|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace League\HTMLToMarkdown\Converter; |
|
4
|
|
|
|
|
5
|
|
|
use League\HTMLToMarkdown\ElementInterface; |
|
6
|
|
|
|
|
7
|
|
|
class ParagraphConverter implements ConverterInterface |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* @param ElementInterface $element |
|
11
|
|
|
* |
|
12
|
|
|
* @return string |
|
13
|
|
|
*/ |
|
14
|
45 |
|
public function convert(ElementInterface $element) |
|
15
|
|
|
{ |
|
16
|
45 |
|
$value = $element->getValue(); |
|
17
|
|
|
|
|
18
|
45 |
|
$markdown = ''; |
|
19
|
|
|
|
|
20
|
45 |
|
$lines = preg_split('/\r\n|\r|\n/', $value); |
|
21
|
45 |
|
foreach ($lines as $line) { |
|
22
|
|
|
/* |
|
23
|
|
|
* Some special characters need to be escaped based on the position that they appear |
|
24
|
|
|
* The following function will deal with those special cases. |
|
25
|
|
|
*/ |
|
26
|
45 |
|
$markdown .= $this->escapeSpecialCharacters($line); |
|
27
|
45 |
|
$markdown .= "\n"; |
|
28
|
30 |
|
} |
|
29
|
|
|
|
|
30
|
45 |
|
return trim($markdown) !== '' ? rtrim($markdown) . "\n\n" : ''; |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @return string[] |
|
35
|
|
|
*/ |
|
36
|
90 |
|
public function getSupportedTags() |
|
37
|
|
|
{ |
|
38
|
90 |
|
return array('p'); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @param string $line |
|
43
|
|
|
* |
|
44
|
|
|
* @return string |
|
45
|
|
|
*/ |
|
46
|
45 |
|
private function escapeSpecialCharacters($line) |
|
47
|
|
|
{ |
|
48
|
45 |
|
$line = $this->escapeFirstCharacters($line); |
|
49
|
45 |
|
$line = $this->escapeOtherCharacters($line); |
|
50
|
45 |
|
$line = $this->escapeOtherCharactersRegex($line); |
|
51
|
|
|
|
|
52
|
45 |
|
return $line; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @param string $line |
|
57
|
|
|
* |
|
58
|
|
|
* @return string |
|
59
|
|
|
*/ |
|
60
|
45 |
|
private function escapeFirstCharacters($line) |
|
61
|
|
|
{ |
|
62
|
|
|
$escapable = array( |
|
63
|
45 |
|
'>', |
|
64
|
30 |
|
'- ', |
|
65
|
30 |
|
'+ ', |
|
66
|
30 |
|
'--', |
|
67
|
30 |
|
'~~~', |
|
68
|
30 |
|
'---', |
|
69
|
|
|
'- - -' |
|
70
|
30 |
|
); |
|
71
|
|
|
|
|
72
|
45 |
|
foreach ($escapable as $i) { |
|
73
|
45 |
|
if (strpos(ltrim($line), $i) === 0) { |
|
74
|
|
|
// Found a character that must be escaped, adding a backslash before |
|
75
|
17 |
|
return '\\' . ltrim($line); |
|
76
|
|
|
} |
|
77
|
30 |
|
} |
|
78
|
|
|
|
|
79
|
45 |
|
return $line; |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* @param string $line |
|
84
|
|
|
* |
|
85
|
|
|
* @return string |
|
86
|
|
|
*/ |
|
87
|
45 |
|
private function escapeOtherCharacters($line) |
|
88
|
|
|
{ |
|
89
|
|
|
$escapable = array( |
|
90
|
15 |
|
'<!--' |
|
91
|
30 |
|
); |
|
92
|
|
|
|
|
93
|
45 |
|
foreach ($escapable as $i) { |
|
94
|
45 |
|
if (strpos($line, $i) !== false) { |
|
95
|
|
|
// Found an escapable character, escaping it |
|
96
|
15 |
|
$line = substr_replace($line, '\\', strpos($line, $i), 0); |
|
97
|
|
|
} |
|
98
|
30 |
|
} |
|
99
|
|
|
|
|
100
|
45 |
|
return $line; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* @param string $line |
|
105
|
|
|
* |
|
106
|
|
|
* @return string |
|
107
|
|
|
*/ |
|
108
|
45 |
|
private function escapeOtherCharactersRegex($line) |
|
109
|
|
|
{ |
|
110
|
|
|
$regExs = array( |
|
111
|
|
|
// Match numbers ending on ')' or '.' that are at the beginning of the line. |
|
112
|
|
|
// They will be escaped if immediately followed by a space or newline. |
|
113
|
15 |
|
'/^[0-9]+(?=(\)|\.)( |$))/' |
|
114
|
30 |
|
); |
|
115
|
|
|
|
|
116
|
45 |
|
foreach ($regExs as $i) { |
|
117
|
45 |
|
if (preg_match($i, $line, $match)) { |
|
118
|
|
|
// Matched an escapable character, adding a backslash on the string before the offending character |
|
119
|
19 |
|
$line = substr_replace($line, '\\', strlen($match[0]), 0); |
|
120
|
4 |
|
} |
|
121
|
30 |
|
} |
|
122
|
|
|
|
|
123
|
45 |
|
return $line; |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
|
thephpleague /
html-to-markdown
