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 Royce 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
|
12 |
|
public function convert(ElementInterface $element) |
23
|
|
|
{ |
24
|
12 |
|
$value = $element->getValue(); |
25
|
|
|
|
26
|
12 |
|
$markdown = ''; |
27
|
|
|
|
28
|
12 |
|
$lines = preg_split('/\r\n|\r|\n/', $value); |
29
|
|
|
|
30
|
12 |
|
foreach ($lines as $line) { |
31
|
|
|
/* |
32
|
|
|
* Some special characters need to be escaped based on the position that they appear |
33
|
|
|
* The following function will deal with those special cases. |
34
|
|
|
*/ |
35
|
12 |
|
$markdown .= $this->escapeSpecialCharacters($line); |
36
|
12 |
|
$markdown .= "\n"; |
37
|
8 |
|
} |
38
|
|
|
|
39
|
12 |
|
return trim($markdown) !== '' ? "\n" . rtrim(ltrim($markdown)) . "\n" : ''; |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Returns the supported HTML tags. |
44
|
|
|
* |
45
|
|
|
* @return string[] |
46
|
|
|
*/ |
47
|
12 |
|
public function getSupportedTags() |
48
|
|
|
{ |
49
|
12 |
|
return array('p'); |
50
|
|
|
} |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* @param string $line |
54
|
|
|
* |
55
|
|
|
* @return string |
56
|
|
|
*/ |
57
|
12 |
|
private function escapeSpecialCharacters($line) |
58
|
|
|
{ |
59
|
12 |
|
$line = $this->escapeFirstCharacters($line); |
60
|
12 |
|
$line = $this->escapeOtherCharacters($line); |
61
|
12 |
|
$line = $this->escapeOtherCharactersRegex($line); |
62
|
|
|
|
63
|
12 |
|
return $line; |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @param string $line |
68
|
|
|
* |
69
|
|
|
* @return string |
70
|
|
|
*/ |
71
|
12 |
|
private function escapeFirstCharacters($line) |
72
|
|
|
{ |
73
|
|
|
$escapable = array( |
74
|
12 |
|
'>', |
75
|
8 |
|
'- ', |
76
|
8 |
|
'+ ', |
77
|
8 |
|
'--', |
78
|
8 |
|
'~~~', |
79
|
8 |
|
'---', |
80
|
|
|
'- - -' |
81
|
8 |
|
); |
82
|
|
|
|
83
|
12 |
|
foreach ($escapable as $i) { |
84
|
12 |
|
if (strpos(ltrim($line), $i) === 0) { |
85
|
|
|
// Found a character that must be escaped, adding a backslash before |
86
|
6 |
|
return '\\' . ltrim($line); |
87
|
|
|
} |
88
|
8 |
|
} |
89
|
|
|
|
90
|
12 |
|
return $line; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* @param string $line |
95
|
|
|
* |
96
|
|
|
* @return string |
97
|
|
|
*/ |
98
|
12 |
|
private function escapeOtherCharacters($line) |
99
|
|
|
{ |
100
|
|
|
$escapable = array( |
101
|
4 |
|
'<!--' |
102
|
8 |
|
); |
103
|
|
|
|
104
|
12 |
|
foreach ($escapable as $i) { |
105
|
12 |
|
if (strpos($line, $i) !== false) { |
106
|
|
|
// Found an escapable character, escaping it |
107
|
6 |
|
$line = substr_replace($line, '\\', strpos($line, $i), 0); |
108
|
2 |
|
} |
109
|
8 |
|
} |
110
|
|
|
|
111
|
12 |
|
return $line; |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* @param string $line |
116
|
|
|
* |
117
|
|
|
* @return string |
118
|
|
|
*/ |
119
|
12 |
|
private function escapeOtherCharactersRegex($line) |
120
|
|
|
{ |
121
|
|
|
$regExs = array( |
122
|
|
|
// Match numbers ending on ')' or '.' that are at the beginning of the line. |
123
|
4 |
|
'/^[0-9]+(?=\)|\.)/' |
124
|
8 |
|
); |
125
|
|
|
|
126
|
12 |
|
foreach ($regExs as $i) { |
127
|
12 |
|
if (preg_match($i, $line, $match)) { |
128
|
|
|
// Matched an escapable character, adding a backslash on the string before the offending character |
129
|
6 |
|
$line = substr_replace($line, '\\', strlen($match[0]), 0); |
130
|
2 |
|
} |
131
|
8 |
|
} |
132
|
|
|
|
133
|
12 |
|
return $line; |
134
|
|
|
} |
135
|
|
|
} |
136
|
|
|
|