1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* File: StatsHtmlScrapper.php |
4
|
|
|
* |
5
|
|
|
* @author Maciej Sławik <[email protected]> |
6
|
|
|
* Github: https://github.com/maciejslawik |
7
|
|
|
*/ |
8
|
|
|
|
9
|
|
|
namespace MSlwk\Otomoto\App\Stats\Scrapper; |
10
|
|
|
|
11
|
|
|
use MSlwk\Otomoto\App\Exception\OffersNotFoundException; |
12
|
|
|
use MSlwk\Otomoto\App\Stats\Data\StatsDTO; |
13
|
|
|
use MSlwk\Otomoto\App\Stats\Data\StatsDTOInterface; |
14
|
|
|
use MSlwk\Otomoto\App\Stats\Scrapper\Data\AverageDataScrapperInterface; |
15
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* Class StatsHtmlScrapper |
19
|
|
|
* @package MSlwk\Otomoto\App\Stats\Scrapper |
20
|
|
|
*/ |
21
|
|
|
class StatsHtmlScrapper implements StatsHtmlScrapperInterface |
22
|
|
|
{ |
23
|
|
|
const HTML_OFFER_SELECTOR = 'article.offer-item.is-row'; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* @var AverageDataScrapperInterface |
27
|
|
|
*/ |
28
|
|
|
private $mileageScrapper; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* @var AverageDataScrapperInterface |
32
|
|
|
*/ |
33
|
|
|
private $yearScrapper; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* @var AverageDataScrapperInterface |
37
|
|
|
*/ |
38
|
|
|
private $priceScrapper; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @var array |
42
|
|
|
*/ |
43
|
|
|
private $mileages = []; |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* @var array |
47
|
|
|
*/ |
48
|
|
|
private $prices = []; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* @var array |
52
|
|
|
*/ |
53
|
|
|
private $years = []; |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* StatsHtmlScrapper constructor. |
57
|
|
|
* @param AverageDataScrapperInterface $mileageScrapper |
58
|
|
|
* @param AverageDataScrapperInterface $yearScrapper |
59
|
|
|
* @param AverageDataScrapperInterface $priceScrapper |
60
|
|
|
*/ |
61
|
3 |
|
public function __construct( |
62
|
|
|
AverageDataScrapperInterface $mileageScrapper, |
63
|
|
|
AverageDataScrapperInterface $yearScrapper, |
64
|
|
|
AverageDataScrapperInterface $priceScrapper |
65
|
|
|
) { |
66
|
3 |
|
$this->mileageScrapper = $mileageScrapper; |
67
|
3 |
|
$this->yearScrapper = $yearScrapper; |
68
|
3 |
|
$this->priceScrapper = $priceScrapper; |
69
|
3 |
|
} |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @param string $html |
73
|
|
|
* @return StatsDTOInterface |
74
|
|
|
*/ |
75
|
2 |
|
public function scrapModelStats(string $html): StatsDTOInterface |
76
|
|
|
{ |
77
|
2 |
|
$crawler = new Crawler($html); |
78
|
2 |
|
$carInstances = $crawler->filter(self::HTML_OFFER_SELECTOR); |
79
|
|
|
|
80
|
|
|
$carInstances->each(function (Crawler $crawler) use (&$mileages, &$prices, &$years) { |
81
|
1 |
|
$this->mileages[] = $this->mileageScrapper->getAverageData($crawler); |
82
|
1 |
|
$this->prices[] = $this->priceScrapper->getAverageData($crawler); |
83
|
1 |
|
$this->years[] = $this->yearScrapper->getAverageData($crawler); |
84
|
2 |
|
}); |
85
|
|
|
|
86
|
2 |
|
$stats = $this->getAverageStats($this->mileages, $this->prices, $this->years); |
87
|
|
|
|
88
|
1 |
|
return $stats; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* @param $mileages |
93
|
|
|
* @param $prices |
94
|
|
|
* @param $years |
95
|
|
|
* @return StatsDTOInterface |
96
|
|
|
* @throws OffersNotFoundException |
97
|
|
|
*/ |
98
|
2 |
|
private function getAverageStats(array $mileages, array $prices, array $years): StatsDTOInterface |
99
|
|
|
{ |
100
|
2 |
|
$stats = new StatsDTO(); |
101
|
|
|
|
102
|
2 |
|
if (!$this->offersFound()) { |
103
|
1 |
|
throw new OffersNotFoundException('No offers were found'); |
104
|
|
|
} |
105
|
1 |
|
$stats->setAverageMileage(array_sum($mileages) / count($mileages)); |
106
|
1 |
|
$stats->setAveragePrice(array_sum($prices) / count($prices)); |
107
|
1 |
|
$stats->setAverageYear(array_sum($years) / count($years)); |
108
|
1 |
|
return $stats; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* @return bool |
113
|
|
|
*/ |
114
|
2 |
|
private function offersFound(): bool |
115
|
|
|
{ |
116
|
2 |
|
return count($this->mileages) && count($this->prices) && count($this->years); |
117
|
|
|
} |
118
|
|
|
} |
119
|
|
|
|