1
|
|
|
<?php |
2
|
|
|
declare(strict_types=1); |
3
|
|
|
|
4
|
|
|
namespace Nelexa\GPlay\Scraper; |
5
|
|
|
|
6
|
|
|
use Nelexa\GPlay\Exception\GooglePlayException; |
7
|
|
|
use Nelexa\GPlay\GPlayApps; |
8
|
|
|
use Nelexa\GPlay\Http\ResponseHandlerInterface; |
9
|
|
|
use Nelexa\GPlay\Model\App; |
10
|
|
|
use Nelexa\GPlay\Model\Developer; |
11
|
|
|
use Nelexa\GPlay\Model\GoogleImage; |
12
|
|
|
use Nelexa\GPlay\Util\ScraperUtil; |
13
|
|
|
use Psr\Http\Message\RequestInterface; |
14
|
|
|
use Psr\Http\Message\ResponseInterface; |
15
|
|
|
use function GuzzleHttp\Psr7\parse_query; |
16
|
|
|
|
17
|
|
|
class ListScraper implements ResponseHandlerInterface |
18
|
|
|
{ |
19
|
|
|
/** |
20
|
|
|
* @param RequestInterface $request |
21
|
|
|
* @param ResponseInterface $response |
22
|
|
|
* @return App[] |
23
|
|
|
* @throws GooglePlayException |
24
|
|
|
*/ |
25
|
|
|
public function __invoke(RequestInterface $request, ResponseInterface $response) |
26
|
|
|
{ |
27
|
|
|
$doc = new \DOMDocument(); |
28
|
|
|
$internalErrors = libxml_use_internal_errors(true); |
29
|
|
|
if (!$doc->loadHTML('<?xml encoding="utf-8" ?>' . $response->getBody()->getContents())) { |
30
|
|
|
throw new \RuntimeException('error load html'); |
31
|
|
|
} |
32
|
|
|
libxml_use_internal_errors($internalErrors); |
33
|
|
|
|
34
|
|
|
$xpath = new \DOMXPath($doc); |
35
|
|
|
$cardNodes = $xpath->query("//div[@class and contains(concat(' ', normalize-space(@class), ' '), ' card ') and @data-docid]"); |
36
|
|
|
|
37
|
|
|
$apps = []; |
38
|
|
|
/** |
39
|
|
|
* @var \DOMElement $cardNode |
40
|
|
|
*/ |
41
|
|
|
foreach ($cardNodes as $cardNode) { |
42
|
|
|
{ |
43
|
|
|
$appId = $cardNode->getAttribute('data-docid'); |
44
|
|
|
} |
45
|
|
|
|
46
|
|
|
{ |
47
|
|
|
$nodeTitle = $xpath->query('.//a[@class="title"]', $cardNode)->item(0); |
48
|
|
|
if ($nodeTitle === null) { |
49
|
|
|
throw (new GooglePlayException('Error parse app list')) |
50
|
|
|
->setUrl($request->getUri()->__toString()); |
51
|
|
|
} |
52
|
|
|
$url = GPlayApps::GOOGLE_PLAY_URL . $nodeTitle->attributes->getNamedItem('href')->textContent; |
53
|
|
|
$name = trim($nodeTitle->attributes->getNamedItem('title')->textContent); |
54
|
|
|
} |
55
|
|
|
|
56
|
|
|
{ |
57
|
|
|
$descriptionNode = $xpath->query('.//div[@class="description"]', $cardNode)->item(0); |
58
|
|
|
$summary = null; |
59
|
|
|
if ($descriptionNode !== null) { |
60
|
|
|
$summary = ScraperUtil::html2text($descriptionNode->textContent); |
61
|
|
|
} |
62
|
|
|
} |
63
|
|
|
|
64
|
|
|
{ |
65
|
|
|
$developerNode = $xpath->query('.//a[@class="subtitle"]', $cardNode)->item(0); |
66
|
|
|
if ($developerNode === null) { |
67
|
|
|
throw (new GooglePlayException('Error parse app list developer node')) |
68
|
|
|
->setUrl($request->getUri()->__toString()); |
69
|
|
|
} |
70
|
|
|
$developerName = trim($developerNode->textContent); |
71
|
|
|
$developerUrl = GPlayApps::GOOGLE_PLAY_URL . $developerNode->attributes->getNamedItem('href')->textContent; |
72
|
|
|
$developerId = parse_query(parse_url($developerUrl, PHP_URL_QUERY))[GPlayApps::REQ_PARAM_APP_ID]; |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
{ |
76
|
|
|
$iconNode = $xpath->query('.//img[@data-cover-large]/@src', $cardNode)->item(0); |
77
|
|
|
if ($iconNode === null) { |
78
|
|
|
throw (new GooglePlayException('Error parse app list icon node')) |
79
|
|
|
->setUrl($request->getUri()->__toString()); |
80
|
|
|
} |
81
|
|
|
$icon = new GoogleImage('https:' . $iconNode->textContent); |
82
|
|
|
$icon->reset(); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
{ |
86
|
|
|
$price = null; |
87
|
|
|
$priceNode = $xpath->query('.//span[@class="display-price"]', $cardNode); |
88
|
|
|
if ($priceNode->length > 0) { |
89
|
|
|
$price = trim($priceNode->item(0)->textContent); |
90
|
|
|
if (empty($price)) { |
91
|
|
|
$price = null; |
92
|
|
|
} |
93
|
|
|
} |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
{ |
97
|
|
|
$score = 0; |
98
|
|
|
$ratingStyleAttr = $xpath->query('.//div[@class="current-rating" and @style]/@style', $cardNode)->item(0); |
99
|
|
|
if ($ratingStyleAttr !== null) { |
100
|
|
|
$ratingStyle = $ratingStyleAttr->textContent; |
101
|
|
|
if (preg_match('/([\d\.]+)%/', $ratingStyle, $match)) { |
102
|
|
|
$score = round($match[1] * 0.05, 1); // percent * 5 star and round result |
103
|
|
|
} |
104
|
|
|
} |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
$locale = parse_query($request->getUri()->getQuery())[GPlayApps::REQ_PARAM_LOCALE] ?? GPlayApps::DEFAULT_LOCALE; |
108
|
|
|
|
109
|
|
|
$apps[] = new App( |
110
|
|
|
App::newBuilder() |
111
|
|
|
->setId($appId) |
112
|
|
|
->setUrl($url) |
113
|
|
|
->setLocale($locale) |
114
|
|
|
->setName($name) |
115
|
|
|
->setSummary($summary) |
116
|
|
|
->setDeveloper( |
117
|
|
|
new Developer( |
118
|
|
|
Developer::newBuilder() |
119
|
|
|
->setId($developerId) |
120
|
|
|
->setUrl($developerUrl) |
121
|
|
|
->setName($developerName) |
122
|
|
|
) |
123
|
|
|
) |
124
|
|
|
->setIcon($icon) |
125
|
|
|
->setScore($score) |
126
|
|
|
->setPriceText($price) |
127
|
|
|
); |
128
|
|
|
} |
129
|
|
|
return $apps; |
130
|
|
|
} |
131
|
|
|
} |
132
|
|
|
|