|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\HttpStatusCheck; |
|
4
|
|
|
|
|
5
|
|
|
use Spatie\Crawler\Crawler; |
|
6
|
|
|
use GuzzleHttp\RequestOptions; |
|
7
|
|
|
use Spatie\Crawler\CrawlAllUrls; |
|
8
|
|
|
use Spatie\Crawler\CrawlInternalUrls; |
|
9
|
|
|
use Symfony\Component\Console\Command\Command; |
|
10
|
|
|
use Symfony\Component\Console\Input\InputOption; |
|
11
|
|
|
use Symfony\Component\Console\Input\InputArgument; |
|
12
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
|
13
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
|
14
|
|
|
use Symfony\Component\Console\Question\ConfirmationQuestion; |
|
15
|
|
|
|
|
16
|
|
|
class ScanCommand extends Command |
|
17
|
|
|
{ |
|
18
|
|
|
protected function configure() |
|
19
|
|
|
{ |
|
20
|
|
|
$this->setName('scan') |
|
21
|
|
|
->setDescription('Check the http status code of all links on a website.') |
|
22
|
|
|
->addArgument( |
|
23
|
|
|
'url', |
|
24
|
|
|
InputArgument::REQUIRED, |
|
25
|
|
|
'The url to check' |
|
26
|
|
|
) |
|
27
|
|
|
->addOption( |
|
28
|
|
|
'concurrency', |
|
29
|
|
|
'c', |
|
30
|
|
|
InputOption::VALUE_REQUIRED, |
|
31
|
|
|
'The amount of concurrent connections to use', |
|
32
|
|
|
10 |
|
33
|
|
|
) |
|
34
|
|
|
->addOption( |
|
35
|
|
|
'output', |
|
36
|
|
|
'o', |
|
37
|
|
|
InputOption::VALUE_REQUIRED, |
|
38
|
|
|
'Log all non-2xx and non-3xx responses in this file' |
|
39
|
|
|
) |
|
40
|
|
|
->addOption( |
|
41
|
|
|
'dont-crawl-external-links', |
|
42
|
|
|
'x', |
|
43
|
|
|
InputOption::VALUE_NONE, |
|
44
|
|
|
'Dont crawl external links' |
|
45
|
|
|
) |
|
46
|
|
|
->addOption( |
|
47
|
|
|
'timeout', |
|
48
|
|
|
't', |
|
49
|
|
|
InputOption::VALUE_OPTIONAL, |
|
50
|
|
|
'The maximum number of seconds the request can take', |
|
51
|
|
|
10 |
|
52
|
|
|
) |
|
53
|
|
|
->addOption( |
|
54
|
|
|
'user-agent', |
|
55
|
|
|
'u', |
|
56
|
|
|
InputOption::VALUE_OPTIONAL, |
|
57
|
|
|
'The User Agent to pass for the request', |
|
58
|
|
|
'' |
|
59
|
|
|
) |
|
60
|
|
|
->addOption( |
|
61
|
|
|
'skip-verification', |
|
62
|
|
|
's', |
|
63
|
|
|
InputOption::VALUE_NONE, |
|
64
|
|
|
'Skips checking the SSL certificate' |
|
65
|
|
|
) |
|
66
|
|
|
->addOption( |
|
67
|
|
|
'options', |
|
68
|
|
|
'opt', |
|
69
|
|
|
InputOption::VALUE_IS_ARRAY | InputOption::VALUE_OPTIONAL, |
|
70
|
|
|
'Additional options to the request', |
|
71
|
|
|
[] |
|
72
|
|
|
); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
/** |
|
76
|
|
|
* @param \Symfony\Component\Console\Input\InputInterface $input |
|
77
|
|
|
* @param \Symfony\Component\Console\Output\OutputInterface $output |
|
78
|
|
|
* |
|
79
|
|
|
* @return int |
|
80
|
|
|
*/ |
|
81
|
|
|
protected function execute(InputInterface $input, OutputInterface $output) |
|
82
|
|
|
{ |
|
83
|
|
|
$baseUrl = $input->getArgument('url'); |
|
84
|
|
|
$crawlProfile = $input->getOption('dont-crawl-external-links') ? new CrawlInternalUrls($baseUrl) : new CrawlAllUrls(); |
|
85
|
|
|
|
|
86
|
|
|
$output->writeln("Start scanning {$baseUrl}"); |
|
87
|
|
|
$output->writeln(''); |
|
88
|
|
|
|
|
89
|
|
|
$crawlLogger = new CrawlLogger($output); |
|
90
|
|
|
|
|
91
|
|
|
if ($input->getOption('output')) { |
|
92
|
|
|
$outputFile = $input->getOption('output'); |
|
93
|
|
|
|
|
94
|
|
|
if (file_exists($outputFile)) { |
|
95
|
|
|
$helper = $this->getHelper('question'); |
|
96
|
|
|
$question = new ConfirmationQuestion( |
|
97
|
|
|
"The output file `{$outputFile}` already exists. Overwrite it? (y/n)", |
|
98
|
|
|
false |
|
99
|
|
|
); |
|
100
|
|
|
|
|
101
|
|
|
if (! $helper->ask($input, $output, $question)) { |
|
102
|
|
|
$output->writeln('Aborting...'); |
|
103
|
|
|
|
|
104
|
|
|
return 0; |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
$crawlLogger->setOutputFile($input->getOption('output')); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
$clientOptions = [ |
|
112
|
|
|
RequestOptions::TIMEOUT => $input->getOption('timeout'), |
|
113
|
|
|
RequestOptions::VERIFY => ! $input->getOption('skip-verification'), |
|
114
|
|
|
]; |
|
115
|
|
|
|
|
116
|
|
|
$clientOptions = array_merge($clientOptions, $input->getOption('options')); |
|
117
|
|
|
|
|
118
|
|
|
if ($input->getOption('user-agent')) { |
|
119
|
|
|
$clientOptions[RequestOptions::HEADERS]['user-agent'] = $input->getOption('user-agent'); |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
Crawler::create($clientOptions) |
|
123
|
|
|
->setConcurrency($input->getOption('concurrency')) |
|
124
|
|
|
->setCrawlObserver($crawlLogger) |
|
125
|
|
|
->setCrawlProfile($crawlProfile) |
|
126
|
|
|
->startCrawling($baseUrl); |
|
127
|
|
|
|
|
128
|
|
|
return 0; |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
|
spatie /
http-status-check
