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
|
|
|
* @var array<int, string> |
78
|
|
|
*/ |
79
|
|
|
private $charCache = []; |
80
|
|
|
|
81
|
|
|
/** |
82
|
|
|
* @param string $line The line being parsed |
83
|
|
|
* @param string $encoding The encoding of that line |
84
|
|
|
*/ |
85
|
2478 |
|
public function __construct(string $line, string $encoding = 'UTF-8') |
86
|
|
|
{ |
87
|
2478 |
|
$this->line = $line; |
88
|
2478 |
|
$this->encoding = $encoding; |
89
|
2478 |
|
$this->length = \mb_strlen($line, $this->encoding); |
90
|
2478 |
|
$this->isMultibyte = $this->length !== \strlen($line); |
91
|
2478 |
|
$this->lineContainsTabs = \preg_match('/\t/', $line) > 0; |
92
|
2478 |
|
} |
93
|
|
|
|
94
|
60 |
|
public function getEncoding(): string |
95
|
|
|
{ |
96
|
60 |
|
return $this->encoding; |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* Returns the position of the next character which is not a space (or tab) |
101
|
|
|
* |
102
|
|
|
* @return int |
103
|
|
|
*/ |
104
|
2208 |
|
public function getNextNonSpacePosition(): int |
105
|
|
|
{ |
106
|
2208 |
|
if ($this->nextNonSpaceCache !== null) { |
107
|
2028 |
|
return $this->nextNonSpaceCache; |
108
|
|
|
} |
109
|
|
|
|
110
|
2208 |
|
$i = $this->currentPosition; |
111
|
2208 |
|
$cols = $this->column; |
112
|
|
|
|
113
|
2208 |
|
while (($c = $this->getCharacter($i)) !== null) { |
114
|
2184 |
|
if ($c === ' ') { |
115
|
516 |
|
$i++; |
116
|
516 |
|
$cols++; |
117
|
2142 |
|
} elseif ($c === "\t") { |
118
|
36 |
|
$i++; |
119
|
36 |
|
$cols += (4 - ($cols % 4)); |
120
|
|
|
} else { |
121
|
2142 |
|
break; |
122
|
|
|
} |
123
|
|
|
} |
124
|
|
|
|
125
|
2208 |
|
$nextNonSpace = ($c === null) ? $this->length : $i; |
126
|
2208 |
|
$this->indent = $cols - $this->column; |
127
|
|
|
|
128
|
2208 |
|
return $this->nextNonSpaceCache = $nextNonSpace; |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* Returns the next character which isn't a space (or tab) |
133
|
|
|
* |
134
|
|
|
* @return string |
135
|
|
|
*/ |
136
|
2001 |
|
public function getNextNonSpaceCharacter(): ?string |
137
|
|
|
{ |
138
|
2001 |
|
return $this->getCharacter($this->getNextNonSpacePosition()); |
139
|
|
|
} |
140
|
|
|
|
141
|
|
|
/** |
142
|
|
|
* Calculates the current indent (number of spaces after current position) |
143
|
|
|
* |
144
|
|
|
* @return int |
145
|
|
|
*/ |
146
|
2088 |
|
public function getIndent(): int |
147
|
|
|
{ |
148
|
2088 |
|
if ($this->nextNonSpaceCache === null) { |
149
|
2088 |
|
$this->getNextNonSpacePosition(); |
150
|
|
|
} |
151
|
|
|
|
152
|
2088 |
|
return $this->indent; |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
/** |
156
|
|
|
* Whether the cursor is indented to INDENT_LEVEL |
157
|
|
|
* |
158
|
|
|
* @return bool |
159
|
|
|
*/ |
160
|
2028 |
|
public function isIndented(): bool |
161
|
|
|
{ |
162
|
2028 |
|
return $this->getIndent() >= self::INDENT_LEVEL; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
/** |
166
|
|
|
* @param int|null $index |
167
|
|
|
* |
168
|
|
|
* @return string|null |
169
|
|
|
*/ |
170
|
2262 |
|
public function getCharacter(?int $index = null): ?string |
171
|
|
|
{ |
172
|
2262 |
|
if ($index === null) { |
173
|
1773 |
|
$index = $this->currentPosition; |
174
|
|
|
} |
175
|
|
|
|
176
|
2262 |
|
if (isset($this->charCache[$index])) { |
177
|
2046 |
|
return $this->charCache[$index]; |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
// Index out-of-bounds, or we're at the end |
181
|
2262 |
|
if ($index < 0 || $index >= $this->length) { |
182
|
1980 |
|
return null; |
183
|
|
|
} |
184
|
|
|
|
185
|
2220 |
|
return $this->charCache[$index] = \mb_substr($this->line, $index, 1, $this->encoding); |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
/** |
189
|
|
|
* Returns the next character (or null, if none) without advancing forwards |
190
|
|
|
* |
191
|
|
|
* @param int $offset |
192
|
|
|
* |
193
|
|
|
* @return string|null |
194
|
|
|
*/ |
195
|
1065 |
|
public function peek(int $offset = 1): ?string |
196
|
|
|
{ |
197
|
1065 |
|
return $this->getCharacter($this->currentPosition + $offset); |
198
|
|
|
} |
199
|
|
|
|
200
|
|
|
/** |
201
|
|
|
* Whether the remainder is blank |
202
|
|
|
* |
203
|
|
|
* @return bool |
204
|
|
|
*/ |
205
|
2046 |
|
public function isBlank(): bool |
206
|
|
|
{ |
207
|
2046 |
|
return $this->nextNonSpaceCache === $this->length || $this->getNextNonSpacePosition() === $this->length; |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
/** |
211
|
|
|
* Move the cursor forwards |
212
|
|
|
*/ |
213
|
846 |
|
public function advance() |
214
|
|
|
{ |
215
|
846 |
|
$this->advanceBy(1); |
216
|
846 |
|
} |
217
|
|
|
|
218
|
|
|
/** |
219
|
|
|
* Move the cursor forwards |
220
|
|
|
* |
221
|
|
|
* @param int $characters Number of characters to advance by |
222
|
|
|
* @param bool $advanceByColumns Whether to advance by columns instead of spaces |
223
|
|
|
*/ |
224
|
2334 |
|
public function advanceBy(int $characters, bool $advanceByColumns = false) |
225
|
|
|
{ |
226
|
2334 |
|
if ($characters === 0) { |
227
|
2058 |
|
$this->previousPosition = $this->currentPosition; |
228
|
|
|
|
229
|
2058 |
|
return; |
230
|
|
|
} |
231
|
|
|
|
232
|
2190 |
|
$this->previousPosition = $this->currentPosition; |
233
|
2190 |
|
$this->nextNonSpaceCache = null; |
234
|
|
|
|
235
|
|
|
// Optimization to avoid tab handling logic if we have no tabs |
236
|
2190 |
|
if (!$this->lineContainsTabs || \preg_match('/\t/', $nextFewChars = mb_substr($this->line, $this->currentPosition, $characters, $this->encoding)) === 0) { |
237
|
2178 |
|
$length = \min($characters, $this->length - $this->currentPosition); |
238
|
2178 |
|
$this->partiallyConsumedTab = false; |
239
|
2178 |
|
$this->currentPosition += $length; |
240
|
2178 |
|
$this->column += $length; |
241
|
|
|
|
242
|
2178 |
|
return; |
243
|
|
|
} |
244
|
|
|
|
245
|
45 |
|
if ($characters === 1 && !empty($nextFewChars)) { |
246
|
18 |
|
$asArray = [$nextFewChars]; |
247
|
|
|
} else { |
248
|
39 |
|
$asArray = \preg_split('//u', $nextFewChars, null, PREG_SPLIT_NO_EMPTY); |
249
|
|
|
} |
250
|
|
|
|
251
|
45 |
|
foreach ($asArray as $relPos => $c) { |
252
|
45 |
|
if ($c === "\t") { |
253
|
45 |
|
$charsToTab = 4 - ($this->column % 4); |
254
|
45 |
|
if ($advanceByColumns) { |
255
|
33 |
|
$this->partiallyConsumedTab = $charsToTab > $characters; |
256
|
33 |
|
$charsToAdvance = $charsToTab > $characters ? $characters : $charsToTab; |
257
|
33 |
|
$this->column += $charsToAdvance; |
258
|
33 |
|
$this->currentPosition += $this->partiallyConsumedTab ? 0 : 1; |
259
|
33 |
|
$characters -= $charsToAdvance; |
260
|
|
|
} else { |
261
|
18 |
|
$this->partiallyConsumedTab = false; |
262
|
18 |
|
$this->column += $charsToTab; |
263
|
18 |
|
$this->currentPosition++; |
264
|
45 |
|
$characters--; |
265
|
|
|
} |
266
|
|
|
} else { |
267
|
12 |
|
$this->partiallyConsumedTab = false; |
268
|
12 |
|
$this->currentPosition++; |
269
|
12 |
|
$this->column++; |
270
|
12 |
|
$characters--; |
271
|
|
|
} |
272
|
|
|
|
273
|
45 |
|
if ($characters <= 0) { |
274
|
45 |
|
break; |
275
|
|
|
} |
276
|
|
|
} |
277
|
45 |
|
} |
278
|
|
|
|
279
|
|
|
/** |
280
|
|
|
* Advances the cursor by a single space or tab, if present |
281
|
|
|
* |
282
|
|
|
* @return bool |
283
|
|
|
*/ |
284
|
348 |
|
public function advanceBySpaceOrTab(): bool |
285
|
|
|
{ |
286
|
348 |
|
$character = $this->getCharacter(); |
287
|
|
|
|
288
|
348 |
|
if ($character === ' ' || $character === "\t") { |
289
|
336 |
|
$this->advanceBy(1, true); |
290
|
|
|
|
291
|
336 |
|
return true; |
292
|
|
|
} |
293
|
|
|
|
294
|
258 |
|
return false; |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* Parse zero or more space/tab characters |
299
|
|
|
* |
300
|
|
|
* @return int Number of positions moved |
301
|
|
|
*/ |
302
|
1935 |
|
public function advanceToNextNonSpaceOrTab(): int |
303
|
|
|
{ |
304
|
1935 |
|
$newPosition = $this->getNextNonSpacePosition(); |
305
|
1935 |
|
$this->advanceBy($newPosition - $this->currentPosition); |
306
|
1935 |
|
$this->partiallyConsumedTab = false; |
307
|
|
|
|
308
|
1935 |
|
return $this->currentPosition - $this->previousPosition; |
309
|
|
|
} |
310
|
|
|
|
311
|
|
|
/** |
312
|
|
|
* Parse zero or more space characters, including at most one newline. |
313
|
|
|
* |
314
|
|
|
* Tab characters are not parsed with this function. |
315
|
|
|
* |
316
|
|
|
* @return int Number of positions moved |
317
|
|
|
*/ |
318
|
477 |
|
public function advanceToNextNonSpaceOrNewline(): int |
319
|
|
|
{ |
320
|
477 |
|
$matches = []; |
321
|
477 |
|
\preg_match('/^ *(?:\n *)?/', $this->getRemainder(), $matches, PREG_OFFSET_CAPTURE); |
322
|
|
|
|
323
|
|
|
// [0][0] contains the matched text |
324
|
|
|
// [0][1] contains the index of that match |
325
|
477 |
|
$increment = $matches[0][1] + \strlen($matches[0][0]); |
326
|
|
|
|
327
|
477 |
|
if ($increment === 0) { |
328
|
336 |
|
return 0; |
329
|
|
|
} |
330
|
|
|
|
331
|
324 |
|
$this->advanceBy($increment); |
332
|
|
|
|
333
|
324 |
|
return $this->currentPosition - $this->previousPosition; |
334
|
|
|
} |
335
|
|
|
|
336
|
|
|
/** |
337
|
|
|
* Move the position to the very end of the line |
338
|
|
|
* |
339
|
|
|
* @return int The number of characters moved |
340
|
|
|
*/ |
341
|
84 |
|
public function advanceToEnd(): int |
342
|
|
|
{ |
343
|
84 |
|
$this->previousPosition = $this->currentPosition; |
344
|
84 |
|
$this->nextNonSpaceCache = null; |
345
|
|
|
|
346
|
84 |
|
$this->currentPosition = $this->length; |
347
|
|
|
|
348
|
84 |
|
return $this->currentPosition - $this->previousPosition; |
349
|
|
|
} |
350
|
|
|
|
351
|
|
|
/** |
352
|
|
|
* @return string |
353
|
|
|
*/ |
354
|
2130 |
|
public function getRemainder(): string |
355
|
|
|
{ |
356
|
2130 |
|
if ($this->currentPosition >= $this->length) { |
357
|
717 |
|
return ''; |
358
|
|
|
} |
359
|
|
|
|
360
|
2118 |
|
$prefix = ''; |
361
|
2118 |
|
$position = $this->currentPosition; |
362
|
2118 |
|
if ($this->partiallyConsumedTab) { |
363
|
15 |
|
$position++; |
364
|
15 |
|
$charsToTab = 4 - ($this->column % 4); |
365
|
15 |
|
$prefix = \str_repeat(' ', $charsToTab); |
366
|
|
|
} |
367
|
|
|
|
368
|
2118 |
|
return $prefix . \mb_substr($this->line, $position, null, $this->encoding); |
369
|
|
|
} |
370
|
|
|
|
371
|
|
|
/** |
372
|
|
|
* @return string |
373
|
|
|
*/ |
374
|
1977 |
|
public function getLine(): string |
375
|
|
|
{ |
376
|
1977 |
|
return $this->line; |
377
|
|
|
} |
378
|
|
|
|
379
|
|
|
/** |
380
|
|
|
* @return bool |
381
|
|
|
*/ |
382
|
447 |
|
public function isAtEnd(): bool |
383
|
|
|
{ |
384
|
447 |
|
return $this->currentPosition >= $this->length; |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
/** |
388
|
|
|
* Try to match a regular expression |
389
|
|
|
* |
390
|
|
|
* Returns the matching text and advances to the end of that match |
391
|
|
|
* |
392
|
|
|
* @param string $regex |
393
|
|
|
* |
394
|
|
|
* @return string|null |
395
|
|
|
*/ |
396
|
1998 |
|
public function match(string $regex): ?string |
397
|
|
|
{ |
398
|
1998 |
|
$subject = $this->getRemainder(); |
399
|
|
|
|
400
|
1998 |
|
if (!\preg_match($regex, $subject, $matches, PREG_OFFSET_CAPTURE)) { |
401
|
1845 |
|
return null; |
402
|
|
|
} |
403
|
|
|
|
404
|
|
|
// $matches[0][0] contains the matched text |
405
|
|
|
// $matches[0][1] contains the index of that match |
406
|
|
|
|
407
|
1869 |
|
if ($this->isMultibyte) { |
408
|
|
|
// PREG_OFFSET_CAPTURE always returns the byte offset, not the char offset, which is annoying |
409
|
54 |
|
$offset = \mb_strlen(\mb_strcut($subject, 0, $matches[0][1], $this->encoding), $this->encoding); |
410
|
54 |
|
$matchLength = \mb_strlen($matches[0][0], $this->encoding); |
411
|
|
|
} else { |
412
|
1815 |
|
$offset = $matches[0][1]; |
413
|
1815 |
|
$matchLength = \strlen($matches[0][0]); |
414
|
|
|
} |
415
|
|
|
|
416
|
|
|
// [0][0] contains the matched text |
417
|
|
|
// [0][1] contains the index of that match |
418
|
1869 |
|
$this->advanceBy($offset + $matchLength); |
419
|
|
|
|
420
|
1869 |
|
return $matches[0][0]; |
421
|
|
|
} |
422
|
|
|
|
423
|
|
|
/** |
424
|
|
|
* Encapsulates the current state of this cursor in case you need to rollback later. |
425
|
|
|
* |
426
|
|
|
* WARNING: Do not parse or use the return value for ANYTHING except for |
427
|
|
|
* passing it back into restoreState(), as the number of values and their |
428
|
|
|
* contents may change in any future release without warning. |
429
|
|
|
* |
430
|
|
|
* @return array |
431
|
|
|
*/ |
432
|
1077 |
|
public function saveState() |
433
|
|
|
{ |
434
|
|
|
return [ |
435
|
1077 |
|
$this->currentPosition, |
436
|
1077 |
|
$this->previousPosition, |
437
|
1077 |
|
$this->nextNonSpaceCache, |
438
|
1077 |
|
$this->indent, |
439
|
1077 |
|
$this->column, |
440
|
1077 |
|
$this->partiallyConsumedTab, |
441
|
|
|
]; |
442
|
|
|
} |
443
|
|
|
|
444
|
|
|
/** |
445
|
|
|
* Restore the cursor to a previous state. |
446
|
|
|
* |
447
|
|
|
* Pass in the value previously obtained by calling saveState(). |
448
|
|
|
* |
449
|
|
|
* @param array $state |
450
|
|
|
*/ |
451
|
843 |
|
public function restoreState($state) |
452
|
|
|
{ |
453
|
|
|
list( |
454
|
843 |
|
$this->currentPosition, |
455
|
843 |
|
$this->previousPosition, |
456
|
843 |
|
$this->nextNonSpaceCache, |
457
|
843 |
|
$this->indent, |
458
|
843 |
|
$this->column, |
459
|
843 |
|
$this->partiallyConsumedTab, |
460
|
843 |
|
) = $state; |
461
|
843 |
|
} |
462
|
|
|
|
463
|
|
|
/** |
464
|
|
|
* @return int |
465
|
|
|
*/ |
466
|
699 |
|
public function getPosition(): int |
467
|
|
|
{ |
468
|
699 |
|
return $this->currentPosition; |
469
|
|
|
} |
470
|
|
|
|
471
|
|
|
/** |
472
|
|
|
* @return string |
473
|
|
|
*/ |
474
|
921 |
|
public function getPreviousText(): string |
475
|
|
|
{ |
476
|
921 |
|
return \mb_substr($this->line, $this->previousPosition, $this->currentPosition - $this->previousPosition, $this->encoding); |
477
|
|
|
} |
478
|
|
|
|
479
|
|
|
/** |
480
|
|
|
* @return int |
481
|
|
|
*/ |
482
|
249 |
|
public function getColumn(): int |
483
|
|
|
{ |
484
|
249 |
|
return $this->column; |
485
|
|
|
} |
486
|
|
|
} |
487
|
|
|
|