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