|
|
throw new WBXMLException("Wrong charset : " . $charsetid); |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
$stringtablesize = $this->getMBUInt(); |
93
|
|
|
if ($stringtablesize !== 0) { |
94
|
|
|
throw new WBXMLException("Wrong string table size : " . $stringtablesize); |
95
|
|
|
} |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
/** |
99
|
|
|
* Returns either start, content or end, and auto-concatenates successive content. |
100
|
|
|
* |
101
|
|
|
* @return element|value |
|
|
|
|
102
|
|
|
*/ |
103
|
|
|
public function getElement() { |
104
|
|
|
$element = $this->getToken(); |
105
|
|
|
if (is_null($element)) { |
106
|
|
|
return false; |
|
|
|
|
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
switch ($element[EN_TYPE]) { |
110
|
|
|
case EN_TYPE_STARTTAG: |
111
|
|
|
return $element; |
112
|
|
|
|
113
|
|
|
case EN_TYPE_ENDTAG: |
114
|
|
|
return $element; |
115
|
|
|
|
116
|
|
|
case EN_TYPE_CONTENT: |
117
|
|
|
WBXMLDecoder::ResetInWhile("decoderGetElement"); |
118
|
|
|
while (WBXMLDecoder::InWhile("decoderGetElement")) { |
119
|
|
|
$next = $this->getToken(); |
120
|
|
|
if ($next == false) { |
121
|
|
|
return false; |
|
|
|
|
122
|
|
|
} |
123
|
|
|
if ($next[EN_TYPE] == EN_CONTENT) { |
124
|
|
|
$element[EN_CONTENT] .= $next[EN_CONTENT]; |
125
|
|
|
} |
126
|
|
|
else { |
127
|
|
|
$this->ungetElement($next); |
128
|
|
|
|
129
|
|
|
break; |
130
|
|
|
} |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
return $element; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
return false; |
|
|
|
|
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
/** |
140
|
|
|
* Get a peek at the next element. |
141
|
|
|
* |
142
|
|
|
* @return element |
143
|
|
|
*/ |
144
|
|
|
public function peek() { |
145
|
|
|
$element = $this->getElement(); |
146
|
|
|
$this->ungetElement($element); |
147
|
|
|
|
148
|
|
|
return $element; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
/** |
152
|
|
|
* Get the element of a StartTag. |
153
|
|
|
* |
154
|
|
|
* @param mixed $tag |
155
|
|
|
* |
156
|
|
|
* @return bool|element returns false if not available |
157
|
|
|
*/ |
158
|
|
|
public function getElementStartTag($tag) { |
159
|
|
|
$element = $this->getToken(); |
160
|
|
|
|
161
|
|
|
if (!$element) { |
|
|
|
|
162
|
|
|
return false; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
if ($element[EN_TYPE] == EN_TYPE_STARTTAG && $element[EN_TAG] == $tag) { |
166
|
|
|
return $element; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
SLog::Write(LOGLEVEL_WBXMLSTACK, sprintf("WBXMLDecoder->getElementStartTag(): unmatched WBXML tag: '%s' matching '%s' type '%s' flags '%s'", $tag, (isset($element[EN_TAG])) ? $element[EN_TAG] : "", (isset($element[EN_TYPE])) ? $element[EN_TYPE] : "", (isset($element[EN_FLAGS])) ? $element[EN_FLAGS] : "")); |
170
|
|
|
$this->ungetElement($element); |
171
|
|
|
|
172
|
|
|
return false; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
/** |
176
|
|
|
* Get the element of a EndTag. |
177
|
|
|
* |
178
|
|
|
* @return bool|element returns false if not available |
179
|
|
|
*/ |
180
|
|
|
public function getElementEndTag() { |
181
|
|
|
$element = $this->getToken(); |
182
|
|
|
|
183
|
|
|
if ($element[EN_TYPE] == EN_TYPE_ENDTAG) { |
184
|
|
|
return $element; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
SLog::Write(LOGLEVEL_WBXMLSTACK, sprintf("WBXMLDecoder->getElementEndTag(): unmatched WBXML tag: '%s' type '%s' flags '%s'", (isset($element[EN_TAG])) ? $element[EN_TAG] : "", (isset($element[EN_TYPE])) ? $element[EN_TYPE] : "", (isset($element[EN_FLAGS])) ? $element[EN_FLAGS] : "")); |
188
|
|
|
|
189
|
|
|
$bt = debug_backtrace(); |
190
|
|
|
SLog::Write(LOGLEVEL_ERROR, sprintf("WBXMLDecoder->getElementEndTag(): could not read end tag in '%s'. Please enable the LOGLEVEL_WBXML and send the log to the grommunio-sync dev team.", $bt[0]["file"] . ":" . $bt[0]["line"])); |
191
|
|
|
|
192
|
|
|
// log the remaining wbxml content |
193
|
|
|
$this->ungetElement($element); |
194
|
|
|
while ($el = $this->getElement()); |
|
|
|
|
195
|
|
|
|
196
|
|
|
return false; |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
/** |
200
|
|
|
* Get the content of an element. |
201
|
|
|
* |
202
|
|
|
* @return bool|string returns false if not available |
203
|
|
|
*/ |
204
|
|
|
public function getElementContent() { |
205
|
|
|
$element = $this->getToken(); |
206
|
|
|
|
207
|
|
|
if ($element[EN_TYPE] == EN_TYPE_CONTENT) { |
208
|
|
|
return $element[EN_CONTENT]; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
SLog::Write(LOGLEVEL_WBXMLSTACK, sprintf("WBXMLDecoder->getElementContent(): unmatched WBXML content: '%s' type '%s' flags '%s'", (isset($element[EN_TAG])) ? $element[EN_TAG] : "", (isset($element[EN_TYPE])) ? $element[EN_TYPE] : "", (isset($element[EN_FLAGS])) ? $element[EN_FLAGS] : "")); |
212
|
|
|
$this->ungetElement($element); |
213
|
|
|
|
214
|
|
|
return false; |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
/** |
218
|
|
|
* 'Ungets' an element writing it into a buffer to be 'get' again. |
219
|
|
|
* |
220
|
|
|
* @param element $element the element to get ungetten |
221
|
|
|
*/ |
222
|
|
|
public function ungetElement($element) { |
223
|
|
|
if ($this->ungetbuffer) { |
224
|
|
|
SLog::Write(LOGLEVEL_ERROR, sprintf("WBXMLDecoder->ungetElement(): WBXML double unget on tag: '%s' type '%s' flags '%s'", (isset($element[EN_TAG])) ? $element[EN_TAG] : "", (isset($element[EN_TYPE])) ? $element[EN_TYPE] : "", (isset($element[EN_FLAGS])) ? $element[EN_FLAGS] : "")); |
225
|
|
|
} |
226
|
|
|
|
227
|
|
|
$this->ungetbuffer = $element; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
/** |
231
|
|
|
* Returns the plain input stream. |
232
|
|
|
* |
233
|
|
|
* @return string |
234
|
|
|
*/ |
235
|
|
|
public function GetPlainInputStream() { |
236
|
|
|
return $this->inputBuffer . stream_get_contents($this->in); |
|
|
|
|
237
|
|
|
} |
238
|
|
|
|
239
|
|
|
/** |
240
|
|
|
* Returns if the input is WBXML. |
241
|
|
|
* |
242
|
|
|
* @return bool |
243
|
|
|
*/ |
244
|
|
|
public function IsWBXML() { |
245
|
|
|
return $this->isWBXML; |
246
|
|
|
} |
247
|
|
|
|
248
|
|
|
/** |
249
|
|
|
* Reads the remaining data from the input stream. |
250
|
|
|
*/ |
251
|
|
|
public function readRemainingData() { |
252
|
|
|
SLog::Write(LOGLEVEL_DEBUG, "WBXMLDecoder->readRemainingData() reading remaining data from input stream"); |
253
|
|
|
while ($this->getElement()); |
254
|
|
|
} |
255
|
|
|
|
256
|
|
|
/*---------------------------------------------------------------------------------------------------------- |
257
|
|
|
* Private WBXMLDecoder stuff |
258
|
|
|
*/ |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* Returns the next token. |
262
|
|
|
* |
263
|
|
|
* @return token |
|
|
|
|
264
|
|
|
*/ |
265
|
|
|
private function getToken() { |
266
|
|
|
// See if there's something in the ungetBuffer |
267
|
|
|
if ($this->ungetbuffer) { |
268
|
|
|
$element = $this->ungetbuffer; |
269
|
|
|
$this->ungetbuffer = false; |
270
|
|
|
|
271
|
|
|
return $element; |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
$el = $this->_getToken(); |
275
|
|
|
if ($this->log && $el) { |
276
|
|
|
$this->logToken($el); |
|
|
|
|
277
|
|
|
} |
278
|
|
|
|
279
|
|
|
return $el; |
|
|
|
|
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
/** |
283
|
|
|
* Log the a token to SLog. |
284
|
|
|
* |
285
|
|
|
* @param string $el token |
286
|
|
|
*/ |
287
|
|
|
private function logToken($el) { |
288
|
|
|
$spaces = str_repeat(" ", count($this->logStack)); |
289
|
|
|
|
290
|
|
|
switch ($el[EN_TYPE]) { |
291
|
|
|
case EN_TYPE_STARTTAG: |
292
|
|
|
if ($el[EN_FLAGS] & EN_FLAGS_CONTENT) { |
293
|
|
|
SLog::Write(LOGLEVEL_WBXML, sprintf("I %s <%s>", $spaces, $el[EN_TAG])); |
294
|
|
|
array_push($this->logStack, $el[EN_TAG]); |
295
|
|
|
} |
296
|
|
|
else { |
297
|
|
|
SLog::Write(LOGLEVEL_WBXML, sprintf("I %s <%s/>", $spaces, $el[EN_TAG])); |
298
|
|
|
} |
299
|
|
|
break; |
300
|
|
|
|
301
|
|
|
case EN_TYPE_ENDTAG: |
302
|
|
|
$tag = array_pop($this->logStack); |
303
|
|
|
SLog::Write(LOGLEVEL_WBXML, sprintf("I %s</%s>", $spaces, $tag)); |
304
|
|
|
break; |
305
|
|
|
|
306
|
|
|
case EN_TYPE_CONTENT: |
307
|
|
|
$messagesize = strlen($el[EN_CONTENT]); |
308
|
|
|
// don't log binary data |
309
|
|
|
if (mb_detect_encoding($el[EN_CONTENT], null, true) === false) { |
310
|
|
|
$content = sprintf("(BINARY DATA: %d bytes long)", $messagesize); |
311
|
|
|
} |
312
|
|
|
// truncate logged data to 10K |
313
|
|
|
elseif ($messagesize > 10240 && !defined('WBXML_DEBUGGING')) { |
314
|
|
|
$content = sprintf("%s (log message with %d bytes truncated)", substr($el[EN_CONTENT], 0, 10240), $messagesize); |
315
|
|
|
} |
316
|
|
|
else { |
317
|
|
|
$content = $el[EN_CONTENT]; |
318
|
|
|
} |
319
|
|
|
|
320
|
|
|
SLog::Write(LOGLEVEL_WBXML, sprintf("I %s %s", $spaces, $content), false); |
321
|
|
|
break; |
322
|
|
|
} |
323
|
|
|
} |
324
|
|
|
|
325
|
|
|
/** |
326
|
|
|
* Returns either a start tag, content or end tag. |
327
|
|
|
*/ |
328
|
|
|
private function _getToken() { |
329
|
|
|
// Get the data from the input stream |
330
|
|
|
$element = []; |
331
|
|
|
|
332
|
|
|
WBXMLDecoder::ResetInWhile("decoderGetToken"); |
333
|
|
|
while (WBXMLDecoder::InWhile("decoderGetToken")) { |
334
|
|
|
$byte = fread($this->in, 1); |
|
|
|
|
335
|
|
|
if ($byte === "" || $byte === false) { |
336
|
|
|
break; |
337
|
|
|
} |
338
|
|
|
$byte = ord($byte); |
339
|
|
|
|
340
|
|
|
switch ($byte) { |
341
|
|
|
case self::WBXML_SWITCH_PAGE: |
342
|
|
|
$this->tagcp = $this->getByte(); |
343
|
|
|
break; |
344
|
|
|
|
345
|
|
|
case self::WBXML_END: |
346
|
|
|
$element[EN_TYPE] = EN_TYPE_ENDTAG; |
347
|
|
|
|
348
|
|
|
return $element; |
349
|
|
|
|
350
|
|
|
case self::WBXML_STR_I: |
351
|
|
|
$element[EN_TYPE] = EN_TYPE_CONTENT; |
352
|
|
|
$element[EN_CONTENT] = $this->getTermStr(); |
353
|
|
|
|
354
|
|
|
return $element; |
355
|
|
|
|
356
|
|
|
case self::WBXML_OPAQUE: |
357
|
|
|
$length = $this->getMBUInt(); |
358
|
|
|
$element[EN_TYPE] = EN_TYPE_CONTENT; |
359
|
|
|
$element[EN_CONTENT] = $this->getOpaque($length); |
360
|
|
|
|
361
|
|
|
return $element; |
362
|
|
|
|
363
|
|
|
case self::WBXML_ENTITY: |
364
|
|
|
case self::WBXML_LITERAL: |
365
|
|
|
case self::WBXML_EXT_I_0: |
366
|
|
|
case self::WBXML_EXT_I_1: |
367
|
|
|
case self::WBXML_EXT_I_2: |
368
|
|
|
case self::WBXML_PI: |
369
|
|
|
case self::WBXML_LITERAL_C: |
370
|
|
|
case self::WBXML_EXT_T_0: |
371
|
|
|
case self::WBXML_EXT_T_1: |
372
|
|
|
case self::WBXML_EXT_T_2: |
373
|
|
|
case self::WBXML_STR_T: |
374
|
|
|
case self::WBXML_LITERAL_A: |
375
|
|
|
case self::WBXML_EXT_0: |
376
|
|
|
case self::WBXML_EXT_1: |
377
|
|
|
case self::WBXML_EXT_2: |
378
|
|
|
case self::WBXML_LITERAL_AC: |
379
|
|
|
throw new WBXMLException("Invalid token :" . $byte); |
380
|
|
|
|
381
|
|
|
default: |
382
|
|
|
if ($byte & self::WBXML_WITH_ATTRIBUTES) { |
383
|
|
|
throw new WBXMLException("Attributes are not allowed :" . $byte); |
384
|
|
|
} |
385
|
|
|
$element[EN_TYPE] = EN_TYPE_STARTTAG; |
386
|
|
|
$element[EN_TAG] = $this->getMapping($this->tagcp, $byte & 0x3F); |
387
|
|
|
$element[EN_FLAGS] = ($byte & self::WBXML_WITH_CONTENT ? EN_FLAGS_CONTENT : 0); |
388
|
|
|
|
389
|
|
|
return $element; |
390
|
|
|
} |
391
|
|
|
} |
392
|
|
|
} |
393
|
|
|
|
394
|
|
|
/** |
395
|
|
|
* Reads from the stream until getting a string terminator. |
396
|
|
|
* |
397
|
|
|
* @return string |
398
|
|
|
*/ |
399
|
|
|
private function getTermStr() { |
400
|
|
|
if (defined('WBXML_DEBUGGING') && WBXML_DEBUGGING === true) { |
|
|
|
|
401
|
|
|
$str = ""; |
402
|
|
|
while (1) { |
403
|
|
|
$in = $this->getByte(); |
404
|
|
|
if ($in == 0) { |
405
|
|
|
break; |
406
|
|
|
} |
407
|
|
|
|
408
|
|
|
$str .= chr($in); |
409
|
|
|
} |
410
|
|
|
|
411
|
|
|
return $str; |
412
|
|
|
} |
413
|
|
|
|
414
|
|
|
// there is no unlimited "length" for stream_get_line, |
415
|
|
|
// so we use a huge value for "length" param (1Gb) |
416
|
|
|
// (0 == PHP_SOCK_CHUNK_SIZE (8192)) |
417
|
|
|
// internally php read at most PHP_SOCK_CHUNK_SIZE at a time, |
418
|
|
|
// so we can use a huge value for "length" without problem |
419
|
|
|
return stream_get_line($this->in, 1073741824, "\0"); |
|
|
|
|
420
|
|
|
} |
421
|
|
|
|
422
|
|
|
/** |
423
|
|
|
* Reads $len from the input stream. |
424
|
|
|
* |
425
|
|
|
* @param int $len |
426
|
|
|
* |
427
|
|
|
* @return string |
428
|
|
|
*/ |
429
|
|
|
private function getOpaque($len) { |
430
|
|
|
$d = stream_get_contents($this->in, $len); |
|
|
|
|
431
|
|
|
if ($d === false) { |
432
|
|
|
throw new HTTPReturnCodeException("WBXMLDecoder->getOpaque(): stream_get_contents === false", HTTP_CODE_500, null, LOGLEVEL_WARN); |
433
|
|
|
} |
434
|
|
|
$l = strlen($d); |
435
|
|
|
if ($l !== $len) { |
436
|
|
|
throw new HTTPReturnCodeException("WBXMLDecoder->getOpaque(): only {$l} byte read instead of {$len}", HTTP_CODE_500, null, LOGLEVEL_WARN); |
437
|
|
|
} |
438
|
|
|
|
439
|
|
|
return $d; |
440
|
|
|
} |
441
|
|
|
|
442
|
|
|
/** |
443
|
|
|
* Reads one byte from the input stream. |
444
|
|
|
* |
445
|
|
|
* @return int|void |
446
|
|
|
*/ |
447
|
|
|
private function getByte() { |
448
|
|
|
$ch = fread($this->in, 1); |
|
|
|
|
449
|
|
|
if (strlen($ch) > 0) { |
450
|
|
|
return ord($ch); |
451
|
|
|
} |
452
|
|
|
} |
453
|
|
|
|
454
|
|
|
/** |
455
|
|
|
* Reads string length from the input stream. |
456
|
|
|
*/ |
457
|
|
|
private function getMBUInt() { |
458
|
|
|
$uint = 0; |
459
|
|
|
|
460
|
|
|
while (1) { |
461
|
|
|
$byte = $this->getByte(); |
462
|
|
|
|
463
|
|
|
$uint |= $byte & 0x7F; |
464
|
|
|
|
465
|
|
|
if ($byte & 0x80) { |
466
|
|
|
$uint = $uint << 7; |
467
|
|
|
} |
468
|
|
|
else { |
469
|
|
|
break; |
470
|
|
|
} |
471
|
|
|
} |
472
|
|
|
|
473
|
|
|
return $uint; |
474
|
|
|
} |
475
|
|
|
|
476
|
|
|
/** |
477
|
|
|
* Returns the mapping for a specified codepage and id. |
478
|
|
|
* |
479
|
|
|
* @param $cp codepage |
|
|
|
|
480
|
|
|
* @param mixed $id |
481
|
|
|
* |
482
|
|
|
* @return string |
483
|
|
|
*/ |
484
|
|
|
private function getMapping($cp, $id) { |
485
|
|
|
if (!isset($this->dtd["codes"][$cp]) || !isset($this->dtd["codes"][$cp][$id])) { |
486
|
|
|
return false; |
|
|
|
|
487
|
|
|
} |
488
|
|
|
|
489
|
|
|
if (isset($this->dtd["namespaces"][$cp])) { |
490
|
|
|
return $this->dtd["namespaces"][$cp] . ":" . $this->dtd["codes"][$cp][$id]; |
491
|
|
|
} |
492
|
|
|
|
493
|
|
|
return $this->dtd["codes"][$cp][$id]; |
494
|
|
|
} |
495
|
|
|
} |
496
|
|
|
|