|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace League\HTMLToMarkdown\Converter; |
|
6
|
|
|
|
|
7
|
|
|
use League\HTMLToMarkdown\Configuration; |
|
8
|
|
|
use League\HTMLToMarkdown\ConfigurationAwareInterface; |
|
9
|
|
|
use League\HTMLToMarkdown\ElementInterface; |
|
10
|
|
|
|
|
11
|
|
|
class EmphasisConverter implements ConverterInterface, ConfigurationAwareInterface |
|
12
|
|
|
{ |
|
13
|
|
|
/** @var Configuration */ |
|
14
|
|
|
protected $config; |
|
15
|
|
|
|
|
16
|
|
|
protected function getNormTag(?ElementInterface $element): string |
|
17
|
|
|
{ |
|
18
|
|
|
if ($element !== null && ! $element->isText()) { |
|
19
|
|
|
$tag = $element->getTagName(); |
|
20
|
|
|
switch($tag) { |
|
21
|
18 |
|
case 'i': |
|
22
|
|
|
case 'em': |
|
23
|
18 |
|
case 'cite': |
|
24
|
18 |
|
case 'dfn': |
|
25
|
18 |
|
case 'var': |
|
26
|
15 |
|
return 'em'; |
|
27
|
15 |
|
case 'b': |
|
28
|
12 |
|
case 'strong': |
|
29
|
|
|
return 'strong'; |
|
30
|
2 |
|
case 'strike': |
|
31
|
18 |
|
case 's': |
|
32
|
|
|
case 'del': |
|
33
|
|
|
return 'del'; |
|
34
|
|
|
case 'sub': |
|
35
|
|
|
return 'sub'; |
|
36
|
|
|
case 'sup': |
|
37
|
99 |
|
return 'sup'; |
|
38
|
|
|
case 'u': |
|
39
|
99 |
|
case 'ins': |
|
40
|
99 |
|
return 'u'; |
|
41
|
|
|
case 'kdb': |
|
42
|
|
|
return 'kbd'; |
|
43
|
|
|
case 'span': |
|
44
|
|
|
case 'small': |
|
45
|
|
|
case 'abbr': |
|
46
|
|
|
return $tag; |
|
47
|
18 |
|
} |
|
48
|
|
|
} |
|
49
|
18 |
|
return ''; |
|
50
|
18 |
|
} |
|
51
|
|
|
|
|
52
|
18 |
|
public function setConfig(Configuration $config): void |
|
53
|
3 |
|
{ |
|
54
|
|
|
$this->config = $config; |
|
55
|
|
|
} |
|
56
|
18 |
|
|
|
57
|
15 |
|
public function convert(ElementInterface $element): string |
|
58
|
10 |
|
{ |
|
59
|
12 |
|
$tag = $this->getNormTag($element); |
|
60
|
|
|
$value = $element->getValue(); |
|
61
|
|
|
|
|
62
|
18 |
|
if (! \trim($value)) { |
|
63
|
18 |
|
return $value; |
|
64
|
|
|
} |
|
65
|
|
|
switch ($tag) { |
|
66
|
|
|
case 'em': |
|
67
|
|
|
$style = $this->config->getOption('italic_style'); |
|
68
|
|
|
break; |
|
69
|
18 |
|
case 'del': |
|
70
|
18 |
|
$style = $this->config->getOption('strikethrough_style'); |
|
71
|
|
|
break; |
|
72
|
18 |
|
case 'sub': |
|
73
|
|
|
$style = $this->config->getOption('subscript_style'); |
|
74
|
|
|
break; |
|
75
|
|
|
case 'sup': |
|
76
|
|
|
$style = $this->config->getOption('superscript_style'); |
|
77
|
|
|
break; |
|
78
|
99 |
|
case 'strong': |
|
79
|
|
|
$style = $this->config->getOption('bold_style'); |
|
80
|
99 |
|
break; |
|
81
|
|
|
case 'u': |
|
82
|
|
|
$style = $this->config->getOption('underline_style'); |
|
83
|
|
|
break; |
|
84
|
|
|
case 'kdb': |
|
85
|
|
|
$style = $this->config->getOption('keyboard_style'); |
|
86
|
|
|
break; |
|
87
|
|
|
default: |
|
88
|
|
|
$style = $this->config->getOption('undefined_style'); |
|
89
|
|
|
break; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
$prefix = \ltrim($value) !== $value ? ' ' : ''; |
|
93
|
|
|
$suffix = \rtrim($value) !== $value ? ' ' : ''; |
|
94
|
|
|
|
|
95
|
|
|
$preStyle = $this->makeDelimiter($element, $tag, $style); |
|
96
|
|
|
$postStyle = $this->makeDelimiter($element, $tag, $style, false); |
|
97
|
|
|
|
|
98
|
|
|
return $prefix . $preStyle . \trim($value) . $postStyle . $suffix; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
/** |
|
102
|
|
|
* @return string[] |
|
103
|
|
|
*/ |
|
104
|
|
|
public function getSupportedTags(): array |
|
105
|
|
|
{ |
|
106
|
|
|
return [ |
|
107
|
|
|
'em', 'i', 'cite', 'dfn', 'var', |
|
108
|
|
|
'strong', 'b', |
|
109
|
|
|
'del', 'strike', 's', |
|
110
|
|
|
'sub', 'sup', |
|
111
|
|
|
'u', 'ins', |
|
112
|
|
|
'kbd', |
|
113
|
|
|
'span', 'small', 'abbr' |
|
114
|
|
|
]; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
protected function makeDelimiter($element, string $tag, $style, bool $prev = true): string |
|
118
|
|
|
{ |
|
119
|
|
|
/* If this node is immediately preceded or followed by one of the same type don't emit |
|
120
|
|
|
* the start or end $style, respectively. This prevents <em>foo</em><em>bar</em> from |
|
121
|
|
|
* being converted to *foo**bar* which is incorrect. We want *foobar* instead. |
|
122
|
|
|
*/ |
|
123
|
|
|
if($prev) { |
|
124
|
|
|
$ignore = $this->getNormTag($element->getPreviousSibling()) === $tag; |
|
125
|
|
|
} else { |
|
126
|
|
|
$ignore = $this->getNormTag($element->getNextSibling()) === $tag; |
|
127
|
|
|
} |
|
128
|
|
|
if (!is_string($style ?? null) || $ignore) return ''; |
|
129
|
|
|
return empty($style) ? "<" . ($prev ? "" : "/") ."{$tag}>" : $style; |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
thephpleague /
html-to-markdown
