1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* @author Matthijs van den Bos <[email protected]> |
4
|
|
|
* @copyright 2013 Matthijs van den Bos |
5
|
|
|
*/ |
6
|
|
|
|
7
|
|
|
namespace VDB\Spider; |
8
|
|
|
|
9
|
|
|
use Symfony\Component\EventDispatcher\EventSubscriberInterface; |
10
|
|
|
use Symfony\Component\EventDispatcher\GenericEvent; |
11
|
|
|
use VDB\Uri\UriInterface; |
12
|
|
|
use VDB\Spider\Event\SpiderEvents; |
13
|
|
|
|
14
|
|
|
class StatsHandler implements EventSubscriberInterface |
15
|
|
|
{ |
16
|
|
|
/** @var string */ |
17
|
|
|
protected $spiderId; |
18
|
|
|
|
19
|
|
|
protected $persisted = array(); |
20
|
|
|
|
21
|
|
|
protected $queued = array(); |
22
|
|
|
|
23
|
|
|
protected $filtered = array(); |
24
|
|
|
|
25
|
|
|
protected $failed = array(); |
26
|
|
|
|
27
|
|
|
public static function getSubscribedEvents() |
28
|
|
|
{ |
29
|
|
|
return array( |
30
|
|
|
SpiderEvents::SPIDER_CRAWL_FILTER_POSTFETCH => 'addToFiltered', |
31
|
|
|
SpiderEvents::SPIDER_CRAWL_FILTER_PREFETCH => 'addToFiltered', |
32
|
|
|
SpiderEvents::SPIDER_CRAWL_POST_ENQUEUE => 'addToQueued', |
33
|
|
|
SpiderEvents::SPIDER_CRAWL_RESOURCE_PERSISTED => 'addToPersisted', |
34
|
|
|
SpiderEvents::SPIDER_CRAWL_ERROR_REQUEST => 'addToFailed' |
35
|
|
|
); |
36
|
|
|
} |
37
|
|
|
|
38
|
|
|
public function setSpiderId($spiderId) |
39
|
|
|
{ |
40
|
|
|
$this->spiderId = $spiderId; |
41
|
|
|
} |
42
|
|
|
|
43
|
|
|
public function getSpiderId() |
44
|
|
|
{ |
45
|
|
|
return $this->spiderId; |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
public function addToQueued(GenericEvent $event) |
49
|
|
|
{ |
50
|
|
|
$this->queued[] = $event->getArgument('uri'); |
51
|
|
|
} |
52
|
|
|
|
53
|
|
|
public function addToPersisted(GenericEvent $event) |
54
|
|
|
{ |
55
|
|
|
$this->persisted[] = $event->getArgument('uri'); |
56
|
|
|
} |
57
|
|
|
|
58
|
|
|
public function addToFiltered(GenericEvent $event) |
59
|
|
|
{ |
60
|
|
|
$this->filtered[] = $event->getArgument('uri'); |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
public function addToFailed(GenericEvent $event) |
64
|
|
|
{ |
65
|
|
|
$this->failed[$event->getArgument('uri')->toString()] = $event->getArgument('message'); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* @return UriInterface[] |
70
|
|
|
*/ |
71
|
|
|
public function getQueued() |
72
|
|
|
{ |
73
|
|
|
return $this->queued; |
74
|
|
|
} |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* @return UriInterface[] |
78
|
|
|
*/ |
79
|
|
|
public function getPersisted() |
80
|
|
|
{ |
81
|
|
|
return $this->persisted; |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @return FilterableInterface[] |
86
|
|
|
*/ |
87
|
|
|
public function getFiltered() |
88
|
|
|
{ |
89
|
|
|
return $this->filtered; |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* @return array of form array($uriString, $reason) |
94
|
|
|
*/ |
95
|
|
|
public function getFailed() |
96
|
|
|
{ |
97
|
|
|
return $this->failed; |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
public function toString() |
101
|
|
|
{ |
102
|
|
|
$spiderId = $this->getSpiderId(); |
103
|
|
|
$queued = $this->getQueued(); |
104
|
|
|
$filtered = $this->getFiltered(); |
105
|
|
|
$failed = $this->getFailed(); |
106
|
|
|
|
107
|
|
|
$string = ''; |
108
|
|
|
|
109
|
|
|
$string .= "\n\nSPIDER ID: " . $spiderId; |
110
|
|
|
$string .= "\n ENQUEUED: " . count($queued); |
111
|
|
|
$string .= "\n SKIPPED: " . count($filtered); |
112
|
|
|
$string .= "\n FAILED: " . count($failed); |
113
|
|
|
|
114
|
|
|
return $string; |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
|