1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\TextFormatter |
5
|
|
|
* @copyright Copyright (c) The s9e authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\TextFormatter; |
9
|
|
|
|
10
|
|
|
use InvalidArgumentException; |
11
|
|
|
use RuntimeException; |
12
|
|
|
use s9e\TextFormatter\Parser\FilterProcessing; |
13
|
|
|
use s9e\TextFormatter\Parser\Logger; |
14
|
|
|
use s9e\TextFormatter\Parser\Tag; |
15
|
|
|
|
16
|
|
|
class Parser |
17
|
|
|
{ |
18
|
|
|
/**#@+ |
19
|
|
|
* Boolean rules bitfield |
20
|
|
|
*/ |
21
|
|
|
const RULE_AUTO_CLOSE = 1 << 0; |
22
|
|
|
const RULE_AUTO_REOPEN = 1 << 1; |
23
|
|
|
const RULE_BREAK_PARAGRAPH = 1 << 2; |
24
|
|
|
const RULE_CREATE_PARAGRAPHS = 1 << 3; |
25
|
|
|
const RULE_DISABLE_AUTO_BR = 1 << 4; |
26
|
|
|
const RULE_ENABLE_AUTO_BR = 1 << 5; |
27
|
|
|
const RULE_IGNORE_TAGS = 1 << 6; |
28
|
|
|
const RULE_IGNORE_TEXT = 1 << 7; |
29
|
|
|
const RULE_IGNORE_WHITESPACE = 1 << 8; |
30
|
|
|
const RULE_IS_TRANSPARENT = 1 << 9; |
31
|
|
|
const RULE_PREVENT_BR = 1 << 10; |
32
|
|
|
const RULE_SUSPEND_AUTO_BR = 1 << 11; |
33
|
|
|
const RULE_TRIM_FIRST_LINE = 1 << 12; |
34
|
|
|
/**#@-*/ |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Bitwise disjunction of rules related to automatic line breaks |
38
|
|
|
*/ |
39
|
|
|
const RULES_AUTO_LINEBREAKS = self::RULE_DISABLE_AUTO_BR | self::RULE_ENABLE_AUTO_BR | self::RULE_SUSPEND_AUTO_BR; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* Bitwise disjunction of rules that are inherited by subcontexts |
43
|
|
|
*/ |
44
|
|
|
const RULES_INHERITANCE = self::RULE_ENABLE_AUTO_BR; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* All the characters that are considered whitespace |
48
|
|
|
*/ |
49
|
|
|
const WHITESPACE = " \n\t"; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var array Number of open tags for each tag name |
53
|
|
|
*/ |
54
|
|
|
protected $cntOpen; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @var array Number of times each tag has been used |
58
|
|
|
*/ |
59
|
|
|
protected $cntTotal; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @var array Current context |
63
|
|
|
*/ |
64
|
|
|
protected $context; |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @var integer How hard the parser has worked on fixing bad markup so far |
68
|
|
|
*/ |
69
|
|
|
protected $currentFixingCost; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @var Tag Current tag being processed |
73
|
|
|
*/ |
74
|
|
|
protected $currentTag; |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* @var bool Whether the output contains "rich" tags, IOW any tag that is not <p> or <br/> |
78
|
|
|
*/ |
79
|
|
|
protected $isRich; |
80
|
|
|
|
81
|
|
|
/** |
82
|
|
|
* @var Logger This parser's logger |
83
|
|
|
*/ |
84
|
|
|
protected $logger; |
85
|
|
|
|
86
|
|
|
/** |
87
|
|
|
* @var integer How hard the parser should work on fixing bad markup |
88
|
|
|
*/ |
89
|
|
|
public $maxFixingCost = 10000; |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* @var array Associative array of namespace prefixes in use in document (prefixes used as key) |
93
|
|
|
*/ |
94
|
|
|
protected $namespaces; |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* @var array Stack of open tags (instances of Tag) |
98
|
|
|
*/ |
99
|
|
|
protected $openTags; |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* @var string This parser's output |
103
|
|
|
*/ |
104
|
|
|
protected $output; |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* @var integer Position of the cursor in the original text |
108
|
|
|
*/ |
109
|
|
|
protected $pos; |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* @var array Array of callbacks, using plugin names as keys |
113
|
|
|
*/ |
114
|
|
|
protected $pluginParsers = []; |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* @var array Associative array of [pluginName => pluginConfig] |
118
|
|
|
*/ |
119
|
|
|
protected $pluginsConfig; |
120
|
|
|
|
121
|
|
|
/** |
122
|
|
|
* @var array Variables registered for use in filters |
123
|
|
|
*/ |
124
|
|
|
public $registeredVars = []; |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* @var array Root context, used at the root of the document |
128
|
|
|
*/ |
129
|
|
|
protected $rootContext; |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* @var array Tags' config |
133
|
|
|
*/ |
134
|
|
|
protected $tagsConfig; |
135
|
|
|
|
136
|
|
|
/** |
137
|
|
|
* @var array Tag storage |
138
|
|
|
*/ |
139
|
|
|
protected $tagStack; |
140
|
|
|
|
141
|
|
|
/** |
142
|
|
|
* @var bool Whether the tags in the stack are sorted |
143
|
|
|
*/ |
144
|
|
|
protected $tagStackIsSorted; |
145
|
|
|
|
146
|
|
|
/** |
147
|
|
|
* @var string Text being parsed |
148
|
|
|
*/ |
149
|
|
|
protected $text; |
150
|
|
|
|
151
|
|
|
/** |
152
|
|
|
* @var integer Length of the text being parsed |
153
|
|
|
*/ |
154
|
|
|
protected $textLen; |
155
|
|
|
|
156
|
|
|
/** |
157
|
|
|
* @var integer Counter incremented everytime the parser is reset. Used to as a canary to detect |
158
|
|
|
* whether the parser was reset during execution |
159
|
|
|
*/ |
160
|
|
|
protected $uid = 0; |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* @var integer Position before which we output text verbatim, without paragraphs or linebreaks |
164
|
|
|
*/ |
165
|
|
|
protected $wsPos; |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* Constructor |
169
|
|
|
*/ |
170
|
191 |
|
public function __construct(array $config) |
171
|
|
|
{ |
172
|
191 |
|
$this->pluginsConfig = $config['plugins']; |
173
|
191 |
|
$this->registeredVars = $config['registeredVars']; |
174
|
191 |
|
$this->rootContext = $config['rootContext']; |
175
|
191 |
|
$this->tagsConfig = $config['tags']; |
176
|
|
|
|
177
|
191 |
|
$this->__wakeup(); |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
/** |
181
|
|
|
* Serializer |
182
|
|
|
* |
183
|
|
|
* Returns the properties that need to persist through serialization. |
184
|
|
|
* |
185
|
|
|
* NOTE: using __sleep() is preferable to implementing Serializable because it leaves the choice |
186
|
|
|
* of the serializer to the user (e.g. igbinary) |
187
|
|
|
* |
188
|
|
|
* @return array |
189
|
|
|
*/ |
190
|
2 |
|
public function __sleep() |
191
|
|
|
{ |
192
|
2 |
|
return ['pluginsConfig', 'registeredVars', 'rootContext', 'tagsConfig']; |
193
|
|
|
} |
194
|
|
|
|
195
|
|
|
/** |
196
|
|
|
* Unserializer |
197
|
|
|
* |
198
|
|
|
* @return void |
199
|
|
|
*/ |
200
|
191 |
|
public function __wakeup() |
201
|
|
|
{ |
202
|
191 |
|
$this->logger = new Logger; |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
|
|
* Reset the parser for a new parsing |
207
|
|
|
* |
208
|
|
|
* @param string $text Text to be parsed |
209
|
|
|
* @return void |
210
|
|
|
*/ |
211
|
183 |
|
protected function reset($text) |
212
|
|
|
{ |
213
|
|
|
// Reject invalid UTF-8 |
214
|
183 |
|
if (!preg_match('//u', $text)) |
215
|
|
|
{ |
216
|
1 |
|
throw new InvalidArgumentException('Invalid UTF-8 input'); |
217
|
|
|
} |
218
|
|
|
|
219
|
|
|
// Normalize CR/CRLF to LF, remove characters that aren't allowed in XML |
220
|
182 |
|
$text = preg_replace('/\\r\\n?/', "\n", $text); |
221
|
182 |
|
$text = preg_replace('/[\\x00-\\x08\\x0B\\x0C\\x0E-\\x1F]|\\xEF\\xBF[\\xBE\\xBF]/', '', $text); |
222
|
|
|
|
223
|
|
|
// Clear the logs |
224
|
182 |
|
$this->logger->clear(); |
225
|
|
|
|
226
|
|
|
// Initialize the rest |
227
|
182 |
|
$this->cntOpen = []; |
228
|
182 |
|
$this->cntTotal = []; |
229
|
182 |
|
$this->currentFixingCost = 0; |
230
|
182 |
|
$this->currentTag = null; |
231
|
182 |
|
$this->isRich = false; |
232
|
182 |
|
$this->namespaces = []; |
233
|
182 |
|
$this->openTags = []; |
234
|
182 |
|
$this->output = ''; |
235
|
182 |
|
$this->pos = 0; |
236
|
182 |
|
$this->tagStack = []; |
237
|
182 |
|
$this->tagStackIsSorted = false; |
238
|
182 |
|
$this->text = $text; |
239
|
182 |
|
$this->textLen = strlen($text); |
240
|
182 |
|
$this->wsPos = 0; |
241
|
|
|
|
242
|
|
|
// Initialize the root context |
243
|
182 |
|
$this->context = $this->rootContext; |
244
|
182 |
|
$this->context['inParagraph'] = false; |
245
|
|
|
|
246
|
|
|
// Bump the UID |
247
|
182 |
|
++$this->uid; |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* Set a tag's option |
252
|
|
|
* |
253
|
|
|
* This method ensures that the tag's config is a value and not a reference, to prevent |
254
|
|
|
* potential side-effects. References contained *inside* the tag's config are left untouched |
255
|
|
|
* |
256
|
|
|
* @param string $tagName Tag's name |
257
|
|
|
* @param string $optionName Option's name |
258
|
|
|
* @param mixed $optionValue Option's value |
259
|
|
|
* @return void |
260
|
|
|
*/ |
261
|
7 |
|
protected function setTagOption($tagName, $optionName, $optionValue) |
262
|
|
|
{ |
263
|
7 |
|
if (isset($this->tagsConfig[$tagName])) |
264
|
|
|
{ |
265
|
|
|
// Copy the tag's config and remove it. That will destroy the reference |
266
|
7 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
267
|
7 |
|
unset($this->tagsConfig[$tagName]); |
268
|
|
|
|
269
|
|
|
// Set the new value and replace the tag's config |
270
|
7 |
|
$tagConfig[$optionName] = $optionValue; |
271
|
7 |
|
$this->tagsConfig[$tagName] = $tagConfig; |
272
|
|
|
} |
273
|
|
|
} |
274
|
|
|
|
275
|
|
|
//========================================================================== |
276
|
|
|
// Public API |
277
|
|
|
//========================================================================== |
278
|
|
|
|
279
|
|
|
/** |
280
|
|
|
* Disable a tag |
281
|
|
|
* |
282
|
|
|
* @param string $tagName Name of the tag |
283
|
|
|
* @return void |
284
|
|
|
*/ |
285
|
3 |
|
public function disableTag($tagName) |
286
|
|
|
{ |
287
|
3 |
|
$this->setTagOption($tagName, 'isDisabled', true); |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
/** |
291
|
|
|
* Enable a tag |
292
|
|
|
* |
293
|
|
|
* @param string $tagName Name of the tag |
294
|
|
|
* @return void |
295
|
|
|
*/ |
296
|
1 |
|
public function enableTag($tagName) |
297
|
|
|
{ |
298
|
1 |
|
if (isset($this->tagsConfig[$tagName])) |
299
|
|
|
{ |
300
|
1 |
|
unset($this->tagsConfig[$tagName]['isDisabled']); |
301
|
|
|
} |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* Get this parser's Logger instance |
306
|
|
|
* |
307
|
|
|
* @return Logger |
308
|
|
|
*/ |
309
|
144 |
|
public function getLogger() |
310
|
|
|
{ |
311
|
144 |
|
return $this->logger; |
312
|
|
|
} |
313
|
|
|
|
314
|
|
|
/** |
315
|
|
|
* Return the last text parsed |
316
|
|
|
* |
317
|
|
|
* This method returns the normalized text, which may be slightly different from the original |
318
|
|
|
* text in that EOLs are normalized to LF and other control codes are stripped. This method is |
319
|
|
|
* meant to be used in support of processing log entries, which contain offsets based on the |
320
|
|
|
* normalized text |
321
|
|
|
* |
322
|
|
|
* @see Parser::reset() |
323
|
|
|
* |
324
|
|
|
* @return string |
325
|
|
|
*/ |
326
|
143 |
|
public function getText() |
327
|
|
|
{ |
328
|
143 |
|
return $this->text; |
329
|
|
|
} |
330
|
|
|
|
331
|
|
|
/** |
332
|
|
|
* Parse a text |
333
|
|
|
* |
334
|
|
|
* @param string $text Text to parse |
335
|
|
|
* @return string XML representation |
336
|
|
|
*/ |
337
|
183 |
|
public function parse($text) |
338
|
|
|
{ |
339
|
|
|
// Reset the parser and save the uid |
340
|
183 |
|
$this->reset($text); |
341
|
182 |
|
$uid = $this->uid; |
342
|
|
|
|
343
|
|
|
// Do the heavy lifting |
344
|
182 |
|
$this->executePluginParsers(); |
345
|
182 |
|
$this->processTags(); |
346
|
|
|
|
347
|
|
|
// Finalize the document |
348
|
182 |
|
$this->finalizeOutput(); |
349
|
|
|
|
350
|
|
|
// Check the uid in case a plugin or a filter reset the parser mid-execution |
351
|
182 |
|
if ($this->uid !== $uid) |
352
|
|
|
{ |
353
|
1 |
|
throw new RuntimeException('The parser has been reset during execution'); |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
// Log a warning if the fixing cost limit was exceeded |
357
|
182 |
|
if ($this->currentFixingCost > $this->maxFixingCost) |
358
|
|
|
{ |
359
|
2 |
|
$this->logger->warn('Fixing cost limit exceeded'); |
360
|
|
|
} |
361
|
|
|
|
362
|
182 |
|
return $this->output; |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
/** |
366
|
|
|
* Change a tag's tagLimit |
367
|
|
|
* |
368
|
|
|
* NOTE: the default tagLimit should generally be set during configuration instead |
369
|
|
|
* |
370
|
|
|
* @param string $tagName The tag's name, in UPPERCASE |
371
|
|
|
* @param integer $tagLimit |
372
|
|
|
* @return void |
373
|
|
|
*/ |
374
|
2 |
|
public function setTagLimit($tagName, $tagLimit) |
375
|
|
|
{ |
376
|
2 |
|
$this->setTagOption($tagName, 'tagLimit', $tagLimit); |
377
|
|
|
} |
378
|
|
|
|
379
|
|
|
/** |
380
|
|
|
* Change a tag's nestingLimit |
381
|
|
|
* |
382
|
|
|
* NOTE: the default nestingLimit should generally be set during configuration instead |
383
|
|
|
* |
384
|
|
|
* @param string $tagName The tag's name, in UPPERCASE |
385
|
|
|
* @param integer $nestingLimit |
386
|
|
|
* @return void |
387
|
|
|
*/ |
388
|
2 |
|
public function setNestingLimit($tagName, $nestingLimit) |
389
|
|
|
{ |
390
|
2 |
|
$this->setTagOption($tagName, 'nestingLimit', $nestingLimit); |
391
|
|
|
} |
392
|
|
|
|
393
|
|
|
//========================================================================== |
394
|
|
|
// Output handling |
395
|
|
|
//========================================================================== |
396
|
|
|
|
397
|
|
|
/** |
398
|
|
|
* Finalize the output by appending the rest of the unprocessed text and create the root node |
399
|
|
|
* |
400
|
|
|
* @return void |
401
|
|
|
*/ |
402
|
182 |
|
protected function finalizeOutput() |
403
|
|
|
{ |
404
|
|
|
// Output the rest of the text and close the last paragraph |
405
|
182 |
|
$this->outputText($this->textLen, 0, true); |
406
|
|
|
|
407
|
|
|
// Remove empty tag pairs, e.g. <I><U></U></I> as well as empty paragraphs |
408
|
|
|
do |
409
|
|
|
{ |
410
|
182 |
|
$this->output = preg_replace('(<([^ />]++)[^>]*></\\1>)', '', $this->output, -1, $cnt); |
411
|
|
|
} |
412
|
182 |
|
while ($cnt > 0); |
413
|
|
|
|
414
|
|
|
// Merge consecutive <i> tags |
415
|
182 |
|
if (strpos($this->output, '</i><i>') !== false) |
416
|
|
|
{ |
417
|
1 |
|
$this->output = str_replace('</i><i>', '', $this->output); |
418
|
|
|
} |
419
|
|
|
|
420
|
|
|
// Remove illegal characters from the output to ensure it's valid XML |
421
|
182 |
|
$this->output = preg_replace('([\\x00-\\x08\\x0B-\\x1F]|\\xEF\\xBF[\\xBE\\xBF])', '', $this->output); |
422
|
|
|
|
423
|
|
|
// Encode Unicode characters that are outside of the BMP |
424
|
182 |
|
$this->output = Utils::encodeUnicodeSupplementaryCharacters($this->output); |
425
|
|
|
|
426
|
|
|
// Use a <r> root if the text is rich, or <t> for plain text (including <p></p> and <br/>) |
427
|
182 |
|
$tagName = ($this->isRich) ? 'r' : 't'; |
428
|
|
|
|
429
|
|
|
// Prepare the root node with all the namespace declarations |
430
|
182 |
|
$tmp = '<' . $tagName; |
431
|
182 |
|
foreach (array_keys($this->namespaces) as $prefix) |
432
|
|
|
{ |
433
|
2 |
|
$tmp .= ' xmlns:' . $prefix . '="urn:s9e:TextFormatter:' . $prefix . '"'; |
434
|
|
|
} |
435
|
|
|
|
436
|
182 |
|
$this->output = $tmp . '>' . $this->output . '</' . $tagName . '>'; |
437
|
|
|
} |
438
|
|
|
|
439
|
|
|
/** |
440
|
|
|
* Append a tag to the output |
441
|
|
|
* |
442
|
|
|
* @param Tag $tag Tag to append |
443
|
|
|
* @return void |
444
|
|
|
*/ |
445
|
139 |
|
protected function outputTag(Tag $tag) |
446
|
|
|
{ |
447
|
139 |
|
$this->isRich = true; |
448
|
|
|
|
449
|
139 |
|
$tagName = $tag->getName(); |
450
|
139 |
|
$tagPos = $tag->getPos(); |
451
|
139 |
|
$tagLen = $tag->getLen(); |
452
|
139 |
|
$tagFlags = $tag->getFlags(); |
453
|
|
|
|
454
|
139 |
|
if ($tagFlags & self::RULE_IGNORE_WHITESPACE) |
455
|
|
|
{ |
456
|
11 |
|
$skipBefore = 1; |
457
|
11 |
|
$skipAfter = ($tag->isEndTag()) ? 2 : 1; |
458
|
|
|
} |
459
|
|
|
else |
460
|
|
|
{ |
461
|
134 |
|
$skipBefore = $skipAfter = 0; |
462
|
|
|
} |
463
|
|
|
|
464
|
|
|
// Current paragraph must end before the tag if: |
465
|
|
|
// - the tag is a start (or self-closing) tag and it breaks paragraphs, or |
466
|
|
|
// - the tag is an end tag (but not self-closing) |
467
|
139 |
|
$closeParagraph = (!$tag->isStartTag() || ($tagFlags & self::RULE_BREAK_PARAGRAPH)); |
468
|
|
|
|
469
|
|
|
// Let the cursor catch up with this tag's position |
470
|
139 |
|
$this->outputText($tagPos, $skipBefore, $closeParagraph); |
471
|
|
|
|
472
|
|
|
// Capture the text consumed by the tag |
473
|
139 |
|
$tagText = ($tagLen) |
474
|
99 |
|
? htmlspecialchars(substr($this->text, $tagPos, $tagLen), ENT_NOQUOTES, 'UTF-8') |
475
|
139 |
|
: ''; |
476
|
|
|
|
477
|
|
|
// Output current tag |
478
|
139 |
|
if ($tag->isStartTag()) |
479
|
|
|
{ |
480
|
|
|
// Handle paragraphs before opening the tag |
481
|
139 |
|
if (!($tagFlags & self::RULE_BREAK_PARAGRAPH)) |
482
|
|
|
{ |
483
|
138 |
|
$this->outputParagraphStart($tagPos); |
484
|
|
|
} |
485
|
|
|
|
486
|
|
|
// Record this tag's namespace, if applicable |
487
|
139 |
|
$colonPos = strpos($tagName, ':'); |
488
|
139 |
|
if ($colonPos) |
489
|
|
|
{ |
490
|
2 |
|
$this->namespaces[substr($tagName, 0, $colonPos)] = 0; |
491
|
|
|
} |
492
|
|
|
|
493
|
|
|
// Open the start tag and add its attributes, but don't close the tag |
494
|
139 |
|
$this->output .= '<' . $tagName; |
495
|
|
|
|
496
|
|
|
// We output the attributes in lexical order. Helps canonicalizing the output and could |
497
|
|
|
// prove useful someday |
498
|
139 |
|
$attributes = $tag->getAttributes(); |
499
|
139 |
|
ksort($attributes); |
500
|
|
|
|
501
|
139 |
|
foreach ($attributes as $attrName => $attrValue) |
502
|
|
|
{ |
503
|
12 |
|
$this->output .= ' ' . $attrName . '="' . str_replace("\n", ' ', htmlspecialchars($attrValue, ENT_COMPAT, 'UTF-8')) . '"'; |
504
|
|
|
} |
505
|
|
|
|
506
|
139 |
|
if ($tag->isSelfClosingTag()) |
507
|
|
|
{ |
508
|
47 |
|
if ($tagLen) |
509
|
|
|
{ |
510
|
36 |
|
$this->output .= '>' . $tagText . '</' . $tagName . '>'; |
511
|
|
|
} |
512
|
|
|
else |
513
|
|
|
{ |
514
|
47 |
|
$this->output .= '/>'; |
515
|
|
|
} |
516
|
|
|
} |
517
|
105 |
|
elseif ($tagLen) |
518
|
|
|
{ |
519
|
71 |
|
$this->output .= '><s>' . $tagText . '</s>'; |
520
|
|
|
} |
521
|
|
|
else |
522
|
|
|
{ |
523
|
139 |
|
$this->output .= '>'; |
524
|
|
|
} |
525
|
|
|
} |
526
|
|
|
else |
527
|
|
|
{ |
528
|
105 |
|
if ($tagLen) |
529
|
|
|
{ |
530
|
59 |
|
$this->output .= '<e>' . $tagText . '</e>'; |
531
|
|
|
} |
532
|
|
|
|
533
|
105 |
|
$this->output .= '</' . $tagName . '>'; |
534
|
|
|
} |
535
|
|
|
|
536
|
|
|
// Move the cursor past the tag |
537
|
139 |
|
$this->pos = $tagPos + $tagLen; |
538
|
|
|
|
539
|
|
|
// Skip newlines (no other whitespace) after this tag |
540
|
139 |
|
$this->wsPos = $this->pos; |
541
|
139 |
|
while ($skipAfter && $this->wsPos < $this->textLen && $this->text[$this->wsPos] === "\n") |
542
|
|
|
{ |
543
|
|
|
// Decrement the number of lines to skip |
544
|
9 |
|
--$skipAfter; |
545
|
|
|
|
546
|
|
|
// Move the cursor past the newline |
547
|
9 |
|
++$this->wsPos; |
548
|
|
|
} |
549
|
|
|
} |
550
|
|
|
|
551
|
|
|
/** |
552
|
|
|
* Output the text between the cursor's position (included) and given position (not included) |
553
|
|
|
* |
554
|
|
|
* @param integer $catchupPos Position we're catching up to |
555
|
|
|
* @param integer $maxLines Maximum number of lines to ignore at the end of the text |
556
|
|
|
* @param bool $closeParagraph Whether to close the paragraph at the end, if applicable |
557
|
|
|
* @return void |
558
|
|
|
*/ |
559
|
182 |
|
protected function outputText($catchupPos, $maxLines, $closeParagraph) |
560
|
|
|
{ |
561
|
182 |
|
if ($closeParagraph) |
562
|
|
|
{ |
563
|
182 |
|
if (!($this->context['flags'] & self::RULE_CREATE_PARAGRAPHS)) |
564
|
|
|
{ |
565
|
171 |
|
$closeParagraph = false; |
566
|
|
|
} |
567
|
|
|
else |
568
|
|
|
{ |
569
|
|
|
// Ignore any number of lines at the end if we're closing a paragraph |
570
|
18 |
|
$maxLines = -1; |
571
|
|
|
} |
572
|
|
|
} |
573
|
|
|
|
574
|
182 |
|
if ($this->pos >= $catchupPos) |
575
|
|
|
{ |
576
|
|
|
// We're already there, close the paragraph if applicable and return |
577
|
134 |
|
if ($closeParagraph) |
578
|
|
|
{ |
579
|
4 |
|
$this->outputParagraphEnd(); |
580
|
|
|
} |
581
|
|
|
|
582
|
134 |
|
return; |
583
|
|
|
} |
584
|
|
|
|
585
|
|
|
// Skip over previously identified whitespace if applicable |
586
|
160 |
|
if ($this->wsPos > $this->pos) |
587
|
|
|
{ |
588
|
9 |
|
$skipPos = min($catchupPos, $this->wsPos); |
589
|
9 |
|
$this->output .= substr($this->text, $this->pos, $skipPos - $this->pos); |
590
|
9 |
|
$this->pos = $skipPos; |
591
|
|
|
|
592
|
9 |
|
if ($this->pos >= $catchupPos) |
593
|
|
|
{ |
594
|
|
|
// Skipped everything. Close the paragraph if applicable and return |
595
|
2 |
|
if ($closeParagraph) |
596
|
|
|
{ |
597
|
1 |
|
$this->outputParagraphEnd(); |
598
|
|
|
} |
599
|
|
|
|
600
|
2 |
|
return; |
601
|
|
|
} |
602
|
|
|
} |
603
|
|
|
|
604
|
|
|
// Test whether we're even supposed to output anything |
605
|
160 |
|
if ($this->context['flags'] & self::RULE_IGNORE_TEXT) |
606
|
|
|
{ |
607
|
5 |
|
$catchupLen = $catchupPos - $this->pos; |
608
|
5 |
|
$catchupText = substr($this->text, $this->pos, $catchupLen); |
609
|
|
|
|
610
|
|
|
// If the catchup text is not entirely composed of whitespace, we put it inside ignore |
611
|
|
|
// tags |
612
|
5 |
|
if (strspn($catchupText, " \n\t") < $catchupLen) |
613
|
|
|
{ |
614
|
5 |
|
$catchupText = '<i>' . htmlspecialchars($catchupText, ENT_NOQUOTES, 'UTF-8') . '</i>'; |
615
|
|
|
} |
616
|
|
|
|
617
|
5 |
|
$this->output .= $catchupText; |
618
|
5 |
|
$this->pos = $catchupPos; |
619
|
|
|
|
620
|
5 |
|
if ($closeParagraph) |
621
|
|
|
{ |
622
|
1 |
|
$this->outputParagraphEnd(); |
623
|
|
|
} |
624
|
|
|
|
625
|
5 |
|
return; |
626
|
|
|
} |
627
|
|
|
|
628
|
|
|
// Compute the amount of text to ignore at the end of the output |
629
|
158 |
|
$ignorePos = $catchupPos; |
630
|
158 |
|
$ignoreLen = 0; |
631
|
|
|
|
632
|
|
|
// Ignore as many lines (including whitespace) as specified |
633
|
158 |
|
while ($maxLines && --$ignorePos >= $this->pos) |
634
|
|
|
{ |
635
|
21 |
|
$c = $this->text[$ignorePos]; |
636
|
21 |
|
if (strpos(self::WHITESPACE, $c) === false) |
637
|
|
|
{ |
638
|
14 |
|
break; |
639
|
|
|
} |
640
|
|
|
|
641
|
12 |
|
if ($c === "\n") |
642
|
|
|
{ |
643
|
10 |
|
--$maxLines; |
644
|
|
|
} |
645
|
|
|
|
646
|
12 |
|
++$ignoreLen; |
647
|
|
|
} |
648
|
|
|
|
649
|
|
|
// Adjust $catchupPos to ignore the text at the end |
650
|
158 |
|
$catchupPos -= $ignoreLen; |
651
|
|
|
|
652
|
|
|
// Break down the text in paragraphs if applicable |
653
|
158 |
|
if ($this->context['flags'] & self::RULE_CREATE_PARAGRAPHS) |
654
|
|
|
{ |
655
|
15 |
|
if (!$this->context['inParagraph']) |
656
|
|
|
{ |
657
|
13 |
|
$this->outputWhitespace($catchupPos); |
658
|
|
|
|
659
|
13 |
|
if ($catchupPos > $this->pos) |
660
|
|
|
{ |
661
|
10 |
|
$this->outputParagraphStart($catchupPos); |
662
|
|
|
} |
663
|
|
|
} |
664
|
|
|
|
665
|
|
|
// Look for a paragraph break in this text |
666
|
15 |
|
$pbPos = strpos($this->text, "\n\n", $this->pos); |
667
|
|
|
|
668
|
15 |
|
while ($pbPos !== false && $pbPos < $catchupPos) |
669
|
|
|
{ |
670
|
3 |
|
$this->outputText($pbPos, 0, true); |
671
|
3 |
|
$this->outputParagraphStart($catchupPos); |
672
|
|
|
|
673
|
3 |
|
$pbPos = strpos($this->text, "\n\n", $this->pos); |
674
|
|
|
} |
675
|
|
|
} |
676
|
|
|
|
677
|
|
|
// Capture, escape and output the text |
678
|
158 |
|
if ($catchupPos > $this->pos) |
679
|
|
|
{ |
680
|
156 |
|
$catchupText = htmlspecialchars( |
681
|
156 |
|
substr($this->text, $this->pos, $catchupPos - $this->pos), |
682
|
156 |
|
ENT_NOQUOTES, |
683
|
156 |
|
'UTF-8' |
684
|
|
|
); |
685
|
|
|
|
686
|
|
|
// Format line breaks if applicable |
687
|
156 |
|
if (($this->context['flags'] & self::RULES_AUTO_LINEBREAKS) === self::RULE_ENABLE_AUTO_BR) |
688
|
|
|
{ |
689
|
21 |
|
$catchupText = str_replace("\n", "<br/>\n", $catchupText); |
690
|
|
|
} |
691
|
|
|
|
692
|
156 |
|
$this->output .= $catchupText; |
693
|
|
|
} |
694
|
|
|
|
695
|
|
|
// Close the paragraph if applicable |
696
|
158 |
|
if ($closeParagraph) |
697
|
|
|
{ |
698
|
14 |
|
$this->outputParagraphEnd(); |
699
|
|
|
} |
700
|
|
|
|
701
|
|
|
// Add the ignored text if applicable |
702
|
158 |
|
if ($ignoreLen) |
703
|
|
|
{ |
704
|
12 |
|
$this->output .= substr($this->text, $catchupPos, $ignoreLen); |
705
|
|
|
} |
706
|
|
|
|
707
|
|
|
// Move the cursor past the text |
708
|
158 |
|
$this->pos = $catchupPos + $ignoreLen; |
709
|
|
|
} |
710
|
|
|
|
711
|
|
|
/** |
712
|
|
|
* Output a linebreak tag |
713
|
|
|
* |
714
|
|
|
* @param Tag $tag |
715
|
|
|
* @return void |
716
|
|
|
*/ |
717
|
6 |
|
protected function outputBrTag(Tag $tag) |
718
|
|
|
{ |
719
|
6 |
|
$this->outputText($tag->getPos(), 0, false); |
720
|
6 |
|
$this->output .= '<br/>'; |
721
|
|
|
} |
722
|
|
|
|
723
|
|
|
/** |
724
|
|
|
* Output an ignore tag |
725
|
|
|
* |
726
|
|
|
* @param Tag $tag |
727
|
|
|
* @return void |
728
|
|
|
*/ |
729
|
19 |
|
protected function outputIgnoreTag(Tag $tag) |
730
|
|
|
{ |
731
|
19 |
|
$tagPos = $tag->getPos(); |
732
|
19 |
|
$tagLen = $tag->getLen(); |
733
|
|
|
|
734
|
|
|
// Capture the text to ignore |
735
|
19 |
|
$ignoreText = substr($this->text, $tagPos, $tagLen); |
736
|
|
|
|
737
|
|
|
// Catch up with the tag's position then output the tag |
738
|
19 |
|
$this->outputText($tagPos, 0, false); |
739
|
19 |
|
$this->output .= '<i>' . htmlspecialchars($ignoreText, ENT_NOQUOTES, 'UTF-8') . '</i>'; |
740
|
19 |
|
$this->isRich = true; |
741
|
|
|
|
742
|
|
|
// Move the cursor past this tag |
743
|
19 |
|
$this->pos = $tagPos + $tagLen; |
744
|
|
|
} |
745
|
|
|
|
746
|
|
|
/** |
747
|
|
|
* Start a paragraph between current position and given position, if applicable |
748
|
|
|
* |
749
|
|
|
* @param integer $maxPos Rightmost position at which the paragraph can be opened |
750
|
|
|
* @return void |
751
|
|
|
*/ |
752
|
145 |
|
protected function outputParagraphStart($maxPos) |
753
|
|
|
{ |
754
|
|
|
// Do nothing if we're already in a paragraph, or if we don't use paragraphs |
755
|
145 |
|
if ($this->context['inParagraph'] |
756
|
145 |
|
|| !($this->context['flags'] & self::RULE_CREATE_PARAGRAPHS)) |
757
|
|
|
{ |
758
|
135 |
|
return; |
759
|
|
|
} |
760
|
|
|
|
761
|
|
|
// Output the whitespace between $this->pos and $maxPos if applicable |
762
|
16 |
|
$this->outputWhitespace($maxPos); |
763
|
|
|
|
764
|
|
|
// Open the paragraph, but only if it's not at the very end of the text |
765
|
16 |
|
if ($this->pos < $this->textLen) |
766
|
|
|
{ |
767
|
16 |
|
$this->output .= '<p>'; |
768
|
16 |
|
$this->context['inParagraph'] = true; |
769
|
|
|
} |
770
|
|
|
} |
771
|
|
|
|
772
|
|
|
/** |
773
|
|
|
* Close current paragraph at current position if applicable |
774
|
|
|
* |
775
|
|
|
* @return void |
776
|
|
|
*/ |
777
|
18 |
|
protected function outputParagraphEnd() |
778
|
|
|
{ |
779
|
|
|
// Do nothing if we're not in a paragraph |
780
|
18 |
|
if (!$this->context['inParagraph']) |
781
|
|
|
{ |
782
|
3 |
|
return; |
783
|
|
|
} |
784
|
|
|
|
785
|
16 |
|
$this->output .= '</p>'; |
786
|
16 |
|
$this->context['inParagraph'] = false; |
787
|
|
|
} |
788
|
|
|
|
789
|
|
|
/** |
790
|
|
|
* Output the content of a verbatim tag |
791
|
|
|
* |
792
|
|
|
* @param Tag $tag |
793
|
|
|
* @return void |
794
|
|
|
*/ |
795
|
4 |
|
protected function outputVerbatim(Tag $tag) |
796
|
|
|
{ |
797
|
4 |
|
$flags = $this->context['flags']; |
798
|
4 |
|
$this->context['flags'] = $tag->getFlags(); |
799
|
4 |
|
$this->outputText($this->currentTag->getPos() + $this->currentTag->getLen(), 0, false); |
800
|
4 |
|
$this->context['flags'] = $flags; |
801
|
|
|
} |
802
|
|
|
|
803
|
|
|
/** |
804
|
|
|
* Skip as much whitespace after current position as possible |
805
|
|
|
* |
806
|
|
|
* @param integer $maxPos Rightmost character to be skipped |
807
|
|
|
* @return void |
808
|
|
|
*/ |
809
|
18 |
|
protected function outputWhitespace($maxPos) |
810
|
|
|
{ |
811
|
18 |
|
if ($maxPos > $this->pos) |
812
|
|
|
{ |
813
|
13 |
|
$spn = strspn($this->text, self::WHITESPACE, $this->pos, $maxPos - $this->pos); |
814
|
|
|
|
815
|
13 |
|
if ($spn) |
816
|
|
|
{ |
817
|
6 |
|
$this->output .= substr($this->text, $this->pos, $spn); |
818
|
6 |
|
$this->pos += $spn; |
819
|
|
|
} |
820
|
|
|
} |
821
|
|
|
} |
822
|
|
|
|
823
|
|
|
//========================================================================== |
824
|
|
|
// Plugins handling |
825
|
|
|
//========================================================================== |
826
|
|
|
|
827
|
|
|
/** |
828
|
|
|
* Disable a plugin |
829
|
|
|
* |
830
|
|
|
* @param string $pluginName Name of the plugin |
831
|
|
|
* @return void |
832
|
|
|
*/ |
833
|
5 |
|
public function disablePlugin($pluginName) |
834
|
|
|
{ |
835
|
5 |
|
if (isset($this->pluginsConfig[$pluginName])) |
836
|
|
|
{ |
837
|
|
|
// Copy the plugin's config to remove the reference |
838
|
4 |
|
$pluginConfig = $this->pluginsConfig[$pluginName]; |
839
|
4 |
|
unset($this->pluginsConfig[$pluginName]); |
840
|
|
|
|
841
|
|
|
// Update the value and replace the plugin's config |
842
|
4 |
|
$pluginConfig['isDisabled'] = true; |
843
|
4 |
|
$this->pluginsConfig[$pluginName] = $pluginConfig; |
844
|
|
|
} |
845
|
|
|
} |
846
|
|
|
|
847
|
|
|
/** |
848
|
|
|
* Enable a plugin |
849
|
|
|
* |
850
|
|
|
* @param string $pluginName Name of the plugin |
851
|
|
|
* @return void |
852
|
|
|
*/ |
853
|
2 |
|
public function enablePlugin($pluginName) |
854
|
|
|
{ |
855
|
2 |
|
if (isset($this->pluginsConfig[$pluginName])) |
856
|
|
|
{ |
857
|
1 |
|
$this->pluginsConfig[$pluginName]['isDisabled'] = false; |
858
|
|
|
} |
859
|
|
|
} |
860
|
|
|
|
861
|
|
|
/** |
862
|
|
|
* Execute given plugin |
863
|
|
|
* |
864
|
|
|
* @param string $pluginName Plugin's name |
865
|
|
|
* @return void |
866
|
|
|
*/ |
867
|
183 |
|
protected function executePluginParser($pluginName) |
868
|
|
|
{ |
869
|
183 |
|
$pluginConfig = $this->pluginsConfig[$pluginName]; |
870
|
183 |
|
if (isset($pluginConfig['quickMatch']) && strpos($this->text, $pluginConfig['quickMatch']) === false) |
871
|
|
|
{ |
872
|
1 |
|
return; |
873
|
|
|
} |
874
|
|
|
|
875
|
182 |
|
$matches = []; |
876
|
182 |
|
if (isset($pluginConfig['regexp'], $pluginConfig['regexpLimit'])) |
877
|
|
|
{ |
878
|
8 |
|
$matches = $this->getMatches($pluginConfig['regexp'], $pluginConfig['regexpLimit']); |
879
|
8 |
|
if (empty($matches)) |
880
|
|
|
{ |
881
|
1 |
|
return; |
882
|
|
|
} |
883
|
|
|
} |
884
|
|
|
|
885
|
|
|
// Execute the plugin's parser, which will add tags via $this->addStartTag() and others |
886
|
181 |
|
call_user_func($this->getPluginParser($pluginName), $this->text, $matches); |
887
|
|
|
} |
888
|
|
|
|
889
|
|
|
/** |
890
|
|
|
* Execute all the plugins |
891
|
|
|
* |
892
|
|
|
* @return void |
893
|
|
|
*/ |
894
|
193 |
|
protected function executePluginParsers() |
895
|
|
|
{ |
896
|
193 |
|
foreach ($this->pluginsConfig as $pluginName => $pluginConfig) |
897
|
|
|
{ |
898
|
184 |
|
if (empty($pluginConfig['isDisabled'])) |
899
|
|
|
{ |
900
|
183 |
|
$this->executePluginParser($pluginName); |
901
|
|
|
} |
902
|
|
|
} |
903
|
|
|
} |
904
|
|
|
|
905
|
|
|
/** |
906
|
|
|
* Execute given regexp and returns as many matches as given limit |
907
|
|
|
* |
908
|
|
|
* @param string $regexp |
909
|
|
|
* @param integer $limit |
910
|
|
|
* @return array |
911
|
|
|
*/ |
912
|
8 |
|
protected function getMatches($regexp, $limit) |
913
|
|
|
{ |
914
|
8 |
|
$cnt = preg_match_all($regexp, $this->text, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE); |
915
|
8 |
|
if ($cnt > $limit) |
916
|
|
|
{ |
917
|
2 |
|
$matches = array_slice($matches, 0, $limit); |
918
|
|
|
} |
919
|
|
|
|
920
|
8 |
|
return $matches; |
921
|
|
|
} |
922
|
|
|
|
923
|
|
|
/** |
924
|
|
|
* Get the cached callback for given plugin's parser |
925
|
|
|
* |
926
|
|
|
* @param string $pluginName Plugin's name |
927
|
|
|
* @return callable |
928
|
|
|
*/ |
929
|
181 |
|
protected function getPluginParser($pluginName) |
930
|
|
|
{ |
931
|
|
|
// Cache a new instance of this plugin's parser if there isn't one already |
932
|
181 |
|
if (!isset($this->pluginParsers[$pluginName])) |
933
|
|
|
{ |
934
|
3 |
|
$pluginConfig = $this->pluginsConfig[$pluginName]; |
935
|
3 |
|
$className = (isset($pluginConfig['className'])) |
936
|
1 |
|
? $pluginConfig['className'] |
937
|
3 |
|
: 's9e\\TextFormatter\\Plugins\\' . $pluginName . '\\Parser'; |
938
|
|
|
|
939
|
|
|
// Register the parser as a callback |
940
|
3 |
|
$this->pluginParsers[$pluginName] = [new $className($this, $pluginConfig), 'parse']; |
941
|
|
|
} |
942
|
|
|
|
943
|
181 |
|
return $this->pluginParsers[$pluginName]; |
944
|
|
|
} |
945
|
|
|
|
946
|
|
|
/** |
947
|
|
|
* Register a parser |
948
|
|
|
* |
949
|
|
|
* Can be used to add a new parser with no plugin config, or pre-generate a parser for an |
950
|
|
|
* existing plugin |
951
|
|
|
* |
952
|
|
|
* @param string $pluginName |
953
|
|
|
* @param callable $parser |
954
|
|
|
* @param string $regexp |
955
|
|
|
* @param integer $limit |
956
|
|
|
* @return void |
957
|
|
|
*/ |
958
|
176 |
|
public function registerParser($pluginName, $parser, $regexp = null, $limit = PHP_INT_MAX) |
959
|
|
|
{ |
960
|
176 |
|
if (!is_callable($parser)) |
961
|
|
|
{ |
962
|
1 |
|
throw new InvalidArgumentException('Argument 1 passed to ' . __METHOD__ . ' must be a valid callback'); |
963
|
|
|
} |
964
|
|
|
// Create an empty config for this plugin to ensure it is executed |
965
|
175 |
|
if (!isset($this->pluginsConfig[$pluginName])) |
966
|
|
|
{ |
967
|
174 |
|
$this->pluginsConfig[$pluginName] = []; |
968
|
|
|
} |
969
|
175 |
|
if (isset($regexp)) |
970
|
|
|
{ |
971
|
2 |
|
$this->pluginsConfig[$pluginName]['regexp'] = $regexp; |
972
|
2 |
|
$this->pluginsConfig[$pluginName]['regexpLimit'] = $limit; |
973
|
|
|
} |
974
|
175 |
|
$this->pluginParsers[$pluginName] = $parser; |
975
|
|
|
} |
976
|
|
|
|
977
|
|
|
//========================================================================== |
978
|
|
|
// Rules handling |
979
|
|
|
//========================================================================== |
980
|
|
|
|
981
|
|
|
/** |
982
|
|
|
* Apply closeAncestor rules associated with given tag |
983
|
|
|
* |
984
|
|
|
* @param Tag $tag Tag |
985
|
|
|
* @return bool Whether a new tag has been added |
986
|
|
|
*/ |
987
|
138 |
|
protected function closeAncestor(Tag $tag) |
988
|
|
|
{ |
989
|
138 |
|
if (!empty($this->openTags)) |
990
|
|
|
{ |
991
|
61 |
|
$tagName = $tag->getName(); |
992
|
61 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
993
|
|
|
|
994
|
61 |
|
if (!empty($tagConfig['rules']['closeAncestor'])) |
995
|
|
|
{ |
996
|
5 |
|
$i = count($this->openTags); |
997
|
|
|
|
998
|
5 |
|
while (--$i >= 0) |
999
|
|
|
{ |
1000
|
5 |
|
$ancestor = $this->openTags[$i]; |
1001
|
5 |
|
$ancestorName = $ancestor->getName(); |
1002
|
|
|
|
1003
|
5 |
|
if (isset($tagConfig['rules']['closeAncestor'][$ancestorName])) |
1004
|
|
|
{ |
1005
|
4 |
|
++$this->currentFixingCost; |
1006
|
|
|
|
1007
|
|
|
// We have to close this ancestor. First we reinsert this tag... |
1008
|
4 |
|
$this->tagStack[] = $tag; |
1009
|
|
|
|
1010
|
|
|
// ...then we add a new end tag for it with a better priority |
1011
|
4 |
|
$this->addMagicEndTag($ancestor, $tag->getPos(), $tag->getSortPriority() - 1); |
1012
|
|
|
|
1013
|
4 |
|
return true; |
1014
|
|
|
} |
1015
|
|
|
} |
1016
|
|
|
} |
1017
|
|
|
} |
1018
|
|
|
|
1019
|
138 |
|
return false; |
1020
|
|
|
} |
1021
|
|
|
|
1022
|
|
|
/** |
1023
|
|
|
* Apply closeParent rules associated with given tag |
1024
|
|
|
* |
1025
|
|
|
* @param Tag $tag Tag |
1026
|
|
|
* @return bool Whether a new tag has been added |
1027
|
|
|
*/ |
1028
|
138 |
|
protected function closeParent(Tag $tag) |
1029
|
|
|
{ |
1030
|
138 |
|
if (!empty($this->openTags)) |
1031
|
|
|
{ |
1032
|
65 |
|
$tagName = $tag->getName(); |
1033
|
65 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
1034
|
|
|
|
1035
|
65 |
|
if (!empty($tagConfig['rules']['closeParent'])) |
1036
|
|
|
{ |
1037
|
7 |
|
$parent = end($this->openTags); |
1038
|
7 |
|
$parentName = $parent->getName(); |
1039
|
|
|
|
1040
|
7 |
|
if (isset($tagConfig['rules']['closeParent'][$parentName])) |
1041
|
|
|
{ |
1042
|
6 |
|
++$this->currentFixingCost; |
1043
|
|
|
|
1044
|
|
|
// We have to close that parent. First we reinsert the tag... |
1045
|
6 |
|
$this->tagStack[] = $tag; |
1046
|
|
|
|
1047
|
|
|
// ...then we add a new end tag for it with a better priority |
1048
|
6 |
|
$this->addMagicEndTag($parent, $tag->getPos(), $tag->getSortPriority() - 1); |
1049
|
|
|
|
1050
|
6 |
|
return true; |
1051
|
|
|
} |
1052
|
|
|
} |
1053
|
|
|
} |
1054
|
|
|
|
1055
|
138 |
|
return false; |
1056
|
|
|
} |
1057
|
|
|
|
1058
|
|
|
/** |
1059
|
|
|
* Apply the createChild rules associated with given tag |
1060
|
|
|
* |
1061
|
|
|
* @param Tag $tag Tag |
1062
|
|
|
* @return void |
1063
|
|
|
*/ |
1064
|
139 |
|
protected function createChild(Tag $tag) |
1065
|
|
|
{ |
1066
|
139 |
|
$tagConfig = $this->tagsConfig[$tag->getName()]; |
1067
|
139 |
|
if (isset($tagConfig['rules']['createChild'])) |
1068
|
|
|
{ |
1069
|
3 |
|
$priority = -1000; |
1070
|
3 |
|
$tagPos = $this->pos + strspn($this->text, " \n\r\t", $this->pos); |
1071
|
3 |
|
foreach ($tagConfig['rules']['createChild'] as $tagName) |
1072
|
|
|
{ |
1073
|
3 |
|
$this->addStartTag($tagName, $tagPos, 0, ++$priority); |
1074
|
|
|
} |
1075
|
|
|
} |
1076
|
|
|
} |
1077
|
|
|
|
1078
|
|
|
/** |
1079
|
|
|
* Apply fosterParent rules associated with given tag |
1080
|
|
|
* |
1081
|
|
|
* NOTE: this rule has the potential for creating an unbounded loop, either if a tag tries to |
1082
|
|
|
* foster itself or two or more tags try to foster each other in a loop. We mitigate the |
1083
|
|
|
* risk by preventing a tag from creating a child of itself (the parent still gets closed) |
1084
|
|
|
* and by checking and increasing the currentFixingCost so that a loop of multiple tags |
1085
|
|
|
* do not run indefinitely. The default tagLimit and nestingLimit also serve to prevent the |
1086
|
|
|
* loop from running indefinitely |
1087
|
|
|
* |
1088
|
|
|
* @param Tag $tag Tag |
1089
|
|
|
* @return bool Whether a new tag has been added |
1090
|
|
|
*/ |
1091
|
138 |
|
protected function fosterParent(Tag $tag) |
1092
|
|
|
{ |
1093
|
138 |
|
if (!empty($this->openTags)) |
1094
|
|
|
{ |
1095
|
68 |
|
$tagName = $tag->getName(); |
1096
|
68 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
1097
|
|
|
|
1098
|
68 |
|
if (!empty($tagConfig['rules']['fosterParent'])) |
1099
|
|
|
{ |
1100
|
13 |
|
$parent = end($this->openTags); |
1101
|
13 |
|
$parentName = $parent->getName(); |
1102
|
|
|
|
1103
|
13 |
|
if (isset($tagConfig['rules']['fosterParent'][$parentName])) |
1104
|
|
|
{ |
1105
|
12 |
|
if ($parentName !== $tagName && $this->currentFixingCost < $this->maxFixingCost) |
1106
|
|
|
{ |
1107
|
11 |
|
$this->addFosterTag($tag, $parent); |
1108
|
|
|
} |
1109
|
|
|
|
1110
|
|
|
// Reinsert current tag |
1111
|
12 |
|
$this->tagStack[] = $tag; |
1112
|
|
|
|
1113
|
|
|
// And finally close its parent with a priority that ensures it is processed |
1114
|
|
|
// before this tag |
1115
|
12 |
|
$this->addMagicEndTag($parent, $tag->getPos(), $tag->getSortPriority() - 1); |
1116
|
|
|
|
1117
|
|
|
// Adjust the fixing cost to account for the additional tags/processing |
1118
|
12 |
|
$this->currentFixingCost += 4; |
1119
|
|
|
|
1120
|
12 |
|
return true; |
1121
|
|
|
} |
1122
|
|
|
} |
1123
|
|
|
} |
1124
|
|
|
|
1125
|
138 |
|
return false; |
1126
|
|
|
} |
1127
|
|
|
|
1128
|
|
|
/** |
1129
|
|
|
* Apply requireAncestor rules associated with given tag |
1130
|
|
|
* |
1131
|
|
|
* @param Tag $tag Tag |
1132
|
|
|
* @return bool Whether this tag has an unfulfilled requireAncestor requirement |
1133
|
|
|
*/ |
1134
|
141 |
|
protected function requireAncestor(Tag $tag) |
1135
|
|
|
{ |
1136
|
141 |
|
$tagName = $tag->getName(); |
1137
|
141 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
1138
|
|
|
|
1139
|
141 |
|
if (isset($tagConfig['rules']['requireAncestor'])) |
1140
|
|
|
{ |
1141
|
3 |
|
foreach ($tagConfig['rules']['requireAncestor'] as $ancestorName) |
1142
|
|
|
{ |
1143
|
3 |
|
if (!empty($this->cntOpen[$ancestorName])) |
1144
|
|
|
{ |
1145
|
1 |
|
return false; |
1146
|
|
|
} |
1147
|
|
|
} |
1148
|
|
|
|
1149
|
2 |
|
$this->logger->err('Tag requires an ancestor', [ |
1150
|
2 |
|
'requireAncestor' => implode(',', $tagConfig['rules']['requireAncestor']), |
1151
|
2 |
|
'tag' => $tag |
1152
|
|
|
]); |
1153
|
|
|
|
1154
|
2 |
|
return true; |
1155
|
|
|
} |
1156
|
|
|
|
1157
|
139 |
|
return false; |
1158
|
|
|
} |
1159
|
|
|
|
1160
|
|
|
//========================================================================== |
1161
|
|
|
// Tag processing |
1162
|
|
|
//========================================================================== |
1163
|
|
|
|
1164
|
|
|
/** |
1165
|
|
|
* Create and add a copy of a tag as a child of a given tag |
1166
|
|
|
* |
1167
|
|
|
* @param Tag $tag Current tag |
1168
|
|
|
* @param Tag $fosterTag Tag to foster |
1169
|
|
|
* @return void |
1170
|
|
|
*/ |
1171
|
11 |
|
protected function addFosterTag(Tag $tag, Tag $fosterTag) |
1172
|
|
|
{ |
1173
|
11 |
|
list($childPos, $childPrio) = $this->getMagicStartCoords($tag->getPos() + $tag->getLen()); |
1174
|
|
|
|
1175
|
|
|
// Add a 0-width copy of the parent tag after this tag and make it depend on this tag |
1176
|
11 |
|
$childTag = $this->addCopyTag($fosterTag, $childPos, 0, $childPrio); |
1177
|
11 |
|
$tag->cascadeInvalidationTo($childTag); |
1178
|
|
|
} |
1179
|
|
|
|
1180
|
|
|
/** |
1181
|
|
|
* Create and add an end tag for given start tag at given position |
1182
|
|
|
* |
1183
|
|
|
* @param Tag $startTag Start tag |
1184
|
|
|
* @param integer $tagPos End tag's position (will be adjusted for whitespace if applicable) |
1185
|
|
|
* @param integer $prio End tag's priority |
1186
|
|
|
* @return Tag |
1187
|
|
|
*/ |
1188
|
35 |
|
protected function addMagicEndTag(Tag $startTag, $tagPos, $prio = 0) |
1189
|
|
|
{ |
1190
|
35 |
|
$tagName = $startTag->getName(); |
1191
|
|
|
|
1192
|
|
|
// Adjust the end tag's position if whitespace is to be minimized |
1193
|
35 |
|
if (($this->currentTag->getFlags() | $startTag->getFlags()) & self::RULE_IGNORE_WHITESPACE) |
1194
|
|
|
{ |
1195
|
3 |
|
$tagPos = $this->getMagicEndPos($tagPos); |
1196
|
|
|
} |
1197
|
|
|
|
1198
|
|
|
// Add a 0-width end tag that is paired with the given start tag |
1199
|
35 |
|
$endTag = $this->addEndTag($tagName, $tagPos, 0, $prio); |
1200
|
35 |
|
$endTag->pairWith($startTag); |
1201
|
|
|
|
1202
|
35 |
|
return $endTag; |
1203
|
|
|
} |
1204
|
|
|
|
1205
|
|
|
/** |
1206
|
|
|
* Compute the position of a magic end tag, adjusted for whitespace |
1207
|
|
|
* |
1208
|
|
|
* @param integer $tagPos Rightmost possible position for the tag |
1209
|
|
|
* @return integer |
1210
|
|
|
*/ |
1211
|
5 |
|
protected function getMagicEndPos($tagPos) |
1212
|
|
|
{ |
1213
|
|
|
// Back up from given position to the cursor's position until we find a character that |
1214
|
|
|
// is not whitespace |
1215
|
5 |
|
while ($tagPos > $this->pos && strpos(self::WHITESPACE, $this->text[$tagPos - 1]) !== false) |
1216
|
|
|
{ |
1217
|
5 |
|
--$tagPos; |
1218
|
|
|
} |
1219
|
|
|
|
1220
|
5 |
|
return $tagPos; |
1221
|
|
|
} |
1222
|
|
|
|
1223
|
|
|
/** |
1224
|
|
|
* Compute the position and priority of a magic start tag, adjusted for whitespace |
1225
|
|
|
* |
1226
|
|
|
* @param integer $tagPos Leftmost possible position for the tag |
1227
|
|
|
* @return integer[] [Tag pos, priority] |
1228
|
|
|
*/ |
1229
|
11 |
|
protected function getMagicStartCoords($tagPos) |
1230
|
|
|
{ |
1231
|
11 |
|
if (empty($this->tagStack)) |
1232
|
|
|
{ |
1233
|
|
|
// Set the next position outside the text boundaries |
1234
|
3 |
|
$nextPos = $this->textLen + 1; |
1235
|
3 |
|
$nextPrio = 0; |
1236
|
|
|
} |
1237
|
|
|
else |
1238
|
|
|
{ |
1239
|
10 |
|
$nextTag = end($this->tagStack); |
1240
|
10 |
|
$nextPos = $nextTag->getPos(); |
1241
|
10 |
|
$nextPrio = $nextTag->getSortPriority(); |
1242
|
|
|
} |
1243
|
|
|
|
1244
|
|
|
// Find the first non-whitespace position before next tag or the end of text |
1245
|
11 |
|
while ($tagPos < $nextPos && strpos(self::WHITESPACE, $this->text[$tagPos]) !== false) |
1246
|
|
|
{ |
1247
|
1 |
|
++$tagPos; |
1248
|
|
|
} |
1249
|
|
|
|
1250
|
|
|
// Set a priority that ensures this tag appears before the next tag |
1251
|
11 |
|
$prio = ($tagPos === $nextPos) ? $nextPrio - 1 : 0; |
1252
|
|
|
|
1253
|
11 |
|
return [$tagPos, $prio]; |
1254
|
|
|
} |
1255
|
|
|
|
1256
|
|
|
/** |
1257
|
|
|
* Test whether given start tag is immediately followed by a closing tag |
1258
|
|
|
* |
1259
|
|
|
* @param Tag $tag Start tag |
1260
|
|
|
* @return bool |
1261
|
|
|
*/ |
1262
|
3 |
|
protected function isFollowedByClosingTag(Tag $tag) |
1263
|
|
|
{ |
1264
|
3 |
|
return (empty($this->tagStack)) ? false : end($this->tagStack)->canClose($tag); |
1265
|
|
|
} |
1266
|
|
|
|
1267
|
|
|
/** |
1268
|
|
|
* Process all tags in the stack |
1269
|
|
|
* |
1270
|
|
|
* @return void |
1271
|
|
|
*/ |
1272
|
182 |
|
protected function processTags() |
1273
|
|
|
{ |
1274
|
182 |
|
if (empty($this->tagStack)) |
1275
|
|
|
{ |
1276
|
24 |
|
return; |
1277
|
|
|
} |
1278
|
|
|
|
1279
|
|
|
// Initialize the count tables |
1280
|
158 |
|
foreach (array_keys($this->tagsConfig) as $tagName) |
1281
|
|
|
{ |
1282
|
144 |
|
$this->cntOpen[$tagName] = 0; |
1283
|
144 |
|
$this->cntTotal[$tagName] = 0; |
1284
|
|
|
} |
1285
|
|
|
|
1286
|
|
|
// Process the tag stack, close tags that were left open and repeat until done |
1287
|
|
|
do |
1288
|
|
|
{ |
1289
|
158 |
|
while (!empty($this->tagStack)) |
1290
|
|
|
{ |
1291
|
158 |
|
if (!$this->tagStackIsSorted) |
1292
|
|
|
{ |
1293
|
158 |
|
$this->sortTags(); |
1294
|
|
|
} |
1295
|
|
|
|
1296
|
158 |
|
$this->currentTag = array_pop($this->tagStack); |
1297
|
158 |
|
$this->processCurrentTag(); |
1298
|
|
|
} |
1299
|
|
|
|
1300
|
|
|
// Close tags that were left open |
1301
|
158 |
|
foreach ($this->openTags as $startTag) |
1302
|
|
|
{ |
1303
|
|
|
// NOTE: we add tags in hierarchical order (ancestors to descendants) but since |
1304
|
|
|
// the stack is processed in LIFO order, it means that tags get closed in |
1305
|
|
|
// the correct order, from descendants to ancestors |
1306
|
19 |
|
$this->addMagicEndTag($startTag, $this->textLen); |
1307
|
|
|
} |
1308
|
|
|
} |
1309
|
158 |
|
while (!empty($this->tagStack)); |
1310
|
|
|
} |
1311
|
|
|
|
1312
|
|
|
/** |
1313
|
|
|
* Process current tag |
1314
|
|
|
* |
1315
|
|
|
* @return void |
1316
|
|
|
*/ |
1317
|
158 |
|
protected function processCurrentTag() |
1318
|
|
|
{ |
1319
|
|
|
// Invalidate current tag if tags are disabled and current tag would not close the last open |
1320
|
|
|
// tag and is not a system tag |
1321
|
158 |
|
if (($this->context['flags'] & self::RULE_IGNORE_TAGS) |
1322
|
158 |
|
&& !$this->currentTag->canClose(end($this->openTags)) |
1323
|
158 |
|
&& !$this->currentTag->isSystemTag()) |
1324
|
|
|
{ |
1325
|
4 |
|
$this->currentTag->invalidate(); |
1326
|
|
|
} |
1327
|
|
|
|
1328
|
158 |
|
$tagPos = $this->currentTag->getPos(); |
1329
|
158 |
|
$tagLen = $this->currentTag->getLen(); |
1330
|
|
|
|
1331
|
|
|
// Test whether the cursor passed this tag's position already |
1332
|
158 |
|
if ($this->pos > $tagPos && !$this->currentTag->isInvalid()) |
1333
|
|
|
{ |
1334
|
|
|
// Test whether this tag is paired with a start tag and this tag is still open |
1335
|
16 |
|
$startTag = $this->currentTag->getStartTag(); |
1336
|
|
|
|
1337
|
16 |
|
if ($startTag && in_array($startTag, $this->openTags, true)) |
1338
|
|
|
{ |
1339
|
|
|
// Create an end tag that matches current tag's start tag, which consumes as much of |
1340
|
|
|
// the same text as current tag and is paired with the same start tag |
1341
|
2 |
|
$this->addEndTag( |
1342
|
2 |
|
$startTag->getName(), |
1343
|
2 |
|
$this->pos, |
1344
|
2 |
|
max(0, $tagPos + $tagLen - $this->pos) |
1345
|
2 |
|
)->pairWith($startTag); |
1346
|
|
|
|
1347
|
|
|
// Note that current tag is not invalidated, it's merely replaced |
1348
|
2 |
|
return; |
1349
|
|
|
} |
1350
|
|
|
|
1351
|
|
|
// If this is an ignore tag, try to ignore as much as the remaining text as possible |
1352
|
14 |
|
if ($this->currentTag->isIgnoreTag()) |
1353
|
|
|
{ |
1354
|
2 |
|
$ignoreLen = $tagPos + $tagLen - $this->pos; |
1355
|
|
|
|
1356
|
2 |
|
if ($ignoreLen > 0) |
1357
|
|
|
{ |
1358
|
|
|
// Create a new ignore tag and move on |
1359
|
1 |
|
$this->addIgnoreTag($this->pos, $ignoreLen); |
1360
|
|
|
|
1361
|
1 |
|
return; |
1362
|
|
|
} |
1363
|
|
|
} |
1364
|
|
|
|
1365
|
|
|
// Skipped tags are invalidated |
1366
|
13 |
|
$this->currentTag->invalidate(); |
1367
|
|
|
} |
1368
|
|
|
|
1369
|
158 |
|
if ($this->currentTag->isInvalid()) |
1370
|
|
|
{ |
1371
|
18 |
|
return; |
1372
|
|
|
} |
1373
|
|
|
|
1374
|
158 |
|
if ($this->currentTag->isIgnoreTag()) |
1375
|
|
|
{ |
1376
|
11 |
|
$this->outputIgnoreTag($this->currentTag); |
1377
|
|
|
} |
1378
|
153 |
|
elseif ($this->currentTag->isBrTag()) |
1379
|
|
|
{ |
1380
|
|
|
// Output the tag if it's allowed, ignore it otherwise |
1381
|
7 |
|
if (!($this->context['flags'] & self::RULE_PREVENT_BR)) |
1382
|
|
|
{ |
1383
|
7 |
|
$this->outputBrTag($this->currentTag); |
1384
|
|
|
} |
1385
|
|
|
} |
1386
|
149 |
|
elseif ($this->currentTag->isParagraphBreak()) |
1387
|
|
|
{ |
1388
|
4 |
|
$this->outputText($this->currentTag->getPos(), 0, true); |
1389
|
|
|
} |
1390
|
146 |
|
elseif ($this->currentTag->isVerbatim()) |
1391
|
|
|
{ |
1392
|
4 |
|
$this->outputVerbatim($this->currentTag); |
1393
|
|
|
} |
1394
|
142 |
|
elseif ($this->currentTag->isStartTag()) |
1395
|
|
|
{ |
1396
|
141 |
|
$this->processStartTag($this->currentTag); |
1397
|
|
|
} |
1398
|
|
|
else |
1399
|
|
|
{ |
1400
|
106 |
|
$this->processEndTag($this->currentTag); |
1401
|
|
|
} |
1402
|
|
|
} |
1403
|
|
|
|
1404
|
|
|
/** |
1405
|
|
|
* Process given start tag (including self-closing tags) at current position |
1406
|
|
|
* |
1407
|
|
|
* @param Tag $tag Start tag (including self-closing) |
1408
|
|
|
* @return void |
1409
|
|
|
*/ |
1410
|
141 |
|
protected function processStartTag(Tag $tag) |
1411
|
|
|
{ |
1412
|
141 |
|
$tagName = $tag->getName(); |
1413
|
141 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
1414
|
|
|
|
1415
|
|
|
// 1. Check that this tag has not reached its global limit tagLimit |
1416
|
|
|
// 2. Execute this tag's filterChain, which will filter/validate its attributes |
1417
|
|
|
// 3. Apply closeParent, closeAncestor and fosterParent rules |
1418
|
|
|
// 4. Check for nestingLimit |
1419
|
|
|
// 5. Apply requireAncestor rules |
1420
|
|
|
// |
1421
|
|
|
// This order ensures that the tag is valid and within the set limits before we attempt to |
1422
|
|
|
// close parents or ancestors. We need to close ancestors before we can check for nesting |
1423
|
|
|
// limits, whether this tag is allowed within current context (the context may change |
1424
|
|
|
// as ancestors are closed) or whether the required ancestors are still there (they might |
1425
|
|
|
// have been closed by a rule.) |
1426
|
141 |
|
if ($this->cntTotal[$tagName] >= $tagConfig['tagLimit']) |
1427
|
|
|
{ |
1428
|
2 |
|
$this->logger->err( |
1429
|
2 |
|
'Tag limit exceeded', |
1430
|
|
|
[ |
1431
|
2 |
|
'tag' => $tag, |
1432
|
2 |
|
'tagName' => $tagName, |
1433
|
2 |
|
'tagLimit' => $tagConfig['tagLimit'] |
1434
|
|
|
] |
1435
|
|
|
); |
1436
|
2 |
|
$tag->invalidate(); |
1437
|
|
|
|
1438
|
2 |
|
return; |
1439
|
|
|
} |
1440
|
|
|
|
1441
|
141 |
|
FilterProcessing::filterTag($tag, $this, $this->tagsConfig, $this->openTags); |
1442
|
141 |
|
if ($tag->isInvalid()) |
1443
|
|
|
{ |
1444
|
1 |
|
return; |
1445
|
|
|
} |
1446
|
|
|
|
1447
|
141 |
|
if ($this->currentFixingCost < $this->maxFixingCost) |
1448
|
|
|
{ |
1449
|
138 |
|
if ($this->fosterParent($tag) || $this->closeParent($tag) || $this->closeAncestor($tag)) |
1450
|
|
|
{ |
1451
|
|
|
// This tag parent/ancestor needs to be closed, we just return (the tag is still valid) |
1452
|
21 |
|
return; |
1453
|
|
|
} |
1454
|
|
|
} |
1455
|
|
|
|
1456
|
141 |
|
if ($this->cntOpen[$tagName] >= $tagConfig['nestingLimit']) |
1457
|
|
|
{ |
1458
|
2 |
|
$this->logger->err( |
1459
|
2 |
|
'Nesting limit exceeded', |
1460
|
|
|
[ |
1461
|
2 |
|
'tag' => $tag, |
1462
|
2 |
|
'tagName' => $tagName, |
1463
|
2 |
|
'nestingLimit' => $tagConfig['nestingLimit'] |
1464
|
|
|
] |
1465
|
|
|
); |
1466
|
2 |
|
$tag->invalidate(); |
1467
|
|
|
|
1468
|
2 |
|
return; |
1469
|
|
|
} |
1470
|
|
|
|
1471
|
141 |
|
if (!$this->tagIsAllowed($tagName)) |
1472
|
|
|
{ |
1473
|
7 |
|
$msg = 'Tag is not allowed in this context'; |
1474
|
7 |
|
$context = ['tag' => $tag, 'tagName' => $tagName]; |
1475
|
7 |
|
if ($tag->getLen() > 0) |
1476
|
|
|
{ |
1477
|
6 |
|
$this->logger->warn($msg, $context); |
1478
|
|
|
} |
1479
|
|
|
else |
1480
|
|
|
{ |
1481
|
1 |
|
$this->logger->debug($msg, $context); |
1482
|
|
|
} |
1483
|
7 |
|
$tag->invalidate(); |
1484
|
|
|
|
1485
|
7 |
|
return; |
1486
|
|
|
} |
1487
|
|
|
|
1488
|
141 |
|
if ($this->requireAncestor($tag)) |
1489
|
|
|
{ |
1490
|
2 |
|
$tag->invalidate(); |
1491
|
|
|
|
1492
|
2 |
|
return; |
1493
|
|
|
} |
1494
|
|
|
|
1495
|
|
|
// If this tag has an autoClose rule and it's not self-closed, paired with an end tag, or |
1496
|
|
|
// immediately followed by an end tag, we replace it with a self-closing tag with the same |
1497
|
|
|
// properties |
1498
|
139 |
|
if ($tag->getFlags() & self::RULE_AUTO_CLOSE |
1499
|
139 |
|
&& !$tag->isSelfClosingTag() |
1500
|
139 |
|
&& !$tag->getEndTag() |
1501
|
139 |
|
&& !$this->isFollowedByClosingTag($tag)) |
1502
|
|
|
{ |
1503
|
2 |
|
$newTag = new Tag(Tag::SELF_CLOSING_TAG, $tagName, $tag->getPos(), $tag->getLen()); |
1504
|
2 |
|
$newTag->setAttributes($tag->getAttributes()); |
1505
|
2 |
|
$newTag->setFlags($tag->getFlags()); |
1506
|
|
|
|
1507
|
2 |
|
$tag = $newTag; |
1508
|
|
|
} |
1509
|
|
|
|
1510
|
139 |
|
if ($tag->getFlags() & self::RULE_TRIM_FIRST_LINE |
1511
|
139 |
|
&& substr($this->text, $tag->getPos() + $tag->getLen(), 1) === "\n") |
1512
|
|
|
{ |
1513
|
2 |
|
$this->addIgnoreTag($tag->getPos() + $tag->getLen(), 1); |
1514
|
|
|
} |
1515
|
|
|
|
1516
|
|
|
// This tag is valid, output it and update the context |
1517
|
139 |
|
$this->outputTag($tag); |
1518
|
139 |
|
$this->pushContext($tag); |
1519
|
|
|
|
1520
|
|
|
// Apply the createChild rules if applicable |
1521
|
139 |
|
$this->createChild($tag); |
1522
|
|
|
} |
1523
|
|
|
|
1524
|
|
|
/** |
1525
|
|
|
* Process given end tag at current position |
1526
|
|
|
* |
1527
|
|
|
* @param Tag $tag end tag |
1528
|
|
|
* @return void |
1529
|
|
|
*/ |
1530
|
106 |
|
protected function processEndTag(Tag $tag) |
1531
|
|
|
{ |
1532
|
106 |
|
$tagName = $tag->getName(); |
1533
|
|
|
|
1534
|
106 |
|
if (empty($this->cntOpen[$tagName])) |
1535
|
|
|
{ |
1536
|
|
|
// This is an end tag with no start tag |
1537
|
9 |
|
return; |
1538
|
|
|
} |
1539
|
|
|
|
1540
|
|
|
/** |
1541
|
|
|
* @var array List of tags need to be closed before given tag |
1542
|
|
|
*/ |
1543
|
105 |
|
$closeTags = []; |
1544
|
|
|
|
1545
|
|
|
// Iterate through all open tags from last to first to find a match for our tag |
1546
|
105 |
|
$i = count($this->openTags); |
1547
|
105 |
|
while (--$i >= 0) |
1548
|
|
|
{ |
1549
|
105 |
|
$openTag = $this->openTags[$i]; |
1550
|
|
|
|
1551
|
105 |
|
if ($tag->canClose($openTag)) |
1552
|
|
|
{ |
1553
|
105 |
|
break; |
1554
|
|
|
} |
1555
|
|
|
|
1556
|
26 |
|
$closeTags[] = $openTag; |
1557
|
26 |
|
++$this->currentFixingCost; |
1558
|
|
|
} |
1559
|
|
|
|
1560
|
105 |
|
if ($i < 0) |
1561
|
|
|
{ |
1562
|
|
|
// Did not find a matching tag |
1563
|
2 |
|
$this->logger->debug('Skipping end tag with no start tag', ['tag' => $tag]); |
1564
|
|
|
|
1565
|
2 |
|
return; |
1566
|
|
|
} |
1567
|
|
|
|
1568
|
|
|
// Accumulate flags to determine whether whitespace should be trimmed |
1569
|
105 |
|
$flags = $tag->getFlags(); |
1570
|
105 |
|
foreach ($closeTags as $openTag) |
1571
|
|
|
{ |
1572
|
25 |
|
$flags |= $openTag->getFlags(); |
1573
|
|
|
} |
1574
|
105 |
|
$ignoreWhitespace = (bool) ($flags & self::RULE_IGNORE_WHITESPACE); |
1575
|
|
|
|
1576
|
|
|
// Only reopen tags if we haven't exceeded our "fixing" budget |
1577
|
105 |
|
$keepReopening = (bool) ($this->currentFixingCost < $this->maxFixingCost); |
1578
|
|
|
|
1579
|
|
|
// Iterate over tags that are being closed, output their end tag and collect tags to be |
1580
|
|
|
// reopened |
1581
|
105 |
|
$reopenTags = []; |
1582
|
105 |
|
foreach ($closeTags as $openTag) |
1583
|
|
|
{ |
1584
|
25 |
|
$openTagName = $openTag->getName(); |
1585
|
|
|
|
1586
|
|
|
// Test whether this tag should be reopened automatically |
1587
|
25 |
|
if ($keepReopening) |
1588
|
|
|
{ |
1589
|
23 |
|
if ($openTag->getFlags() & self::RULE_AUTO_REOPEN) |
1590
|
|
|
{ |
1591
|
12 |
|
$reopenTags[] = $openTag; |
1592
|
|
|
} |
1593
|
|
|
else |
1594
|
|
|
{ |
1595
|
11 |
|
$keepReopening = false; |
1596
|
|
|
} |
1597
|
|
|
} |
1598
|
|
|
|
1599
|
|
|
// Find the earliest position we can close this open tag |
1600
|
25 |
|
$tagPos = $tag->getPos(); |
1601
|
25 |
|
if ($ignoreWhitespace) |
1602
|
|
|
{ |
1603
|
5 |
|
$tagPos = $this->getMagicEndPos($tagPos); |
1604
|
|
|
} |
1605
|
|
|
|
1606
|
|
|
// Output an end tag to close this start tag, then update the context |
1607
|
25 |
|
$endTag = new Tag(Tag::END_TAG, $openTagName, $tagPos, 0); |
1608
|
25 |
|
$endTag->setFlags($openTag->getFlags()); |
1609
|
25 |
|
$this->outputTag($endTag); |
1610
|
25 |
|
$this->popContext(); |
1611
|
|
|
} |
1612
|
|
|
|
1613
|
|
|
// Output our tag, moving the cursor past it, then update the context |
1614
|
105 |
|
$this->outputTag($tag); |
1615
|
105 |
|
$this->popContext(); |
1616
|
|
|
|
1617
|
|
|
// If our fixing budget allows it, peek at upcoming tags and remove end tags that would |
1618
|
|
|
// close tags that are already being closed now. Also, filter our list of tags being |
1619
|
|
|
// reopened by removing those that would immediately be closed |
1620
|
105 |
|
if (!empty($closeTags) && $this->currentFixingCost < $this->maxFixingCost) |
1621
|
|
|
{ |
1622
|
|
|
/** |
1623
|
|
|
* @var integer Rightmost position of the portion of text to ignore |
1624
|
|
|
*/ |
1625
|
23 |
|
$ignorePos = $this->pos; |
1626
|
|
|
|
1627
|
23 |
|
$i = count($this->tagStack); |
1628
|
23 |
|
while (--$i >= 0 && ++$this->currentFixingCost < $this->maxFixingCost) |
1629
|
|
|
{ |
1630
|
15 |
|
$upcomingTag = $this->tagStack[$i]; |
1631
|
|
|
|
1632
|
|
|
// Test whether the upcoming tag is positioned at current "ignore" position and it's |
1633
|
|
|
// strictly an end tag (not a start tag or a self-closing tag) |
1634
|
15 |
|
if ($upcomingTag->getPos() > $ignorePos |
1635
|
15 |
|
|| $upcomingTag->isStartTag()) |
1636
|
|
|
{ |
1637
|
9 |
|
break; |
1638
|
|
|
} |
1639
|
|
|
|
1640
|
|
|
// Test whether this tag would close any of the tags we're about to reopen |
1641
|
10 |
|
$j = count($closeTags); |
1642
|
|
|
|
1643
|
10 |
|
while (--$j >= 0 && ++$this->currentFixingCost < $this->maxFixingCost) |
1644
|
|
|
{ |
1645
|
10 |
|
if ($upcomingTag->canClose($closeTags[$j])) |
1646
|
|
|
{ |
1647
|
|
|
// Remove the tag from the lists and reset the keys |
1648
|
9 |
|
array_splice($closeTags, $j, 1); |
1649
|
|
|
|
1650
|
9 |
|
if (isset($reopenTags[$j])) |
1651
|
|
|
{ |
1652
|
7 |
|
array_splice($reopenTags, $j, 1); |
1653
|
|
|
} |
1654
|
|
|
|
1655
|
|
|
// Extend the ignored text to cover this tag |
1656
|
9 |
|
$ignorePos = max( |
1657
|
9 |
|
$ignorePos, |
1658
|
9 |
|
$upcomingTag->getPos() + $upcomingTag->getLen() |
1659
|
|
|
); |
1660
|
|
|
|
1661
|
9 |
|
break; |
1662
|
|
|
} |
1663
|
|
|
} |
1664
|
|
|
} |
1665
|
|
|
|
1666
|
23 |
|
if ($ignorePos > $this->pos) |
1667
|
|
|
{ |
1668
|
|
|
/** |
1669
|
|
|
* @todo have a method that takes (pos,len) rather than a Tag |
1670
|
|
|
*/ |
1671
|
8 |
|
$this->outputIgnoreTag(new Tag(Tag::SELF_CLOSING_TAG, 'i', $this->pos, $ignorePos - $this->pos)); |
1672
|
|
|
} |
1673
|
|
|
} |
1674
|
|
|
|
1675
|
|
|
// Re-add tags that need to be reopened, at current cursor position |
1676
|
105 |
|
foreach ($reopenTags as $startTag) |
1677
|
|
|
{ |
1678
|
8 |
|
$newTag = $this->addCopyTag($startTag, $this->pos, 0); |
1679
|
|
|
|
1680
|
|
|
// Re-pair the new tag |
1681
|
8 |
|
$endTag = $startTag->getEndTag(); |
1682
|
8 |
|
if ($endTag) |
1683
|
|
|
{ |
1684
|
1 |
|
$newTag->pairWith($endTag); |
1685
|
|
|
} |
1686
|
|
|
} |
1687
|
|
|
} |
1688
|
|
|
|
1689
|
|
|
/** |
1690
|
|
|
* Update counters and replace current context with its parent context |
1691
|
|
|
* |
1692
|
|
|
* @return void |
1693
|
|
|
*/ |
1694
|
105 |
|
protected function popContext() |
1695
|
|
|
{ |
1696
|
105 |
|
$tag = array_pop($this->openTags); |
1697
|
105 |
|
--$this->cntOpen[$tag->getName()]; |
1698
|
105 |
|
$this->context = $this->context['parentContext']; |
1699
|
|
|
} |
1700
|
|
|
|
1701
|
|
|
/** |
1702
|
|
|
* Update counters and replace current context with a new context based on given tag |
1703
|
|
|
* |
1704
|
|
|
* If given tag is a self-closing tag, the context won't change |
1705
|
|
|
* |
1706
|
|
|
* @param Tag $tag Start tag (including self-closing) |
1707
|
|
|
* @return void |
1708
|
|
|
*/ |
1709
|
139 |
|
protected function pushContext(Tag $tag) |
1710
|
|
|
{ |
1711
|
139 |
|
$tagName = $tag->getName(); |
1712
|
139 |
|
$tagFlags = $tag->getFlags(); |
1713
|
139 |
|
$tagConfig = $this->tagsConfig[$tagName]; |
1714
|
|
|
|
1715
|
139 |
|
++$this->cntTotal[$tagName]; |
1716
|
|
|
|
1717
|
|
|
// If this is a self-closing tag, the context remains the same |
1718
|
139 |
|
if ($tag->isSelfClosingTag()) |
1719
|
|
|
{ |
1720
|
47 |
|
return; |
1721
|
|
|
} |
1722
|
|
|
|
1723
|
|
|
// Recompute the allowed tags |
1724
|
105 |
|
$allowed = []; |
1725
|
105 |
|
foreach ($this->context['allowed'] as $k => $v) |
1726
|
|
|
{ |
1727
|
|
|
// If the current tag is not transparent, override the low bits (allowed children) of |
1728
|
|
|
// current context with its high bits (allowed descendants) |
1729
|
105 |
|
if (!($tagFlags & self::RULE_IS_TRANSPARENT)) |
1730
|
|
|
{ |
1731
|
57 |
|
$v = ($v & 0xFF00) | ($v >> 8); |
1732
|
|
|
} |
1733
|
105 |
|
$allowed[] = $tagConfig['allowed'][$k] & $v; |
1734
|
|
|
} |
1735
|
|
|
|
1736
|
|
|
// Use this tag's flags as a base for this context and add inherited rules |
1737
|
105 |
|
$flags = $tagFlags | ($this->context['flags'] & self::RULES_INHERITANCE); |
1738
|
|
|
|
1739
|
|
|
// RULE_DISABLE_AUTO_BR turns off RULE_ENABLE_AUTO_BR |
1740
|
105 |
|
if ($flags & self::RULE_DISABLE_AUTO_BR) |
1741
|
|
|
{ |
1742
|
2 |
|
$flags &= ~self::RULE_ENABLE_AUTO_BR; |
1743
|
|
|
} |
1744
|
|
|
|
1745
|
105 |
|
++$this->cntOpen[$tagName]; |
1746
|
105 |
|
$this->openTags[] = $tag; |
1747
|
105 |
|
$this->context = [ |
1748
|
105 |
|
'allowed' => $allowed, |
1749
|
105 |
|
'flags' => $flags, |
1750
|
|
|
'inParagraph' => false, |
1751
|
105 |
|
'parentContext' => $this->context |
1752
|
|
|
]; |
1753
|
|
|
} |
1754
|
|
|
|
1755
|
|
|
/** |
1756
|
|
|
* Return whether given tag is allowed in current context |
1757
|
|
|
* |
1758
|
|
|
* @param string $tagName |
1759
|
|
|
* @return bool |
1760
|
|
|
*/ |
1761
|
141 |
|
protected function tagIsAllowed($tagName) |
1762
|
|
|
{ |
1763
|
141 |
|
$n = $this->tagsConfig[$tagName]['bitNumber']; |
1764
|
|
|
|
1765
|
141 |
|
return (bool) ($this->context['allowed'][$n >> 3] & (1 << ($n & 7))); |
1766
|
|
|
} |
1767
|
|
|
|
1768
|
|
|
//========================================================================== |
1769
|
|
|
// Tag stack |
1770
|
|
|
//========================================================================== |
1771
|
|
|
|
1772
|
|
|
/** |
1773
|
|
|
* Add a start tag |
1774
|
|
|
* |
1775
|
|
|
* @param string $name Name of the tag |
1776
|
|
|
* @param integer $pos Position of the tag in the text |
1777
|
|
|
* @param integer $len Length of text consumed by the tag |
1778
|
|
|
* @param integer $prio Tag's priority |
1779
|
|
|
* @return Tag |
1780
|
|
|
*/ |
1781
|
124 |
|
public function addStartTag($name, $pos, $len, $prio = 0) |
1782
|
|
|
{ |
1783
|
124 |
|
return $this->addTag(Tag::START_TAG, $name, $pos, $len, $prio); |
1784
|
|
|
} |
1785
|
|
|
|
1786
|
|
|
/** |
1787
|
|
|
* Add an end tag |
1788
|
|
|
* |
1789
|
|
|
* @param string $name Name of the tag |
1790
|
|
|
* @param integer $pos Position of the tag in the text |
1791
|
|
|
* @param integer $len Length of text consumed by the tag |
1792
|
|
|
* @param integer $prio Tag's priority |
1793
|
|
|
* @return Tag |
1794
|
|
|
*/ |
1795
|
111 |
|
public function addEndTag($name, $pos, $len, $prio = 0) |
1796
|
|
|
{ |
1797
|
111 |
|
return $this->addTag(Tag::END_TAG, $name, $pos, $len, $prio); |
1798
|
|
|
} |
1799
|
|
|
|
1800
|
|
|
/** |
1801
|
|
|
* Add a self-closing tag |
1802
|
|
|
* |
1803
|
|
|
* @param string $name Name of the tag |
1804
|
|
|
* @param integer $pos Position of the tag in the text |
1805
|
|
|
* @param integer $len Length of text consumed by the tag |
1806
|
|
|
* @param integer $prio Tag's priority |
1807
|
|
|
* @return Tag |
1808
|
|
|
*/ |
1809
|
68 |
|
public function addSelfClosingTag($name, $pos, $len, $prio = 0) |
1810
|
|
|
{ |
1811
|
68 |
|
return $this->addTag(Tag::SELF_CLOSING_TAG, $name, $pos, $len, $prio); |
1812
|
|
|
} |
1813
|
|
|
|
1814
|
|
|
/** |
1815
|
|
|
* Add a 0-width "br" tag to force a line break at given position |
1816
|
|
|
* |
1817
|
|
|
* @param integer $pos Position of the tag in the text |
1818
|
|
|
* @param integer $prio Tag's priority |
1819
|
|
|
* @return Tag |
1820
|
|
|
*/ |
1821
|
9 |
|
public function addBrTag($pos, $prio = 0) |
1822
|
|
|
{ |
1823
|
9 |
|
return $this->addTag(Tag::SELF_CLOSING_TAG, 'br', $pos, 0, $prio); |
1824
|
|
|
} |
1825
|
|
|
|
1826
|
|
|
/** |
1827
|
|
|
* Add an "ignore" tag |
1828
|
|
|
* |
1829
|
|
|
* @param integer $pos Position of the tag in the text |
1830
|
|
|
* @param integer $len Length of text consumed by the tag |
1831
|
|
|
* @param integer $prio Tag's priority |
1832
|
|
|
* @return Tag |
1833
|
|
|
*/ |
1834
|
13 |
|
public function addIgnoreTag($pos, $len, $prio = 0) |
1835
|
|
|
{ |
1836
|
13 |
|
return $this->addTag(Tag::SELF_CLOSING_TAG, 'i', $pos, min($len, $this->textLen - $pos), $prio); |
1837
|
|
|
} |
1838
|
|
|
|
1839
|
|
|
/** |
1840
|
|
|
* Add a paragraph break at given position |
1841
|
|
|
* |
1842
|
|
|
* Uses a zero-width tag that is actually never output in the result |
1843
|
|
|
* |
1844
|
|
|
* @param integer $pos Position of the tag in the text |
1845
|
|
|
* @param integer $prio Tag's priority |
1846
|
|
|
* @return Tag |
1847
|
|
|
*/ |
1848
|
5 |
|
public function addParagraphBreak($pos, $prio = 0) |
1849
|
|
|
{ |
1850
|
5 |
|
return $this->addTag(Tag::SELF_CLOSING_TAG, 'pb', $pos, 0, $prio); |
1851
|
|
|
} |
1852
|
|
|
|
1853
|
|
|
/** |
1854
|
|
|
* Add a copy of given tag at given position and length |
1855
|
|
|
* |
1856
|
|
|
* @param Tag $tag Original tag |
1857
|
|
|
* @param integer $pos Copy's position |
1858
|
|
|
* @param integer $len Copy's length |
1859
|
|
|
* @param integer $prio Copy's priority (same as original by default) |
1860
|
|
|
* @return Tag Copy tag |
1861
|
|
|
*/ |
1862
|
22 |
|
public function addCopyTag(Tag $tag, $pos, $len, $prio = null) |
1863
|
|
|
{ |
1864
|
22 |
|
if (!isset($prio)) |
1865
|
|
|
{ |
1866
|
11 |
|
$prio = $tag->getSortPriority(); |
1867
|
|
|
} |
1868
|
22 |
|
$copy = $this->addTag($tag->getType(), $tag->getName(), $pos, $len, $prio); |
1869
|
22 |
|
$copy->setAttributes($tag->getAttributes()); |
1870
|
|
|
|
1871
|
22 |
|
return $copy; |
1872
|
|
|
} |
1873
|
|
|
|
1874
|
|
|
/** |
1875
|
|
|
* Add a tag |
1876
|
|
|
* |
1877
|
|
|
* @param integer $type Tag's type |
1878
|
|
|
* @param string $name Name of the tag |
1879
|
|
|
* @param integer $pos Position of the tag in the text |
1880
|
|
|
* @param integer $len Length of text consumed by the tag |
1881
|
|
|
* @param integer $prio Tag's priority |
1882
|
|
|
* @return Tag |
1883
|
|
|
*/ |
1884
|
187 |
|
protected function addTag($type, $name, $pos, $len, $prio) |
1885
|
|
|
{ |
1886
|
|
|
// Create the tag |
1887
|
187 |
|
$tag = new Tag($type, $name, $pos, $len, $prio); |
1888
|
|
|
|
1889
|
|
|
// Set this tag's rules bitfield |
1890
|
187 |
|
if (isset($this->tagsConfig[$name])) |
1891
|
|
|
{ |
1892
|
167 |
|
$tag->setFlags($this->tagsConfig[$name]['rules']['flags']); |
1893
|
|
|
} |
1894
|
|
|
|
1895
|
|
|
// Invalidate this tag if it's an unknown tag, a disabled tag, if either of its length or |
1896
|
|
|
// position is negative or if it's out of bounds |
1897
|
187 |
|
if ((!isset($this->tagsConfig[$name]) && !$tag->isSystemTag()) |
1898
|
187 |
|
|| $this->isInvalidTextSpan($pos, $len)) |
1899
|
|
|
{ |
1900
|
10 |
|
$tag->invalidate(); |
1901
|
|
|
} |
1902
|
179 |
|
elseif (!empty($this->tagsConfig[$name]['isDisabled'])) |
1903
|
|
|
{ |
1904
|
1 |
|
$this->logger->warn( |
1905
|
1 |
|
'Tag is disabled', |
1906
|
|
|
[ |
1907
|
1 |
|
'tag' => $tag, |
1908
|
1 |
|
'tagName' => $name |
1909
|
|
|
] |
1910
|
|
|
); |
1911
|
1 |
|
$tag->invalidate(); |
1912
|
|
|
} |
1913
|
|
|
else |
1914
|
|
|
{ |
1915
|
178 |
|
$this->insertTag($tag); |
1916
|
|
|
} |
1917
|
|
|
|
1918
|
187 |
|
return $tag; |
1919
|
|
|
} |
1920
|
|
|
|
1921
|
|
|
/** |
1922
|
|
|
* Test whether given text span is outside text boundaries or an invalid UTF sequence |
1923
|
|
|
* |
1924
|
|
|
* @param integer $pos Start of text |
1925
|
|
|
* @param integer $len Length of text |
1926
|
|
|
* @return bool |
1927
|
|
|
*/ |
1928
|
185 |
|
protected function isInvalidTextSpan($pos, $len) |
1929
|
|
|
{ |
1930
|
185 |
|
return ($len < 0 || $pos < 0 || $pos + $len > $this->textLen || preg_match('([\\x80-\\xBF])', substr($this->text, $pos, 1) . substr($this->text, $pos + $len, 1))); |
1931
|
|
|
} |
1932
|
|
|
|
1933
|
|
|
/** |
1934
|
|
|
* Insert given tag in the tag stack |
1935
|
|
|
* |
1936
|
|
|
* @param Tag $tag |
1937
|
|
|
* @return void |
1938
|
|
|
*/ |
1939
|
178 |
|
protected function insertTag(Tag $tag) |
1940
|
|
|
{ |
1941
|
178 |
|
if (!$this->tagStackIsSorted) |
1942
|
|
|
{ |
1943
|
178 |
|
$this->tagStack[] = $tag; |
1944
|
|
|
} |
1945
|
|
|
else |
1946
|
|
|
{ |
1947
|
|
|
// Scan the stack and copy every tag to the next slot until we find the correct index |
1948
|
50 |
|
$i = count($this->tagStack); |
1949
|
50 |
|
$key = $this->getSortKey($tag); |
1950
|
50 |
|
while ($i > 0 && $key > $this->getSortKey($this->tagStack[$i - 1])) |
1951
|
|
|
{ |
1952
|
3 |
|
$this->tagStack[$i] = $this->tagStack[$i - 1]; |
1953
|
3 |
|
--$i; |
1954
|
|
|
} |
1955
|
50 |
|
$this->tagStack[$i] = $tag; |
1956
|
|
|
} |
1957
|
|
|
} |
1958
|
|
|
|
1959
|
|
|
/** |
1960
|
|
|
* Add a pair of tags |
1961
|
|
|
* |
1962
|
|
|
* @param string $name Name of the tags |
1963
|
|
|
* @param integer $startPos Position of the start tag |
1964
|
|
|
* @param integer $startLen Length of the start tag |
1965
|
|
|
* @param integer $endPos Position of the start tag |
1966
|
|
|
* @param integer $endLen Length of the start tag |
1967
|
|
|
* @param integer $prio Start tag's priority (the end tag will be set to minus that value) |
1968
|
|
|
* @return Tag Start tag |
1969
|
|
|
*/ |
1970
|
26 |
|
public function addTagPair($name, $startPos, $startLen, $endPos, $endLen, $prio = 0) |
1971
|
|
|
{ |
1972
|
|
|
// NOTE: the end tag is added first to try to keep the stack in the correct order |
1973
|
26 |
|
$endTag = $this->addEndTag($name, $endPos, $endLen, -$prio); |
1974
|
26 |
|
$startTag = $this->addStartTag($name, $startPos, $startLen, $prio); |
1975
|
26 |
|
$startTag->pairWith($endTag); |
1976
|
|
|
|
1977
|
26 |
|
return $startTag; |
1978
|
|
|
} |
1979
|
|
|
|
1980
|
|
|
/** |
1981
|
|
|
* Add a tag that represents a verbatim copy of the original text |
1982
|
|
|
* |
1983
|
|
|
* @param integer $pos Position of the tag in the text |
1984
|
|
|
* @param integer $len Length of text consumed by the tag |
1985
|
|
|
* @param integer $prio Tag's priority |
1986
|
|
|
* @return Tag |
1987
|
|
|
*/ |
1988
|
4 |
|
public function addVerbatim($pos, $len, $prio = 0) |
1989
|
|
|
{ |
1990
|
4 |
|
return $this->addTag(Tag::SELF_CLOSING_TAG, 'v', $pos, $len, $prio); |
1991
|
|
|
} |
1992
|
|
|
|
1993
|
|
|
/** |
1994
|
|
|
* Sort tags by position and precedence |
1995
|
|
|
* |
1996
|
|
|
* @return void |
1997
|
|
|
*/ |
1998
|
165 |
|
protected function sortTags() |
1999
|
|
|
{ |
2000
|
165 |
|
$arr = []; |
2001
|
165 |
|
foreach ($this->tagStack as $i => $tag) |
2002
|
|
|
{ |
2003
|
165 |
|
$key = $this->getSortKey($tag, $i); |
2004
|
165 |
|
$arr[$key] = $tag; |
2005
|
|
|
} |
2006
|
165 |
|
krsort($arr); |
2007
|
|
|
|
2008
|
165 |
|
$this->tagStack = array_values($arr); |
2009
|
165 |
|
$this->tagStackIsSorted = true; |
2010
|
|
|
} |
2011
|
|
|
|
2012
|
|
|
/** |
2013
|
|
|
* Generate a key for given tag that can be used to compare its position using lexical comparisons |
2014
|
|
|
* |
2015
|
|
|
* Tags are sorted by position first, then by priority, then by whether they consume any text, |
2016
|
|
|
* then by length, and finally in order of their creation. |
2017
|
|
|
* |
2018
|
|
|
* The stack's array is in reverse order. Therefore, tags that appear at the start of the text |
2019
|
|
|
* are at the end of the array. |
2020
|
|
|
* |
2021
|
|
|
* @param Tag $tag |
2022
|
|
|
* @param integer $tagIndex |
2023
|
|
|
* @return string |
2024
|
|
|
*/ |
2025
|
165 |
|
protected function getSortKey(Tag $tag, int $tagIndex = 0): string |
2026
|
|
|
{ |
2027
|
|
|
// Ensure that negative values are sorted correctly by flagging them and making them positive |
2028
|
165 |
|
$prioFlag = ($tag->getSortPriority() >= 0); |
2029
|
165 |
|
$prio = $tag->getSortPriority(); |
2030
|
165 |
|
if (!$prioFlag) |
2031
|
|
|
{ |
2032
|
33 |
|
$prio += (1 << 30); |
2033
|
|
|
} |
2034
|
|
|
|
2035
|
|
|
// Sort 0-width tags separately from the rest |
2036
|
165 |
|
$lenFlag = ($tag->getLen() > 0); |
2037
|
165 |
|
if ($lenFlag) |
2038
|
|
|
{ |
2039
|
|
|
// Inverse their length so that longest matches are processed first |
2040
|
118 |
|
$lenOrder = $this->textLen - $tag->getLen(); |
2041
|
|
|
} |
2042
|
|
|
else |
2043
|
|
|
{ |
2044
|
|
|
// Sort self-closing tags in-between start tags and end tags to keep them outside of tag |
2045
|
|
|
// pairs |
2046
|
|
|
$order = [ |
2047
|
88 |
|
Tag::END_TAG => 0, |
2048
|
88 |
|
Tag::SELF_CLOSING_TAG => 1, |
2049
|
88 |
|
Tag::START_TAG => 2 |
2050
|
|
|
]; |
2051
|
88 |
|
$lenOrder = $order[$tag->getType()]; |
2052
|
|
|
} |
2053
|
|
|
|
2054
|
165 |
|
return sprintf('%8x%d%8x%d%8x%8x', $tag->getPos(), $prioFlag, $prio, $lenFlag, $lenOrder, $tagIndex); |
2055
|
|
|
} |
2056
|
|
|
} |