|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Gacela\Container; |
|
6
|
|
|
|
|
7
|
|
|
use function array_filter; |
|
8
|
|
|
use function array_map; |
|
9
|
|
|
use function array_slice; |
|
10
|
|
|
use function count; |
|
11
|
|
|
use function levenshtein; |
|
12
|
|
|
use function max; |
|
13
|
|
|
use function min; |
|
14
|
|
|
use function strlen; |
|
15
|
|
|
use function usort; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Provides fuzzy matching for service names to suggest alternatives. |
|
19
|
|
|
*/ |
|
20
|
|
|
final class FuzzyMatcher |
|
21
|
|
|
{ |
|
22
|
|
|
private const MAX_SUGGESTIONS = 3; |
|
23
|
|
|
private const SIMILARITY_THRESHOLD = 0.6; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Find similar strings from a list of candidates. |
|
27
|
|
|
* |
|
28
|
|
|
* @param list<string> $candidates |
|
29
|
|
|
* |
|
30
|
|
|
* @return list<string> |
|
31
|
|
|
*/ |
|
32
|
11 |
|
public static function findSimilar(string $target, array $candidates): array |
|
33
|
|
|
{ |
|
34
|
11 |
|
if (count($candidates) === 0) { |
|
35
|
3 |
|
return []; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
8 |
|
$scores = array_map( |
|
39
|
8 |
|
static fn (string $candidate): array => [ |
|
40
|
8 |
|
'name' => $candidate, |
|
41
|
8 |
|
'score' => self::calculateSimilarity($target, $candidate), |
|
42
|
8 |
|
], |
|
43
|
8 |
|
$candidates, |
|
44
|
8 |
|
); |
|
45
|
|
|
|
|
46
|
|
|
// Sort by score descending |
|
47
|
8 |
|
usort($scores, static fn (array $a, array $b): int => $b['score'] <=> $a['score']); |
|
48
|
|
|
|
|
49
|
|
|
// Filter by threshold and limit results |
|
50
|
8 |
|
$filtered = array_filter( |
|
51
|
8 |
|
$scores, |
|
52
|
8 |
|
static fn (array $item): bool => $item['score'] >= self::SIMILARITY_THRESHOLD, |
|
53
|
8 |
|
); |
|
54
|
|
|
|
|
55
|
8 |
|
$suggestions = array_map( |
|
56
|
8 |
|
static fn (array $item): string => $item['name'], |
|
57
|
8 |
|
$filtered, |
|
58
|
8 |
|
); |
|
59
|
|
|
|
|
60
|
8 |
|
return array_slice($suggestions, 0, self::MAX_SUGGESTIONS); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* Calculate similarity between two strings (0.0 to 1.0). |
|
65
|
|
|
*/ |
|
66
|
8 |
|
private static function calculateSimilarity(string $a, string $b): float |
|
67
|
|
|
{ |
|
68
|
8 |
|
$maxLength = max(strlen($a), strlen($b)); |
|
69
|
8 |
|
if ($maxLength === 0) { |
|
70
|
|
|
return 1.0; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
8 |
|
$distance = levenshtein($a, $b); |
|
74
|
8 |
|
if ($distance === -1) { |
|
75
|
|
|
// Strings too long for levenshtein, use simple approach |
|
76
|
|
|
return self::simpleSimilarity($a, $b); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
8 |
|
return 1.0 - ($distance / $maxLength); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Fallback similarity calculation for very long strings. |
|
84
|
|
|
*/ |
|
85
|
|
|
private static function simpleSimilarity(string $a, string $b): float |
|
86
|
|
|
{ |
|
87
|
|
|
$maxLength = max(strlen($a), strlen($b)); |
|
88
|
|
|
$minLength = min(strlen($a), strlen($b)); |
|
89
|
|
|
|
|
90
|
|
|
if ($maxLength === 0) { |
|
91
|
|
|
return 1.0; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
// Simple length-based similarity |
|
95
|
|
|
return $minLength / $maxLength; |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
|
|
|
gacela-project /
container
