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
|
|
|
* Original code based on the CommonMark JS reference parser (https://bitly.com/commonmark-js) |
11
|
|
|
* - (c) John MacFarlane |
12
|
|
|
* |
13
|
|
|
* Additional code based on commonmark-java (https://github.com/commonmark/commonmark-java) |
14
|
|
|
* - (c) Atlassian Pty Ltd |
15
|
|
|
* |
16
|
|
|
* For the full copyright and license information, please view the LICENSE |
17
|
|
|
* file that was distributed with this source code. |
18
|
|
|
*/ |
19
|
|
|
|
20
|
|
|
namespace League\CommonMark\Parser; |
21
|
|
|
|
22
|
|
|
use League\CommonMark\Environment\EnvironmentInterface; |
23
|
|
|
use League\CommonMark\Event\DocumentParsedEvent; |
24
|
|
|
use League\CommonMark\Event\DocumentPreParsedEvent; |
25
|
|
|
use League\CommonMark\Input\MarkdownInput; |
26
|
|
|
use League\CommonMark\Node\Block\Document; |
27
|
|
|
use League\CommonMark\Node\Block\Paragraph; |
28
|
|
|
use League\CommonMark\Parser\Block\BlockContinueParserInterface; |
29
|
|
|
use League\CommonMark\Parser\Block\BlockContinueParserWithInlinesInterface; |
30
|
|
|
use League\CommonMark\Parser\Block\BlockStart; |
31
|
|
|
use League\CommonMark\Parser\Block\BlockStartParserInterface; |
32
|
|
|
use League\CommonMark\Parser\Block\DocumentBlockParser; |
33
|
|
|
use League\CommonMark\Parser\Block\ParagraphParser; |
34
|
|
|
use League\CommonMark\Reference\ReferenceInterface; |
35
|
|
|
use League\CommonMark\Reference\ReferenceMap; |
36
|
|
|
use League\CommonMark\Util\RegexHelper; |
37
|
|
|
|
38
|
|
|
final class MarkdownParser implements MarkdownParserInterface |
39
|
|
|
{ |
40
|
|
|
/** |
41
|
|
|
* @var EnvironmentInterface |
42
|
|
|
* |
43
|
|
|
* @psalm-readonly |
44
|
|
|
*/ |
45
|
|
|
private $environment; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* @var int |
49
|
|
|
* |
50
|
|
|
* @psalm-readonly-allow-private-mutation |
51
|
|
|
*/ |
52
|
|
|
private $maxNestingLevel; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* @var ReferenceMap |
56
|
|
|
* |
57
|
|
|
* @psalm-readonly-allow-private-mutation |
58
|
|
|
*/ |
59
|
|
|
private $referenceMap; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @var int |
63
|
|
|
* |
64
|
|
|
* @psalm-readonly-allow-private-mutation |
65
|
|
|
*/ |
66
|
|
|
private $lineNumber = 0; |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* @var Cursor |
70
|
|
|
* |
71
|
|
|
* @psalm-readonly-allow-private-mutation |
72
|
|
|
*/ |
73
|
|
|
private $cursor; |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* @var array<int, BlockContinueParserInterface> |
77
|
|
|
* |
78
|
|
|
* @psalm-readonly-allow-private-mutation |
79
|
|
|
*/ |
80
|
|
|
private $activeBlockParsers = []; |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* @var array<int, BlockContinueParserInterface> |
84
|
|
|
* |
85
|
|
|
* @psalm-readonly-allow-private-mutation |
86
|
|
|
*/ |
87
|
|
|
private $closedBlockParsers = []; |
88
|
|
|
|
89
|
2982 |
|
public function __construct(EnvironmentInterface $environment) |
90
|
|
|
{ |
91
|
2982 |
|
$this->environment = $environment; |
92
|
2982 |
|
} |
93
|
|
|
|
94
|
2964 |
|
private function initialize(): void |
95
|
|
|
{ |
96
|
2964 |
|
$this->referenceMap = new ReferenceMap(); |
97
|
2964 |
|
$this->lineNumber = 0; |
98
|
2964 |
|
$this->activeBlockParsers = []; |
99
|
2964 |
|
$this->closedBlockParsers = []; |
100
|
|
|
|
101
|
2964 |
|
$this->maxNestingLevel = $this->environment->getConfiguration()->get('max_nesting_level'); |
102
|
2952 |
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @throws \RuntimeException |
106
|
|
|
*/ |
107
|
2964 |
|
public function parse(string $input): Document |
108
|
|
|
{ |
109
|
2964 |
|
$this->initialize(); |
110
|
|
|
|
111
|
2952 |
|
$documentParser = new DocumentBlockParser($this->referenceMap); |
112
|
2952 |
|
$this->activateBlockParser($documentParser); |
113
|
|
|
|
114
|
2952 |
|
$preParsedEvent = new DocumentPreParsedEvent($documentParser->getBlock(), new MarkdownInput($input)); |
115
|
2943 |
|
$this->environment->dispatch($preParsedEvent); |
116
|
2940 |
|
$markdownInput = $preParsedEvent->getMarkdown(); |
117
|
|
|
|
118
|
2940 |
|
foreach ($markdownInput->getLines() as $lineNumber => $line) { |
119
|
2940 |
|
$this->lineNumber = $lineNumber; |
120
|
2940 |
|
$this->incorporateLine($line); |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
// finalizeAndProcess |
124
|
2940 |
|
$this->closeBlockParsers(\count($this->activeBlockParsers), $this->lineNumber); |
125
|
2940 |
|
$this->processInlines(); |
126
|
|
|
|
127
|
2940 |
|
$this->environment->dispatch(new DocumentParsedEvent($documentParser->getBlock())); |
128
|
|
|
|
129
|
2940 |
|
return $documentParser->getBlock(); |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
/** |
133
|
|
|
* Analyze a line of text and update the document appropriately. We parse markdown text by calling this on each |
134
|
|
|
* line of input, then finalizing the document. |
135
|
|
|
*/ |
136
|
2940 |
|
private function incorporateLine(string $line): void |
137
|
|
|
{ |
138
|
2940 |
|
$this->cursor = new Cursor($line); |
139
|
|
|
|
140
|
2940 |
|
$matches = 1; |
141
|
2940 |
|
for ($i = 1; $i < \count($this->activeBlockParsers); $i++) { |
|
|
|
|
142
|
1437 |
|
$blockParser = $this->activeBlockParsers[$i]; |
143
|
1437 |
|
$blockContinue = $blockParser->tryContinue(clone $this->cursor, $this->getActiveBlockParser()); |
144
|
1437 |
|
if ($blockContinue === null) { |
145
|
945 |
|
break; |
146
|
|
|
} |
147
|
|
|
|
148
|
963 |
|
if ($blockContinue->isFinalize()) { |
149
|
90 |
|
$this->closeBlockParsers(\count($this->activeBlockParsers) - $i, $this->lineNumber); |
150
|
|
|
|
151
|
90 |
|
return; |
152
|
|
|
} |
153
|
|
|
|
154
|
957 |
|
if (($state = $blockContinue->getCursorState()) !== null) { |
155
|
957 |
|
$this->cursor->restoreState($state); |
156
|
|
|
} |
157
|
|
|
|
158
|
957 |
|
$matches++; |
159
|
|
|
} |
160
|
|
|
|
161
|
2940 |
|
$unmatchedBlocks = \count($this->activeBlockParsers) - $matches; |
162
|
2940 |
|
$blockParser = $this->activeBlockParsers[$matches - 1]; |
163
|
2940 |
|
$startedNewBlock = false; |
164
|
|
|
|
165
|
|
|
// Unless last matched container is a code block, try new container starts |
166
|
2940 |
|
$tryBlockStarts = $blockParser->getBlock() instanceof Paragraph || $blockParser->isContainer(); |
167
|
2940 |
|
while ($tryBlockStarts) { |
168
|
|
|
// this is a little performance optimization |
169
|
2940 |
|
if ($this->cursor->isBlank()) { |
170
|
738 |
|
$this->cursor->advanceToEnd(); |
171
|
738 |
|
break; |
172
|
|
|
} |
173
|
|
|
|
174
|
2940 |
|
if (! $this->cursor->isIndented() && RegexHelper::isLetter($this->cursor->getNextNonSpaceCharacter())) { |
175
|
1395 |
|
$this->cursor->advanceToNextNonSpaceOrTab(); |
176
|
1395 |
|
break; |
177
|
|
|
} |
178
|
|
|
|
179
|
2298 |
|
if ($blockParser->getBlock()->getDepth() >= $this->maxNestingLevel) { |
180
|
3 |
|
break; |
181
|
|
|
} |
182
|
|
|
|
183
|
2298 |
|
$blockStart = $this->findBlockStart($blockParser); |
184
|
2298 |
|
if ($blockStart === null) { |
185
|
1377 |
|
$this->cursor->advanceToNextNonSpaceOrTab(); |
186
|
1377 |
|
break; |
187
|
|
|
} |
188
|
|
|
|
189
|
1080 |
|
if (($state = $blockStart->getCursorState()) !== null) { |
190
|
1080 |
|
$this->cursor->restoreState($state); |
191
|
|
|
} |
192
|
|
|
|
193
|
1080 |
|
$startedNewBlock = true; |
194
|
|
|
|
195
|
|
|
// We're starting a new block. If we have any previous blocks that need to be closed, we need to do it now. |
196
|
1080 |
|
if ($unmatchedBlocks > 0) { |
197
|
258 |
|
$this->closeBlockParsers($unmatchedBlocks, $this->lineNumber - 1); |
198
|
258 |
|
$unmatchedBlocks = 0; |
199
|
|
|
} |
200
|
|
|
|
201
|
1080 |
|
if ($blockStart->isReplaceActiveBlockParser()) { |
202
|
150 |
|
$this->prepareActiveBlockParserForReplacement(); |
203
|
|
|
} |
204
|
|
|
|
205
|
1080 |
|
foreach ($blockStart->getBlockParsers() as $newBlockParser) { |
206
|
1080 |
|
$blockParser = $this->addChild($newBlockParser); |
207
|
1080 |
|
$tryBlockStarts = $newBlockParser->isContainer(); |
208
|
|
|
} |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
// What remains ath the offset is a text line. Add the text to the appropriate block. |
212
|
|
|
|
213
|
|
|
// First check for a lazy paragraph continuation: |
214
|
2940 |
|
if (! $startedNewBlock && ! $this->cursor->isBlank() && $this->getActiveBlockParser()->canHaveLazyContinuationLines()) { |
215
|
447 |
|
$this->getActiveBlockParser()->addLine($this->cursor->getRemainder()); |
216
|
|
|
} else { |
217
|
|
|
// finalize any blocks not matched |
218
|
2940 |
|
if ($unmatchedBlocks > 0) { |
219
|
777 |
|
$this->closeBlockParsers($unmatchedBlocks, $this->lineNumber); |
220
|
|
|
} |
221
|
|
|
|
222
|
2940 |
|
if (! $blockParser->isContainer()) { |
223
|
780 |
|
$this->getActiveBlockParser()->addLine($this->cursor->getRemainder()); |
224
|
2613 |
|
} elseif (! $this->cursor->isBlank()) { |
225
|
2562 |
|
$this->addChild(new ParagraphParser()); |
226
|
2562 |
|
$this->getActiveBlockParser()->addLine($this->cursor->getRemainder()); |
227
|
|
|
} |
228
|
|
|
} |
229
|
2940 |
|
} |
230
|
|
|
|
231
|
2298 |
|
private function findBlockStart(BlockContinueParserInterface $lastMatchedBlockParser): ?BlockStart |
232
|
|
|
{ |
233
|
2298 |
|
$matchedBlockParser = new MarkdownParserState($this->getActiveBlockParser(), $lastMatchedBlockParser); |
234
|
|
|
|
235
|
2298 |
|
foreach ($this->environment->getBlockStartParsers() as $blockStartParser) { |
236
|
|
|
\assert($blockStartParser instanceof BlockStartParserInterface); |
237
|
2295 |
|
if (($result = $blockStartParser->tryStart(clone $this->cursor, $matchedBlockParser)) !== null) { |
238
|
1080 |
|
return $result; |
239
|
|
|
} |
240
|
|
|
} |
241
|
|
|
|
242
|
1377 |
|
return null; |
243
|
|
|
} |
244
|
|
|
|
245
|
2940 |
|
private function closeBlockParsers(int $count, int $endLineNumber): void |
246
|
|
|
{ |
247
|
2940 |
|
for ($i = 0; $i < $count; $i++) { |
248
|
2940 |
|
$blockParser = $this->deactivateBlockParser(); |
249
|
2940 |
|
$this->finalize($blockParser, $endLineNumber); |
250
|
|
|
// Remember for inline parsing |
251
|
2940 |
|
$this->closedBlockParsers[] = $blockParser; |
252
|
|
|
} |
253
|
2940 |
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* Finalize a block. Close it and do any necessary postprocessing, e.g. creating string_content from strings, |
257
|
|
|
* setting the 'tight' or 'loose' status of a list, and parsing the beginnings of paragraphs for reference |
258
|
|
|
* definitions. |
259
|
|
|
*/ |
260
|
2940 |
|
private function finalize(BlockContinueParserInterface $blockParser, int $endLineNumber): void |
261
|
|
|
{ |
262
|
2940 |
|
if ($blockParser instanceof ParagraphParser) { |
263
|
2454 |
|
$this->updateReferenceMap($blockParser->getReferences()); |
264
|
|
|
} |
265
|
|
|
|
266
|
2940 |
|
$blockParser->getBlock()->setEndLine($endLineNumber); |
267
|
2940 |
|
$blockParser->closeBlock(); |
268
|
2940 |
|
} |
269
|
|
|
|
270
|
|
|
/** |
271
|
|
|
* Walk through a block & children recursively, parsing string content into inline content where appropriate. |
272
|
|
|
*/ |
273
|
2940 |
|
private function processInlines(): void |
274
|
|
|
{ |
275
|
2940 |
|
$p = new InlineParserEngine($this->environment, $this->referenceMap); |
276
|
|
|
|
277
|
2940 |
|
foreach ($this->closedBlockParsers as $blockParser) { |
278
|
2940 |
|
if ($blockParser instanceof BlockContinueParserWithInlinesInterface) { |
279
|
2652 |
|
$blockParser->parseInlines($p); |
280
|
|
|
} |
281
|
|
|
} |
282
|
2940 |
|
} |
283
|
|
|
|
284
|
|
|
/** |
285
|
|
|
* Add block of type tag as a child of the tip. If the tip can't accept children, close and finalize it and try |
286
|
|
|
* its parent, and so on til we find a block that can accept children. |
287
|
|
|
*/ |
288
|
2940 |
|
private function addChild(BlockContinueParserInterface $blockParser): BlockContinueParserInterface |
289
|
|
|
{ |
290
|
2940 |
|
$blockParser->getBlock()->setStartLine($this->lineNumber); |
291
|
|
|
|
292
|
2940 |
|
while (! $this->getActiveBlockParser()->canContain($blockParser->getBlock())) { |
293
|
141 |
|
$this->closeBlockParsers(1, $this->lineNumber - 1); |
294
|
|
|
} |
295
|
|
|
|
296
|
2940 |
|
$this->getActiveBlockParser()->getBlock()->appendChild($blockParser->getBlock()); |
297
|
2940 |
|
$this->activateBlockParser($blockParser); |
298
|
|
|
|
299
|
2940 |
|
return $blockParser; |
300
|
|
|
} |
301
|
|
|
|
302
|
2952 |
|
private function activateBlockParser(BlockContinueParserInterface $blockParser): void |
303
|
|
|
{ |
304
|
2952 |
|
$this->activeBlockParsers[] = $blockParser; |
305
|
2952 |
|
} |
306
|
|
|
|
307
|
2940 |
|
private function deactivateBlockParser(): BlockContinueParserInterface |
308
|
|
|
{ |
309
|
2940 |
|
$popped = \array_pop($this->activeBlockParsers); |
310
|
2940 |
|
if ($popped === null) { |
311
|
|
|
throw new \RuntimeException('The last block parser should not be deactivated'); |
312
|
|
|
} |
313
|
|
|
|
314
|
2940 |
|
return $popped; |
315
|
|
|
} |
316
|
|
|
|
317
|
150 |
|
private function prepareActiveBlockParserForReplacement(): void |
318
|
|
|
{ |
319
|
|
|
// Note that we don't want to parse inlines or finalize this block, as it's getting replaced. |
320
|
150 |
|
$old = $this->deactivateBlockParser(); |
321
|
|
|
|
322
|
150 |
|
if ($old instanceof ParagraphParser) { |
323
|
150 |
|
$this->updateReferenceMap($old->getReferences()); |
324
|
|
|
} |
325
|
|
|
|
326
|
150 |
|
$old->getBlock()->detach(); |
327
|
150 |
|
} |
328
|
|
|
|
329
|
|
|
/** |
330
|
|
|
* @param ReferenceInterface[] $references |
331
|
|
|
*/ |
332
|
2562 |
|
private function updateReferenceMap(iterable $references): void |
333
|
|
|
{ |
334
|
2562 |
|
foreach ($references as $reference) { |
335
|
237 |
|
if (! $this->referenceMap->contains($reference->getLabel())) { |
336
|
237 |
|
$this->referenceMap->add($reference); |
337
|
|
|
} |
338
|
|
|
} |
339
|
2562 |
|
} |
340
|
|
|
|
341
|
2940 |
|
public function getActiveBlockParser(): BlockContinueParserInterface |
342
|
|
|
{ |
343
|
2940 |
|
$active = \end($this->activeBlockParsers); |
344
|
2940 |
|
if ($active === false) { |
345
|
|
|
throw new \RuntimeException('No active block parsers are available'); |
346
|
|
|
} |
347
|
|
|
|
348
|
2940 |
|
return $active; |
349
|
|
|
} |
350
|
|
|
} |
351
|
|
|
|
If the size of the collection does not change during the iteration, it is generally a good practice to compute it beforehand, and not on each iteration: