|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
|
|
4
|
|
|
/** |
|
5
|
|
|
* @author Ne-Lexa |
|
6
|
|
|
* @license MIT |
|
7
|
|
|
* @link https://github.com/Ne-Lexa/google-play-scraper |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace Nelexa\GPlay\Scraper; |
|
11
|
|
|
|
|
12
|
|
|
use Nelexa\GPlay\Exception\GooglePlayException; |
|
13
|
|
|
use Nelexa\GPlay\Http\ResponseHandlerInterface; |
|
14
|
|
|
use Nelexa\GPlay\Model\Category; |
|
15
|
|
|
use Nelexa\GPlay\Util\ScraperUtil; |
|
16
|
|
|
use Psr\Http\Message\RequestInterface; |
|
17
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @internal |
|
21
|
|
|
*/ |
|
22
|
|
|
class CategoriesScraper implements ResponseHandlerInterface |
|
23
|
|
|
{ |
|
24
|
|
|
private const CATEGORY_URL_PREFIX = '/store/apps/category/'; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @param RequestInterface $request |
|
28
|
|
|
* @param ResponseInterface $response |
|
29
|
|
|
* @return mixed |
|
30
|
|
|
* @throws GooglePlayException |
|
31
|
|
|
*/ |
|
32
|
3 |
|
public function __invoke(RequestInterface $request, ResponseInterface $response) |
|
33
|
|
|
{ |
|
34
|
3 |
|
$dataCategories = $this->getDataScraper($request, $response); |
|
35
|
|
|
|
|
36
|
|
|
$parseCategories = static function (array $items) use (&$parseCategories) : array { |
|
37
|
|
|
return array_reduce($items, static function ($results, $item) use (&$parseCategories) { |
|
38
|
3 |
|
if (is_array($item)) { |
|
39
|
|
|
if ( |
|
40
|
3 |
|
count($item) === 6 && |
|
41
|
3 |
|
strpos($item[0], self::CATEGORY_URL_PREFIX) === 0 && |
|
42
|
3 |
|
strpos($item[0], '?age=') === false |
|
43
|
|
|
) { |
|
44
|
3 |
|
$id = basename($item[0]); |
|
45
|
3 |
|
$categoryName = $item[1]; |
|
46
|
3 |
|
$results[] = new Category( |
|
47
|
3 |
|
$id, |
|
48
|
3 |
|
$categoryName |
|
49
|
|
|
); |
|
50
|
|
|
} else { |
|
51
|
3 |
|
$results = array_merge($results, $parseCategories($item)); |
|
52
|
|
|
} |
|
53
|
|
|
} |
|
54
|
3 |
|
return $results; |
|
55
|
3 |
|
}, []); |
|
56
|
3 |
|
}; |
|
57
|
3 |
|
return $parseCategories($dataCategories[0][1][0][3]); |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* @param RequestInterface $request |
|
62
|
|
|
* @param ResponseInterface $response |
|
63
|
|
|
* @return array |
|
64
|
|
|
* @throws GooglePlayException |
|
65
|
|
|
*/ |
|
66
|
3 |
|
private function getDataScraper(RequestInterface $request, ResponseInterface $response): array |
|
67
|
|
|
{ |
|
68
|
3 |
|
$scriptData = ScraperUtil::extractScriptData($response->getBody()->getContents()); |
|
69
|
|
|
|
|
70
|
3 |
|
$dataCategories = null; |
|
71
|
3 |
|
foreach ($scriptData as $key => $data) { |
|
72
|
3 |
|
if (isset($data[0][1][0][3][0]) && is_array($data[0][1][0][3][0])) { |
|
73
|
3 |
|
$dataCategories = $data; |
|
74
|
3 |
|
break; |
|
75
|
|
|
} |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
3 |
|
if ($dataCategories === null) { |
|
79
|
|
|
throw (new GooglePlayException('Failed to get the list of categories.')) |
|
80
|
|
|
->setUrl($request->getUri()->__toString()); |
|
81
|
|
|
} |
|
82
|
3 |
|
return $dataCategories; |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
|
Ne-Lexa /
google-play-scraper
