1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/* |
4
|
|
|
* This file is part of the league/commonmark package. |
5
|
|
|
* |
6
|
|
|
* (c) Colin O'Dell <[email protected]> |
7
|
|
|
* |
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
9
|
|
|
* file that was distributed with this source code. |
10
|
|
|
*/ |
11
|
|
|
|
12
|
|
|
namespace League\CommonMark; |
13
|
|
|
|
14
|
|
|
class Cursor |
15
|
|
|
{ |
16
|
|
|
const INDENT_LEVEL = 4; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* @var string |
20
|
|
|
*/ |
21
|
|
|
private $line; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* @var int |
25
|
|
|
*/ |
26
|
|
|
private $length; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* @var int |
30
|
|
|
* |
31
|
|
|
* It's possible for this to be 1 char past the end, meaning we've parsed all chars and have |
32
|
|
|
* reached the end. In this state, any character-returning method MUST return null. |
33
|
|
|
*/ |
34
|
|
|
private $currentPosition = 0; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* @var int |
38
|
|
|
*/ |
39
|
|
|
private $column = 0; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @var int |
43
|
|
|
*/ |
44
|
|
|
private $indent = 0; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* @var int |
48
|
|
|
*/ |
49
|
|
|
private $previousPosition = 0; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var int|null |
53
|
|
|
*/ |
54
|
|
|
private $nextNonSpaceCache; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @var bool |
58
|
|
|
*/ |
59
|
|
|
private $partiallyConsumedTab = false; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @var string |
63
|
|
|
*/ |
64
|
|
|
private $encoding; |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @var bool |
68
|
|
|
*/ |
69
|
|
|
private $lineContainsTabs; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @var bool |
73
|
|
|
*/ |
74
|
|
|
private $isMultibyte; |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* @param string $line |
78
|
|
|
*/ |
79
|
2391 |
|
public function __construct($line) |
80
|
|
|
{ |
81
|
2391 |
|
$this->line = $line; |
82
|
2391 |
|
$this->encoding = mb_detect_encoding($line, 'ASCII,UTF-8', true) ?: 'ISO-8859-1'; |
83
|
2391 |
|
$this->length = mb_strlen($line, $this->encoding); |
84
|
2391 |
|
$this->isMultibyte = $this->length !== strlen($line); |
85
|
2391 |
|
$this->lineContainsTabs = preg_match('/\t/', $line) > 0; |
86
|
2391 |
|
} |
87
|
|
|
|
88
|
|
|
/** |
89
|
|
|
* Returns the position of the next character which is not a space (or tab) |
90
|
|
|
* |
91
|
|
|
* @return int |
92
|
|
|
*/ |
93
|
2118 |
|
public function getNextNonSpacePosition() |
94
|
|
|
{ |
95
|
2118 |
|
if ($this->nextNonSpaceCache !== null) { |
96
|
1938 |
|
return $this->nextNonSpaceCache; |
97
|
|
|
} |
98
|
|
|
|
99
|
2118 |
|
$i = $this->currentPosition; |
100
|
2118 |
|
$cols = $this->column; |
101
|
|
|
|
102
|
2118 |
|
while (($c = $this->getCharacter($i)) !== null) { |
103
|
2094 |
|
if ($c === ' ') { |
104
|
504 |
|
$i++; |
105
|
504 |
|
$cols++; |
106
|
2066 |
|
} elseif ($c === "\t") { |
107
|
36 |
|
$i++; |
108
|
36 |
|
$cols += (4 - ($cols % 4)); |
109
|
12 |
|
} else { |
110
|
2052 |
|
break; |
111
|
|
|
} |
112
|
175 |
|
} |
113
|
|
|
|
114
|
2118 |
|
$nextNonSpace = ($c === null) ? $this->length : $i; |
115
|
2118 |
|
$this->indent = $cols - $this->column; |
116
|
|
|
|
117
|
2118 |
|
return $this->nextNonSpaceCache = $nextNonSpace; |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* Returns the next character which isn't a space (or tab) |
122
|
|
|
* |
123
|
|
|
* @return string |
124
|
|
|
*/ |
125
|
1911 |
|
public function getNextNonSpaceCharacter() |
126
|
|
|
{ |
127
|
1911 |
|
return $this->getCharacter($this->getNextNonSpacePosition()); |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
/** |
131
|
|
|
* Calculates the current indent (number of spaces after current position) |
132
|
|
|
* |
133
|
|
|
* @return int |
134
|
|
|
*/ |
135
|
1890 |
|
public function getIndent() |
136
|
|
|
{ |
137
|
1890 |
|
$this->getNextNonSpacePosition(); |
138
|
|
|
|
139
|
1890 |
|
return $this->indent; |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
/** |
143
|
|
|
* Whether the cursor is indented to INDENT_LEVEL |
144
|
|
|
* |
145
|
|
|
* @return bool |
146
|
|
|
*/ |
147
|
1938 |
|
public function isIndented() |
148
|
|
|
{ |
149
|
1938 |
|
$this->getNextNonSpacePosition(); |
150
|
|
|
|
151
|
1938 |
|
return $this->indent >= self::INDENT_LEVEL; |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* @param int|null $index |
156
|
|
|
* |
157
|
|
|
* @return string|null |
158
|
|
|
*/ |
159
|
2172 |
|
public function getCharacter($index = null) |
160
|
|
|
{ |
161
|
2172 |
|
if ($index === null) { |
162
|
1689 |
|
$index = $this->currentPosition; |
163
|
563 |
|
} |
164
|
|
|
|
165
|
|
|
// Index out-of-bounds, or we're at the end |
166
|
2172 |
|
if ($index < 0 || $index >= $this->length) { |
167
|
1893 |
|
return; |
168
|
|
|
} |
169
|
|
|
|
170
|
2130 |
|
return mb_substr($this->line, $index, 1, $this->encoding); |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
/** |
174
|
|
|
* Returns the next character (or null, if none) without advancing forwards |
175
|
|
|
* |
176
|
|
|
* @param int $offset |
177
|
|
|
* |
178
|
|
|
* @return string|null |
179
|
|
|
*/ |
180
|
1014 |
|
public function peek($offset = 1) |
181
|
|
|
{ |
182
|
1014 |
|
return $this->getCharacter($this->currentPosition + $offset); |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
/** |
186
|
|
|
* Whether the remainder is blank |
187
|
|
|
* |
188
|
|
|
* @return bool |
189
|
|
|
*/ |
190
|
1956 |
|
public function isBlank() |
191
|
|
|
{ |
192
|
1956 |
|
return $this->getNextNonSpacePosition() === $this->length; |
193
|
|
|
} |
194
|
|
|
|
195
|
|
|
/** |
196
|
|
|
* Move the cursor forwards |
197
|
|
|
*/ |
198
|
795 |
|
public function advance() |
199
|
|
|
{ |
200
|
795 |
|
$this->advanceBy(1); |
201
|
795 |
|
} |
202
|
|
|
|
203
|
|
|
/** |
204
|
|
|
* Move the cursor forwards |
205
|
|
|
* |
206
|
|
|
* @param int $characters Number of characters to advance by |
207
|
|
|
* @param bool $advanceByColumns Whether to advance by columns instead of spaces |
208
|
|
|
*/ |
209
|
2247 |
|
public function advanceBy($characters, $advanceByColumns = false) |
210
|
|
|
{ |
211
|
2247 |
|
if ($characters === 0) { |
212
|
1968 |
|
$this->previousPosition = $this->currentPosition; |
213
|
|
|
|
214
|
1968 |
|
return; |
215
|
|
|
} |
216
|
|
|
|
217
|
2106 |
|
$this->previousPosition = $this->currentPosition; |
218
|
2106 |
|
$this->nextNonSpaceCache = null; |
219
|
|
|
|
220
|
2106 |
|
$nextFewChars = mb_substr($this->line, $this->currentPosition, $characters, $this->encoding); |
221
|
|
|
|
222
|
|
|
// Optimization to avoid tab handling logic if we have no tabs |
223
|
2106 |
|
if (!$this->lineContainsTabs || preg_match('/\t/', $nextFewChars) === 0) { |
224
|
2094 |
|
$length = min($characters, $this->length - $this->currentPosition); |
225
|
2094 |
|
$this->partiallyConsumedTab = false; |
226
|
2094 |
|
$this->currentPosition += $length; |
227
|
2094 |
|
$this->column += $length; |
228
|
|
|
|
229
|
2094 |
|
return; |
230
|
|
|
} |
231
|
|
|
|
232
|
45 |
|
if ($characters === 1 && !empty($nextFewChars)) { |
233
|
18 |
|
$asArray = [$nextFewChars]; |
234
|
6 |
|
} else { |
235
|
39 |
|
$asArray = preg_split('//u', $nextFewChars, null, PREG_SPLIT_NO_EMPTY); |
236
|
|
|
} |
237
|
|
|
|
238
|
45 |
|
foreach ($asArray as $relPos => $c) { |
239
|
45 |
|
if ($c === "\t") { |
240
|
45 |
|
$charsToTab = 4 - ($this->column % 4); |
241
|
45 |
|
if ($advanceByColumns) { |
242
|
33 |
|
$this->partiallyConsumedTab = $charsToTab > $characters; |
243
|
33 |
|
$charsToAdvance = $charsToTab > $characters ? $characters : $charsToTab; |
244
|
33 |
|
$this->column += $charsToAdvance; |
245
|
33 |
|
$this->currentPosition += $this->partiallyConsumedTab ? 0 : 1; |
246
|
33 |
|
$characters -= $charsToAdvance; |
247
|
11 |
|
} else { |
248
|
18 |
|
$this->partiallyConsumedTab = false; |
249
|
18 |
|
$this->column += $charsToTab; |
250
|
18 |
|
$this->currentPosition++; |
251
|
36 |
|
$characters--; |
252
|
|
|
} |
253
|
15 |
|
} else { |
254
|
12 |
|
$this->partiallyConsumedTab = false; |
255
|
12 |
|
$this->currentPosition++; |
256
|
12 |
|
$this->column++; |
257
|
12 |
|
$characters--; |
258
|
|
|
} |
259
|
|
|
|
260
|
45 |
|
if ($characters <= 0) { |
261
|
45 |
|
break; |
262
|
|
|
} |
263
|
15 |
|
} |
264
|
45 |
|
} |
265
|
|
|
|
266
|
|
|
/** |
267
|
|
|
* Advances the cursor by a single space or tab, if present |
268
|
|
|
* |
269
|
|
|
* @return bool |
270
|
|
|
*/ |
271
|
339 |
|
public function advanceBySpaceOrTab() |
272
|
|
|
{ |
273
|
339 |
|
$character = $this->getCharacter(); |
274
|
|
|
|
275
|
339 |
|
if ($character === ' ' || $character === "\t") { |
276
|
327 |
|
$this->advanceBy(1, true); |
277
|
|
|
|
278
|
327 |
|
return true; |
279
|
|
|
} |
280
|
|
|
|
281
|
249 |
|
return false; |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
/** |
285
|
|
|
* Parse zero or more space/tab characters |
286
|
|
|
* |
287
|
|
|
* @return int Number of positions moved |
288
|
|
|
*/ |
289
|
1848 |
|
public function advanceToNextNonSpaceOrTab() |
290
|
|
|
{ |
291
|
1848 |
|
$newPosition = $this->getNextNonSpacePosition(); |
292
|
1848 |
|
$this->advanceBy($newPosition - $this->currentPosition); |
293
|
1848 |
|
$this->partiallyConsumedTab = false; |
294
|
|
|
|
295
|
1848 |
|
return $this->currentPosition - $this->previousPosition; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
/** |
299
|
|
|
* Parse zero or more space characters, including at most one newline. |
300
|
|
|
* |
301
|
|
|
* Tab characters are not parsed with this function. |
302
|
|
|
* |
303
|
|
|
* @return int Number of positions moved |
304
|
|
|
*/ |
305
|
441 |
|
public function advanceToNextNonSpaceOrNewline() |
306
|
|
|
{ |
307
|
441 |
|
$matches = []; |
308
|
441 |
|
preg_match('/^ *(?:\n *)?/', $this->getRemainder(), $matches, PREG_OFFSET_CAPTURE); |
309
|
|
|
|
310
|
|
|
// [0][0] contains the matched text |
311
|
|
|
// [0][1] contains the index of that match |
312
|
441 |
|
$increment = $matches[0][1] + strlen($matches[0][0]); |
313
|
|
|
|
314
|
441 |
|
if ($increment === 0) { |
315
|
300 |
|
return 0; |
316
|
|
|
} |
317
|
|
|
|
318
|
300 |
|
$this->advanceBy($increment); |
319
|
|
|
|
320
|
300 |
|
return $this->currentPosition - $this->previousPosition; |
321
|
|
|
} |
322
|
|
|
|
323
|
|
|
/** |
324
|
|
|
* Move the position to the very end of the line |
325
|
|
|
* |
326
|
|
|
* @return int The number of characters moved |
327
|
|
|
*/ |
328
|
84 |
|
public function advanceToEnd() |
329
|
|
|
{ |
330
|
84 |
|
$this->previousPosition = $this->currentPosition; |
331
|
84 |
|
$this->nextNonSpaceCache = null; |
332
|
|
|
|
333
|
84 |
|
$this->currentPosition = $this->length; |
334
|
|
|
|
335
|
84 |
|
return $this->currentPosition - $this->previousPosition; |
336
|
|
|
} |
337
|
|
|
|
338
|
|
|
/** |
339
|
|
|
* @return string |
340
|
|
|
*/ |
341
|
2043 |
|
public function getRemainder() |
342
|
|
|
{ |
343
|
2043 |
|
if ($this->currentPosition >= $this->length) { |
344
|
693 |
|
return ''; |
345
|
|
|
} |
346
|
|
|
|
347
|
2028 |
|
$prefix = ''; |
348
|
2028 |
|
$position = $this->currentPosition; |
349
|
2028 |
|
if ($this->partiallyConsumedTab) { |
350
|
15 |
|
$position++; |
351
|
15 |
|
$charsToTab = 4 - ($this->column % 4); |
352
|
15 |
|
$prefix = str_repeat(' ', $charsToTab); |
353
|
5 |
|
} |
354
|
|
|
|
355
|
2028 |
|
return $prefix . mb_substr($this->line, $position, null, $this->encoding); |
356
|
|
|
} |
357
|
|
|
|
358
|
|
|
/** |
359
|
|
|
* @return string |
360
|
|
|
*/ |
361
|
1887 |
|
public function getLine() |
362
|
|
|
{ |
363
|
1887 |
|
return $this->line; |
364
|
|
|
} |
365
|
|
|
|
366
|
|
|
/** |
367
|
|
|
* @return bool |
368
|
|
|
*/ |
369
|
411 |
|
public function isAtEnd() |
370
|
|
|
{ |
371
|
411 |
|
return $this->currentPosition >= $this->length; |
372
|
|
|
} |
373
|
|
|
|
374
|
|
|
/** |
375
|
|
|
* Try to match a regular expression |
376
|
|
|
* |
377
|
|
|
* Returns the matching text and advances to the end of that match |
378
|
|
|
* |
379
|
|
|
* @param string $regex |
380
|
|
|
* |
381
|
|
|
* @return string|null |
382
|
|
|
*/ |
383
|
1908 |
|
public function match($regex) |
384
|
|
|
{ |
385
|
1908 |
|
$subject = $this->getRemainder(); |
386
|
|
|
|
387
|
1908 |
|
if (!preg_match($regex, $subject, $matches, PREG_OFFSET_CAPTURE)) { |
388
|
1758 |
|
return; |
389
|
|
|
} |
390
|
|
|
|
391
|
|
|
// $matches[0][0] contains the matched text |
392
|
|
|
// $matches[0][1] contains the index of that match |
393
|
|
|
|
394
|
1782 |
|
if ($this->isMultibyte) { |
395
|
|
|
// PREG_OFFSET_CAPTURE always returns the byte offset, not the char offset, which is annoying |
396
|
51 |
|
$offset = mb_strlen(mb_strcut($subject, 0, $matches[0][1], $this->encoding), $this->encoding); |
397
|
17 |
|
} else { |
398
|
1731 |
|
$offset = $matches[0][1]; |
399
|
|
|
} |
400
|
|
|
|
401
|
|
|
// [0][0] contains the matched text |
402
|
|
|
// [0][1] contains the index of that match |
403
|
1782 |
|
$this->advanceBy($offset + mb_strlen($matches[0][0], $this->encoding)); |
404
|
|
|
|
405
|
1782 |
|
return $matches[0][0]; |
406
|
|
|
} |
407
|
|
|
|
408
|
|
|
/** |
409
|
|
|
* Encapsulates the current state of this cursor in case you need to rollback later. |
410
|
|
|
* |
411
|
|
|
* WARNING: Do not parse or use the return value for ANYTHING except for |
412
|
|
|
* passing it back into restoreState(), as the number of values and their |
413
|
|
|
* contents may change in any future release without warning. |
414
|
|
|
* |
415
|
|
|
* @return array |
416
|
|
|
*/ |
417
|
1011 |
|
public function saveState() |
418
|
|
|
{ |
419
|
|
|
return [ |
420
|
1011 |
|
$this->currentPosition, |
421
|
1011 |
|
$this->previousPosition, |
422
|
1011 |
|
$this->nextNonSpaceCache, |
423
|
1011 |
|
$this->indent, |
424
|
1011 |
|
$this->column, |
425
|
1011 |
|
$this->partiallyConsumedTab, |
426
|
337 |
|
]; |
427
|
|
|
} |
428
|
|
|
|
429
|
|
|
/** |
430
|
|
|
* Restore the cursor to a previous state. |
431
|
|
|
* |
432
|
|
|
* Pass in the value previously obtained by calling saveState(). |
433
|
|
|
* |
434
|
|
|
* @param array $state |
435
|
|
|
*/ |
436
|
795 |
|
public function restoreState($state) |
437
|
|
|
{ |
438
|
|
|
list( |
439
|
795 |
|
$this->currentPosition, |
440
|
795 |
|
$this->previousPosition, |
441
|
795 |
|
$this->nextNonSpaceCache, |
442
|
795 |
|
$this->indent, |
443
|
795 |
|
$this->column, |
444
|
795 |
|
$this->partiallyConsumedTab, |
445
|
795 |
|
) = $state; |
446
|
795 |
|
} |
447
|
|
|
|
448
|
|
|
/** |
449
|
|
|
* @return int |
450
|
|
|
*/ |
451
|
648 |
|
public function getPosition() |
452
|
|
|
{ |
453
|
648 |
|
return $this->currentPosition; |
454
|
|
|
} |
455
|
|
|
|
456
|
|
|
/** |
457
|
|
|
* @return string |
458
|
|
|
*/ |
459
|
870 |
|
public function getPreviousText() |
460
|
|
|
{ |
461
|
870 |
|
return mb_substr($this->line, $this->previousPosition, $this->currentPosition - $this->previousPosition, $this->encoding); |
462
|
|
|
} |
463
|
|
|
|
464
|
|
|
/** |
465
|
|
|
* @return int |
466
|
|
|
*/ |
467
|
240 |
|
public function getColumn() |
468
|
|
|
{ |
469
|
240 |
|
return $this->column; |
470
|
|
|
} |
471
|
|
|
} |
472
|
|
|
|