|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Kevintweber\HtmlTokenizer\Tokens; |
|
4
|
|
|
|
|
5
|
|
|
class Text extends AbstractToken |
|
6
|
|
|
{ |
|
7
|
|
|
/** |
|
8
|
|
|
* Constructor |
|
9
|
|
|
* |
|
10
|
|
|
* @param Token|null $parent The parent Token |
|
11
|
|
|
* @param bool $throwOnError |
|
12
|
|
|
* @param string $forcedValue For when contents could be parsed but will be interpreted as TEXT instead. |
|
13
|
|
|
* When forcing a value, do not call parse. |
|
14
|
|
|
*/ |
|
15
|
32 |
|
public function __construct(Token $parent = null, bool $throwOnError = true, string $forcedValue = '') |
|
16
|
|
|
{ |
|
17
|
32 |
|
parent::__construct(Token::TEXT, $parent, $throwOnError); |
|
18
|
|
|
|
|
19
|
32 |
|
$this->value = $forcedValue; |
|
20
|
32 |
|
if ($forcedValue !== '') { |
|
21
|
7 |
|
$this->setTokenPosition($forcedValue); |
|
22
|
|
|
} |
|
23
|
32 |
|
} |
|
24
|
|
|
|
|
25
|
25 |
|
public function parse(string $html) : string |
|
26
|
|
|
{ |
|
27
|
25 |
|
$this->setTokenPosition($html); |
|
28
|
25 |
|
$startingWhitespace = $this->determineStartingWhitespace($html); |
|
29
|
|
|
|
|
30
|
25 |
|
$posOfNextElement = mb_strpos($html, '<'); |
|
31
|
25 |
|
if ($posOfNextElement === false) { |
|
32
|
3 |
|
$this->value = $startingWhitespace . trim($html); |
|
33
|
|
|
|
|
34
|
3 |
|
return ''; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
// Find full length of TEXT. |
|
38
|
22 |
|
$text = mb_substr($html, 0, $posOfNextElement); |
|
39
|
22 |
|
if (trim($text) === '') { |
|
40
|
3 |
|
$this->value = ' '; |
|
41
|
|
|
|
|
42
|
3 |
|
return mb_substr($html, $posOfNextElement); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
21 |
|
$endingWhitespace = $this->determineEndingWhitespace($text); |
|
46
|
21 |
|
$this->value = $startingWhitespace . trim($text) . $endingWhitespace; |
|
47
|
|
|
|
|
48
|
21 |
|
return mb_substr($html, $posOfNextElement); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
14 |
|
public function toArray() : array |
|
52
|
|
|
{ |
|
53
|
|
|
return array( |
|
54
|
14 |
|
'type' => 'text', |
|
55
|
14 |
|
'value' => $this->value, |
|
56
|
14 |
|
'line' => $this->getLine(), |
|
57
|
14 |
|
'position' => $this->getPosition() |
|
58
|
|
|
); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* If whitespace exists at the start of the HTML, collapse it to a single space. |
|
63
|
|
|
* |
|
64
|
|
|
* @param string $html |
|
65
|
|
|
* |
|
66
|
|
|
* @return string |
|
67
|
|
|
*/ |
|
68
|
25 |
|
private function determineStartingWhitespace(string $html) : string |
|
69
|
|
|
{ |
|
70
|
25 |
|
$startingWhitespace = ''; |
|
71
|
25 |
|
if (preg_match("/(^\s)/", $html) === 1) { |
|
72
|
6 |
|
$startingWhitespace = ' '; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
25 |
|
return $startingWhitespace; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* If whitespace exists at the end of the HTML, collapse it to a single space. |
|
80
|
|
|
* |
|
81
|
|
|
* @param string $text |
|
82
|
|
|
* |
|
83
|
|
|
* @return string |
|
84
|
|
|
*/ |
|
85
|
21 |
|
private function determineEndingWhitespace(string $text) : string |
|
86
|
|
|
{ |
|
87
|
21 |
|
$endingWhitespace = ''; |
|
88
|
21 |
|
if (preg_match("/(\s$)/", $text) === 1) { |
|
89
|
8 |
|
$endingWhitespace = ' '; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
21 |
|
return $endingWhitespace; |
|
93
|
|
|
} |
|
94
|
|
|
} |
|
95
|
|
|
|
kevintweber /
HtmlTokenizer
