1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
# |
4
|
|
|
# |
5
|
|
|
# Parsedown |
6
|
|
|
# http://parsedown.org |
7
|
|
|
# |
8
|
|
|
# (c) Emanuil Rusev |
9
|
|
|
# http://erusev.com |
10
|
|
|
# |
11
|
|
|
# For the full license information, please view the LICENSE file that was |
12
|
|
|
# distributed with this source code. |
13
|
|
|
# |
14
|
|
|
# |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* Class Parsedown |
18
|
|
|
*/ |
19
|
|
|
class Parsedown |
20
|
|
|
{ |
21
|
|
|
# |
22
|
|
|
# Multiton (http://en.wikipedia.org/wiki/Multiton_pattern) |
23
|
|
|
# |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* @param string $name |
27
|
|
|
* @return mixed|Parsedown |
28
|
|
|
*/ |
29
|
|
|
public static function instance($name = 'default') |
30
|
|
|
{ |
31
|
|
|
if (isset(self::$instances[$name])) { |
32
|
|
|
return self::$instances[$name]; |
33
|
|
|
} |
34
|
|
|
|
35
|
|
|
$instance = new self(); |
36
|
|
|
|
37
|
|
|
self::$instances[$name] = $instance; |
38
|
|
|
|
39
|
|
|
return $instance; |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
private static $instances = []; |
43
|
|
|
|
44
|
|
|
# |
45
|
|
|
# Setters |
46
|
|
|
# |
47
|
|
|
|
48
|
|
|
private $break_marker = " \n"; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* @param $breaks_enabled |
52
|
|
|
* @return $this |
53
|
|
|
*/ |
54
|
|
|
public function set_breaks_enabled($breaks_enabled) |
55
|
|
|
{ |
56
|
|
|
$this->break_marker = $breaks_enabled ? "\n" : " \n"; |
57
|
|
|
|
58
|
|
|
return $this; |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
# |
62
|
|
|
# Fields |
63
|
|
|
# |
64
|
|
|
|
65
|
|
|
private $reference_map = []; |
66
|
|
|
private $escape_sequence_map = []; |
67
|
|
|
|
68
|
|
|
# |
69
|
|
|
# Public Methods |
70
|
|
|
# |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @param $text |
74
|
|
|
* @return mixed|string |
75
|
|
|
*/ |
76
|
|
|
public function parse($text) |
77
|
|
|
{ |
78
|
|
|
# removes \r characters |
79
|
|
|
$text = str_replace("\r\n", "\n", $text); |
80
|
|
|
$text = str_replace("\r", "\n", $text); |
81
|
|
|
|
82
|
|
|
# replaces tabs with spaces |
83
|
|
|
$text = str_replace("\t", ' ', $text); |
84
|
|
|
|
85
|
|
|
# encodes escape sequences |
86
|
|
|
|
87
|
|
|
if (false !== mb_strpos($text, '\\')) { |
88
|
|
|
$escape_sequences = ['\\\\', '\`', '\*', '\_', '\{', '\}', '\[', '\]', '\(', '\)', '\>', '\#', '\+', '\-', '\.', '\!']; |
89
|
|
|
|
90
|
|
|
foreach ($escape_sequences as $index => $escape_sequence) { |
91
|
|
|
if (false !== mb_strpos($text, $escape_sequence)) { |
92
|
|
|
$code = "\x1A" . '\\' . $index . ';'; |
93
|
|
|
|
94
|
|
|
$text = str_replace($escape_sequence, $code, $text); |
95
|
|
|
|
96
|
|
|
$this->escape_sequence_map[$code] = $escape_sequence; |
97
|
|
|
} |
98
|
|
|
} |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
# ~ |
102
|
|
|
|
103
|
|
|
$text = trim($text, "\n"); |
104
|
|
|
|
105
|
|
|
$lines = explode("\n", $text); |
106
|
|
|
|
107
|
|
|
$text = $this->parse_block_elements($lines); |
108
|
|
|
|
109
|
|
|
# decodes escape sequences |
110
|
|
|
|
111
|
|
|
foreach ($this->escape_sequence_map as $code => $escape_sequence) { |
112
|
|
|
$text = str_replace($code, $escape_sequence[1], $text); |
113
|
|
|
} |
114
|
|
|
|
115
|
|
|
# ~ |
116
|
|
|
|
117
|
|
|
$text = rtrim($text, "\n"); |
118
|
|
|
|
119
|
|
|
return $text; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
# |
123
|
|
|
# Private Methods |
124
|
|
|
# |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* @param array $lines |
128
|
|
|
* @param string $context |
129
|
|
|
* @return string |
130
|
|
|
*/ |
131
|
|
|
private function parse_block_elements(array $lines, $context = '') |
132
|
|
|
{ |
133
|
|
|
$elements = []; |
134
|
|
|
|
135
|
|
|
$element = [ |
136
|
|
|
'type' => '', |
137
|
|
|
]; |
138
|
|
|
|
139
|
|
|
foreach ($lines as $line) { |
140
|
|
|
# fenced elements |
141
|
|
|
|
142
|
|
|
switch ($element['type']) { |
143
|
|
|
case 'fenced block': |
144
|
|
|
|
145
|
|
|
if (!isset($element['closed'])) { |
146
|
|
|
if (preg_match('/^[ ]*' . $element['fence'][0] . '{3,}[ ]*$/', $line)) { |
147
|
|
|
$element['closed'] = true; |
148
|
|
|
} else { |
149
|
|
|
'' !== $element['text'] and $element['text'] .= "\n"; |
150
|
|
|
|
151
|
|
|
$element['text'] .= $line; |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
continue 2; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
break; |
158
|
|
|
case 'block-level markup': |
159
|
|
|
|
160
|
|
|
if (!isset($element['closed'])) { |
161
|
|
|
if (false !== mb_strpos($line, $element['start'])) { # opening tag |
162
|
|
|
$element['depth']++; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
if (false !== mb_strpos($line, $element['end'])) { # closing tag |
166
|
|
|
$element['depth'] > 0 ? $element['depth']-- : $element['closed'] = true; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
$element['text'] .= "\n" . $line; |
170
|
|
|
|
171
|
|
|
continue 2; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
break; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
# * |
178
|
|
|
|
179
|
|
|
$deindented_line = ltrim($line); |
180
|
|
|
|
181
|
|
|
if ('' === $deindented_line) { |
182
|
|
|
$element['interrupted'] = true; |
183
|
|
|
|
184
|
|
|
continue; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
# composite elements |
188
|
|
|
|
189
|
|
|
switch ($element['type']) { |
190
|
|
|
case 'blockquote': |
191
|
|
|
|
192
|
|
|
if (!isset($element['interrupted'])) { |
193
|
|
|
$line = preg_replace('/^[ ]*>[ ]?/', '', $line); |
194
|
|
|
|
195
|
|
|
$element['lines'] [] = $line; |
196
|
|
|
|
197
|
|
|
continue 2; |
198
|
|
|
} |
199
|
|
|
|
200
|
|
|
break; |
201
|
|
|
case 'li': |
202
|
|
|
|
203
|
|
|
if (preg_match('/^([ ]{0,3})(\d+[.]|[*+-])[ ](.*)/', $line, $matches)) { |
204
|
|
|
if ($element['indentation'] !== $matches[1]) { |
205
|
|
|
$element['lines'] [] = $line; |
206
|
|
|
} else { |
207
|
|
|
unset($element['last']); |
208
|
|
|
|
209
|
|
|
$elements [] = $element; |
210
|
|
|
|
211
|
|
|
$element = [ |
212
|
|
|
'type' => 'li', |
213
|
|
|
'indentation' => $matches[1], |
214
|
|
|
'last' => true, |
215
|
|
|
'lines' => [ |
216
|
|
|
preg_replace('/^[ ]{0,4}/', '', $matches[3]), |
217
|
|
|
], |
218
|
|
|
]; |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
continue 2; |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
if (isset($element['interrupted'])) { |
225
|
|
|
if (' ' === $line[0]) { |
226
|
|
|
$element['lines'] [] = ''; |
227
|
|
|
|
228
|
|
|
$line = preg_replace('/^[ ]{0,4}/', '', $line); |
229
|
|
|
|
230
|
|
|
$element['lines'] [] = $line; |
231
|
|
|
|
232
|
|
|
unset($element['interrupted']); |
233
|
|
|
|
234
|
|
|
continue 2; |
235
|
|
|
} |
236
|
|
|
} else { |
237
|
|
|
$line = preg_replace('/^[ ]{0,4}/', '', $line); |
238
|
|
|
|
239
|
|
|
$element['lines'] [] = $line; |
240
|
|
|
|
241
|
|
|
continue 2; |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
break; |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
# indentation sensitive types |
248
|
|
|
|
249
|
|
|
switch ($line[0]) { |
250
|
|
|
case ' ': |
251
|
|
|
|
252
|
|
|
# code block |
253
|
|
|
|
254
|
|
|
if (isset($line[3]) and ' ' === $line[3] and ' ' === $line[2] and ' ' === $line[1]) { |
255
|
|
|
$code_line = mb_substr($line, 4); |
256
|
|
|
|
257
|
|
|
if ('code block' === $element['type']) { |
258
|
|
|
if (isset($element['interrupted'])) { |
259
|
|
|
$element['text'] .= "\n"; |
260
|
|
|
|
261
|
|
|
unset($element['interrupted']); |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
$element['text'] .= "\n" . $code_line; |
265
|
|
|
} else { |
266
|
|
|
$elements [] = $element; |
267
|
|
|
|
268
|
|
|
$element = [ |
269
|
|
|
'type' => 'code block', |
270
|
|
|
'text' => $code_line, |
271
|
|
|
]; |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
continue 2; |
275
|
|
|
} |
276
|
|
|
|
277
|
|
|
break; |
278
|
|
|
case '#': |
279
|
|
|
|
280
|
|
|
# atx heading (#) |
281
|
|
|
|
282
|
|
|
if (isset($line[1])) { |
283
|
|
|
$elements [] = $element; |
284
|
|
|
|
285
|
|
|
$level = 1; |
286
|
|
|
|
287
|
|
|
while (isset($line[$level]) and '#' === $line[$level]) { |
288
|
|
|
++$level; |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
$element = [ |
292
|
|
|
'type' => 'heading', |
293
|
|
|
'text' => trim($line, '# '), |
294
|
|
|
'level' => $level, |
295
|
|
|
]; |
296
|
|
|
|
297
|
|
|
continue 2; |
298
|
|
|
} |
299
|
|
|
|
300
|
|
|
break; |
301
|
|
|
case '-': |
302
|
|
|
case '=': |
303
|
|
|
|
304
|
|
|
# setext heading |
305
|
|
|
|
306
|
|
|
if ('paragraph' === $element['type'] and false === isset($element['interrupted'])) { |
307
|
|
|
$chopped_line = rtrim($line); |
308
|
|
|
|
309
|
|
|
$i = 1; |
310
|
|
|
|
311
|
|
|
while (isset($chopped_line[$i])) { |
312
|
|
|
if ($chopped_line[$i] !== $line[0]) { |
313
|
|
|
break 2; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
++$i; |
317
|
|
|
} |
318
|
|
|
|
319
|
|
|
$element['type'] = 'heading'; |
320
|
|
|
$element['level'] = '-' === $line[0] ? 2 : 1; |
321
|
|
|
|
322
|
|
|
continue 2; |
323
|
|
|
} |
324
|
|
|
|
325
|
|
|
break; |
326
|
|
|
} |
327
|
|
|
|
328
|
|
|
# indentation insensitive types |
329
|
|
|
|
330
|
|
|
switch ($deindented_line[0]) { |
331
|
|
|
case '<': |
332
|
|
|
|
333
|
|
|
$position = mb_strpos($deindented_line, '>'); |
334
|
|
|
|
335
|
|
|
if ($position > 1) { # tag |
336
|
|
|
$name = mb_substr($deindented_line, 1, $position - 1); |
337
|
|
|
$name = rtrim($name); |
338
|
|
|
|
339
|
|
|
if ('/' === mb_substr($name, -1)) { |
340
|
|
|
$self_closing = true; |
341
|
|
|
|
342
|
|
|
$name = mb_substr($name, 0, -1); |
343
|
|
|
} |
344
|
|
|
|
345
|
|
|
$position = mb_strpos($name, ' '); |
346
|
|
|
|
347
|
|
|
if ($position) { |
348
|
|
|
$name = mb_substr($name, 0, $position); |
349
|
|
|
} |
350
|
|
|
|
351
|
|
|
if (!ctype_alpha($name)) { |
352
|
|
|
break; |
353
|
|
|
} |
354
|
|
|
|
355
|
|
|
if (in_array($name, $this->inline_tags, true)) { |
356
|
|
|
break; |
357
|
|
|
} |
358
|
|
|
|
359
|
|
|
$elements [] = $element; |
360
|
|
|
|
361
|
|
|
if (isset($self_closing)) { |
362
|
|
|
$element = [ |
363
|
|
|
'type' => 'self-closing tag', |
364
|
|
|
'text' => $deindented_line, |
365
|
|
|
]; |
366
|
|
|
|
367
|
|
|
unset($self_closing); |
368
|
|
|
|
369
|
|
|
continue 2; |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
$element = [ |
373
|
|
|
'type' => 'block-level markup', |
374
|
|
|
'text' => $deindented_line, |
375
|
|
|
'start' => '<' . $name . '>', |
376
|
|
|
'end' => '</' . $name . '>', |
377
|
|
|
'depth' => 0, |
378
|
|
|
]; |
379
|
|
|
|
380
|
|
|
if (mb_strpos($deindented_line, $element['end'])) { |
381
|
|
|
$element['closed'] = true; |
382
|
|
|
} |
383
|
|
|
|
384
|
|
|
continue 2; |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
break; |
388
|
|
|
case '>': |
389
|
|
|
|
390
|
|
|
# quote |
391
|
|
|
|
392
|
|
|
if (preg_match('/^>[ ]?(.*)/', $deindented_line, $matches)) { |
393
|
|
|
$elements [] = $element; |
394
|
|
|
|
395
|
|
|
$element = [ |
396
|
|
|
'type' => 'blockquote', |
397
|
|
|
'lines' => [ |
398
|
|
|
$matches[1], |
399
|
|
|
], |
400
|
|
|
]; |
401
|
|
|
|
402
|
|
|
continue 2; |
403
|
|
|
} |
404
|
|
|
|
405
|
|
|
break; |
406
|
|
|
case '[': |
407
|
|
|
|
408
|
|
|
# reference |
409
|
|
|
|
410
|
|
|
if (preg_match('/^\[(.+?)\]:[ ]*(.+?)(?:[ ]+[\'"](.+?)[\'"])?[ ]*$/', $deindented_line, $matches)) { |
411
|
|
|
$label = mb_strtolower($matches[1]); |
412
|
|
|
|
413
|
|
|
$this->reference_map[$label] = [ |
414
|
|
|
'»' => trim($matches[2], '<>'), |
415
|
|
|
]; |
416
|
|
|
|
417
|
|
|
if (isset($matches[3])) { |
418
|
|
|
$this->reference_map[$label]['#'] = $matches[3]; |
419
|
|
|
} |
420
|
|
|
|
421
|
|
|
continue 2; |
422
|
|
|
} |
423
|
|
|
|
424
|
|
|
break; |
425
|
|
|
case '`': |
426
|
|
|
case '~': |
427
|
|
|
|
428
|
|
|
# fenced code block |
429
|
|
|
|
430
|
|
|
if (preg_match('/^([`]{3,}|[~]{3,})[ ]*(\S+)?[ ]*$/', $deindented_line, $matches)) { |
431
|
|
|
$elements [] = $element; |
432
|
|
|
|
433
|
|
|
$element = [ |
434
|
|
|
'type' => 'fenced block', |
435
|
|
|
'text' => '', |
436
|
|
|
'fence' => $matches[1], |
437
|
|
|
]; |
438
|
|
|
|
439
|
|
|
isset($matches[2]) and $element['language'] = $matches[2]; |
440
|
|
|
|
441
|
|
|
continue 2; |
442
|
|
|
} |
443
|
|
|
|
444
|
|
|
break; |
445
|
|
|
case '*': |
446
|
|
|
case '+': |
447
|
|
|
case '-': |
448
|
|
|
case '_': |
449
|
|
|
|
450
|
|
|
# hr |
451
|
|
|
|
452
|
|
|
if (preg_match('/^([-*_])([ ]{0,2}\1){2,}[ ]*$/', $deindented_line)) { |
453
|
|
|
$elements [] = $element; |
454
|
|
|
|
455
|
|
|
$element = [ |
456
|
|
|
'type' => 'hr', |
457
|
|
|
]; |
458
|
|
|
|
459
|
|
|
continue 2; |
460
|
|
|
} |
461
|
|
|
|
462
|
|
|
# li |
463
|
|
|
|
464
|
|
|
if (preg_match('/^([ ]*)[*+-][ ](.*)/', $line, $matches)) { |
465
|
|
|
$elements [] = $element; |
466
|
|
|
|
467
|
|
|
$element = [ |
468
|
|
|
'type' => 'li', |
469
|
|
|
'ordered' => false, |
470
|
|
|
'indentation' => $matches[1], |
471
|
|
|
'last' => true, |
472
|
|
|
'lines' => [ |
473
|
|
|
preg_replace('/^[ ]{0,4}/', '', $matches[2]), |
474
|
|
|
], |
475
|
|
|
]; |
476
|
|
|
|
477
|
|
|
continue 2; |
478
|
|
|
} |
479
|
|
|
} |
480
|
|
|
|
481
|
|
|
# li |
482
|
|
|
|
483
|
|
|
if ($deindented_line[0] <= '9' and $deindented_line[0] >= '0' and preg_match('/^([ ]*)\d+[.][ ](.*)/', $line, $matches)) { |
484
|
|
|
$elements [] = $element; |
485
|
|
|
|
486
|
|
|
$element = [ |
487
|
|
|
'type' => 'li', |
488
|
|
|
'ordered' => true, |
489
|
|
|
'indentation' => $matches[1], |
490
|
|
|
'last' => true, |
491
|
|
|
'lines' => [ |
492
|
|
|
preg_replace('/^[ ]{0,4}/', '', $matches[2]), |
493
|
|
|
], |
494
|
|
|
]; |
495
|
|
|
|
496
|
|
|
continue; |
497
|
|
|
} |
498
|
|
|
|
499
|
|
|
# paragraph |
500
|
|
|
|
501
|
|
|
if ('paragraph' === $element['type']) { |
502
|
|
|
if (isset($element['interrupted'])) { |
503
|
|
|
$elements [] = $element; |
504
|
|
|
|
505
|
|
|
$element['text'] = $line; |
506
|
|
|
|
507
|
|
|
unset($element['interrupted']); |
508
|
|
|
} else { |
509
|
|
|
$element['text'] .= "\n" . $line; |
510
|
|
|
} |
511
|
|
|
} else { |
512
|
|
|
$elements [] = $element; |
513
|
|
|
|
514
|
|
|
$element = [ |
515
|
|
|
'type' => 'paragraph', |
516
|
|
|
'text' => $line, |
517
|
|
|
]; |
518
|
|
|
} |
519
|
|
|
} |
520
|
|
|
|
521
|
|
|
$elements [] = $element; |
522
|
|
|
|
523
|
|
|
unset($elements[0]); |
524
|
|
|
|
525
|
|
|
# |
526
|
|
|
# ~ |
527
|
|
|
# |
528
|
|
|
|
529
|
|
|
$markup = ''; |
530
|
|
|
|
531
|
|
|
foreach ($elements as $element) { |
532
|
|
|
switch ($element['type']) { |
533
|
|
|
case 'paragraph': |
534
|
|
|
|
535
|
|
|
$text = $this->parse_span_elements($element['text']); |
536
|
|
|
|
537
|
|
|
if ('li' === $context and '' === $markup) { |
538
|
|
|
if (isset($element['interrupted'])) { |
539
|
|
|
$markup .= "\n" . '<p>' . $text . '</p>' . "\n"; |
540
|
|
|
} else { |
541
|
|
|
$markup .= $text; |
542
|
|
|
} |
543
|
|
|
} else { |
544
|
|
|
$markup .= '<p>' . $text . '</p>' . "\n"; |
545
|
|
|
} |
546
|
|
|
|
547
|
|
|
break; |
548
|
|
|
case 'blockquote': |
549
|
|
|
|
550
|
|
|
$text = $this->parse_block_elements($element['lines']); |
551
|
|
|
|
552
|
|
|
$markup .= '<blockquote>' . "\n" . $text . '</blockquote>' . "\n"; |
553
|
|
|
|
554
|
|
|
break; |
555
|
|
|
case 'code block': |
556
|
|
|
|
557
|
|
|
$text = htmlspecialchars($element['text'], ENT_QUOTES | ENT_NOQUOTES, 'UTF-8'); |
558
|
|
|
|
559
|
|
|
false !== mb_strpos($text, "\x1A\\") and $text = strtr($text, $this->escape_sequence_map); |
560
|
|
|
|
561
|
|
|
$markup .= isset($element['language']) ? '<pre><code class="language-' . $element['language'] . '">' . $text . '</code></pre>' : '<pre><code>' . $text . '</code></pre>'; |
562
|
|
|
|
563
|
|
|
$markup .= "\n"; |
564
|
|
|
|
565
|
|
|
break; |
566
|
|
|
case 'fenced block': |
567
|
|
|
|
568
|
|
|
$text = $element['text']; |
569
|
|
|
|
570
|
|
|
false !== mb_strpos($text, "\x1A\\") and $text = strtr($text, $this->escape_sequence_map); |
571
|
|
|
|
572
|
|
|
$markup .= rex_highlight_string($text, true) . "\n"; |
|
|
|
|
573
|
|
|
|
574
|
|
|
$markup .= "\n"; |
575
|
|
|
|
576
|
|
|
break; |
577
|
|
|
case 'heading': |
578
|
|
|
|
579
|
|
|
$text = $this->parse_span_elements($element['text']); |
580
|
|
|
|
581
|
|
|
$markup .= '<h' . $element['level'] . '>' . $text . '</h' . $element['level'] . '>' . "\n"; |
582
|
|
|
|
583
|
|
|
break; |
584
|
|
|
case 'hr': |
585
|
|
|
|
586
|
|
|
$markup .= '<hr >' . "\n"; |
587
|
|
|
|
588
|
|
|
break; |
589
|
|
|
case 'li': |
590
|
|
|
|
591
|
|
|
if (isset($element['ordered'])) { # first |
592
|
|
|
$list_type = $element['ordered'] ? 'ol' : 'ul'; |
593
|
|
|
|
594
|
|
|
$markup .= '<' . $list_type . '>' . "\n"; |
595
|
|
|
} |
596
|
|
|
|
597
|
|
|
if (isset($element['interrupted']) and !isset($element['last'])) { |
598
|
|
|
$element['lines'] [] = ''; |
599
|
|
|
} |
600
|
|
|
|
601
|
|
|
$text = $this->parse_block_elements($element['lines'], 'li'); |
602
|
|
|
|
603
|
|
|
$markup .= '<li>' . $text . '</li>' . "\n"; |
604
|
|
|
|
605
|
|
|
isset($element['last']) and $markup .= '</' . $list_type . '>' . "\n"; |
|
|
|
|
606
|
|
|
|
607
|
|
|
break; |
608
|
|
|
case 'block-level markup': |
609
|
|
|
|
610
|
|
|
$markup .= $element['text'] . "\n"; |
611
|
|
|
|
612
|
|
|
break; |
613
|
|
|
default: |
614
|
|
|
|
615
|
|
|
$markup .= $element['text'] . "\n"; |
616
|
|
|
} |
617
|
|
|
} |
618
|
|
|
|
619
|
|
|
return $markup; |
620
|
|
|
} |
621
|
|
|
|
622
|
|
|
/** |
623
|
|
|
* @param $text |
624
|
|
|
* @param array $markers |
625
|
|
|
* @return string |
626
|
|
|
*/ |
627
|
|
|
private function parse_span_elements($text, $markers = ['![', '&', '*', '<', '[', '_', '`', 'http', '~~']) |
628
|
|
|
{ |
629
|
|
|
if (false === isset($text[2]) or $markers === []) { |
630
|
|
|
return $text; |
631
|
|
|
} |
632
|
|
|
|
633
|
|
|
# ~ |
634
|
|
|
|
635
|
|
|
$markup = ''; |
636
|
|
|
|
637
|
|
|
while ($markers) { |
|
|
|
|
638
|
|
|
$closest_marker = null; |
639
|
|
|
$closest_marker_index = 0; |
640
|
|
|
$closest_marker_position = null; |
641
|
|
|
|
642
|
|
|
foreach ($markers as $index => $marker) { |
643
|
|
|
$marker_position = mb_strpos($text, $marker); |
644
|
|
|
|
645
|
|
|
if (false === $marker_position) { |
646
|
|
|
unset($markers[$index]); |
647
|
|
|
|
648
|
|
|
continue; |
649
|
|
|
} |
650
|
|
|
|
651
|
|
|
if (null === $closest_marker or $marker_position < $closest_marker_position) { |
652
|
|
|
$closest_marker = $marker; |
653
|
|
|
$closest_marker_index = $index; |
654
|
|
|
$closest_marker_position = $marker_position; |
655
|
|
|
} |
656
|
|
|
} |
657
|
|
|
|
658
|
|
|
# ~ |
659
|
|
|
|
660
|
|
|
if (null === $closest_marker or false === isset($text[$closest_marker_position + 2])) { |
661
|
|
|
$markup .= $text; |
662
|
|
|
|
663
|
|
|
break; |
664
|
|
|
} |
665
|
|
|
$markup .= mb_substr($text, 0, $closest_marker_position); |
666
|
|
|
|
667
|
|
|
$text = mb_substr($text, $closest_marker_position); |
|
|
|
|
668
|
|
|
|
669
|
|
|
# ~ |
670
|
|
|
|
671
|
|
|
unset($markers[$closest_marker_index]); |
672
|
|
|
|
673
|
|
|
# ~ |
674
|
|
|
|
675
|
|
|
switch ($closest_marker) { |
676
|
|
|
case '![': |
677
|
|
|
case '[': |
678
|
|
|
|
679
|
|
|
if (mb_strpos($text, ']') and preg_match('/\[((?:[^][]|(?R))*)\]/', $text, $matches)) { |
680
|
|
|
$element = [ |
681
|
|
|
'!' => '!' === $text[0], |
682
|
|
|
'a' => $matches[1], |
683
|
|
|
]; |
684
|
|
|
|
685
|
|
|
$offset = mb_strlen($matches[0]); |
686
|
|
|
|
687
|
|
|
$element['!'] and ++$offset; |
688
|
|
|
|
689
|
|
|
$remaining_text = mb_substr($text, $offset); |
690
|
|
|
|
691
|
|
|
if ('(' === $remaining_text[0] and preg_match('/\([ ]*(.*?)(?:[ ]+[\'"](.+?)[\'"])?[ ]*\)/', $remaining_text, $matches)) { |
692
|
|
|
$element['»'] = $matches[1]; |
693
|
|
|
|
694
|
|
|
if (isset($matches[2])) { |
695
|
|
|
$element['#'] = $matches[2]; |
696
|
|
|
} |
697
|
|
|
|
698
|
|
|
$offset += mb_strlen($matches[0]); |
699
|
|
|
} elseif ($this->reference_map) { |
|
|
|
|
700
|
|
|
$reference = $element['a']; |
701
|
|
|
|
702
|
|
|
if (preg_match('/^\s*\[(.*?)\]/', $remaining_text, $matches)) { |
703
|
|
|
$reference = $matches[1] ?: $element['a']; |
704
|
|
|
|
705
|
|
|
$offset += mb_strlen($matches[0]); |
706
|
|
|
} |
707
|
|
|
|
708
|
|
|
$reference = mb_strtolower($reference); |
709
|
|
|
|
710
|
|
|
if (isset($this->reference_map[$reference])) { |
711
|
|
|
$element['»'] = $this->reference_map[$reference]['»']; |
712
|
|
|
|
713
|
|
|
if (isset($this->reference_map[$reference]['#'])) { |
714
|
|
|
$element['#'] = $this->reference_map[$reference]['#']; |
715
|
|
|
} |
716
|
|
|
} else { |
717
|
|
|
unset($element); |
718
|
|
|
} |
719
|
|
|
} else { |
720
|
|
|
unset($element); |
721
|
|
|
} |
722
|
|
|
} |
723
|
|
|
|
724
|
|
|
if (isset($element)) { |
725
|
|
|
$element['»'] = str_replace('&', '&', $element['»']); |
726
|
|
|
$element['»'] = str_replace('<', '<', $element['»']); |
727
|
|
|
|
728
|
|
|
if ($element['!']) { |
729
|
|
|
$markup .= '<img alt="' . $element['a'] . '" src="' . $element['»'] . '" >'; |
730
|
|
|
} else { |
731
|
|
|
$element['a'] = $this->parse_span_elements($element['a'], $markers); |
732
|
|
|
|
733
|
|
|
$markup .= isset($element['#']) ? '<a href="' . $element['»'] . '" title="' . $element['#'] . '">' . $element['a'] . '</a>' : '<a href="' . $element['»'] . '">' . $element['a'] . '</a>'; |
734
|
|
|
} |
735
|
|
|
|
736
|
|
|
unset($element); |
737
|
|
|
} else { |
738
|
|
|
$markup .= $closest_marker; |
739
|
|
|
|
740
|
|
|
$offset = '![' === $closest_marker ? 2 : 1; |
741
|
|
|
} |
742
|
|
|
|
743
|
|
|
break; |
744
|
|
|
case '&': |
745
|
|
|
|
746
|
|
|
if (preg_match('/^&#?\w+;/', $text, $matches)) { |
747
|
|
|
$markup .= $matches[0]; |
748
|
|
|
|
749
|
|
|
$offset = mb_strlen($matches[0]); |
750
|
|
|
} else { |
751
|
|
|
$markup .= '&'; |
752
|
|
|
|
753
|
|
|
$offset = 1; |
754
|
|
|
} |
755
|
|
|
|
756
|
|
|
break; |
757
|
|
|
case '*': |
758
|
|
|
case '_': |
759
|
|
|
|
760
|
|
|
if ($text[1] === $closest_marker and preg_match($this->strong_regex[$closest_marker], $text, $matches)) { |
761
|
|
|
$matches[1] = $this->parse_span_elements($matches[1], $markers); |
762
|
|
|
|
763
|
|
|
$markup .= '<strong>' . $matches[1] . '</strong>'; |
764
|
|
|
} elseif (preg_match($this->em_regex[$closest_marker], $text, $matches)) { |
765
|
|
|
$matches[1] = $this->parse_span_elements($matches[1], $markers); |
766
|
|
|
|
767
|
|
|
$markup .= '<em>' . $matches[1] . '</em>'; |
768
|
|
|
} elseif ($text[1] === $closest_marker and preg_match($this->strong_em_regex[$closest_marker], $text, $matches)) { |
769
|
|
|
$matches[2] = $this->parse_span_elements($matches[2], $markers); |
770
|
|
|
|
771
|
|
|
$matches[1] and $matches[1] = $this->parse_span_elements($matches[1], $markers); |
772
|
|
|
$matches[3] and $matches[3] = $this->parse_span_elements($matches[3], $markers); |
773
|
|
|
|
774
|
|
|
$markup .= '<strong>' . $matches[1] . '<em>' . $matches[2] . '</em>' . $matches[3] . '</strong>'; |
775
|
|
|
} elseif (preg_match($this->em_strong_regex[$closest_marker], $text, $matches)) { |
776
|
|
|
$matches[2] = $this->parse_span_elements($matches[2], $markers); |
777
|
|
|
|
778
|
|
|
$matches[1] and $matches[1] = $this->parse_span_elements($matches[1], $markers); |
779
|
|
|
$matches[3] and $matches[3] = $this->parse_span_elements($matches[3], $markers); |
780
|
|
|
|
781
|
|
|
$markup .= '<em>' . $matches[1] . '<strong>' . $matches[2] . '</strong>' . $matches[3] . '</em>'; |
782
|
|
|
} |
783
|
|
|
|
784
|
|
|
if (isset($matches) and $matches) { |
785
|
|
|
$offset = mb_strlen($matches[0]); |
786
|
|
|
} else { |
787
|
|
|
$markup .= $closest_marker; |
788
|
|
|
|
789
|
|
|
$offset = 1; |
790
|
|
|
} |
791
|
|
|
|
792
|
|
|
break; |
793
|
|
|
case '<': |
794
|
|
|
|
795
|
|
|
if (false !== mb_strpos($text, '>')) { |
796
|
|
|
if ('h' === $text[1] and preg_match('/^<(https?:[\/]{2}\S+?)>/i', $text, $matches)) { |
797
|
|
|
$element_url = $matches[1]; |
798
|
|
|
$element_url = str_replace('&', '&', $element_url); |
799
|
|
|
$element_url = str_replace('<', '<', $element_url); |
800
|
|
|
|
801
|
|
|
$markup .= '<a href="' . $element_url . '">' . $element_url . '</a>'; |
802
|
|
|
|
803
|
|
|
$offset = mb_strlen($matches[0]); |
804
|
|
|
} elseif (preg_match('/^<\/?\w.*?>/', $text, $matches)) { |
805
|
|
|
$markup .= $matches[0]; |
806
|
|
|
|
807
|
|
|
$offset = mb_strlen($matches[0]); |
808
|
|
|
} else { |
809
|
|
|
$markup .= '<'; |
810
|
|
|
|
811
|
|
|
$offset = 1; |
812
|
|
|
} |
813
|
|
|
} else { |
814
|
|
|
$markup .= '<'; |
815
|
|
|
|
816
|
|
|
$offset = 1; |
817
|
|
|
} |
818
|
|
|
|
819
|
|
|
break; |
820
|
|
|
case '`': |
821
|
|
|
|
822
|
|
|
if (preg_match('/^`(.+?)`/', $text, $matches)) { |
823
|
|
|
$element_text = $matches[1]; |
824
|
|
|
$element_text = htmlspecialchars($element_text, ENT_QUOTES | ENT_NOQUOTES, 'UTF-8'); |
825
|
|
|
|
826
|
|
|
if ($this->escape_sequence_map and false !== mb_strpos($element_text, "\x1A")) { |
827
|
|
|
$element_text = strtr($element_text, $this->escape_sequence_map); |
828
|
|
|
} |
829
|
|
|
|
830
|
|
|
$markup .= '<code>' . $element_text . '</code>'; |
831
|
|
|
|
832
|
|
|
$offset = mb_strlen($matches[0]); |
833
|
|
|
} else { |
834
|
|
|
$markup .= '`'; |
835
|
|
|
|
836
|
|
|
$offset = 1; |
837
|
|
|
} |
838
|
|
|
|
839
|
|
|
break; |
840
|
|
|
case 'http': |
841
|
|
|
|
842
|
|
|
if (preg_match('/^https?:[\/]{2}\S+\b/i', $text, $matches)) { |
843
|
|
|
$element_url = $matches[0]; |
844
|
|
|
$element_url = str_replace('&', '&', $element_url); |
845
|
|
|
$element_url = str_replace('<', '<', $element_url); |
846
|
|
|
|
847
|
|
|
$markup .= '<a href="' . $element_url . '">' . $element_url . '</a>'; |
848
|
|
|
|
849
|
|
|
$offset = mb_strlen($matches[0]); |
850
|
|
|
} else { |
851
|
|
|
$markup .= 'http'; |
852
|
|
|
|
853
|
|
|
$offset = 4; |
854
|
|
|
} |
855
|
|
|
|
856
|
|
|
break; |
857
|
|
|
case '~~': |
858
|
|
|
|
859
|
|
|
if (preg_match('/^~~(?=\S)(.+?)(?<=\S)~~/', $text, $matches)) { |
860
|
|
|
$matches[1] = $this->parse_span_elements($matches[1], $markers); |
861
|
|
|
|
862
|
|
|
$markup .= '<del>' . $matches[1] . '</del>'; |
863
|
|
|
|
864
|
|
|
$offset = mb_strlen($matches[0]); |
865
|
|
|
} else { |
866
|
|
|
$markup .= '~~'; |
867
|
|
|
|
868
|
|
|
$offset = 2; |
869
|
|
|
} |
870
|
|
|
|
871
|
|
|
break; |
872
|
|
|
} |
873
|
|
|
|
874
|
|
|
if (isset($offset)) { |
875
|
|
|
$text = mb_substr($text, $offset); |
876
|
|
|
} |
877
|
|
|
|
878
|
|
|
$markers[$closest_marker_index] = $closest_marker; |
879
|
|
|
} |
880
|
|
|
|
881
|
|
|
$markup = str_replace($this->break_marker, '<br >' . "\n", $markup); |
882
|
|
|
|
883
|
|
|
return $markup; |
884
|
|
|
} |
885
|
|
|
|
886
|
|
|
# |
887
|
|
|
# Read-only |
888
|
|
|
# |
889
|
|
|
|
890
|
|
|
private $inline_tags = [ |
891
|
|
|
'a', |
892
|
|
|
'abbr', |
893
|
|
|
'acronym', |
894
|
|
|
'b', |
895
|
|
|
'bdo', |
896
|
|
|
'big', |
897
|
|
|
'br', |
898
|
|
|
'button', |
899
|
|
|
'cite', |
900
|
|
|
'code', |
901
|
|
|
'dfn', |
902
|
|
|
'em', |
903
|
|
|
'i', |
904
|
|
|
'img', |
905
|
|
|
'input', |
906
|
|
|
'kbd', |
907
|
|
|
'label', |
908
|
|
|
'map', |
909
|
|
|
'object', |
910
|
|
|
'q', |
911
|
|
|
'samp', |
912
|
|
|
'script', |
913
|
|
|
'select', |
914
|
|
|
'small', |
915
|
|
|
'span', |
916
|
|
|
'strong', |
917
|
|
|
'sub', |
918
|
|
|
'sup', |
919
|
|
|
'textarea', |
920
|
|
|
'tt', |
921
|
|
|
'var', |
922
|
|
|
]; |
923
|
|
|
|
924
|
|
|
# ~ |
925
|
|
|
|
926
|
|
|
private $strong_regex = [ |
927
|
|
|
'*' => '/^[*]{2}([^*]+?)[*]{2}(?![*])/s', |
928
|
|
|
'_' => '/^__([^_]+?)__(?!_)/s', |
929
|
|
|
]; |
930
|
|
|
|
931
|
|
|
private $em_regex = [ |
932
|
|
|
'*' => '/^[*]([^*]+?)[*](?![*])/s', |
933
|
|
|
'_' => '/^_([^_]+?)[_](?![_])\b/s', |
934
|
|
|
]; |
935
|
|
|
|
936
|
|
|
private $strong_em_regex = [ |
937
|
|
|
'*' => '/^[*]{2}(.*?)[*](.+?)[*](.*?)[*]{2}/s', |
938
|
|
|
'_' => '/^__(.*?)_(.+?)_(.*?)__/s', |
939
|
|
|
]; |
940
|
|
|
|
941
|
|
|
private $em_strong_regex = [ |
942
|
|
|
'*' => '/^[*](.*?)[*]{2}(.+?)[*]{2}(.*?)[*]/s', |
943
|
|
|
'_' => '/^_(.*?)__(.+?)__(.*?)_/s', |
944
|
|
|
]; |
945
|
|
|
} |
946
|
|
|
|