|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace TreeHouse\IoBundle\Scrape\EventListener; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\Psr7\Response; |
|
6
|
|
|
use Psr\Log\LoggerInterface; |
|
7
|
|
|
use Symfony\Component\EventDispatcher\EventSubscriberInterface; |
|
8
|
|
|
use TreeHouse\Feeder\Event\FailedItemModificationEvent; |
|
9
|
|
|
use TreeHouse\Feeder\FeedEvents; |
|
10
|
|
|
use TreeHouse\IoBundle\Scrape\Event\FailedItemEvent; |
|
11
|
|
|
use TreeHouse\IoBundle\Scrape\Event\RateLimitEvent; |
|
12
|
|
|
use TreeHouse\IoBundle\Scrape\Event\ScrapeResponseEvent; |
|
13
|
|
|
use TreeHouse\IoBundle\Scrape\Event\ScrapeUrlEvent; |
|
14
|
|
|
use TreeHouse\IoBundle\Scrape\Event\SkippedItemEvent; |
|
15
|
|
|
use TreeHouse\IoBundle\Scrape\Event\SuccessItemEvent; |
|
16
|
|
|
use TreeHouse\IoBundle\Scrape\ScraperEvents; |
|
17
|
|
|
|
|
18
|
|
|
class ScrapeLoggingSubscriber implements EventSubscriberInterface |
|
19
|
|
|
{ |
|
20
|
|
|
/** |
|
21
|
|
|
* @var LoggerInterface |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $logger; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @param LoggerInterface $logger |
|
27
|
|
|
*/ |
|
28
|
4 |
|
public function __construct(LoggerInterface $logger) |
|
29
|
|
|
{ |
|
30
|
4 |
|
$this->logger = $logger; |
|
31
|
4 |
|
} |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @inheritdoc |
|
35
|
|
|
*/ |
|
36
|
4 |
|
public static function getSubscribedEvents() |
|
37
|
|
|
{ |
|
38
|
|
|
return [ |
|
39
|
4 |
|
ScraperEvents::ITEM_SUCCESS => 'onItemSuccess', |
|
40
|
4 |
|
ScraperEvents::ITEM_FAILED => 'onItemFailed', |
|
41
|
4 |
|
ScraperEvents::ITEM_SKIPPED => 'onItemSkipped', |
|
42
|
4 |
|
ScraperEvents::SCRAPE_NEXT_URL => 'onScrapeNextUrl', |
|
43
|
4 |
|
ScraperEvents::RATE_LIMIT_REACHED => 'onRateLimitReached', |
|
44
|
4 |
|
ScraperEvents::SCRAPE_URL_NOT_OK => 'onScrapeUrlNotOk', |
|
45
|
4 |
|
]; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @param SuccessItemEvent $event |
|
50
|
|
|
*/ |
|
51
|
|
|
public function onItemSuccess(SuccessItemEvent $event) |
|
52
|
|
|
{ |
|
53
|
|
|
$this->logger->info(sprintf('✎ updated: %s', (string) $event->getItem())); |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @param SkippedItemEvent $event |
|
58
|
|
|
*/ |
|
59
|
|
|
public function onItemSkipped(SkippedItemEvent $event) |
|
60
|
|
|
{ |
|
61
|
|
|
$this->logger->info(sprintf('# skipped: %s', (string) $event->getItem())); |
|
62
|
|
|
$this->logger->debug(sprintf(' reason: %s', $event->getReason())); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* @param FailedItemEvent $event |
|
67
|
|
|
*/ |
|
68
|
|
|
public function onItemFailed(FailedItemEvent $event) |
|
69
|
|
|
{ |
|
70
|
|
|
$this->logger->warning(sprintf('✘ failed: %s', (string) $event->getItem())); |
|
71
|
|
|
$this->logger->debug(sprintf(' reason: %s', $event->getReason())); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* @param ScrapeUrlEvent $event |
|
76
|
|
|
*/ |
|
77
|
|
|
public function onScrapeNextUrl(ScrapeUrlEvent $event) |
|
78
|
|
|
{ |
|
79
|
|
|
$this->logger->debug(sprintf('⇒ next url: %s', (string) $event->getUrl())); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* @param RateLimitEvent $event |
|
84
|
|
|
*/ |
|
85
|
|
|
public function onRateLimitReached(RateLimitEvent $event) |
|
86
|
|
|
{ |
|
87
|
|
|
$seconds = $event->getRetryDate()->getTimestamp() - time(); |
|
88
|
|
|
$this->logger->debug(sprintf('Rate limit reached, try again after %d seconds', $seconds)); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @param ScrapeResponseEvent $event |
|
93
|
|
|
*/ |
|
94
|
|
|
public function onScrapeUrlNotOk(ScrapeResponseEvent $event) |
|
95
|
|
|
{ |
|
96
|
|
|
$response = $event->getResponse(); |
|
97
|
|
|
|
|
98
|
|
|
$code = $response->getStatusCode(); |
|
99
|
|
|
$text = (new Response($code))->getReasonPhrase(); |
|
100
|
|
|
|
|
101
|
|
|
$this->logger->debug(sprintf('Server replied with response %d (%s)', $code, $text)); |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
treehouselabs /
io-bundle
