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\Parser; |
15
|
|
|
|
16
|
|
|
class Cursor |
17
|
|
|
{ |
18
|
|
|
public const INDENT_LEVEL = 4; |
19
|
|
|
|
20
|
|
|
/** @var string */ |
21
|
|
|
private $line; |
22
|
|
|
|
23
|
|
|
/** @var int */ |
24
|
|
|
private $length; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var int |
28
|
|
|
* |
29
|
|
|
* It's possible for this to be 1 char past the end, meaning we've parsed all chars and have |
30
|
|
|
* reached the end. In this state, any character-returning method MUST return null. |
31
|
|
|
*/ |
32
|
|
|
private $currentPosition = 0; |
33
|
|
|
|
34
|
|
|
/** @var int */ |
35
|
|
|
private $column = 0; |
36
|
|
|
|
37
|
|
|
/** @var int */ |
38
|
|
|
private $indent = 0; |
39
|
|
|
|
40
|
|
|
/** @var int */ |
41
|
|
|
private $previousPosition = 0; |
42
|
|
|
|
43
|
|
|
/** @var int|null */ |
44
|
|
|
private $nextNonSpaceCache; |
45
|
|
|
|
46
|
|
|
/** @var bool */ |
47
|
|
|
private $partiallyConsumedTab = false; |
48
|
|
|
|
49
|
|
|
/** @var bool */ |
50
|
|
|
private $lineContainsTabs; |
51
|
|
|
|
52
|
|
|
/** @var bool */ |
53
|
|
|
private $isMultibyte; |
54
|
|
|
|
55
|
|
|
/** @var array<int, string> */ |
56
|
|
|
private $charCache = []; |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @param string $line The line being parsed (ASCII or UTF-8) |
60
|
|
|
*/ |
61
|
3000 |
|
public function __construct(string $line) |
62
|
|
|
{ |
63
|
3000 |
|
$this->line = $line; |
64
|
3000 |
|
$this->length = \mb_strlen($line, 'UTF-8') ?: 0; |
65
|
3000 |
|
$this->isMultibyte = $this->length !== \strlen($line); |
66
|
3000 |
|
$this->lineContainsTabs = \strpos($line, "\t") !== false; |
67
|
3000 |
|
} |
68
|
|
|
|
69
|
|
|
/** |
70
|
|
|
* Returns the position of the next character which is not a space (or tab) |
71
|
|
|
*/ |
72
|
2700 |
|
public function getNextNonSpacePosition(): int |
73
|
|
|
{ |
74
|
2700 |
|
if ($this->nextNonSpaceCache !== null) { |
75
|
2520 |
|
return $this->nextNonSpaceCache; |
76
|
|
|
} |
77
|
|
|
|
78
|
2700 |
|
$i = $this->currentPosition; |
79
|
2700 |
|
$cols = $this->column; |
80
|
|
|
|
81
|
2700 |
|
while (($c = $this->getCharacter($i)) !== null) { |
82
|
2676 |
|
if ($c === ' ') { |
83
|
813 |
|
$i++; |
84
|
813 |
|
$cols++; |
85
|
2634 |
|
} elseif ($c === "\t") { |
86
|
36 |
|
$i++; |
87
|
36 |
|
$cols += 4 - ($cols % 4); |
88
|
|
|
} else { |
89
|
2634 |
|
break; |
90
|
|
|
} |
91
|
|
|
} |
92
|
|
|
|
93
|
2700 |
|
$nextNonSpace = $c === null ? $this->length : $i; |
94
|
2700 |
|
$this->indent = $cols - $this->column; |
95
|
|
|
|
96
|
2700 |
|
return $this->nextNonSpaceCache = $nextNonSpace; |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* Returns the next character which isn't a space (or tab) |
101
|
|
|
*/ |
102
|
2466 |
|
public function getNextNonSpaceCharacter(): ?string |
103
|
|
|
{ |
104
|
2466 |
|
return $this->getCharacter($this->getNextNonSpacePosition()); |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
/** |
108
|
|
|
* Calculates the current indent (number of spaces after current position) |
109
|
|
|
*/ |
110
|
2580 |
|
public function getIndent(): int |
111
|
|
|
{ |
112
|
2580 |
|
if ($this->nextNonSpaceCache === null) { |
113
|
360 |
|
$this->getNextNonSpacePosition(); |
114
|
|
|
} |
115
|
|
|
|
116
|
2580 |
|
return $this->indent; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* Whether the cursor is indented to INDENT_LEVEL |
121
|
|
|
*/ |
122
|
2520 |
|
public function isIndented(): bool |
123
|
|
|
{ |
124
|
2520 |
|
return $this->getIndent() >= self::INDENT_LEVEL; |
125
|
|
|
} |
126
|
|
|
|
127
|
2760 |
|
public function getCharacter(?int $index = null): ?string |
128
|
|
|
{ |
129
|
2760 |
|
if ($index === null) { |
130
|
2472 |
|
$index = $this->currentPosition; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
// Index out-of-bounds, or we're at the end |
134
|
2760 |
|
if ($index < 0 || $index >= $this->length) { |
135
|
2370 |
|
return null; |
136
|
|
|
} |
137
|
|
|
|
138
|
2718 |
|
if ($this->isMultibyte) { |
139
|
102 |
|
if (isset($this->charCache[$index])) { |
140
|
48 |
|
return $this->charCache[$index]; |
141
|
|
|
} |
142
|
|
|
|
143
|
102 |
|
return $this->charCache[$index] = \mb_substr($this->line, $index, 1, 'UTF-8'); |
144
|
|
|
} |
145
|
|
|
|
146
|
2622 |
|
return $this->line[$index]; |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* Returns the next character (or null, if none) without advancing forwards |
151
|
|
|
*/ |
152
|
1332 |
|
public function peek(int $offset = 1): ?string |
153
|
|
|
{ |
154
|
1332 |
|
return $this->getCharacter($this->currentPosition + $offset); |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* Whether the remainder is blank |
159
|
|
|
*/ |
160
|
2514 |
|
public function isBlank(): bool |
161
|
|
|
{ |
162
|
2514 |
|
return $this->nextNonSpaceCache === $this->length || $this->getNextNonSpacePosition() === $this->length; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
/** |
166
|
|
|
* Move the cursor forwards |
167
|
|
|
*/ |
168
|
513 |
|
public function advance(): void |
169
|
|
|
{ |
170
|
513 |
|
$this->advanceBy(1); |
171
|
513 |
|
} |
172
|
|
|
|
173
|
|
|
/** |
174
|
|
|
* Move the cursor forwards |
175
|
|
|
* |
176
|
|
|
* @param int $characters Number of characters to advance by |
177
|
|
|
* @param bool $advanceByColumns Whether to advance by columns instead of spaces |
178
|
|
|
*/ |
179
|
2868 |
|
public function advanceBy(int $characters, bool $advanceByColumns = false): void |
180
|
|
|
{ |
181
|
2868 |
|
if ($characters === 0) { |
182
|
2577 |
|
$this->previousPosition = $this->currentPosition; |
183
|
|
|
|
184
|
2577 |
|
return; |
185
|
|
|
} |
186
|
|
|
|
187
|
2679 |
|
$this->previousPosition = $this->currentPosition; |
188
|
2679 |
|
$this->nextNonSpaceCache = null; |
189
|
|
|
|
190
|
|
|
// Optimization to avoid tab handling logic if we have no tabs |
191
|
|
|
if ( |
192
|
2679 |
|
! $this->lineContainsTabs || \strpos( |
193
|
51 |
|
$nextFewChars = $this->isMultibyte ? |
194
|
6 |
|
\mb_substr($this->line, $this->currentPosition, $characters, 'UTF-8') : |
195
|
51 |
|
\substr($this->line, $this->currentPosition, $characters), |
196
|
51 |
|
"\t" |
197
|
2679 |
|
) === false |
198
|
|
|
) { |
199
|
2667 |
|
$length = \min($characters, $this->length - $this->currentPosition); |
200
|
2667 |
|
$this->partiallyConsumedTab = false; |
201
|
2667 |
|
$this->currentPosition += $length; |
202
|
2667 |
|
$this->column += $length; |
203
|
|
|
|
204
|
2667 |
|
return; |
205
|
|
|
} |
206
|
|
|
|
207
|
45 |
|
if ($characters === 1 && ! empty($nextFewChars)) { |
208
|
18 |
|
$asArray = [$nextFewChars]; |
209
|
39 |
|
} elseif ($this->isMultibyte) { |
210
|
|
|
/** @var string[] $asArray */ |
211
|
|
|
$asArray = \preg_split('//u', $nextFewChars, -1, \PREG_SPLIT_NO_EMPTY); |
212
|
|
|
} else { |
213
|
39 |
|
$asArray = \str_split($nextFewChars); |
214
|
|
|
} |
215
|
|
|
|
216
|
45 |
|
foreach ($asArray as $relPos => $c) { |
217
|
45 |
|
if ($c === "\t") { |
218
|
45 |
|
$charsToTab = 4 - ($this->column % 4); |
219
|
45 |
|
if ($advanceByColumns) { |
220
|
33 |
|
$this->partiallyConsumedTab = $charsToTab > $characters; |
221
|
33 |
|
$charsToAdvance = $charsToTab > $characters ? $characters : $charsToTab; |
222
|
33 |
|
$this->column += $charsToAdvance; |
223
|
33 |
|
$this->currentPosition += $this->partiallyConsumedTab ? 0 : 1; |
224
|
33 |
|
$characters -= $charsToAdvance; |
225
|
|
|
} else { |
226
|
18 |
|
$this->partiallyConsumedTab = false; |
227
|
18 |
|
$this->column += $charsToTab; |
228
|
18 |
|
$this->currentPosition++; |
229
|
45 |
|
$characters--; |
230
|
|
|
} |
231
|
|
|
} else { |
232
|
12 |
|
$this->partiallyConsumedTab = false; |
233
|
12 |
|
$this->currentPosition++; |
234
|
12 |
|
$this->column++; |
235
|
12 |
|
$characters--; |
236
|
|
|
} |
237
|
|
|
|
238
|
45 |
|
if ($characters <= 0) { |
239
|
45 |
|
break; |
240
|
|
|
} |
241
|
|
|
} |
242
|
45 |
|
} |
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Advances the cursor by a single space or tab, if present |
246
|
|
|
*/ |
247
|
390 |
|
public function advanceBySpaceOrTab(): bool |
248
|
|
|
{ |
249
|
390 |
|
$character = $this->getCharacter(); |
250
|
|
|
|
251
|
390 |
|
if ($character === ' ' || $character === "\t") { |
252
|
378 |
|
$this->advanceBy(1, true); |
253
|
|
|
|
254
|
378 |
|
return true; |
255
|
|
|
} |
256
|
|
|
|
257
|
294 |
|
return false; |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* Parse zero or more space/tab characters |
262
|
|
|
* |
263
|
|
|
* @return int Number of positions moved |
264
|
|
|
*/ |
265
|
2532 |
|
public function advanceToNextNonSpaceOrTab(): int |
266
|
|
|
{ |
267
|
2532 |
|
$newPosition = $this->getNextNonSpacePosition(); |
268
|
2532 |
|
$this->advanceBy($newPosition - $this->currentPosition); |
269
|
2532 |
|
$this->partiallyConsumedTab = false; |
270
|
|
|
|
271
|
2532 |
|
return $this->currentPosition - $this->previousPosition; |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
/** |
275
|
|
|
* Parse zero or more space characters, including at most one newline. |
276
|
|
|
* |
277
|
|
|
* Tab characters are not parsed with this function. |
278
|
|
|
* |
279
|
|
|
* @return int Number of positions moved |
280
|
|
|
*/ |
281
|
249 |
|
public function advanceToNextNonSpaceOrNewline(): int |
282
|
|
|
{ |
283
|
249 |
|
$remainder = $this->getRemainder(); |
284
|
|
|
|
285
|
|
|
// Optimization: Avoid the regex if we know there are no spaces or newlines |
286
|
249 |
|
if (empty($remainder) || ($remainder[0] !== ' ' && $remainder[0] !== "\n")) { |
287
|
216 |
|
$this->previousPosition = $this->currentPosition; |
288
|
|
|
|
289
|
216 |
|
return 0; |
290
|
|
|
} |
291
|
|
|
|
292
|
84 |
|
$matches = []; |
293
|
84 |
|
\preg_match('/^ *(?:\n *)?/', $remainder, $matches, \PREG_OFFSET_CAPTURE); |
294
|
|
|
|
295
|
|
|
// [0][0] contains the matched text |
296
|
|
|
// [0][1] contains the index of that match |
297
|
84 |
|
$increment = $matches[0][1] + \strlen($matches[0][0]); |
298
|
|
|
|
299
|
84 |
|
$this->advanceBy($increment); |
300
|
|
|
|
301
|
84 |
|
return $this->currentPosition - $this->previousPosition; |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* Move the position to the very end of the line |
306
|
|
|
* |
307
|
|
|
* @return int The number of characters moved |
308
|
|
|
*/ |
309
|
798 |
|
public function advanceToEnd(): int |
310
|
|
|
{ |
311
|
798 |
|
$this->previousPosition = $this->currentPosition; |
312
|
798 |
|
$this->nextNonSpaceCache = null; |
313
|
|
|
|
314
|
798 |
|
$this->currentPosition = $this->length; |
315
|
|
|
|
316
|
798 |
|
return $this->currentPosition - $this->previousPosition; |
317
|
|
|
} |
318
|
|
|
|
319
|
2634 |
|
public function getRemainder(): string |
320
|
|
|
{ |
321
|
2634 |
|
if ($this->currentPosition >= $this->length) { |
322
|
555 |
|
return ''; |
323
|
|
|
} |
324
|
|
|
|
325
|
2607 |
|
$prefix = ''; |
326
|
2607 |
|
$position = $this->currentPosition; |
327
|
2607 |
|
if ($this->partiallyConsumedTab) { |
328
|
12 |
|
$position++; |
329
|
12 |
|
$charsToTab = 4 - ($this->column % 4); |
330
|
12 |
|
$prefix = \str_repeat(' ', $charsToTab); |
331
|
|
|
} |
332
|
|
|
|
333
|
2607 |
|
$subString = $this->isMultibyte ? |
334
|
72 |
|
\mb_substr($this->line, $position, null, 'UTF-8') : |
335
|
2607 |
|
\substr($this->line, $position); |
336
|
|
|
|
337
|
2607 |
|
return $prefix . $subString; |
338
|
|
|
} |
339
|
|
|
|
340
|
1965 |
|
public function getLine(): string |
341
|
|
|
{ |
342
|
1965 |
|
return $this->line; |
343
|
|
|
} |
344
|
|
|
|
345
|
2154 |
|
public function isAtEnd(): bool |
346
|
|
|
{ |
347
|
2154 |
|
return $this->currentPosition >= $this->length; |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
/** |
351
|
|
|
* Try to match a regular expression |
352
|
|
|
* |
353
|
|
|
* Returns the matching text and advances to the end of that match |
354
|
|
|
*/ |
355
|
2352 |
|
public function match(string $regex): ?string |
356
|
|
|
{ |
357
|
2352 |
|
$subject = $this->getRemainder(); |
358
|
|
|
|
359
|
2352 |
|
if (! \preg_match($regex, $subject, $matches, \PREG_OFFSET_CAPTURE)) { |
360
|
909 |
|
return null; |
361
|
|
|
} |
362
|
|
|
|
363
|
|
|
// $matches[0][0] contains the matched text |
364
|
|
|
// $matches[0][1] contains the index of that match |
365
|
|
|
|
366
|
2319 |
|
if ($this->isMultibyte) { |
367
|
|
|
// PREG_OFFSET_CAPTURE always returns the byte offset, not the char offset, which is annoying |
368
|
57 |
|
$offset = \mb_strlen(\substr($subject, 0, $matches[0][1]), 'UTF-8'); |
369
|
57 |
|
$matchLength = \mb_strlen($matches[0][0], 'UTF-8'); |
370
|
|
|
} else { |
371
|
2262 |
|
$offset = $matches[0][1]; |
372
|
2262 |
|
$matchLength = \strlen($matches[0][0]); |
373
|
|
|
} |
374
|
|
|
|
375
|
|
|
// [0][0] contains the matched text |
376
|
|
|
// [0][1] contains the index of that match |
377
|
2319 |
|
$this->advanceBy($offset + $matchLength); |
378
|
|
|
|
379
|
2319 |
|
return $matches[0][0]; |
380
|
|
|
} |
381
|
|
|
|
382
|
|
|
/** |
383
|
|
|
* Encapsulates the current state of this cursor in case you need to rollback later. |
384
|
|
|
* |
385
|
|
|
* WARNING: Do not parse or use the return value for ANYTHING except for |
386
|
|
|
* passing it back into restoreState(), as the number of values and their |
387
|
|
|
* contents may change in any future release without warning. |
388
|
|
|
*/ |
389
|
1689 |
|
public function saveState(): CursorState |
390
|
|
|
{ |
391
|
1689 |
|
return new CursorState([ |
392
|
1689 |
|
$this->currentPosition, |
393
|
1689 |
|
$this->previousPosition, |
394
|
1689 |
|
$this->nextNonSpaceCache, |
395
|
1689 |
|
$this->indent, |
396
|
1689 |
|
$this->column, |
397
|
1689 |
|
$this->partiallyConsumedTab, |
398
|
|
|
]); |
399
|
|
|
} |
400
|
|
|
|
401
|
|
|
/** |
402
|
|
|
* Restore the cursor to a previous state. |
403
|
|
|
* |
404
|
|
|
* Pass in the value previously obtained by calling saveState(). |
405
|
|
|
*/ |
406
|
1596 |
|
public function restoreState(CursorState $state): void |
407
|
|
|
{ |
408
|
|
|
[ |
409
|
1596 |
|
$this->currentPosition, |
410
|
1596 |
|
$this->previousPosition, |
411
|
1596 |
|
$this->nextNonSpaceCache, |
412
|
1596 |
|
$this->indent, |
413
|
1596 |
|
$this->column, |
414
|
1596 |
|
$this->partiallyConsumedTab, |
415
|
1596 |
|
] = $state->toArray(); |
416
|
1596 |
|
} |
417
|
|
|
|
418
|
735 |
|
public function getPosition(): int |
419
|
|
|
{ |
420
|
735 |
|
return $this->currentPosition; |
421
|
|
|
} |
422
|
|
|
|
423
|
399 |
|
public function getPreviousText(): string |
424
|
|
|
{ |
425
|
399 |
|
return \mb_substr($this->line, $this->previousPosition, $this->currentPosition - $this->previousPosition, 'UTF-8'); |
426
|
|
|
} |
427
|
|
|
|
428
|
426 |
|
public function getSubstring(int $start, ?int $length = null): string |
429
|
|
|
{ |
430
|
426 |
|
if ($this->isMultibyte) { |
431
|
21 |
|
return \mb_substr($this->line, $start, $length, 'UTF-8'); |
432
|
|
|
} |
433
|
|
|
|
434
|
405 |
|
if ($length !== null) { |
435
|
402 |
|
return \substr($this->line, $start, $length); |
436
|
|
|
} |
437
|
|
|
|
438
|
3 |
|
return \substr($this->line, $start); |
439
|
|
|
} |
440
|
|
|
|
441
|
285 |
|
public function getColumn(): int |
442
|
|
|
{ |
443
|
285 |
|
return $this->column; |
444
|
|
|
} |
445
|
|
|
} |
446
|
|
|
|