|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2016 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Plugins\Autolink; |
|
9
|
|
|
|
|
10
|
|
|
use s9e\TextFormatter\Plugins\ParserBase; |
|
11
|
|
|
|
|
12
|
|
|
class Parser extends ParserBase |
|
13
|
|
|
{ |
|
14
|
|
|
/** |
|
15
|
|
|
* {@inheritdoc} |
|
16
|
|
|
*/ |
|
17
|
31 |
|
public function parse($text, array $matches) |
|
18
|
|
|
{ |
|
19
|
31 |
|
foreach ($matches as $m) |
|
20
|
|
|
{ |
|
21
|
|
|
// Linkify the trimmed URL |
|
22
|
31 |
|
$this->linkifyUrl($m[0][1], $this->trimUrl($m[0][0])); |
|
23
|
31 |
|
} |
|
24
|
31 |
|
} |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Linkify given URL at given position |
|
28
|
|
|
* |
|
29
|
|
|
* @param integer $tagPos URL's position in the text |
|
30
|
|
|
* @param string $url URL |
|
31
|
|
|
* @return void |
|
32
|
|
|
*/ |
|
33
|
31 |
|
protected function linkifyUrl($tagPos, $url) |
|
34
|
|
|
{ |
|
35
|
|
|
// Ensure that the anchor (scheme/www) is still there |
|
36
|
31 |
|
if (!preg_match('/^[^:]+:|^www\\./i', $url)) |
|
37
|
31 |
|
{ |
|
38
|
2 |
|
return; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
// Create a zero-width end tag right after the URL |
|
42
|
29 |
|
$endTag = $this->parser->addEndTag($this->config['tagName'], $tagPos + strlen($url), 0); |
|
43
|
|
|
|
|
44
|
|
|
// If the URL starts with "www." we prepend "http://" |
|
45
|
29 |
|
if ($url[3] === '.') |
|
46
|
29 |
|
{ |
|
47
|
3 |
|
$url = 'http://' . $url; |
|
48
|
3 |
|
} |
|
49
|
|
|
|
|
50
|
|
|
// Create a zero-width start tag right before the URL |
|
51
|
29 |
|
$startTag = $this->parser->addStartTag($this->config['tagName'], $tagPos, 0); |
|
52
|
29 |
|
$startTag->setAttribute($this->config['attrName'], $url); |
|
53
|
|
|
|
|
54
|
|
|
// Give this tag a slightly lower priority than default to allow specialized plugins |
|
55
|
|
|
// to use the URL instead |
|
56
|
29 |
|
$startTag->setSortPriority(1); |
|
57
|
|
|
|
|
58
|
|
|
// Pair the tags together |
|
59
|
29 |
|
$startTag->pairWith($endTag); |
|
60
|
29 |
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Remove trailing punctuation from given URL |
|
64
|
|
|
* |
|
65
|
|
|
* Removes trailing punctuation and right angle brackets. We preserve right parentheses |
|
66
|
|
|
* if there's a balanced number of parentheses in the URL, e.g. |
|
67
|
|
|
* http://en.wikipedia.org/wiki/Mars_(disambiguation) |
|
68
|
|
|
* |
|
69
|
|
|
* @param string $url Original URL |
|
70
|
|
|
* @return string Trimmed URL |
|
71
|
|
|
*/ |
|
72
|
31 |
|
protected function trimUrl($url) |
|
73
|
|
|
{ |
|
74
|
31 |
|
while (1) |
|
75
|
|
|
{ |
|
76
|
|
|
// We remove most ASCII non-letters and Unicode punctuation from the end of the string. |
|
77
|
|
|
// Exceptions: |
|
78
|
|
|
// - dashes (some YouTube URLs end with a dash due to the video ID) |
|
79
|
|
|
// - equal signs (because of "foo?bar="), |
|
80
|
|
|
// - trailing slashes, |
|
81
|
|
|
// - closing parentheses are balanced separately. |
|
82
|
31 |
|
$url = preg_replace('#(?![-=/)])[\\s!-.:-@[-`{-~\\pP]+$#Du', '', $url); |
|
83
|
|
|
|
|
84
|
31 |
|
if (substr($url, -1) === ')' && substr_count($url, '(') < substr_count($url, ')')) |
|
85
|
31 |
|
{ |
|
86
|
2 |
|
$url = substr($url, 0, -1); |
|
87
|
2 |
|
continue; |
|
88
|
|
|
} |
|
89
|
31 |
|
break; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
31 |
|
return $url; |
|
93
|
|
|
} |
|
94
|
|
|
} |
s9e /
TextFormatter
