1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
/* |
6
|
|
|
* This file is part of the league/commonmark package. |
7
|
|
|
* |
8
|
|
|
* (c) Colin O'Dell <[email protected]> |
9
|
|
|
* |
10
|
|
|
* For the full copyright and license information, please view the LICENSE |
11
|
|
|
* file that was distributed with this source code. |
12
|
|
|
*/ |
13
|
|
|
|
14
|
|
|
namespace League\CommonMark; |
15
|
|
|
|
16
|
|
|
class Cursor |
17
|
|
|
{ |
18
|
|
|
public const INDENT_LEVEL = 4; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* @var string |
22
|
|
|
*/ |
23
|
|
|
private $line; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* @var int |
27
|
|
|
*/ |
28
|
|
|
private $length; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* @var int |
32
|
|
|
* |
33
|
|
|
* It's possible for this to be 1 char past the end, meaning we've parsed all chars and have |
34
|
|
|
* reached the end. In this state, any character-returning method MUST return null. |
35
|
|
|
*/ |
36
|
|
|
private $currentPosition = 0; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* @var int |
40
|
|
|
*/ |
41
|
|
|
private $column = 0; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* @var int |
45
|
|
|
*/ |
46
|
|
|
private $indent = 0; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* @var int |
50
|
|
|
*/ |
51
|
|
|
private $previousPosition = 0; |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* @var int|null |
55
|
|
|
*/ |
56
|
|
|
private $nextNonSpaceCache; |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @var bool |
60
|
|
|
*/ |
61
|
|
|
private $partiallyConsumedTab = false; |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* @var string |
65
|
|
|
*/ |
66
|
|
|
private $encoding; |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* @var bool |
70
|
|
|
*/ |
71
|
|
|
private $lineContainsTabs; |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* @var bool |
75
|
|
|
*/ |
76
|
|
|
private $isMultibyte; |
77
|
|
|
|
78
|
|
|
/** |
79
|
|
|
* @var array<int, string> |
80
|
|
|
*/ |
81
|
|
|
private $charCache = []; |
82
|
|
|
|
83
|
|
|
/** |
84
|
|
|
* @param string $line The line being parsed (ASCII or UTF-8) |
85
|
|
|
*/ |
86
|
2526 |
|
public function __construct(string $line) |
87
|
|
|
{ |
88
|
2526 |
|
$this->line = $line; |
89
|
2526 |
|
$this->length = \mb_strlen($line, 'UTF-8') ?: 0; |
90
|
2526 |
|
$this->isMultibyte = $this->length !== \strlen($line); |
91
|
2526 |
|
$this->encoding = $this->isMultibyte ? 'UTF-8' : 'ASCII'; |
92
|
2526 |
|
$this->lineContainsTabs = false !== \strpos($line, "\t"); |
93
|
2526 |
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Returns the position of the next character which is not a space (or tab) |
97
|
|
|
* |
98
|
|
|
* @return int |
99
|
|
|
*/ |
100
|
2244 |
|
public function getNextNonSpacePosition(): int |
101
|
|
|
{ |
102
|
2244 |
|
if ($this->nextNonSpaceCache !== null) { |
103
|
2064 |
|
return $this->nextNonSpaceCache; |
104
|
|
|
} |
105
|
|
|
|
106
|
2244 |
|
$i = $this->currentPosition; |
107
|
2244 |
|
$cols = $this->column; |
108
|
|
|
|
109
|
2244 |
|
while (($c = $this->getCharacter($i)) !== null) { |
110
|
2220 |
|
if ($c === ' ') { |
111
|
510 |
|
$i++; |
112
|
510 |
|
$cols++; |
113
|
2178 |
|
} elseif ($c === "\t") { |
114
|
36 |
|
$i++; |
115
|
36 |
|
$cols += (4 - ($cols % 4)); |
116
|
|
|
} else { |
117
|
2178 |
|
break; |
118
|
|
|
} |
119
|
|
|
} |
120
|
|
|
|
121
|
2244 |
|
$nextNonSpace = ($c === null) ? $this->length : $i; |
122
|
2244 |
|
$this->indent = $cols - $this->column; |
123
|
|
|
|
124
|
2244 |
|
return $this->nextNonSpaceCache = $nextNonSpace; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Returns the next character which isn't a space (or tab) |
129
|
|
|
* |
130
|
|
|
* @return string |
131
|
|
|
*/ |
132
|
2037 |
|
public function getNextNonSpaceCharacter(): ?string |
133
|
|
|
{ |
134
|
2037 |
|
return $this->getCharacter($this->getNextNonSpacePosition()); |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
/** |
138
|
|
|
* Calculates the current indent (number of spaces after current position) |
139
|
|
|
* |
140
|
|
|
* @return int |
141
|
|
|
*/ |
142
|
2124 |
|
public function getIndent(): int |
143
|
|
|
{ |
144
|
2124 |
|
if ($this->nextNonSpaceCache === null) { |
145
|
2124 |
|
$this->getNextNonSpacePosition(); |
146
|
|
|
} |
147
|
|
|
|
148
|
2124 |
|
return $this->indent; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
/** |
152
|
|
|
* Whether the cursor is indented to INDENT_LEVEL |
153
|
|
|
* |
154
|
|
|
* @return bool |
155
|
|
|
*/ |
156
|
2064 |
|
public function isIndented(): bool |
157
|
|
|
{ |
158
|
2064 |
|
return $this->getIndent() >= self::INDENT_LEVEL; |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* @param int|null $index |
163
|
|
|
* |
164
|
|
|
* @return string|null |
165
|
|
|
*/ |
166
|
2298 |
|
public function getCharacter(?int $index = null): ?string |
167
|
|
|
{ |
168
|
2298 |
|
if ($index === null) { |
169
|
2043 |
|
$index = $this->currentPosition; |
170
|
|
|
} |
171
|
|
|
|
172
|
2298 |
|
if (isset($this->charCache[$index])) { |
173
|
2082 |
|
return $this->charCache[$index]; |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
// Index out-of-bounds, or we're at the end |
177
|
2298 |
|
if ($index < 0 || $index >= $this->length) { |
178
|
2148 |
|
return null; |
179
|
|
|
} |
180
|
|
|
|
181
|
2256 |
|
return $this->charCache[$index] = $this->isMultibyte ? |
182
|
99 |
|
\mb_substr($this->line, $index, 1, $this->encoding) : |
183
|
2256 |
|
\substr($this->line, $index, 1); |
184
|
|
|
} |
185
|
|
|
|
186
|
|
|
/** |
187
|
|
|
* Returns the next character (or null, if none) without advancing forwards |
188
|
|
|
* |
189
|
|
|
* @param int $offset |
190
|
|
|
* |
191
|
|
|
* @return string|null |
192
|
|
|
*/ |
193
|
1152 |
|
public function peek(int $offset = 1): ?string |
194
|
|
|
{ |
195
|
1152 |
|
return $this->getCharacter($this->currentPosition + $offset); |
196
|
|
|
} |
197
|
|
|
|
198
|
|
|
/** |
199
|
|
|
* Whether the remainder is blank |
200
|
|
|
* |
201
|
|
|
* @return bool |
202
|
|
|
*/ |
203
|
2082 |
|
public function isBlank(): bool |
204
|
|
|
{ |
205
|
2082 |
|
return $this->nextNonSpaceCache === $this->length || $this->getNextNonSpacePosition() === $this->length; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
/** |
209
|
|
|
* Move the cursor forwards |
210
|
|
|
*/ |
211
|
1011 |
|
public function advance() |
212
|
|
|
{ |
213
|
1011 |
|
$this->advanceBy(1); |
214
|
1011 |
|
} |
215
|
|
|
|
216
|
|
|
/** |
217
|
|
|
* Move the cursor forwards |
218
|
|
|
* |
219
|
|
|
* @param int $characters Number of characters to advance by |
220
|
|
|
* @param bool $advanceByColumns Whether to advance by columns instead of spaces |
221
|
|
|
*/ |
222
|
2370 |
|
public function advanceBy(int $characters, bool $advanceByColumns = false) |
223
|
|
|
{ |
224
|
2370 |
|
if ($characters === 0) { |
225
|
2094 |
|
$this->previousPosition = $this->currentPosition; |
226
|
|
|
|
227
|
2094 |
|
return; |
228
|
|
|
} |
229
|
|
|
|
230
|
2295 |
|
$this->previousPosition = $this->currentPosition; |
231
|
2295 |
|
$this->nextNonSpaceCache = null; |
232
|
|
|
|
233
|
|
|
// Optimization to avoid tab handling logic if we have no tabs |
234
|
2295 |
|
if (!$this->lineContainsTabs || false === \strpos( |
235
|
51 |
|
$nextFewChars = $this->isMultibyte ? |
236
|
6 |
|
\mb_substr($this->line, $this->currentPosition, $characters, $this->encoding) : |
237
|
51 |
|
\substr($this->line, $this->currentPosition, $characters), |
238
|
2295 |
|
"\t")) { |
239
|
2283 |
|
$length = \min($characters, $this->length - $this->currentPosition); |
240
|
2283 |
|
$this->partiallyConsumedTab = false; |
241
|
2283 |
|
$this->currentPosition += $length; |
242
|
2283 |
|
$this->column += $length; |
243
|
|
|
|
244
|
2283 |
|
return; |
245
|
|
|
} |
246
|
|
|
|
247
|
48 |
|
if ($characters === 1 && !empty($nextFewChars)) { |
248
|
18 |
|
$asArray = [$nextFewChars]; |
249
|
42 |
|
} elseif ($this->isMultibyte) { |
250
|
3 |
|
$asArray = \preg_split('//u', $nextFewChars, -1, \PREG_SPLIT_NO_EMPTY); |
251
|
|
|
} else { |
252
|
39 |
|
$asArray = \str_split($nextFewChars); |
253
|
|
|
} |
254
|
|
|
|
255
|
48 |
|
foreach ($asArray as $relPos => $c) { |
256
|
48 |
|
if ($c === "\t") { |
257
|
48 |
|
$charsToTab = 4 - ($this->column % 4); |
258
|
48 |
|
if ($advanceByColumns) { |
259
|
33 |
|
$this->partiallyConsumedTab = $charsToTab > $characters; |
260
|
33 |
|
$charsToAdvance = $charsToTab > $characters ? $characters : $charsToTab; |
261
|
33 |
|
$this->column += $charsToAdvance; |
262
|
33 |
|
$this->currentPosition += $this->partiallyConsumedTab ? 0 : 1; |
263
|
33 |
|
$characters -= $charsToAdvance; |
264
|
|
|
} else { |
265
|
27 |
|
$this->partiallyConsumedTab = false; |
266
|
27 |
|
$this->column += $charsToTab; |
267
|
27 |
|
$this->currentPosition++; |
268
|
48 |
|
$characters--; |
269
|
|
|
} |
270
|
|
|
} else { |
271
|
18 |
|
$this->partiallyConsumedTab = false; |
272
|
18 |
|
$this->currentPosition++; |
273
|
18 |
|
$this->column++; |
274
|
18 |
|
$characters--; |
275
|
|
|
} |
276
|
|
|
|
277
|
48 |
|
if ($characters <= 0) { |
278
|
48 |
|
break; |
279
|
|
|
} |
280
|
|
|
} |
281
|
48 |
|
} |
282
|
|
|
|
283
|
|
|
/** |
284
|
|
|
* Advances the cursor by a single space or tab, if present |
285
|
|
|
* |
286
|
|
|
* @return bool |
287
|
|
|
*/ |
288
|
348 |
|
public function advanceBySpaceOrTab(): bool |
289
|
|
|
{ |
290
|
348 |
|
$character = $this->getCharacter(); |
291
|
|
|
|
292
|
348 |
|
if ($character === ' ' || $character === "\t") { |
293
|
336 |
|
$this->advanceBy(1, true); |
294
|
|
|
|
295
|
336 |
|
return true; |
296
|
|
|
} |
297
|
|
|
|
298
|
258 |
|
return false; |
299
|
|
|
} |
300
|
|
|
|
301
|
|
|
/** |
302
|
|
|
* Parse zero or more space/tab characters |
303
|
|
|
* |
304
|
|
|
* @return int Number of positions moved |
305
|
|
|
*/ |
306
|
1971 |
|
public function advanceToNextNonSpaceOrTab(): int |
307
|
|
|
{ |
308
|
1971 |
|
$newPosition = $this->getNextNonSpacePosition(); |
309
|
1971 |
|
$this->advanceBy($newPosition - $this->currentPosition); |
310
|
1971 |
|
$this->partiallyConsumedTab = false; |
311
|
|
|
|
312
|
1971 |
|
return $this->currentPosition - $this->previousPosition; |
313
|
|
|
} |
314
|
|
|
|
315
|
|
|
/** |
316
|
|
|
* Parse zero or more space characters, including at most one newline. |
317
|
|
|
* |
318
|
|
|
* Tab characters are not parsed with this function. |
319
|
|
|
* |
320
|
|
|
* @return int Number of positions moved |
321
|
|
|
*/ |
322
|
471 |
|
public function advanceToNextNonSpaceOrNewline(): int |
323
|
|
|
{ |
324
|
471 |
|
$matches = []; |
325
|
471 |
|
\preg_match('/^ *(?:\n *)?/', $this->getRemainder(), $matches, \PREG_OFFSET_CAPTURE); |
326
|
|
|
|
327
|
|
|
// [0][0] contains the matched text |
328
|
|
|
// [0][1] contains the index of that match |
329
|
471 |
|
$increment = $matches[0][1] + \strlen($matches[0][0]); |
330
|
|
|
|
331
|
471 |
|
if ($increment === 0) { |
332
|
330 |
|
return 0; |
333
|
|
|
} |
334
|
|
|
|
335
|
324 |
|
$this->advanceBy($increment); |
336
|
|
|
|
337
|
324 |
|
return $this->currentPosition - $this->previousPosition; |
338
|
|
|
} |
339
|
|
|
|
340
|
|
|
/** |
341
|
|
|
* Move the position to the very end of the line |
342
|
|
|
* |
343
|
|
|
* @return int The number of characters moved |
344
|
|
|
*/ |
345
|
84 |
|
public function advanceToEnd(): int |
346
|
|
|
{ |
347
|
84 |
|
$this->previousPosition = $this->currentPosition; |
348
|
84 |
|
$this->nextNonSpaceCache = null; |
349
|
|
|
|
350
|
84 |
|
$this->currentPosition = $this->length; |
351
|
|
|
|
352
|
84 |
|
return $this->currentPosition - $this->previousPosition; |
353
|
|
|
} |
354
|
|
|
|
355
|
|
|
/** |
356
|
|
|
* @return string |
357
|
|
|
*/ |
358
|
2166 |
|
public function getRemainder(): string |
359
|
|
|
{ |
360
|
2166 |
|
if ($this->currentPosition >= $this->length) { |
361
|
711 |
|
return ''; |
362
|
|
|
} |
363
|
|
|
|
364
|
2154 |
|
$prefix = ''; |
365
|
2154 |
|
$position = $this->currentPosition; |
366
|
2154 |
|
if ($this->partiallyConsumedTab) { |
367
|
15 |
|
$position++; |
368
|
15 |
|
$charsToTab = 4 - ($this->column % 4); |
369
|
15 |
|
$prefix = \str_repeat(' ', $charsToTab); |
370
|
|
|
} |
371
|
|
|
|
372
|
2154 |
|
$subString = $this->isMultibyte ? |
373
|
69 |
|
\mb_substr($this->line, $position, null, $this->encoding) : |
374
|
2154 |
|
\substr($this->line, $position); |
375
|
|
|
|
376
|
2154 |
|
return $prefix . $subString; |
377
|
|
|
} |
378
|
|
|
|
379
|
|
|
/** |
380
|
|
|
* @return string |
381
|
|
|
*/ |
382
|
2004 |
|
public function getLine(): string |
383
|
|
|
{ |
384
|
2004 |
|
return $this->line; |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
/** |
388
|
|
|
* @return bool |
389
|
|
|
*/ |
390
|
441 |
|
public function isAtEnd(): bool |
391
|
|
|
{ |
392
|
441 |
|
return $this->currentPosition >= $this->length; |
393
|
|
|
} |
394
|
|
|
|
395
|
|
|
/** |
396
|
|
|
* Try to match a regular expression |
397
|
|
|
* |
398
|
|
|
* Returns the matching text and advances to the end of that match |
399
|
|
|
* |
400
|
|
|
* @param string $regex |
401
|
|
|
* |
402
|
|
|
* @return string|null |
403
|
|
|
*/ |
404
|
2085 |
|
public function match(string $regex): ?string |
405
|
|
|
{ |
406
|
2085 |
|
$subject = $this->getRemainder(); |
407
|
|
|
|
408
|
2085 |
|
if (!\preg_match($regex, $subject, $matches, \PREG_OFFSET_CAPTURE)) { |
409
|
1905 |
|
return null; |
410
|
|
|
} |
411
|
|
|
|
412
|
|
|
// $matches[0][0] contains the matched text |
413
|
|
|
// $matches[0][1] contains the index of that match |
414
|
|
|
|
415
|
2046 |
|
if ($this->isMultibyte) { |
416
|
|
|
// PREG_OFFSET_CAPTURE always returns the byte offset, not the char offset, which is annoying |
417
|
57 |
|
$offset = \mb_strlen(\mb_strcut($subject, 0, $matches[0][1], $this->encoding), $this->encoding); |
418
|
57 |
|
$matchLength = \mb_strlen($matches[0][0], $this->encoding); |
419
|
|
|
} else { |
420
|
1989 |
|
$offset = $matches[0][1]; |
421
|
1989 |
|
$matchLength = \strlen($matches[0][0]); |
422
|
|
|
} |
423
|
|
|
|
424
|
|
|
// [0][0] contains the matched text |
425
|
|
|
// [0][1] contains the index of that match |
426
|
2046 |
|
$this->advanceBy($offset + $matchLength); |
427
|
|
|
|
428
|
2046 |
|
return $matches[0][0]; |
429
|
|
|
} |
430
|
|
|
|
431
|
|
|
/** |
432
|
|
|
* Encapsulates the current state of this cursor in case you need to rollback later. |
433
|
|
|
* |
434
|
|
|
* WARNING: Do not parse or use the return value for ANYTHING except for |
435
|
|
|
* passing it back into restoreState(), as the number of values and their |
436
|
|
|
* contents may change in any future release without warning. |
437
|
|
|
* |
438
|
|
|
* @return array |
439
|
|
|
*/ |
440
|
1089 |
|
public function saveState() |
441
|
|
|
{ |
442
|
|
|
return [ |
443
|
1089 |
|
$this->currentPosition, |
444
|
1089 |
|
$this->previousPosition, |
445
|
1089 |
|
$this->nextNonSpaceCache, |
446
|
1089 |
|
$this->indent, |
447
|
1089 |
|
$this->column, |
448
|
1089 |
|
$this->partiallyConsumedTab, |
449
|
|
|
]; |
450
|
|
|
} |
451
|
|
|
|
452
|
|
|
/** |
453
|
|
|
* Restore the cursor to a previous state. |
454
|
|
|
* |
455
|
|
|
* Pass in the value previously obtained by calling saveState(). |
456
|
|
|
* |
457
|
|
|
* @param array $state |
458
|
|
|
*/ |
459
|
852 |
|
public function restoreState($state) |
460
|
|
|
{ |
461
|
|
|
list( |
462
|
852 |
|
$this->currentPosition, |
463
|
852 |
|
$this->previousPosition, |
464
|
852 |
|
$this->nextNonSpaceCache, |
465
|
852 |
|
$this->indent, |
466
|
852 |
|
$this->column, |
467
|
852 |
|
$this->partiallyConsumedTab, |
468
|
852 |
|
) = $state; |
469
|
852 |
|
} |
470
|
|
|
|
471
|
|
|
/** |
472
|
|
|
* @return int |
473
|
|
|
*/ |
474
|
714 |
|
public function getPosition(): int |
475
|
|
|
{ |
476
|
714 |
|
return $this->currentPosition; |
477
|
|
|
} |
478
|
|
|
|
479
|
|
|
/** |
480
|
|
|
* @return string |
481
|
|
|
*/ |
482
|
999 |
|
public function getPreviousText(): string |
483
|
|
|
{ |
484
|
999 |
|
return \mb_substr($this->line, $this->previousPosition, $this->currentPosition - $this->previousPosition, $this->encoding); |
485
|
|
|
} |
486
|
|
|
|
487
|
|
|
/** |
488
|
|
|
* @param int $start |
489
|
|
|
* @param int|null $length |
490
|
|
|
* |
491
|
|
|
* @return string |
492
|
|
|
*/ |
493
|
405 |
|
public function getSubstring(int $start, ?int $length = null): string |
494
|
|
|
{ |
495
|
405 |
|
if ($this->isMultibyte) { |
496
|
21 |
|
return \mb_substr($this->line, $start, $length, $this->encoding); |
497
|
384 |
|
} elseif ($length !== null) { |
498
|
381 |
|
return \substr($this->line, $start, $length); |
499
|
|
|
} |
500
|
|
|
|
501
|
3 |
|
return \substr($this->line, $start); |
502
|
|
|
} |
503
|
|
|
|
504
|
|
|
/** |
505
|
|
|
* @return int |
506
|
|
|
*/ |
507
|
249 |
|
public function getColumn(): int |
508
|
|
|
{ |
509
|
249 |
|
return $this->column; |
510
|
|
|
} |
511
|
|
|
} |
512
|
|
|
|