|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace WebThumbnailer\Finder; |
|
6
|
|
|
|
|
7
|
|
|
use WebThumbnailer\Application\ConfigManager; |
|
8
|
|
|
use WebThumbnailer\Exception\BadRulesException; |
|
9
|
|
|
use WebThumbnailer\Exception\IOException; |
|
10
|
|
|
use WebThumbnailer\Exception\UnsupportedDomainException; |
|
11
|
|
|
use WebThumbnailer\Utils\DataUtils; |
|
12
|
|
|
use WebThumbnailer\Utils\FileUtils; |
|
13
|
|
|
use WebThumbnailer\Utils\FinderUtils; |
|
14
|
|
|
use WebThumbnailer\Utils\UrlUtils; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Find the appropriate Finder for a given URL, instantiate it and load its rules. |
|
18
|
|
|
*/ |
|
19
|
|
|
class FinderFactory |
|
20
|
|
|
{ |
|
21
|
|
|
/** |
|
22
|
|
|
* Creates a finder object for a given URL. |
|
23
|
|
|
* |
|
24
|
|
|
* @param string $url given URL. |
|
25
|
|
|
* |
|
26
|
|
|
* @return Finder object. |
|
27
|
|
|
* |
|
28
|
|
|
* @throws BadRulesException |
|
29
|
|
|
* @throws IOException |
|
30
|
|
|
*/ |
|
31
|
|
|
public static function getFinder(string $url): Finder |
|
32
|
|
|
{ |
|
33
|
|
|
$domain = UrlUtils::getDomain($url); |
|
34
|
|
|
try { |
|
35
|
|
|
list($domain, $finder, $rules, $options) = static::getThumbnailMeta($domain, $url); |
|
36
|
|
|
|
|
37
|
|
|
$className = '\\WebThumbnailer\\Finder\\' . $finder . 'Finder'; |
|
38
|
|
|
if (!class_exists($className)) { |
|
39
|
|
|
throw new UnsupportedDomainException(); |
|
40
|
|
|
} |
|
41
|
|
|
} catch (UnsupportedDomainException $e) { |
|
42
|
|
|
$className = '\\WebThumbnailer\\Finder\\DefaultFinder'; |
|
43
|
|
|
$rules = []; |
|
44
|
|
|
$options = []; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
return new $className($domain, $url, $rules, $options); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Retrieve JSON metadata for the given domains. |
|
52
|
|
|
* |
|
53
|
|
|
* @param string $inputDomain Domain to search. |
|
54
|
|
|
* @param string $url Complete URL. |
|
55
|
|
|
* |
|
56
|
|
|
* @return mixed[] [domains, finder name, rules, options]. |
|
57
|
|
|
* |
|
58
|
|
|
* @throws UnsupportedDomainException No rules found for the domains. |
|
59
|
|
|
* @throws BadRulesException Mandatory rules not found for the domains. |
|
60
|
|
|
* @throws IOException |
|
61
|
|
|
*/ |
|
62
|
|
|
public static function getThumbnailMeta(string $inputDomain, string $url): array |
|
63
|
|
|
{ |
|
64
|
|
|
// Load JSON rule files. |
|
65
|
|
|
$jsonFiles = ConfigManager::get('settings.rules_filename', ['rules.json']); |
|
66
|
|
|
$allRules = []; |
|
67
|
|
|
foreach ($jsonFiles as $file) { |
|
68
|
|
|
$allRules = array_merge($allRules, DataUtils::loadJson(FileUtils::RESOURCES_PATH . $file)); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
$domain = null; |
|
72
|
|
|
|
|
73
|
|
|
foreach ($allRules as $value) { |
|
74
|
|
|
static::checkMetaFormat($value); |
|
75
|
|
|
|
|
76
|
|
|
$domainFound = false; |
|
77
|
|
|
foreach ($value['domains'] as $domain) { |
|
78
|
|
|
if (strpos($inputDomain, $domain) !== false) { |
|
79
|
|
|
$domainFound = true; |
|
80
|
|
|
break; |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
if (!$domainFound) { |
|
85
|
|
|
continue; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
if (!empty($value['url_exclude'])) { |
|
89
|
|
|
preg_match(FinderUtils::buildRegex($value['url_exclude'], 'i'), $url, $match); |
|
90
|
|
|
if (!empty($match)) { |
|
91
|
|
|
continue; |
|
92
|
|
|
} |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
if (!empty($value['url_require'])) { |
|
96
|
|
|
preg_match(FinderUtils::buildRegex($value['url_require'], 'i'), $url, $match); |
|
97
|
|
|
if (empty($match)) { |
|
98
|
|
|
continue; |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
$value['rules'] = !empty($value['rules']) ? $value['rules'] : []; |
|
103
|
|
|
$value['options'] = !empty($value['options']) ? $value['options'] : []; |
|
104
|
|
|
return [$domain, $value['finder'], $value['rules'], $value['options']]; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
throw new UnsupportedDomainException(); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* Make sure that mandatory directives are present in the metadata. |
|
112
|
|
|
* |
|
113
|
|
|
* @param mixed[] $rules JSON directives. |
|
114
|
|
|
* |
|
115
|
|
|
* @return bool True if the check is successful. |
|
116
|
|
|
* |
|
117
|
|
|
* @throws BadRulesException Mandatory rules not found for the domains. |
|
118
|
|
|
*/ |
|
119
|
|
|
public static function checkMetaFormat(array $rules): bool |
|
120
|
|
|
{ |
|
121
|
|
|
$mandatoryDirectives = ['domains', 'finder']; |
|
122
|
|
|
foreach ($mandatoryDirectives as $mandatoryDirective) { |
|
123
|
|
|
if (empty($rules[$mandatoryDirective])) { |
|
124
|
|
|
throw new BadRulesException(); |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
return true; |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
|