|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* User: ms |
|
4
|
|
|
* Date: 12.10.15 |
|
5
|
|
|
* Time: 22:16 |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace Amazon; |
|
9
|
|
|
|
|
10
|
|
|
use Amazon\Exceptions\InvalidDomainException; |
|
11
|
|
|
|
|
12
|
|
|
abstract class Parser |
|
13
|
|
|
{ |
|
14
|
|
|
/** |
|
15
|
|
|
* @var string |
|
16
|
|
|
*/ |
|
17
|
|
|
private $tld = null; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var array |
|
21
|
|
|
*/ |
|
22
|
|
|
private $unusedDomainParts = array( |
|
23
|
|
|
'amazon' |
|
24
|
|
|
, 'amzn' |
|
25
|
|
|
, 'www' |
|
26
|
|
|
); |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @var array |
|
30
|
|
|
*/ |
|
31
|
|
|
private $amazonDomains = array( |
|
32
|
|
|
'amazon' |
|
33
|
|
|
, 'amzn' |
|
34
|
|
|
); |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $url = null; |
|
40
|
|
|
|
|
41
|
|
|
public function __construct($url) |
|
42
|
|
|
{ |
|
43
|
|
|
|
|
44
|
|
|
$this->url = $url; |
|
45
|
|
|
//Question Marks |
|
46
|
|
|
if (preg_match('/%3F/', $url)) { |
|
47
|
|
|
$this->url = substr($url, 0, strpos($url, "%3F")); |
|
48
|
|
|
|
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
$urlParameter = parse_url($this->getUrl()); |
|
52
|
|
|
$this->processUrl($urlParameter); |
|
53
|
|
|
$this->processTld($urlParameter['host']); |
|
54
|
|
|
|
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* @param string $url |
|
59
|
|
|
*/ |
|
60
|
|
|
protected function setUrl($url) |
|
61
|
|
|
{ |
|
62
|
|
|
$this->url = $url; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* @return string |
|
67
|
|
|
*/ |
|
68
|
|
|
protected function getUrl() |
|
69
|
|
|
{ |
|
70
|
|
|
return $this->url; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @param $urlParameter |
|
75
|
|
|
* @throws InvalidDomainException |
|
76
|
|
|
*/ |
|
77
|
|
|
protected function processUrl($urlParameter) |
|
78
|
|
|
{ |
|
79
|
|
|
if (false === array_key_exists('path', $urlParameter)) { |
|
80
|
|
|
throw new InvalidDomainException(sprintf('Url %s has no path', $this->getUrl())); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
//Check if we have an Amazon Domain |
|
84
|
|
|
$isAmazonDomain = false; |
|
85
|
|
|
//ShortUrl |
|
86
|
|
|
foreach ($this->amazonDomains as $domain) { |
|
87
|
|
|
if (preg_match('/' . $domain . '/', $this->getUrl())) { |
|
88
|
|
|
$isAmazonDomain = true; |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
if (false === $isAmazonDomain) { |
|
92
|
|
|
throw new InvalidDomainException(sprintf('Url %s does not belong to Amazon', $this->getUrl())); |
|
93
|
|
|
} |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* @param $host |
|
98
|
|
|
* @return string |
|
99
|
|
|
*/ |
|
100
|
|
|
protected function processTld($host) |
|
101
|
|
|
{ |
|
102
|
|
|
$tldStrings = []; |
|
103
|
|
|
$parts = explode('.', $host); |
|
104
|
|
|
foreach ($parts as $part) { |
|
105
|
|
|
|
|
106
|
|
|
if (false === in_array($part, $this->unusedDomainParts)) { |
|
107
|
|
|
$tldStrings[] = $part; |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
$this->setTld(implode('.', $tldStrings)); |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* @return string |
|
116
|
|
|
*/ |
|
117
|
|
|
public function getTld() |
|
118
|
|
|
{ |
|
119
|
|
|
return $this->tld; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* @param string $tld |
|
124
|
|
|
*/ |
|
125
|
|
|
protected function setTld($tld) |
|
126
|
|
|
{ |
|
127
|
|
|
$this->tld = $tld; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
} |
schmiddim /
amazon-asin-parser
