|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Matecat\Finder; |
|
4
|
|
|
|
|
5
|
|
|
use Matecat\Finder\Helper\RegexEscaper; |
|
6
|
|
|
use Matecat\Finder\Helper\Strings; |
|
7
|
|
|
use Matecat\Finder\Helper\Replacer; |
|
8
|
|
|
|
|
9
|
|
|
class WholeTextFinder |
|
10
|
|
|
{ |
|
11
|
|
|
/** |
|
12
|
|
|
* @param string $haystack |
|
13
|
|
|
* @param string $needle |
|
14
|
|
|
* @param bool $skipHtmlEntities |
|
15
|
|
|
* @param bool $exactMatch |
|
16
|
|
|
* @param bool $caseSensitive |
|
17
|
|
|
* @param bool $preserveNbsps |
|
18
|
|
|
* |
|
19
|
|
|
* @return array |
|
20
|
|
|
*/ |
|
21
|
|
|
public static function find($haystack, $needle, $skipHtmlEntities = true, $exactMatch = false, $caseSensitive = false, $preserveNbsps = false) |
|
22
|
|
|
{ |
|
23
|
|
|
$patternAndHaystack = self::getPatternAndHaystack($haystack, $needle, $skipHtmlEntities, $exactMatch, $caseSensitive, $preserveNbsps); |
|
24
|
|
|
|
|
25
|
|
|
preg_match_all($patternAndHaystack['pattern'], $patternAndHaystack['haystack'], $matches, PREG_OFFSET_CAPTURE); |
|
26
|
|
|
|
|
27
|
|
|
if($skipHtmlEntities === true){ |
|
28
|
|
|
$patternAndHaystack['haystack'] = Strings::unprotectHTMLTags($patternAndHaystack['haystack']); |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
self::mbCorrectMatchPositions($patternAndHaystack['haystack'], $matches); |
|
32
|
|
|
|
|
33
|
|
|
return $matches[0]; |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Correct position for multi byte strings |
|
38
|
|
|
* |
|
39
|
|
|
* @param string $haystack |
|
40
|
|
|
* @param array $matches |
|
41
|
|
|
* |
|
42
|
|
|
* @return mixed |
|
43
|
|
|
*/ |
|
44
|
|
|
private static function mbCorrectMatchPositions( $haystack, &$matches) |
|
45
|
|
|
{ |
|
46
|
|
|
if(!Strings::isMultibyte($haystack) ){ |
|
47
|
|
|
return $matches[0]; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
foreach ($matches[0] as $index => $match){ |
|
51
|
|
|
$word = $match[0]; |
|
52
|
|
|
$position = $match[1]; |
|
53
|
|
|
|
|
54
|
|
|
$correctPosition = self::mbFindTheCorrectPosition($haystack, $word, $position); |
|
55
|
|
|
$matches[0][$index][1] = $correctPosition; |
|
56
|
|
|
} |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* @param string $haystack |
|
61
|
|
|
* @param string $word |
|
62
|
|
|
* @param int $position |
|
63
|
|
|
* |
|
64
|
|
|
* @return int |
|
65
|
|
|
*/ |
|
66
|
|
|
private static function mbFindTheCorrectPosition( $haystack, $word, &$position) |
|
67
|
|
|
{ |
|
68
|
|
|
$wordCheck = mb_substr($haystack, $position, mb_strlen($word)); |
|
69
|
|
|
|
|
70
|
|
|
if($wordCheck !== $word){ |
|
71
|
|
|
$position = $position - 1; |
|
72
|
|
|
|
|
73
|
|
|
self::mbFindTheCorrectPosition($haystack, $word, $position); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return $position; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param string $haystack |
|
81
|
|
|
* @param string $needle |
|
82
|
|
|
* @param bool $skipHtmlEntities |
|
83
|
|
|
* @param bool $exactMatch |
|
84
|
|
|
* @param bool $caseSensitive |
|
85
|
|
|
* @param bool $preserveNbsps |
|
86
|
|
|
* |
|
87
|
|
|
* @return array |
|
88
|
|
|
*/ |
|
89
|
|
|
private static function getPatternAndHaystack($haystack, $needle, $skipHtmlEntities = true, $exactMatch = false, $caseSensitive = false, $preserveNbsps = false) |
|
90
|
|
|
{ |
|
91
|
|
|
$pattern = self::getSearchPattern($needle, $skipHtmlEntities, $exactMatch, $caseSensitive, $preserveNbsps); |
|
92
|
|
|
$haystack = ($skipHtmlEntities) ? Strings::htmlEntityDecode(Strings::protectHTMLTags($haystack)) : $haystack; |
|
93
|
|
|
$haystack = (false === $preserveNbsps) ? Strings::cutNbsps($haystack) : $haystack; |
|
94
|
|
|
|
|
95
|
|
|
return [ |
|
96
|
|
|
'pattern' => $pattern, |
|
97
|
|
|
'haystack' => $haystack, |
|
98
|
|
|
]; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
/** |
|
102
|
|
|
* @param string $needle |
|
103
|
|
|
* @param bool $skipHtmlEntities |
|
104
|
|
|
* @param bool $exactMatch |
|
105
|
|
|
* @param bool $caseSensitive |
|
106
|
|
|
* @param bool $preserveNbsps |
|
107
|
|
|
* |
|
108
|
|
|
* @return string |
|
109
|
|
|
*/ |
|
110
|
|
|
private static function getSearchPattern($needle, $skipHtmlEntities = true, $exactMatch = false, $caseSensitive = false, $preserveNbsps = false) |
|
111
|
|
|
{ |
|
112
|
|
|
$needle = (false === $preserveNbsps) ? Strings::cutNbsps($needle) : $needle; |
|
113
|
|
|
$needle = ($skipHtmlEntities) ? Strings::htmlEntityDecode($needle) : $needle; |
|
114
|
|
|
|
|
115
|
|
|
$pattern = '/'; |
|
116
|
|
|
$pattern .= ($exactMatch) ? RegexEscaper::escapeWholeTextPattern($needle) : RegexEscaper::escapeRegularPattern($needle); |
|
117
|
|
|
$pattern .= '/'; |
|
118
|
|
|
$pattern .= (false === $caseSensitive) ? 'i' : ''; |
|
119
|
|
|
$pattern .= ($skipHtmlEntities) ? 'u' : ''; |
|
120
|
|
|
|
|
121
|
|
|
return $pattern; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
/** |
|
125
|
|
|
* @param string $haystack |
|
126
|
|
|
* @param string $needle |
|
127
|
|
|
* @param string $replacement |
|
128
|
|
|
* @param bool $skipHtmlEntities |
|
129
|
|
|
* @param bool $exactMatch |
|
130
|
|
|
* @param bool $caseSensitive |
|
131
|
|
|
* @param bool $preserveNbsps |
|
132
|
|
|
* |
|
133
|
|
|
* @return array |
|
134
|
|
|
*/ |
|
135
|
|
|
public static function findAndReplace($haystack, $needle, $replacement, $skipHtmlEntities = true, $exactMatch = false, $caseSensitive = false, $preserveNbsps = false) |
|
136
|
|
|
{ |
|
137
|
|
|
$patternAndHaystack = self::getPatternAndHaystack($haystack, $needle, $skipHtmlEntities, $exactMatch, $caseSensitive, $preserveNbsps); |
|
138
|
|
|
$replacement = Replacer::replace($patternAndHaystack['pattern'], $replacement, $patternAndHaystack['haystack']); |
|
139
|
|
|
|
|
140
|
|
|
if($skipHtmlEntities === true){ |
|
141
|
|
|
$replacement = Strings::unprotectHTMLTags($replacement); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
return [ |
|
145
|
|
|
'replacement' => $replacement, |
|
146
|
|
|
'occurrencies' => self::find($haystack, $needle, $skipHtmlEntities, $exactMatch, $caseSensitive, $preserveNbsps), |
|
147
|
|
|
]; |
|
148
|
|
|
} |
|
149
|
|
|
} |
|
150
|
|
|
|
matecat /
whole-text-finder
