|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PiedWeb\SeoPocketCrawler; |
|
4
|
|
|
|
|
5
|
|
|
class ExtractExternalLinks |
|
6
|
|
|
{ |
|
7
|
|
|
protected $id; |
|
8
|
|
|
protected $dir; |
|
9
|
|
|
protected $filter; |
|
10
|
|
|
protected $filterType; |
|
11
|
|
|
protected $external = []; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
3 |
|
* @var CrawlerConfig |
|
15
|
|
|
*/ |
|
16
|
3 |
|
protected $config; |
|
17
|
3 |
|
|
|
18
|
|
|
protected function __construct(string $id, ?string $dataDirectory = null) |
|
19
|
3 |
|
{ |
|
20
|
|
|
$this->config = CrawlerConfig::loadFrom($id, $dataDirectory); |
|
21
|
|
|
$this->dir = $this->config->getDataFolder().'/links'; |
|
22
|
3 |
|
} |
|
23
|
3 |
|
|
|
24
|
|
|
protected function filter($filename) |
|
25
|
3 |
|
{ |
|
26
|
|
|
return 0 === strpos($filename, 'From_'); |
|
27
|
3 |
|
} |
|
28
|
|
|
|
|
29
|
|
|
public static function scan(string $id) |
|
30
|
3 |
|
{ |
|
31
|
|
|
$self = new self($id); |
|
32
|
3 |
|
$self->scanLinksDir(); |
|
33
|
3 |
|
|
|
34
|
|
|
return $self->getExternals(); |
|
35
|
3 |
|
} |
|
36
|
|
|
|
|
37
|
|
|
protected function scanLinksDir() |
|
38
|
3 |
|
{ |
|
39
|
|
|
if ($resource = opendir($this->dir)) { |
|
40
|
3 |
|
while (false !== ($filename = readdir($resource))) { |
|
41
|
3 |
|
if ($this->filter($filename)) { |
|
42
|
3 |
|
$this->harvestExternalLinks( |
|
43
|
2 |
|
trim(file_get_contents($this->dir.'/'.$filename)), |
|
44
|
|
|
$this->config->getUrlFromId(substr($filename, strlen('From_'))) |
|
45
|
|
|
); |
|
46
|
3 |
|
} |
|
47
|
|
|
} |
|
48
|
3 |
|
closedir($resource); |
|
49
|
|
|
} |
|
50
|
2 |
|
} |
|
51
|
|
|
|
|
52
|
2 |
|
protected function harvestExternalLinks(string $strUrlsLinked, $from) |
|
53
|
|
|
{ |
|
54
|
2 |
|
if (empty($strUrlsLinked)) { |
|
55
|
2 |
|
return; |
|
56
|
2 |
|
} |
|
57
|
|
|
|
|
58
|
|
|
$lines = explode(chr(10), $strUrlsLinked); |
|
59
|
2 |
|
|
|
60
|
|
|
foreach ($lines as $line) { |
|
61
|
3 |
|
if (0 !== strpos($line, $this->config->getBase())) { |
|
62
|
|
|
if (! isset($this->external[$line])) { |
|
63
|
3 |
|
$this->external[$line] = []; |
|
64
|
|
|
} |
|
65
|
|
|
$this->external[$line][] = $from; |
|
66
|
|
|
} |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
public function getExternals() |
|
71
|
|
|
{ |
|
72
|
|
|
return $this->external; |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
|
PiedWeb /
SeoPocketCrawler
