|
1
|
|
|
<?php |
|
2
|
|
|
/* vim: set expandtab tabstop=4 shiftwidth=4: */ |
|
3
|
|
|
// |
|
4
|
|
|
// +----------------------------------------------------------------------+ |
|
5
|
|
|
// | PHP Version 4 | |
|
6
|
|
|
// +----------------------------------------------------------------------+ |
|
7
|
|
|
// | Copyright (c) 1997-2002 The PHP Group | |
|
8
|
|
|
// +----------------------------------------------------------------------+ |
|
9
|
|
|
// | This source file is subject to version 2.02 of the PHP license, | |
|
10
|
|
|
// | that is bundled with this package in the file LICENSE, and is | |
|
11
|
|
|
// | available at through the world-wide-web at | |
|
12
|
|
|
// | http://www.php.net/license/3_0.txt. | |
|
13
|
|
|
// | If you did not receive a copy of the PHP license and are unable to | |
|
14
|
|
|
// | obtain it through the world-wide-web, please send a note to | |
|
15
|
|
|
// | [email protected] so we can mail you a copy immediately. | |
|
16
|
|
|
// +----------------------------------------------------------------------+ |
|
17
|
|
|
// | Authors: Alexander Zhukov <[email protected]> Original port from Python | |
|
18
|
|
|
// | Authors: Harry Fuecks <[email protected]> Port to PEAR + more | |
|
19
|
|
|
// | Authors: Many @ Sitepointforums Advanced PHP Forums | |
|
20
|
|
|
// +----------------------------------------------------------------------+ |
|
21
|
|
|
// |
|
22
|
|
|
// $Id: HTMLSax3.php 3188 2012-07-12 12:13:23Z ctrlaltca $ |
|
23
|
|
|
// |
|
24
|
|
|
/** |
|
25
|
|
|
* Main parser components |
|
26
|
|
|
* @package System.Security.SafeHtml |
|
27
|
|
|
*/ |
|
28
|
|
|
|
|
29
|
|
|
namespace Prado\Vendor\SafeHtml\HTMLSax3; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* Required classes |
|
33
|
|
|
*/ |
|
34
|
|
|
require_once(dirname(__FILE__).'/States.php'); |
|
35
|
|
|
require_once(dirname(__FILE__).'/Decorators.php'); |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Base State Parser |
|
39
|
|
|
* @package System.Security.SafeHtml |
|
40
|
|
|
* @access protected |
|
41
|
|
|
* @abstract |
|
42
|
|
|
*/ |
|
43
|
|
|
class TSax3_StateParser { |
|
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Instance of user front end class to be passed to callbacks |
|
46
|
|
|
* @var TSax3 |
|
47
|
|
|
* @access private |
|
48
|
|
|
*/ |
|
49
|
|
|
public $htmlsax; |
|
50
|
|
|
/** |
|
51
|
|
|
* User defined object for handling elements |
|
52
|
|
|
* @var object |
|
53
|
|
|
* @access private |
|
54
|
|
|
*/ |
|
55
|
|
|
public $handler_object_element; |
|
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* User defined open tag handler method |
|
58
|
|
|
* @var string |
|
59
|
|
|
* @access private |
|
60
|
|
|
*/ |
|
61
|
|
|
public $handler_method_opening; |
|
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* User defined close tag handler method |
|
64
|
|
|
* @var string |
|
65
|
|
|
* @access private |
|
66
|
|
|
*/ |
|
67
|
|
|
public $handler_method_closing; |
|
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* User defined object for handling data in elements |
|
70
|
|
|
* @var object |
|
71
|
|
|
* @access private |
|
72
|
|
|
*/ |
|
73
|
|
|
public $handler_object_data; |
|
74
|
|
|
/** |
|
75
|
|
|
* User defined data handler method |
|
76
|
|
|
* @var string |
|
77
|
|
|
* @access private |
|
78
|
|
|
*/ |
|
79
|
|
|
public $handler_method_data; |
|
80
|
|
|
/** |
|
81
|
|
|
* User defined object for handling processing instructions |
|
82
|
|
|
* @var object |
|
83
|
|
|
* @access private |
|
84
|
|
|
*/ |
|
85
|
|
|
public $handler_object_pi; |
|
86
|
|
|
/** |
|
87
|
|
|
* User defined processing instruction handler method |
|
88
|
|
|
* @var string |
|
89
|
|
|
* @access private |
|
90
|
|
|
*/ |
|
91
|
|
|
public $handler_method_pi; |
|
92
|
|
|
/** |
|
93
|
|
|
* User defined object for handling JSP/ASP tags |
|
94
|
|
|
* @var object |
|
95
|
|
|
* @access private |
|
96
|
|
|
*/ |
|
97
|
|
|
public $handler_object_jasp; |
|
98
|
|
|
/** |
|
99
|
|
|
* User defined JSP/ASP handler method |
|
100
|
|
|
* @var string |
|
101
|
|
|
* @access private |
|
102
|
|
|
*/ |
|
103
|
|
|
public $handler_method_jasp; |
|
104
|
|
|
/** |
|
105
|
|
|
* User defined object for handling XML escapes |
|
106
|
|
|
* @var object |
|
107
|
|
|
* @access private |
|
108
|
|
|
*/ |
|
109
|
|
|
public $handler_object_escape; |
|
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* User defined XML escape handler method |
|
112
|
|
|
* @var string |
|
113
|
|
|
* @access private |
|
114
|
|
|
*/ |
|
115
|
|
|
public $handler_method_escape; |
|
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* User defined handler object or NullHandler |
|
118
|
|
|
* @var object |
|
119
|
|
|
* @access private |
|
120
|
|
|
*/ |
|
121
|
|
|
public $handler_default; |
|
122
|
|
|
/** |
|
123
|
|
|
* Parser options determining parsing behavior |
|
124
|
|
|
* @var array |
|
125
|
|
|
* @access private |
|
126
|
|
|
*/ |
|
127
|
|
|
protected $parser_options = array(); |
|
128
|
|
|
/** |
|
129
|
|
|
* XML document being parsed |
|
130
|
|
|
* @var string |
|
131
|
|
|
* @access private |
|
132
|
|
|
*/ |
|
133
|
|
|
protected $rawtext; |
|
134
|
|
|
/** |
|
135
|
|
|
* Position in XML document relative to start (0) |
|
136
|
|
|
* @var int |
|
137
|
|
|
* @access private |
|
138
|
|
|
*/ |
|
139
|
|
|
protected $position; |
|
140
|
|
|
/** |
|
141
|
|
|
* Length of the XML document in characters |
|
142
|
|
|
* @var int |
|
143
|
|
|
* @access private |
|
144
|
|
|
*/ |
|
145
|
|
|
protected $length; |
|
146
|
|
|
/** |
|
147
|
|
|
* Array of state objects |
|
148
|
|
|
* @var array |
|
149
|
|
|
* @access private |
|
150
|
|
|
*/ |
|
151
|
|
|
protected $State = array(); |
|
152
|
|
|
|
|
153
|
|
|
const TSAX3_STATE_STOP = 0; |
|
154
|
|
|
const TSAX3_STATE_START = 1; |
|
155
|
|
|
const TSAX3_STATE_TAG = 2; |
|
156
|
|
|
const TSAX3_STATE_OPENING_TAG = 3; |
|
157
|
|
|
const TSAX3_STATE_CLOSING_TAG = 4; |
|
158
|
|
|
const TSAX3_STATE_ESCAPE = 6; |
|
159
|
|
|
const TSAX3_STATE_JASP = 7; |
|
160
|
|
|
const TSAX3_STATE_PI = 8; |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* Constructs TSax3_StateParser setting up states |
|
164
|
|
|
* @var TSax3 instance of user front end class |
|
165
|
|
|
* @access protected |
|
166
|
|
|
*/ |
|
167
|
|
|
protected function __construct($htmlsax) { |
|
168
|
|
|
$this->htmlsax = $htmlsax; |
|
169
|
|
|
$this->State[self::TSAX3_STATE_START] = new TSax3_StartingState(); |
|
170
|
|
|
|
|
171
|
|
|
$this->State[self::TSAX3_STATE_CLOSING_TAG] = new TSax3_ClosingTagState(); |
|
172
|
|
|
$this->State[self::TSAX3_STATE_TAG] = new TSax3_TagState(); |
|
173
|
|
|
$this->State[self::TSAX3_STATE_OPENING_TAG] = new TSax3_OpeningTagState(); |
|
174
|
|
|
|
|
175
|
|
|
$this->State[self::TSAX3_STATE_PI] = new TSax3_PiState(); |
|
176
|
|
|
$this->State[self::TSAX3_STATE_JASP] = new TSax3_JaspState(); |
|
177
|
|
|
$this->State[self::TSAX3_STATE_ESCAPE] = new TSax3_EscapeState(); |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
/** |
|
181
|
|
|
* Moves the position back one character |
|
182
|
|
|
* @access protected |
|
183
|
|
|
* @return void |
|
184
|
|
|
*/ |
|
185
|
|
|
function unscanCharacter() { |
|
186
|
|
|
$this->position -= 1; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
/** |
|
190
|
|
|
* Moves the position forward one character |
|
191
|
|
|
* @access protected |
|
192
|
|
|
* @return void |
|
193
|
|
|
*/ |
|
194
|
|
|
function ignoreCharacter() { |
|
195
|
|
|
$this->position += 1; |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
/** |
|
199
|
|
|
* Returns the next character from the XML document or void if at end |
|
200
|
|
|
* @access protected |
|
201
|
|
|
* @return mixed |
|
|
|
|
|
|
202
|
|
|
*/ |
|
203
|
|
|
function scanCharacter() { |
|
204
|
|
|
if ($this->position < $this->length) { |
|
205
|
|
|
return $this->rawtext{$this->position++}; |
|
206
|
|
|
} |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
/** |
|
210
|
|
|
* Returns a string from the current position to the next occurance |
|
211
|
|
|
* of the supplied string |
|
212
|
|
|
* @param string string to search until |
|
213
|
|
|
* @access protected |
|
214
|
|
|
* @return string |
|
215
|
|
|
*/ |
|
216
|
|
|
function scanUntilString($string) { |
|
217
|
|
|
$start = $this->position; |
|
218
|
|
|
$this->position = strpos($this->rawtext, $string, $start); |
|
219
|
|
|
if ($this->position === FALSE) { |
|
220
|
|
|
$this->position = $this->length; |
|
221
|
|
|
} |
|
222
|
|
|
return substr($this->rawtext, $start, $this->position - $start); |
|
223
|
|
|
} |
|
224
|
|
|
|
|
225
|
|
|
/** |
|
226
|
|
|
* Returns a string from the current position until the first instance of |
|
227
|
|
|
* one of the characters in the supplied string argument |
|
228
|
|
|
* @param string string to search until |
|
229
|
|
|
* @access protected |
|
230
|
|
|
* @return string |
|
231
|
|
|
* @abstract |
|
232
|
|
|
*/ |
|
233
|
|
|
function scanUntilCharacters($string) {} |
|
234
|
|
|
|
|
235
|
|
|
/** |
|
236
|
|
|
* Moves the position forward past any whitespace characters |
|
237
|
|
|
* @access protected |
|
238
|
|
|
* @return void |
|
239
|
|
|
* @abstract |
|
240
|
|
|
*/ |
|
241
|
|
|
function ignoreWhitespace() {} |
|
242
|
|
|
|
|
243
|
|
|
/** |
|
244
|
|
|
* Begins the parsing operation, setting up any decorators, depending on |
|
245
|
|
|
* parse options invoking _parse() to execute parsing |
|
246
|
|
|
* @param string XML document to parse |
|
247
|
|
|
* @access protected |
|
248
|
|
|
* @return void |
|
249
|
|
|
*/ |
|
250
|
|
|
function parse($data) { |
|
251
|
|
|
if ($this->parser_options['XML_OPTION_TRIM_DATA_NODES']==1) { |
|
252
|
|
|
$decorator = new TSax3_Trim( |
|
253
|
|
|
$this->handler_object_data, |
|
254
|
|
|
$this->handler_method_data); |
|
255
|
|
|
$this->handler_object_data =& $decorator; |
|
256
|
|
|
$this->handler_method_data = 'trimData'; |
|
257
|
|
|
} |
|
258
|
|
|
if ($this->parser_options['XML_OPTION_CASE_FOLDING']==1) { |
|
259
|
|
|
$open_decor = new TSax3_CaseFolding( |
|
260
|
|
|
$this->handler_object_element, |
|
261
|
|
|
$this->handler_method_opening, |
|
262
|
|
|
$this->handler_method_closing); |
|
263
|
|
|
$this->handler_object_element =& $open_decor; |
|
264
|
|
|
$this->handler_method_opening ='foldOpen'; |
|
265
|
|
|
$this->handler_method_closing ='foldClose'; |
|
266
|
|
|
} |
|
267
|
|
|
if ($this->parser_options['XML_OPTION_LINEFEED_BREAK']==1) { |
|
268
|
|
|
$decorator = new TSax3_Linefeed( |
|
269
|
|
|
$this->handler_object_data, |
|
270
|
|
|
$this->handler_method_data); |
|
271
|
|
|
$this->handler_object_data =& $decorator; |
|
272
|
|
|
$this->handler_method_data = 'breakData'; |
|
273
|
|
|
} |
|
274
|
|
|
if ($this->parser_options['XML_OPTION_TAB_BREAK']==1) { |
|
275
|
|
|
$decorator = new TSax3_Tab( |
|
276
|
|
|
$this->handler_object_data, |
|
277
|
|
|
$this->handler_method_data); |
|
278
|
|
|
$this->handler_object_data =& $decorator; |
|
279
|
|
|
$this->handler_method_data = 'breakData'; |
|
280
|
|
|
} |
|
281
|
|
|
if ($this->parser_options['XML_OPTION_ENTITIES_UNPARSED']==1) { |
|
282
|
|
|
$decorator = new TSax3_Entities_Unparsed( |
|
283
|
|
|
$this->handler_object_data, |
|
284
|
|
|
$this->handler_method_data); |
|
285
|
|
|
$this->handler_object_data =& $decorator; |
|
286
|
|
|
$this->handler_method_data = 'breakData'; |
|
287
|
|
|
} |
|
288
|
|
|
if ($this->parser_options['XML_OPTION_ENTITIES_PARSED']==1) { |
|
289
|
|
|
$decorator = new TSax3_Entities_Parsed( |
|
290
|
|
|
$this->handler_object_data, |
|
291
|
|
|
$this->handler_method_data); |
|
292
|
|
|
$this->handler_object_data =& $decorator; |
|
293
|
|
|
$this->handler_method_data = 'breakData'; |
|
294
|
|
|
} |
|
295
|
|
|
// Note switched on by default |
|
296
|
|
|
if ($this->parser_options['XML_OPTION_STRIP_ESCAPES']==1) { |
|
297
|
|
|
$decorator = new TSax3_Escape_Stripper( |
|
298
|
|
|
$this->handler_object_escape, |
|
299
|
|
|
$this->handler_method_escape); |
|
300
|
|
|
$this->handler_object_escape =& $decorator; |
|
301
|
|
|
$this->handler_method_escape = 'strip'; |
|
302
|
|
|
} |
|
303
|
|
|
$this->rawtext = $data; |
|
304
|
|
|
$this->length = strlen($data); |
|
305
|
|
|
$this->position = 0; |
|
306
|
|
|
$this->_parse(); |
|
307
|
|
|
} |
|
308
|
|
|
|
|
309
|
|
|
/** |
|
310
|
|
|
* Performs the parsing itself, delegating calls to a specific parser |
|
311
|
|
|
* state |
|
312
|
|
|
* @param constant state object to parse with |
|
313
|
|
|
* @access protected |
|
314
|
|
|
* @return void |
|
315
|
|
|
*/ |
|
316
|
|
|
function _parse($state = self::TSAX3_STATE_START) { |
|
317
|
|
|
do { |
|
318
|
|
|
$state = $this->State[$state]->parse($this); |
|
319
|
|
|
} while ($state != self::TSAX3_STATE_STOP && |
|
320
|
|
|
$this->position < $this->length); |
|
321
|
|
|
} |
|
322
|
|
|
} |
|
323
|
|
|
|
|
324
|
|
|
/** |
|
325
|
|
|
* Parser for PHP Versions below 4.3.0. Uses a slower parsing mechanism than |
|
326
|
|
|
* the equivalent PHP 4.3.0+ subclass of StateParser |
|
327
|
|
|
* @package System.Security.SafeHtml |
|
328
|
|
|
* @access protected |
|
329
|
|
|
* @see TSax3_StateParser_Gtet430 |
|
330
|
|
|
*/ |
|
331
|
|
|
class TSax3_StateParser_Lt430 extends TSax3_StateParser { |
|
|
|
|
|
|
332
|
|
|
/** |
|
333
|
|
|
* Constructs TSax3_StateParser_Lt430 defining available |
|
334
|
|
|
* parser options |
|
335
|
|
|
* @var TSax3 instance of user front end class |
|
336
|
|
|
* @access protected |
|
337
|
|
|
*/ |
|
338
|
|
|
function __construct(& $htmlsax) { |
|
339
|
|
|
parent::__construct($htmlsax); |
|
340
|
|
|
$this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; |
|
341
|
|
|
$this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; |
|
342
|
|
|
$this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; |
|
343
|
|
|
$this->parser_options['XML_OPTION_TAB_BREAK'] = 0; |
|
344
|
|
|
$this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; |
|
345
|
|
|
$this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; |
|
346
|
|
|
$this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; |
|
347
|
|
|
//var_dump($this->parser_options); |
|
348
|
|
|
} |
|
349
|
|
|
|
|
350
|
|
|
/** |
|
351
|
|
|
* Returns a string from the current position until the first instance of |
|
352
|
|
|
* one of the characters in the supplied string argument |
|
353
|
|
|
* @param string string to search until |
|
354
|
|
|
* @access protected |
|
355
|
|
|
* @return string |
|
356
|
|
|
*/ |
|
357
|
|
|
function scanUntilCharacters($string) { |
|
358
|
|
|
$startpos = $this->position; |
|
359
|
|
|
while ($this->position < $this->length && strpos($string, $this->rawtext{$this->position}) === FALSE) { |
|
360
|
|
|
$this->position++; |
|
361
|
|
|
} |
|
362
|
|
|
return substr($this->rawtext, $startpos, $this->position - $startpos); |
|
363
|
|
|
} |
|
364
|
|
|
|
|
365
|
|
|
/** |
|
366
|
|
|
* Moves the position forward past any whitespace characters |
|
367
|
|
|
* @access protected |
|
368
|
|
|
* @return void |
|
369
|
|
|
*/ |
|
370
|
|
|
function ignoreWhitespace() { |
|
371
|
|
|
while ($this->position < $this->length && |
|
372
|
|
|
strpos(" \n\r\t", $this->rawtext{$this->position}) !== FALSE) { |
|
373
|
|
|
$this->position++; |
|
374
|
|
|
} |
|
375
|
|
|
} |
|
376
|
|
|
|
|
377
|
|
|
/** |
|
378
|
|
|
* Begins the parsing operation, setting up the unparsed XML entities |
|
379
|
|
|
* decorator if necessary then delegating further work to parent |
|
380
|
|
|
* @param string XML document to parse |
|
381
|
|
|
* @access protected |
|
382
|
|
|
* @return void |
|
383
|
|
|
*/ |
|
384
|
|
|
function parse($data) { |
|
385
|
|
|
parent::parse($data); |
|
386
|
|
|
} |
|
387
|
|
|
} |
|
388
|
|
|
|
|
389
|
|
|
/** |
|
390
|
|
|
* Parser for PHP Versions equal to or greater than 4.3.0. Uses a faster |
|
391
|
|
|
* parsing mechanism than the equivalent PHP < 4.3.0 subclass of StateParser |
|
392
|
|
|
* @package System.Security.SafeHtml |
|
393
|
|
|
* @access protected |
|
394
|
|
|
* @see TSax3_StateParser_Lt430 |
|
395
|
|
|
*/ |
|
396
|
|
|
class TSax3_StateParser_Gtet430 extends TSax3_StateParser { |
|
|
|
|
|
|
397
|
|
|
/** |
|
398
|
|
|
* Constructs TSax3_StateParser_Gtet430 defining available |
|
399
|
|
|
* parser options |
|
400
|
|
|
* @var TSax3 instance of user front end class |
|
401
|
|
|
* @access protected |
|
402
|
|
|
*/ |
|
403
|
|
|
function __construct(& $htmlsax) { |
|
404
|
|
|
parent::__construct($htmlsax); |
|
405
|
|
|
$this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; |
|
406
|
|
|
$this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; |
|
407
|
|
|
$this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; |
|
408
|
|
|
$this->parser_options['XML_OPTION_TAB_BREAK'] = 0; |
|
409
|
|
|
$this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; |
|
410
|
|
|
$this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; |
|
411
|
|
|
$this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; |
|
412
|
|
|
} |
|
413
|
|
|
/** |
|
414
|
|
|
* Returns a string from the current position until the first instance of |
|
415
|
|
|
* one of the characters in the supplied string argument. |
|
416
|
|
|
* @param string string to search until |
|
417
|
|
|
* @access protected |
|
418
|
|
|
* @return string |
|
419
|
|
|
*/ |
|
420
|
|
|
function scanUntilCharacters($string) { |
|
421
|
|
|
$startpos = $this->position; |
|
422
|
|
|
$length = strcspn($this->rawtext, $string, $startpos); |
|
423
|
|
|
$this->position += $length; |
|
424
|
|
|
return substr($this->rawtext, $startpos, $length); |
|
425
|
|
|
} |
|
426
|
|
|
|
|
427
|
|
|
/** |
|
428
|
|
|
* Moves the position forward past any whitespace characters |
|
429
|
|
|
* @access protected |
|
430
|
|
|
* @return void |
|
431
|
|
|
*/ |
|
432
|
|
|
function ignoreWhitespace() { |
|
433
|
|
|
$this->position += strspn($this->rawtext, " \n\r\t", $this->position); |
|
434
|
|
|
} |
|
435
|
|
|
|
|
436
|
|
|
/** |
|
437
|
|
|
* Begins the parsing operation, setting up the parsed and unparsed |
|
438
|
|
|
* XML entity decorators if necessary then delegating further work |
|
439
|
|
|
* to parent |
|
440
|
|
|
* @param string XML document to parse |
|
441
|
|
|
* @access protected |
|
442
|
|
|
* @return void |
|
443
|
|
|
*/ |
|
444
|
|
|
function parse($data) { |
|
445
|
|
|
parent::parse($data); |
|
446
|
|
|
} |
|
447
|
|
|
} |
|
448
|
|
|
|
|
449
|
|
|
/** |
|
450
|
|
|
* Default NullHandler for methods which were not set by user |
|
451
|
|
|
* @package System.Security.SafeHtml |
|
452
|
|
|
* @access protected |
|
453
|
|
|
*/ |
|
454
|
|
|
class TSax3_NullHandler { |
|
|
|
|
|
|
455
|
|
|
/** |
|
456
|
|
|
* Generic handler method which does nothing |
|
457
|
|
|
* @access protected |
|
458
|
|
|
* @return void |
|
459
|
|
|
*/ |
|
460
|
|
|
function DoNothing() { |
|
461
|
|
|
} |
|
462
|
|
|
} |
|
463
|
|
|
|
|
464
|
|
|
/** |
|
465
|
|
|
* User interface class. All user calls should only be made to this class |
|
466
|
|
|
* @package System.Security.SafeHtml |
|
467
|
|
|
* @access public |
|
468
|
|
|
*/ |
|
469
|
|
|
class TSax3 { |
|
470
|
|
|
/** |
|
471
|
|
|
* Instance of concrete subclass of TSax3_StateParser |
|
472
|
|
|
* @var TSax3_StateParser |
|
473
|
|
|
* @access private |
|
474
|
|
|
*/ |
|
475
|
|
|
private $state_parser; |
|
476
|
|
|
|
|
477
|
|
|
/** |
|
478
|
|
|
* Constructs TSax3 selecting concrete StateParser subclass |
|
479
|
|
|
* depending on PHP version being used as well as setting the default |
|
480
|
|
|
* NullHandler for all callbacks<br /> |
|
481
|
|
|
* <b>Example:</b> |
|
482
|
|
|
* <pre> |
|
483
|
|
|
* $myHandler = & new MyHandler(); |
|
484
|
|
|
* $parser = new TSax3(); |
|
485
|
|
|
* $parser->set_object($myHandler); |
|
486
|
|
|
* $parser->set_option('XML_OPTION_CASE_FOLDING'); |
|
487
|
|
|
* $parser->set_element_handler('myOpenHandler','myCloseHandler'); |
|
488
|
|
|
* $parser->set_data_handler('myDataHandler'); |
|
489
|
|
|
* $parser->parser($xml); |
|
490
|
|
|
* </pre> |
|
491
|
|
|
* @access public |
|
492
|
|
|
*/ |
|
493
|
|
|
function __construct() { |
|
494
|
|
|
if (version_compare(phpversion(), '4.3', 'ge')) { |
|
495
|
|
|
$this->state_parser = new TSax3_StateParser_Gtet430($this); |
|
496
|
|
|
} else { |
|
497
|
|
|
$this->state_parser = new TSax3_StateParser_Lt430($this); |
|
498
|
|
|
} |
|
499
|
|
|
$nullhandler = new TSax3_NullHandler(); |
|
500
|
|
|
$this->set_object($nullhandler); |
|
501
|
|
|
$this->set_element_handler('DoNothing', 'DoNothing'); |
|
502
|
|
|
$this->set_data_handler('DoNothing'); |
|
503
|
|
|
$this->set_pi_handler('DoNothing'); |
|
504
|
|
|
$this->set_jasp_handler('DoNothing'); |
|
505
|
|
|
$this->set_escape_handler('DoNothing'); |
|
506
|
|
|
} |
|
507
|
|
|
|
|
508
|
|
|
/** |
|
509
|
|
|
* Sets the user defined handler object. Returns a PEAR Error |
|
510
|
|
|
* if supplied argument is not an object. |
|
511
|
|
|
* @param object handler object containing SAX callback methods |
|
512
|
|
|
* @access public |
|
513
|
|
|
* @return mixed |
|
|
|
|
|
|
514
|
|
|
*/ |
|
515
|
|
|
function set_object(&$object) { |
|
516
|
|
|
if ( is_object($object) ) { |
|
517
|
|
|
$this->state_parser->handler_default =& $object; |
|
518
|
|
|
return true; |
|
519
|
|
|
} else { |
|
520
|
|
|
require_once('PEAR.php'); |
|
521
|
|
|
PEAR::raiseError('TSax3::set_object requires '. |
|
522
|
|
|
'an object instance'); |
|
523
|
|
|
} |
|
524
|
|
|
} |
|
525
|
|
|
|
|
526
|
|
|
/** |
|
527
|
|
|
* Sets a parser option. By default all options are switched off. |
|
528
|
|
|
* Returns a PEAR Error if option is invalid<br /> |
|
529
|
|
|
* <b>Available options:</b> |
|
530
|
|
|
* <ul> |
|
531
|
|
|
* <li>XML_OPTION_TRIM_DATA_NODES: trim whitespace off the beginning |
|
532
|
|
|
* and end of data passed to the data handler</li> |
|
533
|
|
|
* <li>XML_OPTION_LINEFEED_BREAK: linefeeds result in additional data |
|
534
|
|
|
* handler calls</li> |
|
535
|
|
|
* <li>XML_OPTION_TAB_BREAK: tabs result in additional data handler |
|
536
|
|
|
* calls</li> |
|
537
|
|
|
* <li>XML_OPTION_ENTITIES_UNPARSED: XML entities are returned as |
|
538
|
|
|
* seperate data handler calls in unparsed form</li> |
|
539
|
|
|
* <li>XML_OPTION_ENTITIES_PARSED: (PHP 4.3.0+ only) XML entities are |
|
540
|
|
|
* returned as seperate data handler calls and are parsed with |
|
541
|
|
|
* PHP's html_entity_decode() function</li> |
|
542
|
|
|
* <li>XML_OPTION_STRIP_ESCAPES: strips out the -- -- comment markers |
|
543
|
|
|
* or CDATA markup inside an XML escape, if found.</li> |
|
544
|
|
|
* </ul> |
|
545
|
|
|
* To get HTMLSax to behave in the same way as the native PHP SAX parser, |
|
546
|
|
|
* using it's default state, you need to switch on XML_OPTION_LINEFEED_BREAK, |
|
547
|
|
|
* XML_OPTION_ENTITIES_PARSED and XML_OPTION_CASE_FOLDING |
|
548
|
|
|
* @param string name of parser option |
|
549
|
|
|
* @param int (optional) 1 to switch on, 0 for off |
|
550
|
|
|
* @access public |
|
551
|
|
|
* @return boolean |
|
552
|
|
|
*/ |
|
553
|
|
|
function set_option($name, $value=1) { |
|
554
|
|
|
if ( array_key_exists($name,$this->state_parser->parser_options) ) { |
|
|
|
|
|
|
555
|
|
|
$this->state_parser->parser_options[$name] = $value; |
|
|
|
|
|
|
556
|
|
|
return true; |
|
557
|
|
|
} else { |
|
558
|
|
|
require_once('PEAR.php'); |
|
559
|
|
|
PEAR::raiseError('TSax3::set_option('.$name.') illegal'); |
|
560
|
|
|
} |
|
561
|
|
|
} |
|
562
|
|
|
|
|
563
|
|
|
/** |
|
564
|
|
|
* Sets the data handler method which deals with the contents of XML |
|
565
|
|
|
* elements.<br /> |
|
566
|
|
|
* The handler method must accept two arguments, the first being an |
|
567
|
|
|
* instance of TSax3 and the second being the contents of an |
|
568
|
|
|
* XML element e.g. |
|
569
|
|
|
* <pre> |
|
570
|
|
|
* function myDataHander(& $parser,$data){} |
|
571
|
|
|
* </pre> |
|
572
|
|
|
* @param string name of method |
|
573
|
|
|
* @access public |
|
574
|
|
|
* @return void |
|
575
|
|
|
* @see set_object |
|
576
|
|
|
*/ |
|
577
|
|
|
function set_data_handler($data_method) { |
|
|
|
|
|
|
578
|
|
|
$this->state_parser->handler_object_data =& $this->state_parser->handler_default; |
|
579
|
|
|
$this->state_parser->handler_method_data = $data_method; |
|
580
|
|
|
} |
|
581
|
|
|
|
|
582
|
|
|
/** |
|
583
|
|
|
* Sets the open and close tag handlers |
|
584
|
|
|
* <br />The open handler method must accept three arguments; the parser, |
|
585
|
|
|
* the tag name and an array of attributes e.g. |
|
586
|
|
|
* <pre> |
|
587
|
|
|
* function myOpenHander(& $parser,$tagname,$attrs=array()){} |
|
588
|
|
|
* </pre> |
|
589
|
|
|
* The close handler method must accept two arguments; the parser and |
|
590
|
|
|
* the tag name e.g. |
|
591
|
|
|
* <pre> |
|
592
|
|
|
* function myCloseHander(& $parser,$tagname){} |
|
593
|
|
|
* </pre> |
|
594
|
|
|
* @param string name of open method |
|
595
|
|
|
* @param string name of close method |
|
596
|
|
|
* @access public |
|
597
|
|
|
* @return void |
|
598
|
|
|
* @see set_object |
|
599
|
|
|
*/ |
|
600
|
|
|
function set_element_handler($opening_method, $closing_method) { |
|
|
|
|
|
|
601
|
|
|
$this->state_parser->handler_object_element =& $this->state_parser->handler_default; |
|
602
|
|
|
$this->state_parser->handler_method_opening = $opening_method; |
|
603
|
|
|
$this->state_parser->handler_method_closing = $closing_method; |
|
604
|
|
|
} |
|
605
|
|
|
|
|
606
|
|
|
/** |
|
607
|
|
|
* Sets the processing instruction handler method e.g. for PHP open |
|
608
|
|
|
* and close tags<br /> |
|
609
|
|
|
* The handler method must accept three arguments; the parser, the |
|
610
|
|
|
* PI target and data inside the PI |
|
611
|
|
|
* <pre> |
|
612
|
|
|
* function myPIHander(& $parser,$target, $data){} |
|
613
|
|
|
* </pre> |
|
614
|
|
|
* @param string name of method |
|
615
|
|
|
* @access public |
|
616
|
|
|
* @return void |
|
617
|
|
|
* @see set_object |
|
618
|
|
|
*/ |
|
619
|
|
|
function set_pi_handler($pi_method) { |
|
|
|
|
|
|
620
|
|
|
$this->state_parser->handler_object_pi =& $this->state_parser->handler_default; |
|
621
|
|
|
$this->state_parser->handler_method_pi = $pi_method; |
|
622
|
|
|
} |
|
623
|
|
|
|
|
624
|
|
|
/** |
|
625
|
|
|
* Sets the XML escape handler method e.g. for comments and doctype |
|
626
|
|
|
* declarations<br /> |
|
627
|
|
|
* The handler method must accept two arguments; the parser and the |
|
628
|
|
|
* contents of the escaped section |
|
629
|
|
|
* <pre> |
|
630
|
|
|
* function myEscapeHander(& $parser, $data){} |
|
631
|
|
|
* </pre> |
|
632
|
|
|
* @param string name of method |
|
633
|
|
|
* @access public |
|
634
|
|
|
* @return void |
|
635
|
|
|
* @see set_object |
|
636
|
|
|
*/ |
|
637
|
|
|
function set_escape_handler($escape_method) { |
|
|
|
|
|
|
638
|
|
|
$this->state_parser->handler_object_escape =& $this->state_parser->handler_default; |
|
639
|
|
|
$this->state_parser->handler_method_escape = $escape_method; |
|
640
|
|
|
} |
|
641
|
|
|
|
|
642
|
|
|
/** |
|
643
|
|
|
* Sets the JSP/ASP markup handler<br /> |
|
644
|
|
|
* The handler method must accept two arguments; the parser and |
|
645
|
|
|
* body of the JASP tag |
|
646
|
|
|
* <pre> |
|
647
|
|
|
* function myJaspHander(& $parser, $data){} |
|
648
|
|
|
* </pre> |
|
649
|
|
|
* @param string name of method |
|
650
|
|
|
* @access public |
|
651
|
|
|
* @return void |
|
652
|
|
|
* @see set_object |
|
653
|
|
|
*/ |
|
654
|
|
|
function set_jasp_handler ($jasp_method) { |
|
|
|
|
|
|
655
|
|
|
$this->state_parser->handler_object_jasp =& $this->state_parser->handler_default; |
|
656
|
|
|
$this->state_parser->handler_method_jasp = $jasp_method; |
|
657
|
|
|
} |
|
658
|
|
|
|
|
659
|
|
|
/** |
|
660
|
|
|
* Returns the current string position of the "cursor" inside the XML |
|
661
|
|
|
* document |
|
662
|
|
|
* <br />Intended for use from within a user defined handler called |
|
663
|
|
|
* via the $parser reference e.g. |
|
664
|
|
|
* <pre> |
|
665
|
|
|
* function myDataHandler(& $parser,$data) { |
|
666
|
|
|
* echo( 'Current position: '.$parser->get_current_position() ); |
|
667
|
|
|
* } |
|
668
|
|
|
* </pre> |
|
669
|
|
|
* @access public |
|
670
|
|
|
* @return int |
|
671
|
|
|
* @see get_length |
|
672
|
|
|
*/ |
|
673
|
|
|
function get_current_position() { |
|
674
|
|
|
return $this->state_parser->position; |
|
|
|
|
|
|
675
|
|
|
} |
|
676
|
|
|
|
|
677
|
|
|
/** |
|
678
|
|
|
* Returns the string length of the XML document being parsed |
|
679
|
|
|
* @access public |
|
680
|
|
|
* @return int |
|
681
|
|
|
*/ |
|
682
|
|
|
function get_length() { |
|
683
|
|
|
return $this->state_parser->length; |
|
|
|
|
|
|
684
|
|
|
} |
|
685
|
|
|
|
|
686
|
|
|
/** |
|
687
|
|
|
* Start parsing some XML |
|
688
|
|
|
* @param string XML document |
|
689
|
|
|
* @access public |
|
690
|
|
|
* @return void |
|
691
|
|
|
*/ |
|
692
|
|
|
function parse($data) { |
|
693
|
|
|
$this->state_parser->parse($data); |
|
694
|
|
|
} |
|
695
|
|
|
} |
|
696
|
|
|
|
Classes in PHP are usually named in CamelCase.
In camelCase names are written without any punctuation, the start of each new word being marked by a capital letter. The whole name starts with a capital letter as well.
Thus the name database provider becomes
DatabaseProvider.