| 148 |
lars |
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 <colinodell@gmail.com>
|
|
|
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\Util;
|
|
|
18 |
|
|
|
19 |
use League\CommonMark\Parser\Cursor;
|
|
|
20 |
|
|
|
21 |
/**
|
|
|
22 |
* @psalm-immutable
|
|
|
23 |
*/
|
|
|
24 |
final class LinkParserHelper
|
|
|
25 |
{
|
|
|
26 |
/**
|
|
|
27 |
* Attempt to parse link destination
|
|
|
28 |
*
|
|
|
29 |
* @return string|null The string, or null if no match
|
|
|
30 |
*/
|
|
|
31 |
public static function parseLinkDestination(Cursor $cursor): ?string
|
|
|
32 |
{
|
|
|
33 |
if ($res = $cursor->match(RegexHelper::REGEX_LINK_DESTINATION_BRACES)) {
|
|
|
34 |
// Chop off surrounding <..>:
|
|
|
35 |
return UrlEncoder::unescapeAndEncode(
|
|
|
36 |
RegexHelper::unescape(\substr($res, 1, -1))
|
|
|
37 |
);
|
|
|
38 |
}
|
|
|
39 |
|
|
|
40 |
if ($cursor->getCurrentCharacter() === '<') {
|
|
|
41 |
return null;
|
|
|
42 |
}
|
|
|
43 |
|
|
|
44 |
$destination = self::manuallyParseLinkDestination($cursor);
|
|
|
45 |
if ($destination === null) {
|
|
|
46 |
return null;
|
|
|
47 |
}
|
|
|
48 |
|
|
|
49 |
return UrlEncoder::unescapeAndEncode(
|
|
|
50 |
RegexHelper::unescape($destination)
|
|
|
51 |
);
|
|
|
52 |
}
|
|
|
53 |
|
|
|
54 |
public static function parseLinkLabel(Cursor $cursor): int
|
|
|
55 |
{
|
|
|
56 |
$match = $cursor->match('/^\[(?:[^\\\\\[\]]|\\\\.){0,1000}\]/');
|
|
|
57 |
if ($match === null) {
|
|
|
58 |
return 0;
|
|
|
59 |
}
|
|
|
60 |
|
|
|
61 |
$length = \mb_strlen($match, 'UTF-8');
|
|
|
62 |
|
|
|
63 |
if ($length > 1001) {
|
|
|
64 |
return 0;
|
|
|
65 |
}
|
|
|
66 |
|
|
|
67 |
return $length;
|
|
|
68 |
}
|
|
|
69 |
|
|
|
70 |
public static function parsePartialLinkLabel(Cursor $cursor): ?string
|
|
|
71 |
{
|
|
|
72 |
return $cursor->match('/^(?:[^\\\\\[\]]+|\\\\.?)*/');
|
|
|
73 |
}
|
|
|
74 |
|
|
|
75 |
/**
|
|
|
76 |
* Attempt to parse link title (sans quotes)
|
|
|
77 |
*
|
|
|
78 |
* @return string|null The string, or null if no match
|
|
|
79 |
*/
|
|
|
80 |
public static function parseLinkTitle(Cursor $cursor): ?string
|
|
|
81 |
{
|
|
|
82 |
if ($title = $cursor->match('/' . RegexHelper::PARTIAL_LINK_TITLE . '/')) {
|
|
|
83 |
// Chop off quotes from title and unescape
|
|
|
84 |
return RegexHelper::unescape(\substr($title, 1, -1));
|
|
|
85 |
}
|
|
|
86 |
|
|
|
87 |
return null;
|
|
|
88 |
}
|
|
|
89 |
|
|
|
90 |
public static function parsePartialLinkTitle(Cursor $cursor, string $endDelimiter): ?string
|
|
|
91 |
{
|
|
|
92 |
$endDelimiter = \preg_quote($endDelimiter, '/');
|
|
|
93 |
$regex = \sprintf('/(%s|[^%s\x00])*(?:%s)?/', RegexHelper::PARTIAL_ESCAPED_CHAR, $endDelimiter, $endDelimiter);
|
|
|
94 |
if (($partialTitle = $cursor->match($regex)) === null) {
|
|
|
95 |
return null;
|
|
|
96 |
}
|
|
|
97 |
|
|
|
98 |
return RegexHelper::unescape($partialTitle);
|
|
|
99 |
}
|
|
|
100 |
|
|
|
101 |
private static function manuallyParseLinkDestination(Cursor $cursor): ?string
|
|
|
102 |
{
|
|
|
103 |
$oldPosition = $cursor->getPosition();
|
|
|
104 |
$oldState = $cursor->saveState();
|
|
|
105 |
|
|
|
106 |
$openParens = 0;
|
|
|
107 |
while (($c = $cursor->getCurrentCharacter()) !== null) {
|
|
|
108 |
if ($c === '\\' && ($peek = $cursor->peek()) !== null && RegexHelper::isEscapable($peek)) {
|
|
|
109 |
$cursor->advanceBy(2);
|
|
|
110 |
} elseif ($c === '(') {
|
|
|
111 |
$cursor->advanceBy(1);
|
|
|
112 |
$openParens++;
|
|
|
113 |
} elseif ($c === ')') {
|
|
|
114 |
if ($openParens < 1) {
|
|
|
115 |
break;
|
|
|
116 |
}
|
|
|
117 |
|
|
|
118 |
$cursor->advanceBy(1);
|
|
|
119 |
$openParens--;
|
|
|
120 |
} elseif (\preg_match(RegexHelper::REGEX_WHITESPACE_CHAR, $c)) {
|
|
|
121 |
break;
|
|
|
122 |
} else {
|
|
|
123 |
$cursor->advanceBy(1);
|
|
|
124 |
}
|
|
|
125 |
}
|
|
|
126 |
|
|
|
127 |
if ($openParens !== 0) {
|
|
|
128 |
return null;
|
|
|
129 |
}
|
|
|
130 |
|
|
|
131 |
if ($cursor->getPosition() === $oldPosition && (! isset($c) || $c !== ')')) {
|
|
|
132 |
return null;
|
|
|
133 |
}
|
|
|
134 |
|
|
|
135 |
$newPos = $cursor->getPosition();
|
|
|
136 |
$cursor->restoreState($oldState);
|
|
|
137 |
|
|
|
138 |
$cursor->advanceBy($newPos - $cursor->getPosition());
|
|
|
139 |
|
|
|
140 |
return $cursor->getPreviousText();
|
|
|
141 |
}
|
|
|
142 |
}
|