1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Pilipinews\Common\Converters; |
4
|
|
|
|
5
|
|
|
use League\HTMLToMarkdown\Converter\ConverterInterface; |
6
|
|
|
use League\HTMLToMarkdown\ElementInterface; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* Paragraph Converter |
10
|
|
|
* |
11
|
|
|
* @package Pilipinews |
12
|
|
|
* @author Rougin Gutib <[email protected]> |
13
|
|
|
*/ |
14
|
|
|
class ParagraphConverter implements ConverterInterface |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* Converts the specified element into a parsed string. |
18
|
|
|
* |
19
|
|
|
* @param \League\HTMLToMarkdown\ElementInterface $element |
20
|
|
|
* @return string |
21
|
|
|
*/ |
22
|
15 |
|
public function convert(ElementInterface $element) |
23
|
|
|
{ |
24
|
15 |
|
$value = $element->getValue(); |
25
|
|
|
|
26
|
15 |
|
$markdown = ''; |
27
|
|
|
|
28
|
15 |
|
$lines = preg_split('/\r\n|\r|\n/', $value); |
29
|
|
|
|
30
|
15 |
|
foreach ($lines as $line) |
31
|
|
|
{ |
32
|
|
|
/* |
33
|
|
|
* Some special characters need to be escaped based on the position that they appear |
34
|
|
|
* The following function will deal with those special cases. |
35
|
|
|
*/ |
36
|
|
|
|
37
|
15 |
|
$markdown .= $this->escapeSpecialCharacters($line); |
38
|
15 |
|
$markdown .= "\n"; |
39
|
5 |
|
} |
40
|
|
|
|
41
|
15 |
|
return trim($markdown) !== '' ? "\n" . rtrim(ltrim($markdown)) . "\n" : ''; |
42
|
|
|
} |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* Returns the supported HTML tags. |
46
|
|
|
* |
47
|
|
|
* @return string[] |
48
|
|
|
*/ |
49
|
15 |
|
public function getSupportedTags() |
50
|
|
|
{ |
51
|
15 |
|
return array('p'); |
52
|
|
|
} |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* @param string $line |
56
|
|
|
* |
57
|
|
|
* @return string |
58
|
|
|
*/ |
59
|
15 |
|
private function escapeSpecialCharacters($line) |
60
|
|
|
{ |
61
|
15 |
|
$line = $this->escapeFirstCharacters($line); |
62
|
|
|
|
63
|
15 |
|
$line = $this->escapeOtherCharacters($line); |
64
|
|
|
|
65
|
15 |
|
return $this->escapeOtherCharactersRegex($line); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* @param string $line |
70
|
|
|
* |
71
|
|
|
* @return string |
72
|
|
|
*/ |
73
|
15 |
|
private function escapeFirstCharacters($line) |
74
|
|
|
{ |
75
|
15 |
|
$escapable = array('>', '- ', '+ ', '--', '~~~', '---', '- - -'); |
76
|
|
|
|
77
|
15 |
|
foreach ($escapable as $i) |
78
|
|
|
{ |
79
|
15 |
|
if (strpos(ltrim($line), $i) === 0) |
80
|
5 |
|
{ |
81
|
|
|
// Found a character that must be escaped, adding a backslash before |
82
|
7 |
|
return '\\' . ltrim($line); |
83
|
|
|
} |
84
|
5 |
|
} |
85
|
|
|
|
86
|
15 |
|
return $line; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* @param string $line |
91
|
|
|
* |
92
|
|
|
* @return string |
93
|
|
|
*/ |
94
|
15 |
|
private function escapeOtherCharacters($line) |
95
|
|
|
{ |
96
|
15 |
|
$escapable = array('<!--'); |
97
|
|
|
|
98
|
15 |
|
foreach ($escapable as $i) |
99
|
|
|
{ |
100
|
15 |
|
if (strpos($line, $i) !== false) |
101
|
5 |
|
{ |
102
|
|
|
// Found an escapable character, escaping it |
103
|
5 |
|
$line = substr_replace($line, '\\', strpos($line, $i), 0); |
104
|
|
|
} |
105
|
5 |
|
} |
106
|
|
|
|
107
|
15 |
|
return $line; |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
/** |
111
|
|
|
* @param string $line |
112
|
|
|
* |
113
|
|
|
* @return string |
114
|
|
|
*/ |
115
|
15 |
|
private function escapeOtherCharactersRegex($line) |
116
|
|
|
{ |
117
|
|
|
// Match numbers ending on ')' or '.' that are at the beginning of the line. |
118
|
15 |
|
$regExs = array('/^[0-9]+(?=\)|\.)/'); |
119
|
|
|
|
120
|
15 |
|
foreach ($regExs as $i) |
121
|
|
|
{ |
122
|
15 |
|
if (preg_match($i, $line, $match)) |
123
|
5 |
|
{ |
124
|
|
|
// Matched an escapable character, adding a backslash on |
125
|
|
|
// the string before the offending character |
126
|
|
|
|
127
|
7 |
|
$line = substr_replace($line, '\\', strlen($match[0]), 0); |
128
|
1 |
|
} |
129
|
5 |
|
} |
130
|
|
|
|
131
|
15 |
|
return $line; |
132
|
|
|
} |
133
|
|
|
} |
134
|
|
|
|