|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace League\HTMLToMarkdown\Converter; |
|
4
|
|
|
|
|
5
|
|
|
use League\HTMLToMarkdown\ElementInterface; |
|
6
|
|
|
|
|
7
|
|
|
class CodeConverter implements ConverterInterface |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* @param ElementInterface $element |
|
11
|
|
|
* |
|
12
|
|
|
* @return string |
|
13
|
|
|
*/ |
|
14
|
12 |
|
public function convert(ElementInterface $element) |
|
15
|
|
|
{ |
|
16
|
12 |
|
$language = ''; |
|
17
|
|
|
|
|
18
|
|
|
// Checking for language class on the code block |
|
19
|
12 |
|
$classes = $element->getAttribute('class'); |
|
20
|
|
|
|
|
21
|
12 |
|
if ($classes) { |
|
22
|
|
|
// Since tags can have more than one class, we need to find the one that starts with 'language-' |
|
23
|
3 |
|
$classes = explode(' ', $classes); |
|
24
|
3 |
|
foreach ($classes as $class) { |
|
25
|
3 |
|
if (strpos($class, 'language-') !== false) { |
|
26
|
|
|
// Found one, save it as the selected language and stop looping over the classes. |
|
27
|
3 |
|
$language = str_replace('language-', '', $class); |
|
28
|
3 |
|
break; |
|
29
|
|
|
} |
|
30
|
2 |
|
} |
|
31
|
2 |
|
} |
|
32
|
|
|
|
|
33
|
12 |
|
$markdown = ''; |
|
34
|
12 |
|
$code = html_entity_decode($element->getChildrenAsString()); |
|
35
|
|
|
|
|
36
|
|
|
// In order to remove the code tags we need to search for them and, in the case of the opening tag |
|
37
|
|
|
// use a regular expression to find the tag and the other attributes it might have |
|
38
|
12 |
|
$code = preg_replace('/<code\b[^>]*>/', '', $code); |
|
39
|
12 |
|
$code = str_replace('</code>', '', $code); |
|
40
|
|
|
|
|
41
|
|
|
// Checking if it's a code block or span |
|
42
|
12 |
|
if ($this->shouldBeBlock($element, $code)) { |
|
43
|
|
|
// Code block detected, newlines will be added in parent |
|
44
|
6 |
|
$markdown .= '```' . $language . "\n" . $code . "\n" . '```'; |
|
45
|
4 |
|
} else { |
|
46
|
|
|
// One line of code, wrapping it on one backtick, removing new lines |
|
47
|
9 |
|
$markdown .= '`' . preg_replace('/\r\n|\r|\n/', '', $code) . '`'; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
12 |
|
return $markdown; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* @return string[] |
|
55
|
|
|
*/ |
|
56
|
84 |
|
public function getSupportedTags() |
|
57
|
|
|
{ |
|
58
|
84 |
|
return array('code'); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* @param ElementInterface $element |
|
63
|
|
|
* @param string $code |
|
64
|
|
|
* |
|
65
|
|
|
* @return bool |
|
66
|
|
|
*/ |
|
67
|
12 |
|
private function shouldBeBlock(ElementInterface $element, $code) |
|
68
|
|
|
{ |
|
69
|
12 |
|
if ($element->getParent()->getTagName() == 'pre') { |
|
70
|
6 |
|
return true; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
9 |
|
if (preg_match('/[^\s]` `/', $code)) { |
|
74
|
3 |
|
return true; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
9 |
|
return false; |
|
78
|
|
|
} |
|
79
|
|
|
} |
|
80
|
|
|
|
thephpleague /
html-to-markdown
