1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace HumanDirect\Socially; |
4
|
|
|
|
5
|
|
|
use HumanDirect\Socially\Normalizer\DefaultNormalizer; |
6
|
|
|
use HumanDirect\Socially\Normalizer\NormalizerInterface; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* Class Parser. |
10
|
|
|
*/ |
11
|
|
|
class Parser implements ParserInterface |
12
|
|
|
{ |
13
|
|
|
/** |
14
|
|
|
* @var NormalizerInterface[] |
15
|
|
|
*/ |
16
|
|
|
private $normalizers; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* {@inheritdoc} |
20
|
|
|
*/ |
21
|
|
|
public function parse(string $url): ResultInterface |
22
|
|
|
{ |
23
|
|
|
if (!$this->isSocialMediaProfile($url)) { |
24
|
|
|
throw new NotSupportedException('Supplied URL is not a social media profile URL.'); |
25
|
|
|
} |
26
|
|
|
|
27
|
|
|
return Result::createFromUrl($url); |
28
|
|
|
} |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* {@inheritdoc} |
32
|
|
|
*/ |
33
|
|
|
public function normalize(string $url): string |
34
|
|
|
{ |
35
|
|
|
if (null === $this->normalizers) { |
36
|
|
|
$this->loadNormalizers(); |
37
|
|
|
} |
38
|
|
|
|
39
|
|
|
try { |
40
|
|
|
$platform = $this->identifyPlatform($url); |
41
|
|
|
} catch (NotSupportedException | \Exception $e) { |
42
|
|
|
return $url; |
43
|
|
|
} |
44
|
|
|
|
45
|
|
|
$normalizer = $this->getNormalizer($platform); |
46
|
|
|
|
47
|
|
|
return $normalizer->normalize($url); |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* {@inheritdoc} |
52
|
|
|
*/ |
53
|
|
|
public function isSocialMediaProfile(string $url): bool |
54
|
|
|
{ |
55
|
|
|
if (!Util::isValidUrl($url)) { |
56
|
|
|
return false; |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
try { |
60
|
|
|
$this->identifyPlatform($url); |
61
|
|
|
} catch (NotSupportedException | \Exception $e) { |
62
|
|
|
return false; |
63
|
|
|
} |
64
|
|
|
|
65
|
|
|
return true; |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* {@inheritdoc} |
70
|
|
|
*/ |
71
|
|
|
public function identifyPlatform(string $url): string |
72
|
|
|
{ |
73
|
|
|
foreach (self::SOCIAL_MEDIA_PATTERNS as $platform => $regex) { |
74
|
|
|
foreach ($regex as $prefix => $pattern) { |
75
|
|
|
if (preg_match('#' . $pattern . '#i', rawurldecode($url))) { |
76
|
|
|
return \is_string($prefix) ? sprintf('%s.%s', $platform, $prefix) : $platform; |
77
|
|
|
} |
78
|
|
|
} |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
throw new NotSupportedException('Could not identify platform for URL.'); |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* Load normalizers. |
86
|
|
|
*/ |
87
|
|
|
private function loadNormalizers(): void |
88
|
|
|
{ |
89
|
|
|
$normalizerNS = '\\HumanDirect\\Socially\\Normalizer\\'; |
90
|
|
|
$directory = new \RecursiveDirectoryIterator( |
91
|
|
|
__DIR__ . '/Normalizer', |
92
|
|
|
\RecursiveDirectoryIterator::SKIP_DOTS |
93
|
|
|
); |
94
|
|
|
$files = new \RecursiveCallbackFilterIterator($directory, function ($current, $key, $iterator) use ($normalizerNS) { |
|
|
|
|
95
|
|
|
$className = str_replace('.php', '', $current->getFilename()); |
96
|
|
|
$isNormalizer = 'Normalizer' === substr($className, -10); |
97
|
|
|
|
98
|
|
|
if (!$isNormalizer || $current->isDir() || $iterator->hasChildren()) { |
99
|
|
|
return false; |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
$r = new \ReflectionClass($normalizerNS . $className); |
103
|
|
|
|
104
|
|
|
return $r->isInstantiable(); |
105
|
|
|
}); |
106
|
|
|
|
107
|
|
|
foreach ($files as $file) { |
108
|
|
|
$className = $normalizerNS . str_replace('.php', '', $file->getFilename()); |
109
|
|
|
$this->normalizers[] = new $className(); |
110
|
|
|
} |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Get a normalizer for a specific platform or a default normalizer. |
115
|
|
|
* |
116
|
|
|
* @param string $platform |
117
|
|
|
* |
118
|
|
|
* @return NormalizerInterface |
119
|
|
|
*/ |
120
|
|
|
private function getNormalizer(string $platform): NormalizerInterface |
121
|
|
|
{ |
122
|
|
|
foreach ($this->normalizers as $normalizer) { |
123
|
|
|
if ($normalizer->supports($platform)) { |
124
|
|
|
return $normalizer; |
125
|
|
|
} |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
return new DefaultNormalizer(); |
129
|
|
|
} |
130
|
|
|
} |
131
|
|
|
|