|
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 CategoryAppsScraper 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
|
|
|
$locale = parse_query($request->getUri()->getQuery())[GPlayApps::REQ_PARAM_LOCALE] ?? GPlayApps::DEFAULT_LOCALE; |
|
28
|
|
|
|
|
29
|
|
|
$xpath = $this->getXPath($response); |
|
30
|
|
|
$cardNodes = $xpath->query("//div[@class and contains(concat(' ', normalize-space(@class), ' '), ' card ') and @data-docid]"); |
|
31
|
|
|
|
|
32
|
|
|
$apps = []; |
|
33
|
|
|
foreach ($cardNodes as $cardNode) { |
|
34
|
|
|
$apps[] = $this->extractApps($request, $xpath, $cardNode, $locale); |
|
35
|
|
|
} |
|
36
|
|
|
return $apps; |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @param ResponseInterface $response |
|
41
|
|
|
* @return \DOMXPath |
|
42
|
|
|
*/ |
|
43
|
|
|
private function getXPath(ResponseInterface $response): \DOMXPath |
|
44
|
|
|
{ |
|
45
|
|
|
$doc = new \DOMDocument(); |
|
46
|
|
|
$internalErrors = libxml_use_internal_errors(true); |
|
47
|
|
|
if (!$doc->loadHTML('<?xml encoding="utf-8" ?>' . $response->getBody()->getContents())) { |
|
48
|
|
|
throw new \RuntimeException('error load html'); |
|
49
|
|
|
} |
|
50
|
|
|
libxml_use_internal_errors($internalErrors); |
|
51
|
|
|
|
|
52
|
|
|
return new \DOMXPath($doc); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @param RequestInterface $request |
|
57
|
|
|
* @param \DOMXPath $xpath |
|
58
|
|
|
* @param \DOMElement $cardNode |
|
59
|
|
|
* @param string $locale |
|
60
|
|
|
* @return App |
|
61
|
|
|
* @throws GooglePlayException |
|
62
|
|
|
*/ |
|
63
|
|
|
private function extractApps( |
|
64
|
|
|
RequestInterface $request, |
|
65
|
|
|
\DOMXPath $xpath, |
|
66
|
|
|
\DOMElement $cardNode, |
|
67
|
|
|
string $locale |
|
68
|
|
|
): App { |
|
69
|
|
|
{ |
|
70
|
|
|
$appId = $cardNode->getAttribute('data-docid'); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
{ |
|
74
|
|
|
$nodeTitle = $xpath->query('.//a[@class="title"]', $cardNode)->item(0); |
|
75
|
|
|
if ($nodeTitle === null) { |
|
76
|
|
|
throw (new GooglePlayException('Error parse app list')) |
|
77
|
|
|
->setUrl($request->getUri()->__toString()); |
|
78
|
|
|
} |
|
79
|
|
|
$url = GPlayApps::GOOGLE_PLAY_URL . $nodeTitle->attributes->getNamedItem('href')->textContent; |
|
80
|
|
|
$name = trim($nodeTitle->attributes->getNamedItem('title')->textContent); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
$summary = $this->extractSummary($xpath, $cardNode); |
|
84
|
|
|
$developer = $this->extractDeveloper($request, $xpath, $cardNode); |
|
85
|
|
|
$icon = $this->extractIcon($request, $xpath, $cardNode); |
|
86
|
|
|
$price = $this->extractPrice($xpath, $cardNode); |
|
87
|
|
|
$score = $this->extractScore($xpath, $cardNode); |
|
88
|
|
|
|
|
89
|
|
|
return new App( |
|
90
|
|
|
App::newBuilder() |
|
91
|
|
|
->setId($appId) |
|
92
|
|
|
->setUrl($url) |
|
93
|
|
|
->setLocale($locale) |
|
94
|
|
|
->setName($name) |
|
95
|
|
|
->setSummary($summary) |
|
96
|
|
|
->setDeveloper($developer) |
|
97
|
|
|
->setIcon($icon) |
|
98
|
|
|
->setScore($score) |
|
99
|
|
|
->setPriceText($price) |
|
100
|
|
|
); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* @param \DOMXPath $xpath |
|
105
|
|
|
* @param \DOMElement $cardNode |
|
106
|
|
|
* @return string|null |
|
107
|
|
|
*/ |
|
108
|
|
|
private function extractSummary(\DOMXPath $xpath, \DOMElement $cardNode): ?string |
|
109
|
|
|
{ |
|
110
|
|
|
$descriptionNode = $xpath->query('.//div[@class="description"]', $cardNode)->item(0); |
|
111
|
|
|
if ($descriptionNode !== null) { |
|
112
|
|
|
return ScraperUtil::html2text($descriptionNode->textContent); |
|
113
|
|
|
} |
|
114
|
|
|
return null; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @param RequestInterface $request |
|
119
|
|
|
* @param \DOMXPath $xpath |
|
120
|
|
|
* @param \DOMElement $cardNode |
|
121
|
|
|
* @return Developer |
|
122
|
|
|
* @throws GooglePlayException |
|
123
|
|
|
*/ |
|
124
|
|
|
private function extractDeveloper(RequestInterface $request, \DOMXPath $xpath, \DOMElement $cardNode): Developer |
|
125
|
|
|
{ |
|
126
|
|
|
$developerNode = $xpath->query('.//a[@class="subtitle"]', $cardNode)->item(0); |
|
127
|
|
|
if ($developerNode === null) { |
|
128
|
|
|
throw (new GooglePlayException('Error parse app list developer node')) |
|
129
|
|
|
->setUrl($request->getUri()->__toString()); |
|
130
|
|
|
} |
|
131
|
|
|
$developerName = trim($developerNode->textContent); |
|
132
|
|
|
$developerUrl = GPlayApps::GOOGLE_PLAY_URL . $developerNode->attributes->getNamedItem('href')->textContent; |
|
133
|
|
|
$developerId = parse_query(parse_url($developerUrl, PHP_URL_QUERY))[GPlayApps::REQ_PARAM_ID]; |
|
134
|
|
|
$developer = new Developer( |
|
135
|
|
|
Developer::newBuilder() |
|
136
|
|
|
->setId($developerId) |
|
137
|
|
|
->setUrl($developerUrl) |
|
138
|
|
|
->setName($developerName) |
|
139
|
|
|
); |
|
140
|
|
|
return $developer; |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
/** |
|
144
|
|
|
* @param RequestInterface $request |
|
145
|
|
|
* @param \DOMXPath $xpath |
|
146
|
|
|
* @param \DOMElement $cardNode |
|
147
|
|
|
* @return GoogleImage |
|
148
|
|
|
* @throws GooglePlayException |
|
149
|
|
|
*/ |
|
150
|
|
|
private function extractIcon(RequestInterface $request, \DOMXPath $xpath, \DOMElement $cardNode): GoogleImage |
|
151
|
|
|
{ |
|
152
|
|
|
$iconNode = $xpath->query('.//img[@data-cover-large]/@src', $cardNode)->item(0); |
|
153
|
|
|
if ($iconNode === null) { |
|
154
|
|
|
throw (new GooglePlayException('Error parse app list icon node')) |
|
155
|
|
|
->setUrl($request->getUri()->__toString()); |
|
156
|
|
|
} |
|
157
|
|
|
$icon = new GoogleImage('https:' . $iconNode->textContent); |
|
158
|
|
|
$icon->reset(); |
|
159
|
|
|
return $icon; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* @param \DOMXPath $xpath |
|
164
|
|
|
* @param \DOMElement $cardNode |
|
165
|
|
|
* @return string|null |
|
166
|
|
|
*/ |
|
167
|
|
|
private function extractPrice(\DOMXPath $xpath, \DOMElement $cardNode): ?string |
|
168
|
|
|
{ |
|
169
|
|
|
$priceNode = $xpath->query('.//span[@class="display-price"]', $cardNode); |
|
170
|
|
|
if ($priceNode->length > 0) { |
|
171
|
|
|
$price = trim($priceNode->item(0)->textContent); |
|
172
|
|
|
if (!empty($price)) { |
|
173
|
|
|
return $price; |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
return null; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
/** |
|
180
|
|
|
* @param \DOMXPath $xpath |
|
181
|
|
|
* @param \DOMElement $cardNode |
|
182
|
|
|
* @return float |
|
183
|
|
|
*/ |
|
184
|
|
|
private function extractScore(\DOMXPath $xpath, \DOMElement $cardNode): float |
|
185
|
|
|
{ |
|
186
|
|
|
$ratingStyleAttr = $xpath->query('.//div[@class="current-rating" and @style]/@style', $cardNode)->item(0); |
|
187
|
|
|
if ($ratingStyleAttr !== null) { |
|
188
|
|
|
$ratingStyle = $ratingStyleAttr->textContent; |
|
189
|
|
|
if (preg_match('/([\d\.]+)%/', $ratingStyle, $match)) { |
|
190
|
|
|
return round($match[1] * 0.05, 1); // percent * 5 star and round result |
|
191
|
|
|
} |
|
192
|
|
|
} |
|
193
|
|
|
return 0; |
|
194
|
|
|
} |
|
195
|
|
|
} |
|
196
|
|
|
|
Ne-Lexa /
google-play-scraper
