1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace League\HTMLToMarkdown\Converter; |
6
|
|
|
|
7
|
|
|
use League\HTMLToMarkdown\ElementInterface; |
8
|
|
|
|
9
|
|
|
class ParagraphConverter implements ConverterInterface |
10
|
|
|
{ |
11
|
|
|
public function convert(ElementInterface $element): string |
12
|
|
|
{ |
13
|
|
|
$value = $element->getValue(); |
14
|
51 |
|
|
15
|
|
|
$markdown = ''; |
16
|
51 |
|
|
17
|
|
|
$lines = \preg_split('/\r\n|\r|\n/', $value); |
18
|
51 |
|
\assert($lines !== false); |
19
|
|
|
|
20
|
51 |
|
foreach ($lines as $line) { |
21
|
51 |
|
/* |
22
|
|
|
* Some special characters need to be escaped based on the position that they appear |
23
|
|
|
* The following function will deal with those special cases. |
24
|
|
|
*/ |
25
|
|
|
$markdown .= $this->escapeSpecialCharacters($line); |
26
|
51 |
|
$markdown .= "\n"; |
27
|
51 |
|
} |
28
|
34 |
|
|
29
|
|
|
return \trim($markdown) !== '' ? \rtrim($markdown) . "\n\n" : ''; |
30
|
51 |
|
} |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @return string[] |
34
|
|
|
*/ |
35
|
|
|
public function getSupportedTags(): array |
36
|
99 |
|
{ |
37
|
|
|
return ['p']; |
38
|
99 |
|
} |
39
|
|
|
|
40
|
|
|
private function escapeSpecialCharacters(string $line): string |
41
|
|
|
{ |
42
|
|
|
$line = $this->escapeFirstCharacters($line); |
43
|
|
|
$line = $this->escapeOtherCharacters($line); |
44
|
|
|
$line = $this->escapeOtherCharactersRegex($line); |
45
|
|
|
|
46
|
51 |
|
return $line; |
47
|
|
|
} |
48
|
51 |
|
|
49
|
51 |
|
private function escapeFirstCharacters(string $line): string |
50
|
51 |
|
{ |
51
|
|
|
$escapable = [ |
52
|
51 |
|
'>', |
53
|
|
|
'- ', |
54
|
|
|
'+ ', |
55
|
|
|
'--', |
56
|
|
|
'~~~', |
57
|
|
|
'---', |
58
|
|
|
'- - -', |
59
|
|
|
]; |
60
|
51 |
|
|
61
|
|
|
foreach ($escapable as $i) { |
62
|
|
|
if (\strpos(\ltrim($line), $i) === 0) { |
63
|
51 |
|
// Found a character that must be escaped, adding a backslash before |
64
|
34 |
|
return '\\' . \ltrim($line); |
65
|
34 |
|
} |
66
|
34 |
|
} |
67
|
34 |
|
|
68
|
34 |
|
return $line; |
69
|
|
|
} |
70
|
34 |
|
|
71
|
|
|
private function escapeOtherCharacters(string $line): string |
72
|
51 |
|
{ |
73
|
51 |
|
$escapable = [ |
74
|
|
|
'<!--', |
75
|
19 |
|
]; |
76
|
|
|
|
77
|
34 |
|
foreach ($escapable as $i) { |
78
|
|
|
if (($pos = \strpos($line, $i)) === false) { |
79
|
51 |
|
continue; |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
// Found an escapable character, escaping it |
83
|
|
|
$line = \substr_replace($line, '\\', $pos, 0); |
84
|
|
|
} |
85
|
|
|
|
86
|
|
|
return $line; |
|
|
|
|
87
|
51 |
|
} |
88
|
|
|
|
89
|
|
|
private function escapeOtherCharactersRegex(string $line): string |
90
|
17 |
|
{ |
91
|
34 |
|
$regExs = [ |
92
|
|
|
// Match numbers ending on ')' or '.' that are at the beginning of the line. |
93
|
51 |
|
// They will be escaped if immediately followed by a space or newline. |
94
|
51 |
|
'/^[0-9]+(?=(\)|\.)( |$))/', |
95
|
|
|
]; |
96
|
17 |
|
|
97
|
|
|
foreach ($regExs as $i) { |
98
|
34 |
|
if (! \preg_match($i, $line, $match)) { |
99
|
|
|
continue; |
100
|
51 |
|
} |
101
|
|
|
|
102
|
|
|
// Matched an escapable character, adding a backslash on the string before the offending character |
103
|
|
|
$line = \substr_replace($line, '\\', \strlen($match[0]), 0); |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
return $line; |
|
|
|
|
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
|