|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Pilipinews\Common; |
|
4
|
|
|
|
|
5
|
|
|
use League\HTMLToMarkdown\Environment; |
|
6
|
|
|
use League\HTMLToMarkdown\Converter\CodeConverter; |
|
7
|
|
|
use League\HTMLToMarkdown\Converter\CommentConverter; |
|
8
|
|
|
use League\HTMLToMarkdown\Converter\ConverterInterface; |
|
9
|
|
|
use League\HTMLToMarkdown\Converter\DefaultConverter; |
|
10
|
|
|
use League\HTMLToMarkdown\Converter\DivConverter; |
|
11
|
|
|
use League\HTMLToMarkdown\Converter\HardBreakConverter; |
|
12
|
|
|
use League\HTMLToMarkdown\Converter\HorizontalRuleConverter; |
|
13
|
|
|
use League\HTMLToMarkdown\Converter\PreformattedConverter; |
|
14
|
|
|
use League\HTMLToMarkdown\Converter\TextConverter; |
|
15
|
|
|
use League\HTMLToMarkdown\HtmlConverter; |
|
16
|
|
|
use Pilipinews\Common\Converters\BlockquoteConverter; |
|
17
|
|
|
use Pilipinews\Common\Converters\EmphasisConverter; |
|
18
|
|
|
use Pilipinews\Common\Converters\HeaderConverter; |
|
19
|
|
|
use Pilipinews\Common\Converters\ImageConverter; |
|
20
|
|
|
use Pilipinews\Common\Converters\LinkConverter; |
|
21
|
|
|
use Pilipinews\Common\Converters\ListBlockConverter; |
|
22
|
|
|
use Pilipinews\Common\Converters\ListItemConverter; |
|
23
|
|
|
use Pilipinews\Common\Converters\ParagraphConverter; |
|
24
|
|
|
use Pilipinews\Common\Converters\TableBlockConverter; |
|
25
|
|
|
use Pilipinews\Common\Converters\TableDataConverter; |
|
26
|
|
|
use Pilipinews\Common\Converters\TableRowConverter; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Converter |
|
30
|
|
|
* |
|
31
|
|
|
* @package Pilipinews |
|
32
|
|
|
* @author Rougin Gutib <[email protected]> |
|
33
|
|
|
*/ |
|
34
|
|
|
class Converter extends HtmlConverter |
|
35
|
|
|
{ |
|
36
|
|
|
/** |
|
37
|
|
|
* Initializes the converter instance |
|
38
|
|
|
*/ |
|
39
|
15 |
|
public function __construct() |
|
40
|
|
|
{ |
|
41
|
15 |
|
$environment = new Environment; |
|
42
|
|
|
|
|
43
|
15 |
|
$environment->addConverter(new BlockquoteConverter); |
|
44
|
15 |
|
$environment->addConverter(new CodeConverter); |
|
45
|
15 |
|
$environment->addConverter(new CommentConverter); |
|
46
|
15 |
|
$environment->addConverter(new DivConverter); |
|
47
|
15 |
|
$environment->addConverter(new EmphasisConverter); |
|
48
|
15 |
|
$environment->addConverter(new HardBreakConverter); |
|
49
|
15 |
|
$environment->addConverter(new HeaderConverter); |
|
50
|
15 |
|
$environment->addConverter(new HorizontalRuleConverter); |
|
51
|
15 |
|
$environment->addConverter(new ImageConverter); |
|
52
|
15 |
|
$environment->addConverter(new LinkConverter); |
|
53
|
15 |
|
$environment->addConverter(new ListBlockConverter); |
|
54
|
15 |
|
$environment->addConverter(new ListItemConverter); |
|
55
|
15 |
|
$environment->addConverter(new ParagraphConverter); |
|
56
|
15 |
|
$environment->addConverter(new PreformattedConverter); |
|
57
|
15 |
|
$environment->addConverter(new TextConverter); |
|
58
|
15 |
|
$environment->addConverter(new TableBlockConverter); |
|
59
|
15 |
|
$environment->addConverter(new TableDataConverter); |
|
60
|
15 |
|
$environment->addConverter(new TableRowConverter); |
|
61
|
|
|
|
|
62
|
15 |
|
$environment->getConfig()->setOption('bold_style', ''); |
|
63
|
15 |
|
$environment->getConfig()->setOption('header_style', 'atx'); |
|
64
|
15 |
|
$environment->getConfig()->setOption('italic_style', ''); |
|
65
|
15 |
|
$environment->getConfig()->setOption('strip_tags', true); |
|
66
|
15 |
|
$environment->getConfig()->setOption('suppress_errors', true); |
|
67
|
15 |
|
$environment->getConfig()->setOption('hard_break', true); |
|
68
|
|
|
|
|
69
|
15 |
|
parent::__construct($environment); |
|
70
|
15 |
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* Parses the HTML into Markdown format. |
|
74
|
|
|
* |
|
75
|
|
|
* @param string $html |
|
76
|
|
|
* @return string |
|
77
|
|
|
*/ |
|
78
|
15 |
|
public function convert($html) |
|
79
|
|
|
{ |
|
80
|
15 |
|
$html = parent::convert((string) $html); |
|
81
|
|
|
|
|
82
|
15 |
|
$html = $this->encode('--', '-', $html, false); |
|
83
|
15 |
|
$html = $this->encode('\#', '#', $html); |
|
84
|
15 |
|
$html = $this->encode('\*', '*', $html); |
|
85
|
15 |
|
$html = $this->encode('\_', '_', $html); |
|
86
|
15 |
|
$html = $this->encode('•', '*', $html); |
|
87
|
15 |
|
$html = $this->encode('…', '...', $html); |
|
88
|
|
|
|
|
89
|
15 |
|
$html = $this->encode('\[', '[', (string) $html); |
|
90
|
15 |
|
$html = $this->encode('\]', ']', (string) $html); |
|
91
|
15 |
|
$html = str_replace(array('—', '–', '\-'), '-', $html); |
|
92
|
15 |
|
$html = str_replace(array('“', '”'), '"', $html); |
|
93
|
15 |
|
$html = str_replace(array('’', '‘'), "'", $html); |
|
94
|
15 |
|
$html = str_replace("\n\n\n", "\n\n", $html); |
|
95
|
15 |
|
$html = str_replace("\n ", "\n", (string) $html); |
|
96
|
|
|
|
|
97
|
15 |
|
$html = htmlspecialchars_decode($html); |
|
98
|
|
|
|
|
99
|
15 |
|
return str_replace('\.', '.', (string) $html); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* Converts the specified string. |
|
104
|
|
|
* |
|
105
|
|
|
* @param string $from |
|
106
|
|
|
* @param string $text |
|
107
|
|
|
* @param string $html |
|
108
|
|
|
* @param boolean $whole |
|
109
|
|
|
* @return string |
|
110
|
|
|
*/ |
|
111
|
15 |
|
protected function encode($from, $text, $html, $whole = true) |
|
112
|
|
|
{ |
|
113
|
15 |
|
$html = str_replace($from . ' ', $text . ' ', $html); |
|
114
|
|
|
|
|
115
|
15 |
|
$html = str_replace(' ' . $from, ' ' . $text, $html); |
|
116
|
|
|
|
|
117
|
15 |
|
$html = str_replace('>' . $from, '>' . $text, $html); |
|
118
|
|
|
|
|
119
|
15 |
|
$html = str_replace($from . '<', $text . '<', $html); |
|
120
|
|
|
|
|
121
|
15 |
|
return $whole ? str_replace($from, $text, $html) : $html; |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|