|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\LinkChecker\Reporters; |
|
4
|
|
|
|
|
5
|
|
|
use Illuminate\Contracts\Logging\Log; |
|
6
|
|
|
use Spatie\Crawler\Url; |
|
7
|
|
|
|
|
8
|
|
|
class LogBrokenLinks extends BaseReporter |
|
9
|
|
|
{ |
|
10
|
|
|
/** |
|
11
|
|
|
* @var \Illuminate\Contracts\Logging\Log |
|
12
|
|
|
*/ |
|
13
|
|
|
protected $log; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @param \Illuminate\Contracts\Logging\Log $log |
|
17
|
|
|
*/ |
|
18
|
|
|
public function __construct(Log $log) |
|
19
|
|
|
{ |
|
20
|
|
|
$this->log = $log; |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* Called when the crawler has crawled the given url. |
|
25
|
|
|
* |
|
26
|
|
|
* @param \Spatie\Crawler\Url $url |
|
27
|
|
|
* @param \Psr\Http\Message\ResponseInterface|null $response |
|
28
|
|
|
* @param \Spatie\Crawler\Url $foundOnUrl |
|
29
|
|
|
* |
|
30
|
|
|
* @return string |
|
31
|
|
|
*/ |
|
32
|
|
|
public function hasBeenCrawled(Url $url, $response, Url $foundOnUrl = null) |
|
33
|
|
|
{ |
|
34
|
|
|
$statusCode = parent::hasBeenCrawled($url, $response); |
|
35
|
|
|
|
|
36
|
|
|
if ($this->isSuccessOrRedirect($statusCode)) { |
|
37
|
|
|
return; |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
if ($this->excludeStatusCode($statusCode)) { |
|
41
|
|
|
return; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
$reason = $response ? $response->getReasonPhrase() : ''; |
|
45
|
|
|
|
|
46
|
|
|
$logMessage = "{$statusCode} {$reason} - {$url}"; |
|
47
|
|
|
|
|
48
|
|
|
if ($foundOnUrl) { |
|
49
|
|
|
$logMessage .= " (found on {$foundOnUrl}"; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
$this->log->warning($logMessage); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Called when the crawl has ended. |
|
57
|
|
|
*/ |
|
58
|
|
|
public function finishedCrawling() |
|
59
|
|
|
{ |
|
60
|
|
|
$this->log->info('link checker summary'); |
|
61
|
|
|
|
|
62
|
|
|
collect($this->urlsGroupedByStatusCode) |
|
63
|
|
|
->each(function ($urls, $statusCode) { |
|
64
|
|
|
if ($this->isSuccessOrRedirect($statusCode)) { |
|
65
|
|
|
return; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
$count = count($urls); |
|
69
|
|
|
|
|
70
|
|
|
if ($statusCode == static::UNRESPONSIVE_HOST) { |
|
71
|
|
|
$this->log->warning("{$count} url(s) did have unresponsive host(s)"); |
|
72
|
|
|
|
|
73
|
|
|
return; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
$this->log->warning("Crawled {$count} url(s) with statuscode {$statusCode}"); |
|
77
|
|
|
|
|
78
|
|
|
}); |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
|