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
|
|
|
/** |
21
|
|
|
* @var string |
22
|
|
|
* |
23
|
|
|
* @psalm-readonly |
24
|
|
|
*/ |
25
|
|
|
private $line; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* @var int |
29
|
|
|
* |
30
|
|
|
* @psalm-readonly |
31
|
|
|
*/ |
32
|
|
|
private $length; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* @var int |
36
|
|
|
* |
37
|
|
|
* It's possible for this to be 1 char past the end, meaning we've parsed all chars and have |
38
|
|
|
* reached the end. In this state, any character-returning method MUST return null. |
39
|
|
|
*/ |
40
|
|
|
private $currentPosition = 0; |
41
|
|
|
|
42
|
|
|
/** @var int */ |
43
|
|
|
private $column = 0; |
44
|
|
|
|
45
|
|
|
/** @var int */ |
46
|
|
|
private $indent = 0; |
47
|
|
|
|
48
|
|
|
/** @var int */ |
49
|
|
|
private $previousPosition = 0; |
50
|
|
|
|
51
|
|
|
/** @var int|null */ |
52
|
|
|
private $nextNonSpaceCache; |
53
|
|
|
|
54
|
|
|
/** @var bool */ |
55
|
|
|
private $partiallyConsumedTab = false; |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* @var bool |
59
|
|
|
* |
60
|
|
|
* @psalm-readonly |
61
|
|
|
*/ |
62
|
|
|
private $lineContainsTabs; |
63
|
|
|
|
64
|
|
|
/** |
65
|
|
|
* @var bool |
66
|
|
|
* |
67
|
|
|
* @psalm-readonly |
68
|
|
|
*/ |
69
|
|
|
private $isMultibyte; |
70
|
|
|
|
71
|
|
|
/** @var array<int, string> */ |
72
|
|
|
private $charCache = []; |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @param string $line The line being parsed (ASCII or UTF-8) |
76
|
|
|
*/ |
77
|
3000 |
|
public function __construct(string $line) |
78
|
|
|
{ |
79
|
3000 |
|
$this->line = $line; |
80
|
3000 |
|
$this->length = \mb_strlen($line, 'UTF-8') ?: 0; |
81
|
3000 |
|
$this->isMultibyte = $this->length !== \strlen($line); |
82
|
3000 |
|
$this->lineContainsTabs = \strpos($line, "\t") !== false; |
83
|
3000 |
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Returns the position of the next character which is not a space (or tab) |
87
|
|
|
*/ |
88
|
2700 |
|
public function getNextNonSpacePosition(): int |
89
|
|
|
{ |
90
|
2700 |
|
if ($this->nextNonSpaceCache !== null) { |
91
|
2520 |
|
return $this->nextNonSpaceCache; |
92
|
|
|
} |
93
|
|
|
|
94
|
2700 |
|
$c = null; |
|
|
|
|
95
|
2700 |
|
$i = $this->currentPosition; |
96
|
2700 |
|
$cols = $this->column; |
97
|
|
|
|
98
|
2700 |
|
while (($c = $this->getCharacter($i)) !== null) { |
99
|
2676 |
|
if ($c === ' ') { |
100
|
813 |
|
$i++; |
101
|
813 |
|
$cols++; |
102
|
2634 |
|
} elseif ($c === "\t") { |
103
|
36 |
|
$i++; |
104
|
36 |
|
$cols += 4 - ($cols % 4); |
105
|
|
|
} else { |
106
|
2634 |
|
break; |
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
|
110
|
2700 |
|
$nextNonSpace = $c === null ? $this->length : $i; |
111
|
2700 |
|
$this->indent = $cols - $this->column; |
112
|
|
|
|
113
|
2700 |
|
return $this->nextNonSpaceCache = $nextNonSpace; |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* Returns the next character which isn't a space (or tab) |
118
|
|
|
*/ |
119
|
2466 |
|
public function getNextNonSpaceCharacter(): ?string |
120
|
|
|
{ |
121
|
2466 |
|
return $this->getCharacter($this->getNextNonSpacePosition()); |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* Calculates the current indent (number of spaces after current position) |
126
|
|
|
*/ |
127
|
2580 |
|
public function getIndent(): int |
128
|
|
|
{ |
129
|
2580 |
|
if ($this->nextNonSpaceCache === null) { |
130
|
360 |
|
$this->getNextNonSpacePosition(); |
131
|
|
|
} |
132
|
|
|
|
133
|
2580 |
|
return $this->indent; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
/** |
137
|
|
|
* Whether the cursor is indented to INDENT_LEVEL |
138
|
|
|
*/ |
139
|
2520 |
|
public function isIndented(): bool |
140
|
|
|
{ |
141
|
2520 |
|
return $this->getIndent() >= self::INDENT_LEVEL; |
142
|
|
|
} |
143
|
|
|
|
144
|
2760 |
|
public function getCharacter(?int $index = null): ?string |
145
|
|
|
{ |
146
|
2760 |
|
if ($index === null) { |
147
|
2472 |
|
$index = $this->currentPosition; |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
// Index out-of-bounds, or we're at the end |
151
|
2760 |
|
if ($index < 0 || $index >= $this->length) { |
152
|
2370 |
|
return null; |
153
|
|
|
} |
154
|
|
|
|
155
|
2718 |
|
if ($this->isMultibyte) { |
156
|
102 |
|
if (isset($this->charCache[$index])) { |
157
|
48 |
|
return $this->charCache[$index]; |
158
|
|
|
} |
159
|
|
|
|
160
|
102 |
|
return $this->charCache[$index] = \mb_substr($this->line, $index, 1, 'UTF-8'); |
161
|
|
|
} |
162
|
|
|
|
163
|
2622 |
|
return $this->line[$index]; |
164
|
|
|
} |
165
|
|
|
|
166
|
|
|
/** |
167
|
|
|
* Returns the next character (or null, if none) without advancing forwards |
168
|
|
|
*/ |
169
|
1332 |
|
public function peek(int $offset = 1): ?string |
170
|
|
|
{ |
171
|
1332 |
|
return $this->getCharacter($this->currentPosition + $offset); |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
/** |
175
|
|
|
* Whether the remainder is blank |
176
|
|
|
*/ |
177
|
2514 |
|
public function isBlank(): bool |
178
|
|
|
{ |
179
|
2514 |
|
return $this->nextNonSpaceCache === $this->length || $this->getNextNonSpacePosition() === $this->length; |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* Move the cursor forwards |
184
|
|
|
*/ |
185
|
513 |
|
public function advance(): void |
186
|
|
|
{ |
187
|
513 |
|
$this->advanceBy(1); |
188
|
513 |
|
} |
189
|
|
|
|
190
|
|
|
/** |
191
|
|
|
* Move the cursor forwards |
192
|
|
|
* |
193
|
|
|
* @param int $characters Number of characters to advance by |
194
|
|
|
* @param bool $advanceByColumns Whether to advance by columns instead of spaces |
195
|
|
|
*/ |
196
|
2868 |
|
public function advanceBy(int $characters, bool $advanceByColumns = false): void |
197
|
|
|
{ |
198
|
2868 |
|
if ($characters === 0) { |
199
|
2577 |
|
$this->previousPosition = $this->currentPosition; |
200
|
|
|
|
201
|
2577 |
|
return; |
202
|
|
|
} |
203
|
|
|
|
204
|
2679 |
|
$this->previousPosition = $this->currentPosition; |
205
|
2679 |
|
$this->nextNonSpaceCache = null; |
206
|
|
|
|
207
|
|
|
// Optimization to avoid tab handling logic if we have no tabs |
208
|
2679 |
|
if (! $this->lineContainsTabs) { |
209
|
2652 |
|
$this->advanceWithoutTabCharacters($characters); |
210
|
|
|
|
211
|
2652 |
|
return; |
212
|
|
|
} |
213
|
|
|
|
214
|
51 |
|
$nextFewChars = $this->isMultibyte ? |
215
|
6 |
|
\mb_substr($this->line, $this->currentPosition, $characters, 'UTF-8') : |
216
|
51 |
|
\substr($this->line, $this->currentPosition, $characters); |
217
|
|
|
|
218
|
|
|
// Optimization to avoid tab handling logic if we have no tabs |
219
|
51 |
|
if (\strpos($nextFewChars, "\t") === false) { |
220
|
24 |
|
$this->advanceWithoutTabCharacters($characters); |
221
|
|
|
|
222
|
24 |
|
return; |
223
|
|
|
} |
224
|
|
|
|
225
|
45 |
|
if ($characters === 1 && ! empty($nextFewChars)) { |
226
|
18 |
|
$asArray = [$nextFewChars]; |
227
|
39 |
|
} elseif ($this->isMultibyte) { |
228
|
|
|
/** @var string[] $asArray */ |
229
|
|
|
$asArray = \preg_split('//u', $nextFewChars, -1, \PREG_SPLIT_NO_EMPTY); |
230
|
|
|
} else { |
231
|
39 |
|
$asArray = \str_split($nextFewChars); |
232
|
|
|
} |
233
|
|
|
|
234
|
45 |
|
foreach ($asArray as $relPos => $c) { |
235
|
45 |
|
if ($c === "\t") { |
236
|
45 |
|
$charsToTab = 4 - ($this->column % 4); |
237
|
45 |
|
if ($advanceByColumns) { |
238
|
33 |
|
$this->partiallyConsumedTab = $charsToTab > $characters; |
239
|
33 |
|
$charsToAdvance = $charsToTab > $characters ? $characters : $charsToTab; |
240
|
33 |
|
$this->column += $charsToAdvance; |
241
|
33 |
|
$this->currentPosition += $this->partiallyConsumedTab ? 0 : 1; |
242
|
33 |
|
$characters -= $charsToAdvance; |
243
|
|
|
} else { |
244
|
18 |
|
$this->partiallyConsumedTab = false; |
245
|
18 |
|
$this->column += $charsToTab; |
246
|
18 |
|
$this->currentPosition++; |
247
|
45 |
|
$characters--; |
248
|
|
|
} |
249
|
|
|
} else { |
250
|
12 |
|
$this->partiallyConsumedTab = false; |
251
|
12 |
|
$this->currentPosition++; |
252
|
12 |
|
$this->column++; |
253
|
12 |
|
$characters--; |
254
|
|
|
} |
255
|
|
|
|
256
|
45 |
|
if ($characters <= 0) { |
257
|
45 |
|
break; |
258
|
|
|
} |
259
|
|
|
} |
260
|
45 |
|
} |
261
|
|
|
|
262
|
2667 |
|
private function advanceWithoutTabCharacters(int $characters): void |
263
|
|
|
{ |
264
|
2667 |
|
$length = \min($characters, $this->length - $this->currentPosition); |
265
|
2667 |
|
$this->partiallyConsumedTab = false; |
266
|
2667 |
|
$this->currentPosition += $length; |
267
|
2667 |
|
$this->column += $length; |
268
|
|
|
|
269
|
2667 |
|
return; |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
/** |
273
|
|
|
* Advances the cursor by a single space or tab, if present |
274
|
|
|
*/ |
275
|
390 |
|
public function advanceBySpaceOrTab(): bool |
276
|
|
|
{ |
277
|
390 |
|
$character = $this->getCharacter(); |
278
|
|
|
|
279
|
390 |
|
if ($character === ' ' || $character === "\t") { |
280
|
378 |
|
$this->advanceBy(1, true); |
281
|
|
|
|
282
|
378 |
|
return true; |
283
|
|
|
} |
284
|
|
|
|
285
|
294 |
|
return false; |
286
|
|
|
} |
287
|
|
|
|
288
|
|
|
/** |
289
|
|
|
* Parse zero or more space/tab characters |
290
|
|
|
* |
291
|
|
|
* @return int Number of positions moved |
292
|
|
|
*/ |
293
|
2532 |
|
public function advanceToNextNonSpaceOrTab(): int |
294
|
|
|
{ |
295
|
2532 |
|
$newPosition = $this->getNextNonSpacePosition(); |
296
|
2532 |
|
$this->advanceBy($newPosition - $this->currentPosition); |
297
|
2532 |
|
$this->partiallyConsumedTab = false; |
298
|
|
|
|
299
|
2532 |
|
return $this->currentPosition - $this->previousPosition; |
300
|
|
|
} |
301
|
|
|
|
302
|
|
|
/** |
303
|
|
|
* Parse zero or more space characters, including at most one newline. |
304
|
|
|
* |
305
|
|
|
* Tab characters are not parsed with this function. |
306
|
|
|
* |
307
|
|
|
* @return int Number of positions moved |
308
|
|
|
*/ |
309
|
249 |
|
public function advanceToNextNonSpaceOrNewline(): int |
310
|
|
|
{ |
311
|
249 |
|
$remainder = $this->getRemainder(); |
312
|
|
|
|
313
|
|
|
// Optimization: Avoid the regex if we know there are no spaces or newlines |
314
|
249 |
|
if (empty($remainder) || ($remainder[0] !== ' ' && $remainder[0] !== "\n")) { |
315
|
216 |
|
$this->previousPosition = $this->currentPosition; |
316
|
|
|
|
317
|
216 |
|
return 0; |
318
|
|
|
} |
319
|
|
|
|
320
|
84 |
|
$matches = []; |
321
|
84 |
|
\preg_match('/^ *(?:\n *)?/', $remainder, $matches, \PREG_OFFSET_CAPTURE); |
322
|
|
|
|
323
|
|
|
// [0][0] contains the matched text |
324
|
|
|
// [0][1] contains the index of that match |
325
|
84 |
|
$increment = $matches[0][1] + \strlen($matches[0][0]); |
326
|
|
|
|
327
|
84 |
|
$this->advanceBy($increment); |
328
|
|
|
|
329
|
84 |
|
return $this->currentPosition - $this->previousPosition; |
330
|
|
|
} |
331
|
|
|
|
332
|
|
|
/** |
333
|
|
|
* Move the position to the very end of the line |
334
|
|
|
* |
335
|
|
|
* @return int The number of characters moved |
336
|
|
|
*/ |
337
|
798 |
|
public function advanceToEnd(): int |
338
|
|
|
{ |
339
|
798 |
|
$this->previousPosition = $this->currentPosition; |
340
|
798 |
|
$this->nextNonSpaceCache = null; |
341
|
|
|
|
342
|
798 |
|
$this->currentPosition = $this->length; |
343
|
|
|
|
344
|
798 |
|
return $this->currentPosition - $this->previousPosition; |
345
|
|
|
} |
346
|
|
|
|
347
|
2634 |
|
public function getRemainder(): string |
348
|
|
|
{ |
349
|
2634 |
|
if ($this->currentPosition >= $this->length) { |
350
|
555 |
|
return ''; |
351
|
|
|
} |
352
|
|
|
|
353
|
2607 |
|
$prefix = ''; |
354
|
2607 |
|
$position = $this->currentPosition; |
355
|
2607 |
|
if ($this->partiallyConsumedTab) { |
356
|
12 |
|
$position++; |
357
|
12 |
|
$charsToTab = 4 - ($this->column % 4); |
358
|
12 |
|
$prefix = \str_repeat(' ', $charsToTab); |
359
|
|
|
} |
360
|
|
|
|
361
|
2607 |
|
$subString = $this->isMultibyte ? |
362
|
72 |
|
\mb_substr($this->line, $position, null, 'UTF-8') : |
363
|
2607 |
|
\substr($this->line, $position); |
364
|
|
|
|
365
|
2607 |
|
return $prefix . $subString; |
366
|
|
|
} |
367
|
|
|
|
368
|
1965 |
|
public function getLine(): string |
369
|
|
|
{ |
370
|
1965 |
|
return $this->line; |
371
|
|
|
} |
372
|
|
|
|
373
|
2154 |
|
public function isAtEnd(): bool |
374
|
|
|
{ |
375
|
2154 |
|
return $this->currentPosition >= $this->length; |
376
|
|
|
} |
377
|
|
|
|
378
|
|
|
/** |
379
|
|
|
* Try to match a regular expression |
380
|
|
|
* |
381
|
|
|
* Returns the matching text and advances to the end of that match |
382
|
|
|
*/ |
383
|
2352 |
|
public function match(string $regex): ?string |
384
|
|
|
{ |
385
|
2352 |
|
$subject = $this->getRemainder(); |
386
|
|
|
|
387
|
2352 |
|
if (! \preg_match($regex, $subject, $matches, \PREG_OFFSET_CAPTURE)) { |
388
|
909 |
|
return null; |
389
|
|
|
} |
390
|
|
|
|
391
|
|
|
// $matches[0][0] contains the matched text |
392
|
|
|
// $matches[0][1] contains the index of that match |
393
|
|
|
|
394
|
2319 |
|
if ($this->isMultibyte) { |
395
|
|
|
// PREG_OFFSET_CAPTURE always returns the byte offset, not the char offset, which is annoying |
396
|
57 |
|
$offset = \mb_strlen(\substr($subject, 0, $matches[0][1]), 'UTF-8'); |
397
|
57 |
|
$matchLength = \mb_strlen($matches[0][0], 'UTF-8'); |
398
|
|
|
} else { |
399
|
2262 |
|
$offset = $matches[0][1]; |
400
|
2262 |
|
$matchLength = \strlen($matches[0][0]); |
401
|
|
|
} |
402
|
|
|
|
403
|
|
|
// [0][0] contains the matched text |
404
|
|
|
// [0][1] contains the index of that match |
405
|
2319 |
|
$this->advanceBy($offset + $matchLength); |
406
|
|
|
|
407
|
2319 |
|
return $matches[0][0]; |
408
|
|
|
} |
409
|
|
|
|
410
|
|
|
/** |
411
|
|
|
* Encapsulates the current state of this cursor in case you need to rollback later. |
412
|
|
|
* |
413
|
|
|
* WARNING: Do not parse or use the return value for ANYTHING except for |
414
|
|
|
* passing it back into restoreState(), as the number of values and their |
415
|
|
|
* contents may change in any future release without warning. |
416
|
|
|
*/ |
417
|
1689 |
|
public function saveState(): CursorState |
418
|
|
|
{ |
419
|
1689 |
|
return new CursorState([ |
420
|
1689 |
|
$this->currentPosition, |
421
|
1689 |
|
$this->previousPosition, |
422
|
1689 |
|
$this->nextNonSpaceCache, |
423
|
1689 |
|
$this->indent, |
424
|
1689 |
|
$this->column, |
425
|
1689 |
|
$this->partiallyConsumedTab, |
426
|
|
|
]); |
427
|
|
|
} |
428
|
|
|
|
429
|
|
|
/** |
430
|
|
|
* Restore the cursor to a previous state. |
431
|
|
|
* |
432
|
|
|
* Pass in the value previously obtained by calling saveState(). |
433
|
|
|
*/ |
434
|
1596 |
|
public function restoreState(CursorState $state): void |
435
|
|
|
{ |
436
|
|
|
[ |
437
|
1596 |
|
$this->currentPosition, |
438
|
1596 |
|
$this->previousPosition, |
439
|
1596 |
|
$this->nextNonSpaceCache, |
440
|
1596 |
|
$this->indent, |
441
|
1596 |
|
$this->column, |
442
|
1596 |
|
$this->partiallyConsumedTab, |
443
|
1596 |
|
] = $state->toArray(); |
444
|
1596 |
|
} |
445
|
|
|
|
446
|
735 |
|
public function getPosition(): int |
447
|
|
|
{ |
448
|
735 |
|
return $this->currentPosition; |
449
|
|
|
} |
450
|
|
|
|
451
|
399 |
|
public function getPreviousText(): string |
452
|
|
|
{ |
453
|
399 |
|
return \mb_substr($this->line, $this->previousPosition, $this->currentPosition - $this->previousPosition, 'UTF-8'); |
454
|
|
|
} |
455
|
|
|
|
456
|
426 |
|
public function getSubstring(int $start, ?int $length = null): string |
457
|
|
|
{ |
458
|
426 |
|
if ($this->isMultibyte) { |
459
|
21 |
|
return \mb_substr($this->line, $start, $length, 'UTF-8'); |
460
|
|
|
} |
461
|
|
|
|
462
|
405 |
|
if ($length !== null) { |
463
|
402 |
|
return \substr($this->line, $start, $length); |
464
|
|
|
} |
465
|
|
|
|
466
|
3 |
|
return \substr($this->line, $start); |
467
|
|
|
} |
468
|
|
|
|
469
|
285 |
|
public function getColumn(): int |
470
|
|
|
{ |
471
|
285 |
|
return $this->column; |
472
|
|
|
} |
473
|
|
|
} |
474
|
|
|
|
This check looks for variable assignements that are either overwritten by other assignments or where the variable is not used subsequently.
Both the
$myVar
assignment in line 1 and the$higher
assignment in line 2 are dead. The first because$myVar
is never used and the second because$higher
is always overwritten for every possible time line.