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
|
|
|
* For the full copyright and license information, please view the LICENSE |
14
|
|
|
* file that was distributed with this source code. |
15
|
|
|
*/ |
16
|
|
|
|
17
|
|
|
namespace League\CommonMark\Parser; |
18
|
|
|
|
19
|
|
|
use League\CommonMark\Delimiter\Delimiter; |
20
|
|
|
use League\CommonMark\Delimiter\Processor\DelimiterProcessorInterface; |
21
|
|
|
use League\CommonMark\Environment\EnvironmentInterface; |
22
|
|
|
use League\CommonMark\Node\Block\AbstractBlock; |
23
|
|
|
use League\CommonMark\Node\Inline\AdjacentTextMerger; |
24
|
|
|
use League\CommonMark\Node\Inline\Text; |
25
|
|
|
use League\CommonMark\Node\Node; |
26
|
|
|
use League\CommonMark\Reference\ReferenceMapInterface; |
27
|
|
|
use League\CommonMark\Util\RegexHelper; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* @internal |
31
|
|
|
*/ |
32
|
|
|
final class InlineParserEngine implements InlineParserEngineInterface |
33
|
|
|
{ |
34
|
|
|
/** |
35
|
|
|
* @var EnvironmentInterface |
36
|
|
|
* |
37
|
|
|
* @psalm-readonly |
38
|
|
|
*/ |
39
|
|
|
private $environment; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @var ReferenceMapInterface |
43
|
|
|
* |
44
|
|
|
* @psalm-readonly |
45
|
|
|
*/ |
46
|
|
|
private $referenceMap; |
47
|
|
|
|
48
|
2496 |
|
public function __construct(EnvironmentInterface $environment, ReferenceMapInterface $referenceMap) |
49
|
|
|
{ |
50
|
2496 |
|
$this->environment = $environment; |
51
|
2496 |
|
$this->referenceMap = $referenceMap; |
52
|
2496 |
|
} |
53
|
|
|
|
54
|
2208 |
|
public function parse(string $contents, AbstractBlock $block): void |
55
|
|
|
{ |
56
|
2208 |
|
$inlineParserContext = new InlineParserContext($contents, $block, $this->referenceMap); |
57
|
2208 |
|
$cursor = $inlineParserContext->getCursor(); |
58
|
2208 |
|
while (($character = $cursor->getCharacter()) !== null) { |
59
|
2205 |
|
if (! $this->parseCharacter($character, $inlineParserContext)) { |
60
|
2067 |
|
$this->addPlainText($character, $block, $inlineParserContext); |
61
|
|
|
} |
62
|
|
|
} |
63
|
|
|
|
64
|
2208 |
|
$delimiterStack = $inlineParserContext->getDelimiterStack(); |
65
|
2208 |
|
$delimiterStack->processDelimiters(null, $this->environment->getDelimiterProcessors()); |
66
|
2208 |
|
$delimiterStack->removeAll(); |
67
|
|
|
|
68
|
2208 |
|
AdjacentTextMerger::mergeChildNodes($block); |
69
|
2208 |
|
} |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @return bool Whether we successfully parsed a character at that position |
73
|
|
|
*/ |
74
|
2205 |
|
private function parseCharacter(string $character, InlineParserContext $inlineParserContext): bool |
75
|
|
|
{ |
76
|
2205 |
|
foreach ($this->environment->getInlineParsersForCharacter($character) as $parser) { |
77
|
1125 |
|
if ($parser->parse($inlineParserContext)) { |
78
|
1053 |
|
return true; |
79
|
|
|
} |
80
|
|
|
} |
81
|
|
|
|
82
|
2076 |
|
if ($delimiterProcessor = $this->environment->getDelimiterProcessors()->getDelimiterProcessor($character)) { |
83
|
675 |
|
return $this->parseDelimiters($delimiterProcessor, $inlineParserContext); |
84
|
|
|
} |
85
|
|
|
|
86
|
2067 |
|
return false; |
87
|
|
|
} |
88
|
|
|
|
89
|
675 |
|
private function parseDelimiters(DelimiterProcessorInterface $delimiterProcessor, InlineParserContext $inlineContext): bool |
90
|
|
|
{ |
91
|
675 |
|
$cursor = $inlineContext->getCursor(); |
92
|
675 |
|
$character = $cursor->getCharacter(); |
93
|
675 |
|
$numDelims = 0; |
94
|
|
|
|
95
|
675 |
|
if ($character === null) { |
96
|
|
|
throw new \RuntimeException('Cannot parse delimiters without a valid character'); |
97
|
|
|
} |
98
|
|
|
|
99
|
675 |
|
$charBefore = $cursor->peek(-1); |
100
|
675 |
|
if ($charBefore === null) { |
101
|
408 |
|
$charBefore = "\n"; |
102
|
|
|
} |
103
|
|
|
|
104
|
675 |
|
while ($cursor->peek($numDelims) === $character) { |
105
|
675 |
|
++$numDelims; |
106
|
|
|
} |
107
|
|
|
|
108
|
675 |
|
if ($numDelims < $delimiterProcessor->getMinLength()) { |
109
|
6 |
|
return false; |
110
|
|
|
} |
111
|
|
|
|
112
|
669 |
|
$cursor->advanceBy($numDelims); |
113
|
|
|
|
114
|
669 |
|
$charAfter = $cursor->getCharacter(); |
115
|
669 |
|
if ($charAfter === null) { |
116
|
414 |
|
$charAfter = "\n"; |
117
|
|
|
} |
118
|
|
|
|
119
|
669 |
|
[$canOpen, $canClose] = self::determineCanOpenOrClose($charBefore, $charAfter, $character, $delimiterProcessor); |
|
|
|
|
120
|
|
|
|
121
|
669 |
|
$node = new Text(\str_repeat($character, $numDelims), [ |
122
|
669 |
|
'delim' => true, |
123
|
|
|
]); |
124
|
669 |
|
$inlineContext->getContainer()->appendChild($node); |
125
|
|
|
|
126
|
|
|
// Add entry to stack to this opener |
127
|
669 |
|
if ($canOpen || $canClose) { |
128
|
606 |
|
$delimiter = new Delimiter($character, $numDelims, $node, $canOpen, $canClose); |
129
|
606 |
|
$inlineContext->getDelimiterStack()->push($delimiter); |
130
|
|
|
} |
131
|
|
|
|
132
|
669 |
|
return true; |
133
|
|
|
} |
134
|
|
|
|
135
|
2067 |
|
private function addPlainText(string $character, Node $container, InlineParserContext $inlineParserContext): void |
136
|
|
|
{ |
137
|
|
|
// We reach here if none of the parsers can handle the input |
138
|
|
|
// Attempt to match multiple non-special characters at once |
139
|
2067 |
|
$text = $inlineParserContext->getCursor()->match($this->environment->getInlineParserCharacterRegex()); |
140
|
|
|
// This might fail if we're currently at a special character which wasn't parsed; if so, just add that character |
141
|
2067 |
|
if ($text === null) { |
142
|
285 |
|
$inlineParserContext->getCursor()->advanceBy(1); |
143
|
285 |
|
$text = $character; |
144
|
|
|
} |
145
|
|
|
|
146
|
2067 |
|
$lastInline = $container->lastChild(); |
147
|
2067 |
|
if ($lastInline instanceof Text && ! isset($lastInline->data['delim'])) { |
148
|
330 |
|
$lastInline->append($text); |
149
|
|
|
} else { |
150
|
2031 |
|
$container->appendChild(new Text($text)); |
151
|
|
|
} |
152
|
2067 |
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* @return bool[] |
156
|
|
|
*/ |
157
|
669 |
|
private static function determineCanOpenOrClose(string $charBefore, string $charAfter, string $character, DelimiterProcessorInterface $delimiterProcessor): array |
158
|
|
|
{ |
159
|
669 |
|
$afterIsWhitespace = \preg_match(RegexHelper::REGEX_UNICODE_WHITESPACE_CHAR, $charAfter); |
160
|
669 |
|
$afterIsPunctuation = \preg_match(RegexHelper::REGEX_PUNCTUATION, $charAfter); |
161
|
669 |
|
$beforeIsWhitespace = \preg_match(RegexHelper::REGEX_UNICODE_WHITESPACE_CHAR, $charBefore); |
162
|
669 |
|
$beforeIsPunctuation = \preg_match(RegexHelper::REGEX_PUNCTUATION, $charBefore); |
163
|
|
|
|
164
|
669 |
|
$leftFlanking = ! $afterIsWhitespace && (! $afterIsPunctuation || $beforeIsWhitespace || $beforeIsPunctuation); |
165
|
669 |
|
$rightFlanking = ! $beforeIsWhitespace && (! $beforeIsPunctuation || $afterIsWhitespace || $afterIsPunctuation); |
166
|
|
|
|
167
|
669 |
|
if ($character === '_') { |
168
|
234 |
|
$canOpen = $leftFlanking && (! $rightFlanking || $beforeIsPunctuation); |
169
|
234 |
|
$canClose = $rightFlanking && (! $leftFlanking || $afterIsPunctuation); |
170
|
|
|
} else { |
171
|
471 |
|
$canOpen = $leftFlanking && $character === $delimiterProcessor->getOpeningCharacter(); |
172
|
471 |
|
$canClose = $rightFlanking && $character === $delimiterProcessor->getClosingCharacter(); |
173
|
|
|
} |
174
|
|
|
|
175
|
669 |
|
return [$canOpen, $canClose]; |
176
|
|
|
} |
177
|
|
|
} |
178
|
|
|
|
This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug.