|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace VDB\Spider\Discoverer; |
|
4
|
|
|
|
|
5
|
|
|
use VDB\Spider\Resource; |
|
6
|
|
|
use VDB\Spider\Filter\PreFetchFilter; |
|
7
|
|
|
use VDB\Spider\Uri\FilterableUri; |
|
8
|
|
|
|
|
9
|
|
|
class DiscovererSet implements \IteratorAggregate |
|
10
|
|
|
{ |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* @var Discoverer[] |
|
14
|
|
|
*/ |
|
15
|
|
|
private $discoverers = array(); |
|
16
|
|
|
|
|
17
|
|
|
/** @var Filter[] */ |
|
18
|
|
|
private $filters = array(); |
|
19
|
|
|
|
|
20
|
|
|
public function __construct(array $discoverers = array()) |
|
21
|
|
|
{ |
|
22
|
|
|
foreach ($discoverers as $alias => $discoverer) { |
|
23
|
|
|
$this->set($discoverer, is_int($alias) ? null : $alias); |
|
24
|
|
|
} |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @param Resource $resource |
|
29
|
|
|
* @return UriInterface[] |
|
30
|
|
|
*/ |
|
31
|
|
|
public function discover(Resource $resource) |
|
32
|
|
|
{ |
|
33
|
|
|
$discoveredUris = array(); |
|
34
|
|
|
|
|
35
|
|
|
foreach ($this->discoverers as $discoverer) { |
|
36
|
|
|
$discoveredUris = array_merge($discoveredUris, $discoverer->discover($resource)); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
// TODO: perf improvement: do al this in one loop instead of three |
|
40
|
|
|
$this->normalize($discoveredUris); |
|
41
|
|
|
$this->removeDuplicates($discoveredUris); |
|
42
|
|
|
$this->filter($discoveredUris); |
|
43
|
|
|
|
|
44
|
|
|
return $discoveredUris; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Sets a discoverer. |
|
49
|
|
|
* |
|
50
|
|
|
* @param discovererInterface $discoverer The discoverer instance |
|
51
|
|
|
* @param string $alias An alias |
|
52
|
|
|
*/ |
|
53
|
|
|
public function set(DiscovererInterface $discoverer, $alias = null) |
|
54
|
|
|
{ |
|
55
|
|
|
$this->discoverers[$discoverer->getName()] = $discoverer; |
|
56
|
|
|
if (null !== $alias) { |
|
57
|
|
|
$this->discoverers[$alias] = $discoverer; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
$discoverer->setDiscovererSet($this); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @param Filter $filter |
|
65
|
|
|
*/ |
|
66
|
|
|
public function addFilter(PreFetchFilter $filter) |
|
67
|
|
|
{ |
|
68
|
|
|
$this->filters[] = $filter; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Returns true if the discoverer is defined. |
|
73
|
|
|
* |
|
74
|
|
|
* @param string $name The discoverer name |
|
75
|
|
|
* |
|
76
|
|
|
* @return bool true if the discoverer is defined, false otherwise |
|
77
|
|
|
*/ |
|
78
|
|
|
public function has($name) |
|
79
|
|
|
{ |
|
80
|
|
|
return isset($this->discoverers[$name]); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* Gets a discoverer. |
|
85
|
|
|
* |
|
86
|
|
|
* @param string $name The discoverer name |
|
87
|
|
|
* |
|
88
|
|
|
* @return Discoverer The discoverer instance |
|
89
|
|
|
* |
|
90
|
|
|
* @throws InvalidArgumentException if the discoverer is not defined |
|
91
|
|
|
*/ |
|
92
|
|
|
public function get($name) |
|
93
|
|
|
{ |
|
94
|
|
|
if (!$this->has($name)) { |
|
95
|
|
|
throw new \InvalidArgumentException(sprintf('The discoverer "%s" is not defined.', $name)); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
return $this->discoverers[$name]; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
public function getIterator() |
|
102
|
|
|
{ |
|
103
|
|
|
return new \ArrayIterator($this->discoverers); |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* @param UriInterface[] $discoveredUris |
|
108
|
|
|
*/ |
|
109
|
|
|
private function normalize(array &$discoveredUris) |
|
110
|
|
|
{ |
|
111
|
|
|
foreach ($discoveredUris as &$uri) { |
|
112
|
|
|
$uri->normalize(); |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* @param UriInterface[] $discoveredUris |
|
118
|
|
|
*/ |
|
119
|
|
|
private function filter(array &$discoveredUris) |
|
120
|
|
|
{ |
|
121
|
|
|
foreach ($discoveredUris as $k => &$uri) { |
|
122
|
|
|
$uri = new FilterableUri($uri->toString()); |
|
123
|
|
|
foreach ($this->filters as $filter) { |
|
124
|
|
|
if ($filter->match($uri)) { |
|
125
|
|
|
unset($discoveredUris[$k]); |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
/** |
|
132
|
|
|
* @param UriInterface[] $discoveredUris |
|
133
|
|
|
*/ |
|
134
|
|
|
private function removeDuplicates(array &$discoveredUris) |
|
135
|
|
|
{ |
|
136
|
|
|
// make sure there are no duplicates in the list |
|
137
|
|
|
$tmp = array(); |
|
138
|
|
|
/** @var Uri $uri */ |
|
139
|
|
|
foreach ($discoveredUris as $k => $uri) { |
|
140
|
|
|
$tmp[$k] = $uri->toString(); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
// Find duplicates in temporary array |
|
144
|
|
|
$tmp = array_unique($tmp); |
|
145
|
|
|
|
|
146
|
|
|
// Remove the duplicates from original array |
|
147
|
|
|
foreach ($discoveredUris as $k => $uri) { |
|
148
|
|
|
if (!array_key_exists($k, $tmp)) { |
|
149
|
|
|
unset($discoveredUris[$k]); |
|
150
|
|
|
} |
|
151
|
|
|
} |
|
152
|
|
|
} |
|
153
|
|
|
} |
|
154
|
|
|
|
mvdbos /
php-spider
