1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\TextFormatter |
5
|
|
|
* @copyright Copyright (c) 2010-2017 The s9e Authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\TextFormatter\Plugins\BBCodes; |
9
|
|
|
|
10
|
|
|
use RuntimeException; |
11
|
|
|
use s9e\TextFormatter\Parser\Tag; |
12
|
|
|
use s9e\TextFormatter\Plugins\ParserBase; |
13
|
|
|
|
14
|
|
|
class Parser extends ParserBase |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* @var array Attributes of the BBCode being parsed |
18
|
|
|
*/ |
19
|
|
|
protected $attributes; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @var array Configuration for the BBCode being parsed |
23
|
|
|
*/ |
24
|
|
|
protected $bbcodeConfig; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var string Name of the BBCode being parsed |
28
|
|
|
*/ |
29
|
|
|
protected $bbcodeName; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var string Suffix of the BBCode being parsed, including its colon |
33
|
|
|
*/ |
34
|
|
|
protected $bbcodeSuffix; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* @var integer Position of the cursor in the original text |
38
|
|
|
*/ |
39
|
|
|
protected $pos; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @var integer Position of the start of the BBCode being parsed |
43
|
|
|
*/ |
44
|
|
|
protected $startPos; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* @var string Text being parsed |
48
|
|
|
*/ |
49
|
|
|
protected $text; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var integer Length of the text being parsed |
53
|
|
|
*/ |
54
|
|
|
protected $textLen; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @var string Text being parsed, normalized to uppercase |
58
|
|
|
*/ |
59
|
|
|
protected $uppercaseText; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* {@inheritdoc} |
63
|
|
|
*/ |
64
|
50 |
|
public function parse($text, array $matches) |
65
|
|
|
{ |
66
|
50 |
|
$this->text = $text; |
67
|
50 |
|
$this->textLen = strlen($text); |
68
|
50 |
|
$this->uppercaseText = ''; |
69
|
50 |
|
foreach ($matches as $m) |
70
|
|
|
{ |
71
|
50 |
|
$this->bbcodeName = strtoupper($m[1][0]); |
72
|
50 |
|
if (!isset($this->config['bbcodes'][$this->bbcodeName])) |
73
|
|
|
{ |
74
|
3 |
|
continue; |
75
|
|
|
} |
76
|
50 |
|
$this->bbcodeConfig = $this->config['bbcodes'][$this->bbcodeName]; |
77
|
50 |
|
$this->startPos = $m[0][1]; |
78
|
50 |
|
$this->pos = $this->startPos + strlen($m[0][0]); |
79
|
|
|
|
80
|
|
|
try |
81
|
|
|
{ |
82
|
50 |
|
$this->parseBBCode(); |
83
|
|
|
} |
84
|
7 |
|
catch (RuntimeException $e) |
85
|
50 |
|
{ |
86
|
|
|
// Do nothing |
87
|
|
|
} |
88
|
|
|
} |
89
|
50 |
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* Add the end tag that matches current BBCode |
93
|
|
|
* |
94
|
|
|
* @return Tag |
95
|
|
|
*/ |
96
|
22 |
|
protected function addBBCodeEndTag() |
97
|
|
|
{ |
98
|
22 |
|
return $this->parser->addEndTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos); |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* Add the self-closing tag that matches current BBCode |
103
|
|
|
* |
104
|
|
|
* @return Tag |
105
|
|
|
*/ |
106
|
13 |
|
protected function addBBCodeSelfClosingTag() |
107
|
|
|
{ |
108
|
13 |
|
$tag = $this->parser->addSelfClosingTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos); |
109
|
13 |
|
$tag->setAttributes($this->attributes); |
110
|
|
|
|
111
|
13 |
|
return $tag; |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* Add the start tag that matches current BBCode |
116
|
|
|
* |
117
|
|
|
* @return Tag |
118
|
|
|
*/ |
119
|
23 |
|
protected function addBBCodeStartTag() |
120
|
|
|
{ |
121
|
23 |
|
$tag = $this->parser->addStartTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos); |
122
|
23 |
|
$tag->setAttributes($this->attributes); |
123
|
|
|
|
124
|
23 |
|
return $tag; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Parse the end tag that matches given BBCode name and suffix starting at current position |
129
|
|
|
* |
130
|
|
|
* @return Tag|null |
131
|
|
|
*/ |
132
|
11 |
|
protected function captureEndTag() |
133
|
|
|
{ |
134
|
11 |
|
if (empty($this->uppercaseText)) |
135
|
|
|
{ |
136
|
11 |
|
$this->uppercaseText = strtoupper($this->text); |
137
|
|
|
} |
138
|
11 |
|
$match = '[/' . $this->bbcodeName . $this->bbcodeSuffix . ']'; |
139
|
11 |
|
$endTagPos = strpos($this->uppercaseText, $match, $this->pos); |
140
|
11 |
|
if ($endTagPos === false) |
141
|
|
|
{ |
142
|
5 |
|
return; |
143
|
|
|
} |
144
|
|
|
|
145
|
7 |
|
return $this->parser->addEndTag($this->getTagName(), $endTagPos, strlen($match)); |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* Get the tag name for current BBCode |
150
|
|
|
* |
151
|
|
|
* @return string |
152
|
|
|
*/ |
153
|
37 |
|
protected function getTagName() |
154
|
|
|
{ |
155
|
|
|
// Use the configured tagName if available, or reuse the BBCode's name otherwise |
156
|
37 |
|
return (isset($this->bbcodeConfig['tagName'])) |
157
|
1 |
|
? $this->bbcodeConfig['tagName'] |
158
|
37 |
|
: $this->bbcodeName; |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* Parse attributes starting at current position |
163
|
|
|
* |
164
|
|
|
* @return array Associative array of [name => value] |
165
|
|
|
*/ |
166
|
49 |
|
protected function parseAttributes() |
167
|
|
|
{ |
168
|
49 |
|
$firstPos = $this->pos; |
169
|
49 |
|
$this->attributes = []; |
170
|
49 |
|
while ($this->pos < $this->textLen) |
171
|
|
|
{ |
172
|
49 |
|
$c = $this->text[$this->pos]; |
173
|
49 |
|
if (strpos(" \n\t", $c) !== false) |
174
|
|
|
{ |
175
|
26 |
|
++$this->pos; |
176
|
26 |
|
continue; |
177
|
|
|
} |
178
|
48 |
|
if (strpos('/]', $c) !== false) |
179
|
|
|
{ |
180
|
41 |
|
return; |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
// Capture the attribute name |
184
|
28 |
|
$spn = strspn($this->text, 'abcdefghijklmnopqrstuvwxyz_0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ-', $this->pos); |
185
|
28 |
|
if ($spn) |
186
|
|
|
{ |
187
|
20 |
|
$attrName = strtolower(substr($this->text, $this->pos, $spn)); |
188
|
20 |
|
$this->pos += $spn; |
189
|
20 |
|
if ($this->pos >= $this->textLen) |
190
|
|
|
{ |
191
|
|
|
// The attribute name extends to the end of the text |
192
|
1 |
|
throw new RuntimeException; |
193
|
|
|
} |
194
|
19 |
|
if ($this->text[$this->pos] !== '=') |
195
|
|
|
{ |
196
|
|
|
// It's an attribute name not followed by an equal sign, ignore it |
197
|
19 |
|
continue; |
198
|
|
|
} |
199
|
|
|
} |
200
|
9 |
|
elseif ($c === '=' && $this->pos === $firstPos) |
201
|
|
|
{ |
202
|
|
|
// This is the default param, e.g. [quote=foo] |
203
|
8 |
|
$attrName = (isset($this->bbcodeConfig['defaultAttribute'])) |
204
|
2 |
|
? $this->bbcodeConfig['defaultAttribute'] |
205
|
8 |
|
: strtolower($this->bbcodeName); |
206
|
|
|
} |
207
|
|
|
else |
208
|
|
|
{ |
209
|
1 |
|
throw new RuntimeException; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
// Move past the = and make sure we're not at the end of the text |
213
|
25 |
|
if (++$this->pos >= $this->textLen) |
214
|
|
|
{ |
215
|
1 |
|
throw new RuntimeException; |
216
|
|
|
} |
217
|
|
|
|
218
|
24 |
|
$this->attributes[$attrName] = $this->parseAttributeValue(); |
219
|
|
|
} |
220
|
1 |
|
} |
221
|
|
|
|
222
|
|
|
/** |
223
|
|
|
* Parse the attribute value starting at current position |
224
|
|
|
* |
225
|
|
|
* @return string |
226
|
|
|
*/ |
227
|
24 |
|
protected function parseAttributeValue() |
228
|
|
|
{ |
229
|
|
|
// Test whether the value is in quotes |
230
|
24 |
|
if ($this->text[$this->pos] === '"' || $this->text[$this->pos] === "'") |
231
|
|
|
{ |
232
|
14 |
|
return $this->parseQuotedAttributeValue(); |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
// Capture everything up to whichever comes first: |
236
|
|
|
// - an endline |
237
|
|
|
// - whitespace followed by a slash and a closing bracket |
238
|
|
|
// - a closing bracket, optionally preceded by whitespace |
239
|
|
|
// - whitespace followed by another attribute (name followed by equal sign) |
240
|
|
|
// |
241
|
|
|
// NOTE: this is for compatibility with some forums (such as vBulletin it seems) |
242
|
|
|
// that do not put attribute values in quotes, e.g. |
243
|
|
|
// [quote=John Smith;123456] (quoting "John Smith" from post #123456) |
244
|
10 |
|
if (!preg_match('#[^\\]\\n]*?(?=\\s*(?:\\s/)?\\]|\\s+[-\\w]+=)#', $this->text, $m, null, $this->pos)) |
245
|
|
|
{ |
246
|
1 |
|
throw new RuntimeException; |
247
|
|
|
} |
248
|
|
|
|
249
|
9 |
|
$attrValue = $m[0]; |
250
|
9 |
|
$this->pos += strlen($attrValue); |
251
|
|
|
|
252
|
9 |
|
return $attrValue; |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* Parse current BBCode |
257
|
|
|
* |
258
|
|
|
* @return void |
259
|
|
|
*/ |
260
|
50 |
|
protected function parseBBCode() |
261
|
|
|
{ |
262
|
50 |
|
$this->parseBBCodeSuffix(); |
263
|
|
|
|
264
|
|
|
// Test whether this is an end tag |
265
|
50 |
|
if ($this->text[$this->startPos + 1] === '/') |
266
|
|
|
{ |
267
|
|
|
// Test whether the tag is properly closed and whether this tag has an identifier. |
268
|
|
|
// We skip end tags that carry an identifier because they're automatically added |
269
|
|
|
// when their start tag is processed |
270
|
25 |
|
if (substr($this->text, $this->pos, 1) === ']' && $this->bbcodeSuffix === '') |
271
|
|
|
{ |
272
|
22 |
|
++$this->pos; |
273
|
22 |
|
$this->addBBCodeEndTag(); |
274
|
|
|
} |
275
|
|
|
|
276
|
25 |
|
return; |
277
|
|
|
} |
278
|
|
|
|
279
|
|
|
// Parse attributes |
280
|
49 |
|
$this->parseAttributes(); |
281
|
|
|
|
282
|
|
|
// Test whether the tag is properly closed |
283
|
42 |
|
if (substr($this->text, $this->pos, 1) === ']') |
284
|
|
|
{ |
285
|
26 |
|
++$this->pos; |
286
|
|
|
} |
287
|
|
|
else |
288
|
|
|
{ |
289
|
|
|
// Test whether this is a self-closing tag |
290
|
16 |
|
if (substr($this->text, $this->pos, 2) === '/]') |
291
|
|
|
{ |
292
|
13 |
|
$this->pos += 2; |
293
|
13 |
|
$this->addBBCodeSelfClosingTag(); |
294
|
|
|
} |
295
|
|
|
|
296
|
16 |
|
return; |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
// Record the names of attributes that need the content of this tag |
300
|
26 |
|
$contentAttributes = []; |
301
|
26 |
|
if (isset($this->bbcodeConfig['contentAttributes'])) |
302
|
|
|
{ |
303
|
4 |
|
foreach ($this->bbcodeConfig['contentAttributes'] as $attrName) |
304
|
|
|
{ |
305
|
4 |
|
if (!isset($this->attributes[$attrName])) |
306
|
|
|
{ |
307
|
4 |
|
$contentAttributes[] = $attrName; |
308
|
|
|
} |
309
|
|
|
} |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
// Look ahead and parse the end tag that matches this tag, if applicable |
313
|
26 |
|
$requireEndTag = ($this->bbcodeSuffix || !empty($this->bbcodeConfig['forceLookahead'])); |
314
|
26 |
|
$endTag = ($requireEndTag || !empty($contentAttributes)) ? $this->captureEndTag() : null; |
315
|
26 |
|
if (isset($endTag)) |
316
|
|
|
{ |
317
|
7 |
|
foreach ($contentAttributes as $attrName) |
318
|
|
|
{ |
319
|
7 |
|
$this->attributes[$attrName] = substr($this->text, $this->pos, $endTag->getPos() - $this->pos); |
320
|
|
|
} |
321
|
|
|
} |
322
|
21 |
|
elseif ($requireEndTag) |
323
|
|
|
{ |
324
|
4 |
|
return; |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
// Create this start tag |
328
|
23 |
|
$tag = $this->addBBCodeStartTag(); |
329
|
|
|
|
330
|
|
|
// If an end tag was created, pair it with this start tag |
331
|
23 |
|
if (isset($endTag)) |
332
|
|
|
{ |
333
|
7 |
|
$tag->pairWith($endTag); |
334
|
|
|
} |
335
|
23 |
|
} |
336
|
|
|
|
337
|
|
|
/** |
338
|
|
|
* Parse the BBCode suffix starting at current position |
339
|
|
|
* |
340
|
|
|
* Used to explicitly pair specific tags together, e.g. |
341
|
|
|
* [code:123][code]type your code here[/code][/code:123] |
342
|
|
|
* |
343
|
|
|
* @return void |
344
|
|
|
*/ |
345
|
50 |
|
protected function parseBBCodeSuffix() |
346
|
|
|
{ |
347
|
50 |
|
$this->bbcodeSuffix = ''; |
348
|
50 |
|
if ($this->text[$this->pos] === ':') |
349
|
|
|
{ |
350
|
|
|
// Capture the colon and the (0 or more) digits following it |
351
|
7 |
|
$spn = 1 + strspn($this->text, '0123456789', 1 + $this->pos); |
352
|
7 |
|
$this->bbcodeSuffix = substr($this->text, $this->pos, $spn); |
353
|
|
|
|
354
|
|
|
// Move past the suffix |
355
|
7 |
|
$this->pos += $spn; |
356
|
|
|
} |
357
|
50 |
|
} |
358
|
|
|
|
359
|
|
|
/** |
360
|
|
|
* Parse a quoted attribute value that starts at current offset |
361
|
|
|
* |
362
|
|
|
* @return string |
363
|
|
|
*/ |
364
|
14 |
|
protected function parseQuotedAttributeValue() |
365
|
|
|
{ |
366
|
14 |
|
$quote = $this->text[$this->pos]; |
367
|
14 |
|
$valuePos = $this->pos + 1; |
368
|
14 |
|
while (1) |
369
|
|
|
{ |
370
|
|
|
// Look for the next quote |
371
|
14 |
|
$this->pos = strpos($this->text, $quote, $this->pos + 1); |
372
|
14 |
|
if ($this->pos === false) |
373
|
|
|
{ |
374
|
|
|
// No matching quote. Apparently that string never ends... |
375
|
3 |
|
throw new RuntimeException; |
376
|
|
|
} |
377
|
|
|
|
378
|
|
|
// Test for an odd number of backslashes before this character |
379
|
11 |
|
$n = 0; |
380
|
|
|
do |
381
|
|
|
{ |
382
|
11 |
|
++$n; |
383
|
|
|
} |
384
|
11 |
|
while ($this->text[$this->pos - $n] === '\\'); |
385
|
|
|
|
386
|
11 |
|
if ($n % 2) |
387
|
|
|
{ |
388
|
|
|
// If $n is odd, it means there's an even number of backslashes. We can exit this |
389
|
|
|
// loop |
390
|
11 |
|
break; |
391
|
|
|
} |
392
|
|
|
} |
393
|
|
|
|
394
|
|
|
// Unescape special characters ' " and \ |
395
|
11 |
|
$attrValue = preg_replace( |
396
|
11 |
|
'#\\\\([\\\\\'"])#', |
397
|
11 |
|
'$1', |
398
|
11 |
|
substr($this->text, $valuePos, $this->pos - $valuePos) |
399
|
|
|
); |
400
|
|
|
|
401
|
|
|
// Skip past the closing quote |
402
|
11 |
|
++$this->pos; |
403
|
|
|
|
404
|
11 |
|
return $attrValue; |
405
|
|
|
} |
406
|
|
|
} |