|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* @file |
|
4
|
|
|
*/ |
|
5
|
|
|
|
|
6
|
|
|
namespace CultuurNet\UDB3\StringFilter; |
|
7
|
|
|
|
|
8
|
|
|
use Stringy\Stringy as Stringy; |
|
9
|
|
|
|
|
10
|
|
|
class TruncateStringFilter implements StringFilterInterface |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var bool |
|
14
|
|
|
*/ |
|
15
|
|
|
protected $wordSafe = false; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* @var int |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $minWordSafeLength; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var bool |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $addEllipsis = false; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var bool |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $spaceBeforeEllipsis = false; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var bool |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $sentenceFriendly = false; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var int |
|
39
|
|
|
*/ |
|
40
|
|
|
protected $maxLength; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @param int $maxLength |
|
44
|
|
|
*/ |
|
45
|
|
|
public function __construct($maxLength) |
|
46
|
|
|
{ |
|
47
|
|
|
$this->setMaxLength($maxLength); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* @param int $maxLength |
|
52
|
|
|
*/ |
|
53
|
|
|
public function setMaxLength($maxLength) |
|
54
|
|
|
{ |
|
55
|
|
|
$this->maxLength = $maxLength; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @param bool $toggle |
|
60
|
|
|
*/ |
|
61
|
|
|
public function addEllipsis($toggle = true) |
|
62
|
|
|
{ |
|
63
|
|
|
$this->addEllipsis = $toggle; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* @param bool $toggle |
|
68
|
|
|
*/ |
|
69
|
|
|
public function spaceBeforeEllipsis($toggle = true) |
|
70
|
|
|
{ |
|
71
|
|
|
$this->spaceBeforeEllipsis = $toggle; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* @param int $minWordSafeLength |
|
76
|
|
|
*/ |
|
77
|
|
|
public function turnOnWordSafe($minWordSafeLength = 1) |
|
78
|
|
|
{ |
|
79
|
|
|
$this->wordSafe = true; |
|
80
|
|
|
$this->minWordSafeLength = $minWordSafeLength; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* When turned on, the filter will try not to truncate in the middle of a sentence. |
|
85
|
|
|
*/ |
|
86
|
|
|
public function beSentenceFriendly() |
|
87
|
|
|
{ |
|
88
|
|
|
$this->sentenceFriendly = true; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @inheritdoc |
|
93
|
|
|
*/ |
|
94
|
|
|
public function filter($string) |
|
95
|
|
|
{ |
|
96
|
|
|
// Maximum length and minimum length to enable word-safe truncating should always be greater than zero. |
|
97
|
|
|
$maxLength = max($this->maxLength, 0); |
|
98
|
|
|
$minWordSafeLength = max($this->minWordSafeLength, 0); |
|
99
|
|
|
|
|
100
|
|
|
// Do not attempt word-safe truncating if the maximum length is smaller than the minimum length to do |
|
101
|
|
|
// word-safe truncating. |
|
102
|
|
|
$wordSafe = $this->wordSafe && $maxLength >= $minWordSafeLength; |
|
103
|
|
|
|
|
104
|
|
|
// Define the suffix of the truncated string. |
|
105
|
|
|
$suffix = ''; |
|
106
|
|
|
if ($this->addEllipsis) { |
|
107
|
|
|
$ellipsis = '...'; |
|
108
|
|
|
if ($this->spaceBeforeEllipsis) { |
|
109
|
|
|
$ellipsis = ' ...'; |
|
110
|
|
|
} |
|
111
|
|
|
$suffix = Stringy::create($ellipsis, 'UTF-8'); |
|
112
|
|
|
|
|
113
|
|
|
// If the ellipsis is longer or equal to the maximum length, simply truncate the ellipsis so it fits in |
|
114
|
|
|
// the maximum length and return it. |
|
115
|
|
|
if ($suffix->length() >= $maxLength) { |
|
116
|
|
|
return (string) $suffix->truncate($maxLength); |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
$stringy = Stringy::create($string, 'UTF-8'); |
|
121
|
|
|
|
|
122
|
|
|
$sentencePattern = '/(.*[.!?])(?:\\s|\\h|$|\\\u00a0).*/su'; |
|
123
|
|
|
$trunc = (string) $stringy->first($maxLength); |
|
124
|
|
|
$hasEndingSymbolInRange = preg_match($sentencePattern, $trunc); |
|
125
|
|
|
|
|
126
|
|
|
if ($this->sentenceFriendly && $hasEndingSymbolInRange === 1) { |
|
127
|
|
|
$sentenceTruncated = preg_replace($sentencePattern, "$1".$suffix, $trunc); |
|
128
|
|
|
$truncated = Stringy::create($sentenceTruncated, 'UTF-8'); |
|
129
|
|
|
} elseif ($wordSafe) { |
|
130
|
|
|
$truncated = $stringy->safeTruncate($maxLength, $suffix); |
|
|
|
|
|
|
131
|
|
|
} else { |
|
132
|
|
|
$truncated = $stringy->truncate($maxLength, $suffix); |
|
|
|
|
|
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
if ($this->addEllipsis) { |
|
136
|
|
|
// Make sure the string does not end in more than 3 dots. The pattern looks for a sequence of |
|
137
|
|
|
// 4 or more ("{4,}") dots ("(\\.)") at the end of the string ("$"). |
|
138
|
|
|
$pattern = '(\\.){4,}$'; |
|
139
|
|
|
$truncated = $truncated->regexReplace($pattern, $suffix); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
return (string) $truncated; |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
If a method or function can return multiple different values and unless you are sure that you only can receive a single value in this context, we recommend to add an additional type check:
If this a common case that PHP Analyzer should handle natively, please let us know by opening an issue.