|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PiedWeb\SeoPocketCrawler; |
|
4
|
|
|
|
|
5
|
|
|
use League\Csv\Reader; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* Not used anymore... |
|
9
|
|
|
*/ |
|
10
|
|
|
class LinksVisualizer |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var CrawlerConfig |
|
14
|
|
|
*/ |
|
15
|
|
|
protected $config; |
|
16
|
|
|
|
|
17
|
|
|
protected $results = ['nodes' => [], 'links' => []]; |
|
18
|
|
|
|
|
19
|
|
|
public function __construct(string $id, ?string $dataDirectory = null) |
|
20
|
|
|
{ |
|
21
|
|
|
$this->config = CrawlerConfig::loadFrom($id, $dataDirectory); |
|
22
|
|
|
|
|
23
|
|
|
//$this->loadNodes(); |
|
24
|
|
|
//$this->loadLinks(); |
|
25
|
|
|
|
|
26
|
|
|
file_put_contents( |
|
27
|
|
|
$this->config->getDataFolder().'/pagerank.html', |
|
28
|
|
|
file_get_contents(dirname(__FILE__).'/Resources/PageRankVisualizer.html') |
|
29
|
|
|
); |
|
30
|
|
|
/** |
|
31
|
|
|
file_put_contents( |
|
32
|
|
|
$this->config->getDataFolder().Recorder::LINKS_DIR.'/data.json', |
|
33
|
|
|
json_encode($this->results, JSON_PRETTY_PRINT) |
|
34
|
|
|
);**/ |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
protected function loadLinks() |
|
38
|
|
|
{ |
|
39
|
|
|
$csv = Reader::createFromPath($this->config->getDataFolder().Recorder::LINKS_DIR.'/Index.csv', 'r'); |
|
40
|
|
|
$csv->setHeaderOffset(0); |
|
41
|
|
|
$records = $csv->getRecords(); |
|
42
|
|
|
foreach ($records as $r) { |
|
43
|
|
|
if ( |
|
44
|
|
|
$r['To'] > 0 // pas de liens externe |
|
45
|
|
|
&& isset($this->results['nodes'][$r['From']]) && isset ($this->results['nodes'][$r['To']]) |
|
46
|
|
|
) |
|
47
|
|
|
$this->results['links'][] = ['target' => $r['From'], 'source' => $r['To']]; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
$this->results['nodes'] = array_values($this->results['nodes']); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
protected function loadNodes() |
|
54
|
|
|
{ |
|
55
|
|
|
$urls = $this->config->getDataFromPreviousCrawl()['urls']; |
|
56
|
|
|
|
|
57
|
|
|
foreach ($urls as $url) { |
|
58
|
|
|
if ($url->mime_type == 1) //seulement html |
|
59
|
|
|
$this->results['nodes'][$url->id] = ['id' => $url->id, 'pagerank' => $url->pagerank, 'uri' => $url->uri]; |
|
60
|
|
|
} |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|