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
|
48 |
|
public function convert(ElementInterface $element) |
15
|
|
|
{ |
16
|
48 |
|
$value = $element->getValue(); |
17
|
|
|
|
18
|
48 |
|
$markdown = ''; |
19
|
|
|
|
20
|
48 |
|
$lines = preg_split('/\r\n|\r|\n/', $value); |
21
|
48 |
|
foreach ($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
|
48 |
|
$markdown .= $this->escapeSpecialCharacters($line); |
27
|
48 |
|
$markdown .= "\n"; |
28
|
|
|
} |
29
|
|
|
|
30
|
48 |
|
return trim($markdown) !== '' ? rtrim($markdown) . "\n\n" : ''; |
31
|
|
|
} |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* @return string[] |
35
|
|
|
*/ |
36
|
93 |
|
public function getSupportedTags() |
37
|
|
|
{ |
38
|
93 |
|
return array('p'); |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @param string $line |
43
|
|
|
* |
44
|
|
|
* @return string |
45
|
|
|
*/ |
46
|
48 |
|
private function escapeSpecialCharacters($line) |
47
|
|
|
{ |
48
|
48 |
|
$line = $this->escapeFirstCharacters($line); |
49
|
48 |
|
$line = $this->escapeOtherCharacters($line); |
50
|
48 |
|
$line = $this->escapeOtherCharactersRegex($line); |
51
|
|
|
|
52
|
48 |
|
return $line; |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* @param string $line |
57
|
|
|
* |
58
|
|
|
* @return string |
59
|
|
|
*/ |
60
|
48 |
|
private function escapeFirstCharacters($line) |
61
|
|
|
{ |
62
|
|
|
$escapable = array( |
63
|
48 |
|
'>', |
64
|
|
|
'- ', |
65
|
|
|
'+ ', |
66
|
|
|
'--', |
67
|
|
|
'~~~', |
68
|
|
|
'---', |
69
|
|
|
'- - -' |
70
|
|
|
); |
71
|
|
|
|
72
|
48 |
|
foreach ($escapable as $i) { |
73
|
48 |
|
if (strpos(ltrim($line), $i) === 0) { |
74
|
|
|
// Found a character that must be escaped, adding a backslash before |
75
|
3 |
|
return '\\' . ltrim($line); |
76
|
|
|
} |
77
|
|
|
} |
78
|
|
|
|
79
|
48 |
|
return $line; |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* @param string $line |
84
|
|
|
* |
85
|
|
|
* @return string |
86
|
|
|
*/ |
87
|
48 |
|
private function escapeOtherCharacters($line) |
88
|
|
|
{ |
89
|
|
|
$escapable = array( |
90
|
48 |
|
'<!--' |
91
|
|
|
); |
92
|
|
|
|
93
|
48 |
|
foreach ($escapable as $i) { |
94
|
48 |
|
if (strpos($line, $i) !== false) { |
95
|
|
|
// Found an escapable character, escaping it |
96
|
|
|
$line = substr_replace($line, '\\', strpos($line, $i), 0); |
97
|
|
|
} |
98
|
|
|
} |
99
|
|
|
|
100
|
48 |
|
return $line; |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* @param string $line |
105
|
|
|
* |
106
|
|
|
* @return string |
107
|
|
|
*/ |
108
|
48 |
|
private function escapeOtherCharactersRegex($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
|
48 |
|
'/^[0-9]+(?=(\)|\.)( |$))/' |
114
|
|
|
); |
115
|
|
|
|
116
|
48 |
|
foreach ($regExs as $i) { |
117
|
48 |
|
if (preg_match($i, $line, $match)) { |
118
|
|
|
// Matched an escapable character, adding a backslash on the string before the offending character |
119
|
6 |
|
$line = substr_replace($line, '\\', strlen($match[0]), 0); |
120
|
|
|
} |
121
|
|
|
} |
122
|
|
|
|
123
|
48 |
|
return $line; |
124
|
|
|
} |
125
|
|
|
} |
126
|
|
|
|