|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\HttpStatusCheck; |
|
4
|
|
|
|
|
5
|
|
|
use Spatie\Crawler\Url; |
|
6
|
|
|
use Spatie\Crawler\CrawlObserver; |
|
7
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
|
8
|
|
|
|
|
9
|
|
|
class CrawlLogger implements CrawlObserver |
|
10
|
|
|
{ |
|
11
|
|
|
const UNRESPONSIVE_HOST = 'Host did not respond'; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* @var \Symfony\Component\Console\Output\OutputInterface |
|
15
|
|
|
*/ |
|
16
|
|
|
protected $output; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* @var array |
|
20
|
|
|
*/ |
|
21
|
|
|
protected $crawledUrls = []; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @var string|null |
|
25
|
|
|
*/ |
|
26
|
|
|
protected $outputFile = null; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @var bool |
|
30
|
|
|
*/ |
|
31
|
|
|
protected $overwriteOutputFile = false; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @param \Symfony\Component\Console\Output\OutputInterface $output |
|
35
|
|
|
*/ |
|
36
|
|
|
public function __construct(OutputInterface $output) |
|
37
|
|
|
{ |
|
38
|
|
|
$this->output = $output; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Called when the crawl will crawl the url. |
|
43
|
|
|
* |
|
44
|
|
|
* @param \Spatie\Crawler\Url $url |
|
45
|
|
|
*/ |
|
46
|
|
|
public function willCrawl(Url $url) |
|
47
|
|
|
{ |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Called when the crawler has crawled the given url. |
|
52
|
|
|
* |
|
53
|
|
|
* @param \Spatie\Crawler\Url $url |
|
54
|
|
|
* @param \Psr\Http\Message\ResponseInterface|null $response |
|
55
|
|
|
* @param \Spatie\Crawler\Url $foundOn |
|
56
|
|
|
*/ |
|
57
|
|
|
public function hasBeenCrawled(Url $url, $response, Url $foundOn = null) |
|
58
|
|
|
{ |
|
59
|
|
|
$statusCode = $response ? $response->getStatusCode() : self::UNRESPONSIVE_HOST; |
|
60
|
|
|
|
|
61
|
|
|
$reason = $response ? $response->getReasonPhrase() : ''; |
|
62
|
|
|
|
|
63
|
|
|
$colorTag = $this->getColorTagForStatusCode($statusCode); |
|
64
|
|
|
|
|
65
|
|
|
$timestamp = date('Y-m-d H:i:s'); |
|
66
|
|
|
|
|
67
|
|
|
$message = (string) $url; |
|
68
|
|
|
|
|
69
|
|
|
if ($foundOn && $colorTag === 'error') { |
|
70
|
|
|
$message .= " (found on {$foundOn})"; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
$this->output->writeln("<{$colorTag}>[{$timestamp}] {$statusCode} {$reason} - {$message}</{$colorTag}>"); |
|
74
|
|
|
|
|
75
|
|
|
$this->crawledUrls[$statusCode][] = $url; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* Called when the crawl has ended. |
|
80
|
|
|
*/ |
|
81
|
|
|
public function finishedCrawling() |
|
82
|
|
|
{ |
|
83
|
|
|
$outputFileData = []; |
|
84
|
|
|
$outputFileData[] = ''; |
|
85
|
|
|
$outputFileData[] = 'Crawling summary'; |
|
86
|
|
|
$outputFileData[] = '----------------'; |
|
87
|
|
|
|
|
88
|
|
|
$this->output->writeln(''); |
|
89
|
|
|
$this->output->writeln('Crawling summary'); |
|
90
|
|
|
$this->output->writeln('----------------'); |
|
91
|
|
|
|
|
92
|
|
|
ksort($this->crawledUrls); |
|
93
|
|
|
|
|
94
|
|
|
foreach ($this->crawledUrls as $statusCode => $urls) { |
|
95
|
|
|
$colorTag = $this->getColorTagForStatusCode($statusCode); |
|
96
|
|
|
|
|
97
|
|
|
$count = count($urls); |
|
98
|
|
|
|
|
99
|
|
|
if (is_numeric($statusCode)) { |
|
100
|
|
|
$this->output->writeln("<{$colorTag}>Crawled {$count} url(s) with statuscode {$statusCode}</{$colorTag}>"); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
if ($statusCode == static::UNRESPONSIVE_HOST) { |
|
104
|
|
|
$this->output->writeln("<{$colorTag}>{$count} url(s) did have unresponsive host(s)</{$colorTag}>"); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
if ($this->outputFile !== null && $colorTag !== 'info') { |
|
108
|
|
|
$outputFileData[] = "Status: {$statusCode}"; |
|
109
|
|
|
$outputFileData = array_merge($outputFileData, $urls); |
|
110
|
|
|
} |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
$outputFileData[] = ''; |
|
114
|
|
|
$this->output->writeln(''); |
|
115
|
|
|
|
|
116
|
|
|
if ($this->outputFile !== null) { |
|
117
|
|
|
if ($this->overwriteOutputFile === false) { |
|
118
|
|
|
file_put_contents($this->outputFile, implode(\PHP_EOL, $outputFileData), \FILE_APPEND); |
|
119
|
|
|
} else { |
|
120
|
|
|
file_put_contents($this->outputFile, implode(\PHP_EOL, $outputFileData)); |
|
121
|
|
|
} |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
protected function getColorTagForStatusCode(string $code): string |
|
126
|
|
|
{ |
|
127
|
|
|
if ($this->startsWith($code, '2')) { |
|
128
|
|
|
return 'info'; |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
if ($this->startsWith($code, '3')) { |
|
132
|
|
|
return 'comment'; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
return 'error'; |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
/** |
|
139
|
|
|
* @param string|null $haystack |
|
140
|
|
|
* @param string|array $needles |
|
141
|
|
|
* |
|
142
|
|
|
* @return bool |
|
143
|
|
|
*/ |
|
144
|
|
|
public function startsWith($haystack, $needles): bool |
|
145
|
|
|
{ |
|
146
|
|
|
foreach ((array) $needles as $needle) { |
|
147
|
|
|
if ($needle != '' && substr($haystack, 0, strlen($needle)) === (string) $needle) { |
|
148
|
|
|
return true; |
|
149
|
|
|
} |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
return false; |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
/** |
|
156
|
|
|
* Set the filename to write the output log. |
|
157
|
|
|
* |
|
158
|
|
|
* @param string $filename |
|
159
|
|
|
*/ |
|
160
|
|
|
public function setOutputFile($filename) |
|
161
|
|
|
{ |
|
162
|
|
|
$this->outputFile = $filename; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
/** |
|
166
|
|
|
* Set the state indicating if the output file should be overwritten. |
|
167
|
|
|
* |
|
168
|
|
|
* @param bool $flag |
|
169
|
|
|
*/ |
|
170
|
|
|
public function setOverwriteOutputFile($flag) |
|
171
|
|
|
{ |
|
172
|
|
|
$this->overwriteOutputFile = (bool) $flag; |
|
173
|
|
|
} |
|
174
|
|
|
} |
|
175
|
|
|
|
spatie /
http-status-check
