|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Sesame\Crawler; |
|
4
|
|
|
|
|
5
|
|
|
use Sesame\Model\Article; |
|
6
|
|
|
use \Symfony\Component\DomCrawler\Crawler as DomCrawler; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class ArticleCrawler |
|
10
|
|
|
* |
|
11
|
|
|
* @package Sesame\Crawler |
|
12
|
|
|
*/ |
|
13
|
|
|
class ArticleCrawler extends Crawler |
|
14
|
|
|
{ |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* ArticleCrawler constructor. |
|
18
|
|
|
*/ |
|
19
|
|
|
public function __construct() |
|
20
|
|
|
{ |
|
21
|
|
|
parent::__construct(); |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* crawlArticle |
|
26
|
|
|
* |
|
27
|
|
|
* @param string $responseBody |
|
28
|
|
|
* @param bool $crawlVariations |
|
29
|
|
|
* @return Article |
|
30
|
|
|
*/ |
|
31
|
|
|
public function crawlArticle(string $responseBody, bool $crawlVariations): Article |
|
|
|
|
|
|
32
|
|
|
{ |
|
33
|
|
|
$articleCrawler = new DomCrawler($responseBody); |
|
34
|
|
|
$article = new Article(); |
|
35
|
|
|
|
|
36
|
|
|
// extract name of the article |
|
37
|
|
|
$nameHtml = $articleCrawler->filter('h1.product-name'); |
|
38
|
|
|
$article->setName($nameHtml->text()); |
|
39
|
|
|
|
|
40
|
|
|
// extract ammount of total orders |
|
41
|
|
|
$article->setOrders($this->extractOrders($articleCrawler)); |
|
42
|
|
|
|
|
43
|
|
|
// extract amount of articles in stock |
|
44
|
|
|
$article->setStock($this->extractStock($articleCrawler)); |
|
45
|
|
|
|
|
46
|
|
|
return $article; |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* @param DomCrawler $articleCrawler |
|
51
|
|
|
* @return int |
|
52
|
|
|
*/ |
|
53
|
|
|
protected function extractOrders(DomCrawler $articleCrawler): int |
|
54
|
|
|
{ |
|
55
|
|
|
$ordersHtml = $articleCrawler->filter('#j-order-num'); |
|
56
|
|
|
$ordersString = $ordersHtml->text(); |
|
57
|
|
|
$ordersStringParts = explode(' ', $ordersString); |
|
58
|
|
|
|
|
59
|
|
|
return (int) $ordersStringParts[0]; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* @param DomCrawler $articleCrawler |
|
64
|
|
|
* @return int |
|
65
|
|
|
*/ |
|
66
|
|
|
protected function extractStock(DomCrawler $articleCrawler): int |
|
67
|
|
|
{ |
|
68
|
|
|
$stockAmountHtml = $articleCrawler->filter('#j-sell-stock-num'); |
|
69
|
|
|
$stockAmountString = $stockAmountHtml->text(); |
|
70
|
|
|
$stockAmountStringParts = explode(' ', $stockAmountString); |
|
71
|
|
|
|
|
72
|
|
|
return (int) $stockAmountStringParts[0]; |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
|
This check looks for parameters that have been defined for a function or method, but which are not used in the method body.