|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
|
|
4
|
|
|
namespace Nelexa\GPlay\Scraper; |
|
5
|
|
|
|
|
6
|
|
|
use function GuzzleHttp\Psr7\parse_query; |
|
7
|
|
|
use Nelexa\GPlay\GPlayApps; |
|
8
|
|
|
use Nelexa\GPlay\Http\ResponseHandlerInterface; |
|
9
|
|
|
use Nelexa\GPlay\Model\GoogleImage; |
|
10
|
|
|
use Nelexa\GPlay\Model\ReplyReview; |
|
11
|
|
|
use Nelexa\GPlay\Model\Review; |
|
12
|
|
|
use Nelexa\GPlay\Util\DateStringFormatter; |
|
13
|
|
|
use Nelexa\GPlay\Util\LocaleHelper; |
|
14
|
|
|
use Psr\Http\Message\RequestInterface; |
|
15
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
16
|
|
|
|
|
17
|
|
|
class AppReviewScraper implements ResponseHandlerInterface |
|
18
|
|
|
{ |
|
19
|
|
|
/** |
|
20
|
|
|
* @param RequestInterface $request |
|
21
|
|
|
* @param ResponseInterface $response |
|
22
|
|
|
* @return Review[] |
|
23
|
|
|
*/ |
|
24
|
|
|
public function __invoke(RequestInterface $request, ResponseInterface $response): array |
|
25
|
|
|
{ |
|
26
|
|
|
$contents = substr($response->getBody()->getContents(), 6); |
|
27
|
|
|
$json = \GuzzleHttp\json_decode($contents, true); |
|
28
|
|
|
$html = $json[0][2]; |
|
29
|
|
|
|
|
30
|
|
|
$doc = new \DOMDocument(); |
|
31
|
|
|
$internalErrors = libxml_use_internal_errors(true); |
|
32
|
|
|
if (!$doc->loadHTML('<?xml encoding="utf-8" ?>' . $html)) { |
|
33
|
|
|
throw new \RuntimeException('error load html'); |
|
34
|
|
|
} |
|
35
|
|
|
libxml_use_internal_errors($internalErrors); |
|
36
|
|
|
|
|
37
|
|
|
$locale = parse_query($request->getUri()->getQuery())[GPlayApps::REQ_PARAM_LOCALE] ?? GPlayApps::DEFAULT_LOCALE; |
|
38
|
|
|
|
|
39
|
|
|
$reviews = []; |
|
40
|
|
|
|
|
41
|
|
|
$xpath = new \DOMXPath($doc); |
|
42
|
|
|
$reviewNodes = $xpath->query('//div[@class="single-review"]'); |
|
43
|
|
|
/** |
|
44
|
|
|
* @var \DOMElement $reviewNode |
|
45
|
|
|
*/ |
|
46
|
|
|
foreach ($reviewNodes as $reviewNode) { |
|
47
|
|
|
// review id |
|
48
|
|
|
{ |
|
49
|
|
|
$reviewIdNode = $xpath->query('.//div[@data-reviewid]', $reviewNode)->item(0); |
|
50
|
|
|
if ($reviewIdNode === null) { |
|
51
|
|
|
continue; |
|
52
|
|
|
} |
|
53
|
|
|
$reviewId = $reviewIdNode->attributes->getNamedItem('data-reviewid')->textContent; |
|
54
|
|
|
|
|
55
|
|
|
// url |
|
56
|
|
|
{ |
|
57
|
|
|
$reviewUrlNode = $xpath->query('.//a[@class="reviews-permalink"]/@href', $reviewIdNode); |
|
58
|
|
|
if ($reviewUrlNode === null) { |
|
59
|
|
|
continue; |
|
60
|
|
|
} |
|
61
|
|
|
$reviewUrl = GPlayApps::GOOGLE_PLAY_URL . $reviewUrlNode->item(0)->textContent; |
|
62
|
|
|
} |
|
63
|
|
|
// user name |
|
64
|
|
|
{ |
|
65
|
|
|
$userNameNode = $xpath->query('.//span[@class="author-name"]', $reviewIdNode)->item(0); |
|
66
|
|
|
if ($userNameNode === null) { |
|
67
|
|
|
continue; |
|
68
|
|
|
} |
|
69
|
|
|
$userName = trim($userNameNode->textContent); |
|
70
|
|
|
} |
|
71
|
|
|
// review date |
|
72
|
|
|
{ |
|
73
|
|
|
$reviewDateNode = $xpath->query('.//span[@class="review-date"]', $reviewIdNode)->item(0); |
|
74
|
|
|
if ($reviewDateNode === null) { |
|
75
|
|
|
continue; |
|
76
|
|
|
} |
|
77
|
|
|
$reviewDate = trim($reviewDateNode->textContent); |
|
78
|
|
|
$reviewDateTime = DateStringFormatter::formatted($locale, $reviewDate); |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
// avatar |
|
82
|
|
|
{ |
|
83
|
|
|
$avatar = null; |
|
84
|
|
|
$avatarAttr = $xpath->query('.//span[@class="responsive-img-hdpi"]/span/@style', $reviewNode)->item(0); |
|
85
|
|
|
if ($avatarAttr !== null) { |
|
86
|
|
|
$style = $avatarAttr->textContent; |
|
87
|
|
|
if (preg_match('~url\s*\((.*?)\)~', $style, $match)) { |
|
88
|
|
|
$avatarImage = $match[1]; |
|
89
|
|
|
$avatarImage = str_replace(['"', "'"], '', $avatarImage); |
|
90
|
|
|
$avatar = new GoogleImage($avatarImage); |
|
91
|
|
|
} |
|
92
|
|
|
} |
|
93
|
|
|
} |
|
94
|
|
|
// rating |
|
95
|
|
|
{ |
|
96
|
|
|
$score = 0; |
|
97
|
|
|
$ratingStyleAttr = $xpath->query('.//div[@class="current-rating" and @style]/@style', $reviewNode)->item(0); |
|
98
|
|
|
if ($ratingStyleAttr !== null) { |
|
99
|
|
|
$ratingStyle = $ratingStyleAttr->textContent; |
|
100
|
|
|
if (preg_match('/([\d]+)%/', $ratingStyle, $match)) { |
|
101
|
|
|
$score = (int)($match[1] * 0.05); // percent * 5 star |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
// text |
|
106
|
|
|
{ |
|
107
|
|
|
$reviewTextNode = $xpath->query(".//div[@class and contains(concat(' ', normalize-space(@class), ' '), ' review-body ')]", $reviewNode)->item(0); |
|
108
|
|
|
if ($reviewTextNode === null) { |
|
109
|
|
|
continue; |
|
110
|
|
|
} |
|
111
|
|
|
$nodeReviewLinkNode = $xpath->query(".//div[@class='review-link']", $reviewTextNode); |
|
112
|
|
|
if ($nodeReviewLinkNode->length > 0) { |
|
113
|
|
|
$reviewTextNode->removeChild($nodeReviewLinkNode->item(0)); |
|
114
|
|
|
} |
|
115
|
|
|
$reviewText = trim($reviewTextNode->textContent); |
|
116
|
|
|
} |
|
117
|
|
|
// reply |
|
118
|
|
|
{ |
|
119
|
|
|
$reply = null; |
|
120
|
|
|
if ($reviewNode->nextSibling !== null && $reviewNode->nextSibling->nextSibling->getAttribute('class') === 'developer-reply') { |
|
121
|
|
|
$replyDateTime = null; |
|
122
|
|
|
$replyNode = $reviewNode->nextSibling->nextSibling; |
|
123
|
|
|
// reply date |
|
124
|
|
|
{ |
|
125
|
|
|
$replyDateNode = $xpath->query('.//span[@class="review-date"]', $replyNode)->item(0); |
|
126
|
|
|
if ($replyDateNode !== null) { |
|
127
|
|
|
$replyDate = trim($replyDateNode->textContent); |
|
128
|
|
|
$replyDateTime = DateStringFormatter::formatted($locale, $replyDate); |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
// reply text |
|
132
|
|
|
{ |
|
133
|
|
|
$replyText = null; |
|
134
|
|
|
$replyTextNode = $replyNode->childNodes->item(2); |
|
135
|
|
|
if ($replyTextNode !== null) { |
|
136
|
|
|
$replyText = trim($replyTextNode->textContent); |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
if ($replyDateTime !== null && $replyText !== null) { |
|
140
|
|
|
$reply = new ReplyReview( |
|
141
|
|
|
$replyDateTime, |
|
142
|
|
|
$replyText |
|
143
|
|
|
); |
|
144
|
|
|
} |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
$reviews[] = new Review( |
|
149
|
|
|
$reviewId, |
|
150
|
|
|
$reviewUrl, |
|
151
|
|
|
$userName, |
|
152
|
|
|
$reviewText, |
|
153
|
|
|
$avatar, |
|
|
|
|
|
|
154
|
|
|
$reviewDateTime, |
|
155
|
|
|
$score, |
|
156
|
|
|
0, |
|
157
|
|
|
$reply |
|
158
|
|
|
); |
|
159
|
|
|
} |
|
160
|
|
|
return $reviews; |
|
161
|
|
|
} |
|
162
|
|
|
} |
|
163
|
|
|
|