|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Caxy\HtmlDiff\Strategy; |
|
4
|
|
|
|
|
5
|
|
|
use Caxy\HtmlDiff\Preprocessor; |
|
6
|
|
|
use Caxy\HtmlDiff\Util\MbStringUtil; |
|
7
|
|
|
|
|
8
|
|
|
class ListItemMatchStrategy implements MatchStrategyInterface |
|
9
|
|
|
{ |
|
10
|
|
|
/** |
|
11
|
|
|
* @var MbStringUtil |
|
12
|
|
|
*/ |
|
13
|
|
|
protected $stringUtil; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @var int |
|
17
|
|
|
*/ |
|
18
|
|
|
protected $similarityThreshold; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var float |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $lengthRatioThreshold; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var float |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $commonTextRatioThreshold; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* ListItemMatchStrategy constructor. |
|
32
|
|
|
* |
|
33
|
|
|
* @param MbStringUtil $stringUtil |
|
34
|
|
|
* @param int $similarityThreshold |
|
35
|
|
|
* @param float $lengthRatioThreshold |
|
36
|
|
|
* @param float $commonTextRatioThreshold |
|
37
|
|
|
*/ |
|
38
|
7 |
|
public function __construct($stringUtil, $similarityThreshold = 80, $lengthRatioThreshold = 0.1, $commonTextRatioThreshold = 0.6) |
|
39
|
|
|
{ |
|
40
|
7 |
|
$this->stringUtil = $stringUtil; |
|
41
|
7 |
|
$this->similarityThreshold = $similarityThreshold; |
|
42
|
7 |
|
$this->lengthRatioThreshold = $lengthRatioThreshold; |
|
43
|
7 |
|
$this->commonTextRatioThreshold = $commonTextRatioThreshold; |
|
44
|
7 |
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @param string $a |
|
48
|
|
|
* @param string $b |
|
49
|
|
|
* |
|
50
|
|
|
* @return bool |
|
51
|
|
|
*/ |
|
52
|
7 |
|
public function isMatch($a, $b) |
|
53
|
|
|
{ |
|
54
|
7 |
|
$percentage = null; |
|
55
|
|
|
|
|
56
|
|
|
// Strip tags and check similarity |
|
57
|
7 |
|
$aStripped = strip_tags($a); |
|
58
|
7 |
|
$bStripped = strip_tags($b); |
|
59
|
7 |
|
similar_text($aStripped, $bStripped, $percentage); |
|
60
|
|
|
|
|
61
|
7 |
|
if ($percentage >= $this->similarityThreshold) { |
|
62
|
7 |
|
return true; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
// Check w/o stripped tags |
|
66
|
6 |
|
similar_text($a, $b, $percentage); |
|
67
|
6 |
|
if ($percentage >= $this->similarityThreshold) { |
|
68
|
|
|
return true; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
// Check common prefix/ suffix length |
|
72
|
6 |
|
$aCleaned = trim($aStripped); |
|
73
|
6 |
|
$bCleaned = trim($bStripped); |
|
74
|
6 |
|
if ($this->stringUtil->strlen($aCleaned) === 0 || $this->stringUtil->strlen($bCleaned) === 0) { |
|
75
|
1 |
|
$aCleaned = $a; |
|
76
|
1 |
|
$bCleaned = $b; |
|
77
|
|
|
} |
|
78
|
6 |
|
if ($this->stringUtil->strlen($aCleaned) === 0 || $this->stringUtil->strlen($bCleaned) === 0) { |
|
79
|
|
|
return false; |
|
80
|
|
|
} |
|
81
|
6 |
|
$prefixIndex = Preprocessor::diffCommonPrefix($aCleaned, $bCleaned, $this->stringUtil); |
|
82
|
6 |
|
$suffixIndex = Preprocessor::diffCommonSuffix($aCleaned, $bCleaned, $this->stringUtil); |
|
83
|
|
|
|
|
84
|
|
|
// Use shorter string, and see how much of it is leftover |
|
85
|
6 |
|
$len = min($this->stringUtil->strlen($aCleaned), $this->stringUtil->strlen($bCleaned)); |
|
86
|
6 |
|
$remaining = $len - ($prefixIndex + $suffixIndex); |
|
87
|
6 |
|
$strLengthPercent = $len / max($this->stringUtil->strlen($a), $this->stringUtil->strlen($b)); |
|
88
|
|
|
|
|
89
|
6 |
|
if ($remaining === 0 && $strLengthPercent > $this->lengthRatioThreshold) { |
|
|
|
|
|
|
90
|
|
|
return true; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
6 |
|
$percentCommon = ($prefixIndex + $suffixIndex) / $len; |
|
94
|
|
|
|
|
95
|
6 |
|
if ($strLengthPercent > 0.1 && $percentCommon > $this->commonTextRatioThreshold) { |
|
96
|
2 |
|
return true; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
6 |
|
return false; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
caxy /
php-htmldiff
Loading history...