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 $firstNonSpaceCache; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @param string $line |
58
|
|
|
*/ |
59
|
2292 |
|
public function __construct($line) |
60
|
|
|
{ |
61
|
2292 |
|
$this->line = $line; |
62
|
2292 |
|
$this->length = mb_strlen($line, 'utf-8'); |
63
|
2292 |
|
} |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* Returns the position of the next non-space character |
67
|
|
|
* |
68
|
|
|
* @return int |
69
|
|
|
*/ |
70
|
1980 |
|
public function getFirstNonSpacePosition() |
71
|
|
|
{ |
72
|
1980 |
|
if ($this->firstNonSpaceCache !== null) { |
73
|
1854 |
|
return $this->firstNonSpaceCache; |
74
|
|
|
} |
75
|
|
|
|
76
|
1980 |
|
$i = $this->currentPosition; |
77
|
1980 |
|
$cols = $this->column; |
78
|
|
|
|
79
|
1980 |
|
while (($c = $this->getCharacter($i)) !== null) { |
80
|
1965 |
|
if ($c === ' ') { |
81
|
474 |
|
$i++; |
82
|
474 |
|
$cols++; |
83
|
1965 |
|
} elseif ($c === "\t") { |
84
|
18 |
|
$i++; |
85
|
18 |
|
$cols += (4 - ($cols % 4)); |
86
|
18 |
|
} else { |
87
|
1932 |
|
break; |
88
|
|
|
} |
89
|
480 |
|
} |
90
|
|
|
|
91
|
1980 |
|
$nextNonSpace = ($c === null) ? $this->length : $i; |
92
|
1980 |
|
$this->indent = $cols - $this->column; |
93
|
|
|
|
94
|
1980 |
|
return $this->firstNonSpaceCache = $nextNonSpace; |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
/** |
98
|
|
|
* Returns the next character which isn't a space |
99
|
|
|
* |
100
|
|
|
* @return string |
101
|
|
|
*/ |
102
|
1827 |
|
public function getFirstNonSpaceCharacter() |
103
|
|
|
{ |
104
|
1827 |
|
return $this->getCharacter($this->getFirstNonSpacePosition()); |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
/** |
108
|
|
|
* Calculates the current indent (number of spaces after current position) |
109
|
|
|
* |
110
|
|
|
* @return int |
111
|
|
|
*/ |
112
|
1914 |
|
public function getIndent() |
113
|
|
|
{ |
114
|
1914 |
|
$this->getFirstNonSpacePosition(); |
115
|
|
|
|
116
|
1914 |
|
return $this->indent; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* Whether the cursor is indented to INDENT_LEVEL |
121
|
|
|
* |
122
|
|
|
* @return bool |
123
|
|
|
*/ |
124
|
1854 |
|
public function isIndented() |
125
|
|
|
{ |
126
|
1854 |
|
return $this->getIndent() >= self::INDENT_LEVEL; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* @param int|null $index |
131
|
|
|
* |
132
|
|
|
* @return string|null |
133
|
|
|
*/ |
134
|
2073 |
|
public function getCharacter($index = null) |
135
|
|
|
{ |
136
|
2073 |
|
if ($index === null) { |
137
|
1608 |
|
$index = $this->currentPosition; |
138
|
1608 |
|
} |
139
|
|
|
|
140
|
|
|
// Index out-of-bounds, or we're at the end |
141
|
2073 |
|
if ($index < 0 || $index >= $this->length) { |
142
|
1800 |
|
return; |
143
|
|
|
} |
144
|
|
|
|
145
|
2040 |
|
return mb_substr($this->line, $index, 1, 'utf-8'); |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* Returns the next character (or null, if none) without advancing forwards |
150
|
|
|
* |
151
|
|
|
* @param int $offset |
152
|
|
|
* |
153
|
|
|
* @return string|null |
154
|
|
|
*/ |
155
|
978 |
|
public function peek($offset = 1) |
156
|
|
|
{ |
157
|
978 |
|
return $this->getCharacter($this->currentPosition + $offset); |
158
|
|
|
} |
159
|
|
|
|
160
|
|
|
/** |
161
|
|
|
* Whether the remainder is blank |
162
|
|
|
* |
163
|
|
|
* @return bool |
164
|
|
|
*/ |
165
|
1872 |
|
public function isBlank() |
166
|
|
|
{ |
167
|
1872 |
|
return $this->getFirstNonSpacePosition() === $this->length; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* Move the cursor forwards |
172
|
|
|
*/ |
173
|
756 |
|
public function advance() |
174
|
|
|
{ |
175
|
756 |
|
$this->advanceBy(1); |
176
|
756 |
|
} |
177
|
|
|
|
178
|
|
|
/** |
179
|
|
|
* Move the cursor forwards |
180
|
|
|
* |
181
|
|
|
* @param int $characters Number of characters to advance by |
182
|
|
|
*/ |
183
|
2115 |
|
public function advanceBy($characters, $advanceByColumns = false) |
184
|
|
|
{ |
185
|
2115 |
|
$this->firstNonSpaceCache = null; |
186
|
|
|
|
187
|
2115 |
|
$cols = 0; |
188
|
2115 |
|
$relPos = -1; |
189
|
|
|
|
190
|
2115 |
|
$nextFewChars = mb_substr($this->line, $this->currentPosition, $characters, 'utf-8'); |
191
|
2115 |
|
if ($characters === 1) { |
192
|
1389 |
|
$asArray = [$nextFewChars]; |
193
|
1389 |
|
} else { |
194
|
1986 |
|
$asArray = preg_split('//u', $nextFewChars, null, PREG_SPLIT_NO_EMPTY); |
195
|
|
|
} |
196
|
|
|
|
197
|
2115 |
|
foreach ($asArray as $relPos => $char) { |
198
|
2040 |
|
if ($char === "\t") { |
199
|
27 |
|
$cols += (4 - (($this->column + $cols) % 4)); |
200
|
27 |
|
} else { |
201
|
2037 |
|
$cols++; |
202
|
|
|
} |
203
|
|
|
|
204
|
2040 |
|
if ($advanceByColumns && $cols >= $characters) { |
205
|
327 |
|
break; |
206
|
|
|
} |
207
|
2115 |
|
} |
208
|
|
|
|
209
|
2115 |
|
$i = $advanceByColumns ? $relPos + 1 : $characters; |
210
|
|
|
|
211
|
2115 |
|
$this->previousPosition = $this->currentPosition; |
212
|
2115 |
|
$newPosition = $this->currentPosition + $i; |
213
|
|
|
|
214
|
2115 |
|
$this->column += $cols; |
215
|
|
|
|
216
|
2115 |
|
if ($newPosition >= $this->length) { |
217
|
1821 |
|
$this->currentPosition = $this->length; |
218
|
1821 |
|
} else { |
219
|
1839 |
|
$this->currentPosition = $newPosition; |
|
|
|
|
220
|
|
|
} |
221
|
2115 |
|
} |
222
|
|
|
|
223
|
|
|
/** |
224
|
|
|
* Advances the cursor while the given character is matched |
225
|
|
|
* |
226
|
|
|
* @param string $character Character to match |
227
|
|
|
* @param int|null $maximumCharactersToAdvance Maximum number of characters to advance before giving up |
228
|
|
|
* |
229
|
|
|
* @return int Number of positions moved (0 if unsuccessful) |
230
|
|
|
*/ |
231
|
144 |
|
public function advanceWhileMatches($character, $maximumCharactersToAdvance = null) |
232
|
|
|
{ |
233
|
|
|
// Calculate how far to advance |
234
|
144 |
|
$start = $this->currentPosition; |
235
|
144 |
|
$newIndex = $start; |
236
|
144 |
|
if ($maximumCharactersToAdvance === null) { |
237
|
18 |
|
$maximumCharactersToAdvance = $this->length; |
238
|
18 |
|
} |
239
|
|
|
|
240
|
144 |
|
$max = min($start + $maximumCharactersToAdvance, $this->length); |
241
|
|
|
|
242
|
144 |
|
while ($newIndex < $max && $this->getCharacter($newIndex) === $character) { |
243
|
45 |
|
++$newIndex; |
244
|
45 |
|
} |
245
|
|
|
|
246
|
144 |
|
if ($newIndex <= $start) { |
247
|
108 |
|
return 0; |
248
|
|
|
} |
249
|
|
|
|
250
|
45 |
|
$this->advanceBy($newIndex - $start); |
251
|
|
|
|
252
|
45 |
|
return $this->currentPosition - $this->previousPosition; |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* Parse zero or more space characters, including at most one newline |
257
|
|
|
* |
258
|
|
|
* @return int Number of positions moved |
259
|
|
|
*/ |
260
|
1863 |
|
public function advanceToFirstNonSpace() |
261
|
|
|
{ |
262
|
1863 |
|
$matches = []; |
263
|
1863 |
|
preg_match('/^ *(?:\n *)?/', $this->getRemainder(), $matches, PREG_OFFSET_CAPTURE); |
264
|
|
|
|
265
|
|
|
// [0][0] contains the matched text |
266
|
|
|
// [0][1] contains the index of that match |
267
|
1863 |
|
$increment = $matches[0][1] + strlen($matches[0][0]); |
268
|
|
|
|
269
|
1863 |
|
if ($increment === 0) { |
270
|
1800 |
|
return 0; |
271
|
|
|
} |
272
|
|
|
|
273
|
495 |
|
$this->advanceBy($increment); |
274
|
|
|
|
275
|
495 |
|
return $this->currentPosition - $this->previousPosition; |
276
|
|
|
} |
277
|
|
|
|
278
|
|
|
/** |
279
|
|
|
* @return string |
280
|
|
|
*/ |
281
|
1956 |
|
public function getRemainder() |
282
|
|
|
{ |
283
|
1956 |
|
if ($this->isAtEnd()) { |
284
|
657 |
|
return ''; |
285
|
|
|
} else { |
286
|
1944 |
|
return mb_substr($this->line, $this->currentPosition, null, 'utf-8'); |
287
|
|
|
} |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
/** |
291
|
|
|
* @return string |
292
|
|
|
*/ |
293
|
1812 |
|
public function getLine() |
294
|
|
|
{ |
295
|
1812 |
|
return $this->line; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
/** |
299
|
|
|
* @return bool |
300
|
|
|
*/ |
301
|
1977 |
|
public function isAtEnd() |
302
|
|
|
{ |
303
|
1977 |
|
return $this->currentPosition >= $this->length; |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
/** |
307
|
|
|
* Try to match a regular expression |
308
|
|
|
* |
309
|
|
|
* Returns the matching text and advances to the end of that match |
310
|
|
|
* |
311
|
|
|
* @param string $regex |
312
|
|
|
* |
313
|
|
|
* @return string|null |
314
|
|
|
*/ |
315
|
1827 |
|
public function match($regex) |
316
|
|
|
{ |
317
|
1827 |
|
$subject = $this->getRemainder(); |
318
|
|
|
|
319
|
1827 |
|
$matches = []; |
320
|
1827 |
|
if (!preg_match($regex, $subject, $matches, PREG_OFFSET_CAPTURE)) { |
321
|
1689 |
|
return; |
322
|
|
|
} |
323
|
|
|
|
324
|
|
|
// PREG_OFFSET_CAPTURE always returns the byte offset, not the char offset, which is annoying |
325
|
1713 |
|
$offset = mb_strlen(mb_strcut($subject, 0, $matches[0][1], 'utf-8'), 'utf-8'); |
326
|
|
|
|
327
|
|
|
// [0][0] contains the matched text |
328
|
|
|
// [0][1] contains the index of that match |
329
|
1713 |
|
$this->advanceBy($offset + mb_strlen($matches[0][0], 'utf-8')); |
330
|
|
|
|
331
|
1713 |
|
return $matches[0][0]; |
332
|
|
|
} |
333
|
|
|
|
334
|
|
|
/** |
335
|
|
|
* @return CursorState |
336
|
|
|
*/ |
337
|
1770 |
|
public function saveState() |
338
|
|
|
{ |
339
|
1770 |
|
return new CursorState( |
340
|
1770 |
|
$this->line, |
341
|
1770 |
|
$this->length, |
342
|
1770 |
|
$this->currentPosition, |
343
|
1770 |
|
$this->previousPosition, |
344
|
1770 |
|
$this->firstNonSpaceCache, |
345
|
1770 |
|
$this->indent, |
346
|
1770 |
|
$this->column |
347
|
1770 |
|
); |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
/** |
351
|
|
|
* @param CursorState $state |
352
|
|
|
*/ |
353
|
1689 |
|
public function restoreState(CursorState $state) |
354
|
|
|
{ |
355
|
1689 |
|
$this->line = $state->getLine(); |
356
|
1689 |
|
$this->length = $state->getLength(); |
357
|
1689 |
|
$this->currentPosition = $state->getCurrentPosition(); |
358
|
1689 |
|
$this->previousPosition = $state->getPreviousPosition(); |
359
|
1689 |
|
$this->firstNonSpaceCache = $state->getFirstNonSpaceCache(); |
360
|
1689 |
|
$this->column = $state->getColumn(); |
361
|
1689 |
|
$this->indent = $state->getIndent(); |
362
|
1689 |
|
} |
363
|
|
|
|
364
|
|
|
/** |
365
|
|
|
* @return int |
366
|
|
|
*/ |
367
|
588 |
|
public function getPosition() |
368
|
|
|
{ |
369
|
588 |
|
return $this->currentPosition; |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
/** |
373
|
|
|
* @return string |
374
|
|
|
*/ |
375
|
810 |
|
public function getPreviousText() |
376
|
|
|
{ |
377
|
810 |
|
return mb_substr($this->line, $this->previousPosition, $this->currentPosition - $this->previousPosition, 'utf-8'); |
378
|
|
|
} |
379
|
|
|
|
380
|
|
|
/** |
381
|
|
|
* @return int |
382
|
|
|
*/ |
383
|
234 |
|
public function getColumn() |
384
|
|
|
{ |
385
|
234 |
|
return $this->column; |
386
|
|
|
} |
387
|
|
|
} |
388
|
|
|
|
This check looks for assignments to scalar types that may be of the wrong type.
To ensure the code behaves as expected, it may be a good idea to add an explicit type cast.