|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
/* |
|
6
|
|
|
* Copyright (c) Ne-Lexa |
|
7
|
|
|
* |
|
8
|
|
|
* For the full copyright and license information, please view |
|
9
|
|
|
* the LICENSE file that was distributed with this source code. |
|
10
|
|
|
* |
|
11
|
|
|
* @see https://github.com/Ne-Lexa/google-play-scraper |
|
12
|
|
|
*/ |
|
13
|
|
|
|
|
14
|
|
|
namespace Nelexa\GPlay\Scraper; |
|
15
|
|
|
|
|
16
|
|
|
use GuzzleHttp\Psr7\Query; |
|
17
|
|
|
use Nelexa\GPlay\Exception\GooglePlayException; |
|
18
|
|
|
use Nelexa\GPlay\GPlayApps; |
|
19
|
|
|
use Nelexa\GPlay\HttpClient\ParseHandlerInterface; |
|
20
|
|
|
use Nelexa\GPlay\Model\AppId; |
|
21
|
|
|
use Nelexa\GPlay\Model\AppInfo; |
|
22
|
|
|
use Nelexa\GPlay\Model\Category; |
|
23
|
|
|
use Nelexa\GPlay\Model\Developer; |
|
24
|
|
|
use Nelexa\GPlay\Model\GoogleImage; |
|
25
|
|
|
use Nelexa\GPlay\Model\HistogramRating; |
|
26
|
|
|
use Nelexa\GPlay\Model\Review; |
|
27
|
|
|
use Nelexa\GPlay\Model\Video; |
|
28
|
|
|
use Nelexa\GPlay\Scraper\Extractor\ReviewsExtractor; |
|
29
|
|
|
use Nelexa\GPlay\Util\DateStringFormatter; |
|
30
|
|
|
use Nelexa\GPlay\Util\ScraperUtil; |
|
31
|
|
|
use Psr\Http\Message\RequestInterface; |
|
32
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @internal |
|
36
|
|
|
*/ |
|
37
|
|
|
class AppInfoScraper implements ParseHandlerInterface |
|
38
|
|
|
{ |
|
39
|
|
|
/** |
|
40
|
|
|
* @param RequestInterface $request |
|
41
|
|
|
* @param ResponseInterface $response |
|
42
|
|
|
* @param array $options |
|
43
|
|
|
* |
|
44
|
|
|
* @throws \Nelexa\GPlay\Exception\GooglePlayException |
|
45
|
|
|
* |
|
46
|
|
|
* @return AppInfo |
|
47
|
|
|
*/ |
|
48
|
10 |
|
public function __invoke(RequestInterface $request, ResponseInterface $response, array &$options = []): AppInfo |
|
49
|
|
|
{ |
|
50
|
10 |
|
$scriptData = ScraperUtil::extractScriptData($response->getBody()->getContents()); |
|
51
|
|
|
|
|
52
|
10 |
|
$appInfo = null; |
|
53
|
10 |
|
$editorsChoice = false; |
|
54
|
|
|
|
|
55
|
10 |
|
foreach ($scriptData as $data) { |
|
56
|
10 |
|
if (isset($data[1][2][72][0][1])) { |
|
57
|
10 |
|
$appInfo = $data[1][2]; |
|
58
|
10 |
|
} elseif (isset($data[1][2][136][0][1][0])) { |
|
59
|
|
|
$editorsChoice = (bool) $data[1][2][136][0][1][0]; |
|
60
|
|
|
} |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
10 |
|
if (!\is_array($appInfo)) { |
|
64
|
|
|
throw (new GooglePlayException('Unable to get data for this application.'))->setUrl( |
|
65
|
|
|
$request->getUri()->__toString() |
|
66
|
|
|
); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
10 |
|
$query = Query::parse($request->getUri()->getQuery()); |
|
70
|
10 |
|
$id = $query[GPlayApps::REQ_PARAM_ID]; |
|
71
|
10 |
|
$locale = $query[GPlayApps::REQ_PARAM_LOCALE] ?? GPlayApps::DEFAULT_LOCALE; |
|
72
|
10 |
|
$country = $query[GPlayApps::REQ_PARAM_COUNTRY] ?? GPlayApps::DEFAULT_COUNTRY; |
|
73
|
|
|
|
|
74
|
10 |
|
$name = $appInfo[0][0]; |
|
75
|
10 |
|
$description = ScraperUtil::html2text($appInfo[72][0][1]); |
|
76
|
10 |
|
$developer = $this->extractDeveloper($appInfo); |
|
77
|
|
|
|
|
78
|
10 |
|
$category = $this->extractCategory($appInfo[79][0][0] ?? []); |
|
79
|
10 |
|
$summary = $this->extractSummary($appInfo); |
|
80
|
10 |
|
$installsText = $appInfo[13][0] ?? null; |
|
81
|
10 |
|
$installs = $appInfo[13][2] ?? 0; |
|
82
|
10 |
|
$score = (float) ($appInfo[51][0][1] ?? 0); |
|
83
|
10 |
|
$numberVoters = (int) ($appInfo[51][2][1] ?? 0); |
|
84
|
10 |
|
$numberReviews = (int) ($appInfo[51][3][1] ?? 0); |
|
85
|
10 |
|
$histogramRating = null; |
|
86
|
10 |
|
if (isset($appInfo[51][1])) { |
|
87
|
10 |
|
$histogramRating = $this->extractHistogramRating($appInfo[51][1]); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
10 |
|
$scriptDataPrice = $appInfo[57][0][0][0][0][1] ?? null; |
|
91
|
10 |
|
$price = $scriptDataPrice !== null ? $this->extractPrice($scriptDataPrice) : 0.0; |
|
92
|
10 |
|
$currency = $scriptDataPrice[0][1] ?? 'USD'; |
|
93
|
10 |
|
$priceText = $scriptDataPrice[0][2] ?? null; |
|
94
|
|
|
|
|
95
|
10 |
|
$offersIAPCost = $appInfo[19][0] ?? null; |
|
96
|
10 |
|
$containsAds = (bool) ($appInfo[48][0] ?? false); |
|
97
|
|
|
|
|
98
|
10 |
|
$androidVersion = $appInfo[140][1][1][0][0][1] ?? null; |
|
99
|
10 |
|
$appVersion = $appInfo[140][0][0][0] ?? null; |
|
100
|
|
|
|
|
101
|
10 |
|
if ($androidVersion !== null) { |
|
102
|
4 |
|
$minAndroidVersion = preg_replace('~.*?(\d+(\.\d+)*).*~', '$1', $androidVersion); |
|
103
|
|
|
} else { |
|
104
|
7 |
|
$minAndroidVersion = null; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
10 |
|
$privacyPoliceUrl = $appInfo[99][0][5][2] ?? ''; |
|
108
|
10 |
|
$categoryFamily = $this->extractCategory($appInfo[118][0][0][0] ?? []); |
|
109
|
10 |
|
$icon = $this->extractIcon($appInfo); |
|
110
|
10 |
|
$cover = $this->extractCover($appInfo); |
|
111
|
10 |
|
$screenshots = $this->extractScreenshots($appInfo); |
|
112
|
10 |
|
$video = $this->extractVideo($appInfo); |
|
113
|
10 |
|
$contentRating = $appInfo[111][1] ?? ''; |
|
114
|
10 |
|
$released = $this->convertDate($appInfo[10][1][0] ?? null); |
|
115
|
10 |
|
$updated = $this->convertDate($appInfo[145][0][1][0] ?? null); |
|
116
|
10 |
|
$recentChanges = $this->extractRecentChanges($appInfo); |
|
117
|
|
|
|
|
118
|
10 |
|
$reviews = $this->extractReviews(new AppId($id, $locale, $country), $scriptData); |
|
119
|
|
|
|
|
120
|
10 |
|
return AppInfo::newBuilder() |
|
121
|
10 |
|
->setId($id) |
|
122
|
10 |
|
->setLocale($locale) |
|
123
|
10 |
|
->setCountry($country) |
|
124
|
10 |
|
->setName($name) |
|
125
|
10 |
|
->setDescription($description) |
|
126
|
10 |
|
->setSummary($summary) |
|
127
|
10 |
|
->setIcon($icon) |
|
128
|
10 |
|
->setCover($cover) |
|
129
|
10 |
|
->setScreenshots($screenshots) |
|
130
|
10 |
|
->setDeveloper($developer) |
|
131
|
10 |
|
->setCategory($category) |
|
132
|
10 |
|
->setCategoryFamily($categoryFamily) |
|
133
|
10 |
|
->setVideo($video) |
|
134
|
10 |
|
->setRecentChanges($recentChanges) |
|
135
|
10 |
|
->setEditorsChoice($editorsChoice) |
|
136
|
10 |
|
->setPrivacyPoliceUrl($privacyPoliceUrl) |
|
137
|
10 |
|
->setInstalls($installs) |
|
138
|
10 |
|
->setInstallsText($installsText) |
|
139
|
10 |
|
->setScore($score) |
|
140
|
10 |
|
->setRecentChanges($recentChanges) |
|
141
|
10 |
|
->setEditorsChoice($editorsChoice) |
|
142
|
10 |
|
->setPrivacyPoliceUrl($privacyPoliceUrl) |
|
143
|
10 |
|
->setInstalls($installs) |
|
144
|
10 |
|
->setScore($score) |
|
145
|
10 |
|
->setNumberVoters($numberVoters) |
|
146
|
10 |
|
->setHistogramRating($histogramRating) |
|
147
|
10 |
|
->setPrice($price) |
|
148
|
10 |
|
->setCurrency($currency) |
|
149
|
10 |
|
->setPriceText($priceText) |
|
150
|
10 |
|
->setOffersIAPCost($offersIAPCost) |
|
151
|
10 |
|
->setContainsAds($containsAds) |
|
152
|
10 |
|
->setAppVersion($appVersion) |
|
153
|
10 |
|
->setAndroidVersion($androidVersion) |
|
154
|
10 |
|
->setMinAndroidVersion($minAndroidVersion) |
|
155
|
10 |
|
->setContentRating($contentRating) |
|
156
|
10 |
|
->setReleased($released) |
|
157
|
10 |
|
->setUpdated($updated) |
|
158
|
10 |
|
->setNumberReviews($numberReviews) |
|
159
|
10 |
|
->setReviews($reviews) |
|
160
|
10 |
|
->buildDetailInfo() |
|
161
|
|
|
; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
/** |
|
165
|
|
|
* @param array $appInfo |
|
166
|
|
|
* |
|
167
|
|
|
* @return string|null |
|
168
|
|
|
*/ |
|
169
|
10 |
|
private function extractSummary(array $appInfo): ?string |
|
170
|
|
|
{ |
|
171
|
10 |
|
return empty($appInfo[73][0][1]) |
|
172
|
|
|
? null |
|
173
|
10 |
|
: ScraperUtil::html2text(str_replace("\n", ' ', $appInfo[73][0][1])); |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
/** |
|
177
|
|
|
* @param array $appInfo |
|
178
|
|
|
* |
|
179
|
|
|
* @return Developer |
|
180
|
|
|
*/ |
|
181
|
10 |
|
private function extractDeveloper(array $appInfo): Developer |
|
182
|
|
|
{ |
|
183
|
10 |
|
$developerPage = GPlayApps::GOOGLE_PLAY_URL . $appInfo[68][1][4][2]; |
|
184
|
10 |
|
$developerId = Query::parse(parse_url($developerPage, \PHP_URL_QUERY))[GPlayApps::REQ_PARAM_ID]; |
|
185
|
10 |
|
$developerName = $appInfo[68][0]; |
|
186
|
10 |
|
$developerEmail = $appInfo[69][1][0] ?? null; |
|
187
|
10 |
|
$developerWebsite = $appInfo[69][0][5][2] ?? null; |
|
188
|
10 |
|
$developerAddress = $appInfo[69][2][0] ?? null; |
|
189
|
|
|
|
|
190
|
10 |
|
return new Developer( |
|
191
|
10 |
|
Developer::newBuilder() |
|
192
|
10 |
|
->setId($developerId) |
|
193
|
10 |
|
->setUrl($developerPage) |
|
194
|
10 |
|
->setName($developerName) |
|
195
|
10 |
|
->setEmail($developerEmail) |
|
196
|
10 |
|
->setAddress($developerAddress) |
|
197
|
10 |
|
->setWebsite($developerWebsite) |
|
198
|
|
|
); |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
/** |
|
202
|
|
|
* @param array $data |
|
203
|
|
|
* |
|
204
|
|
|
* @return Category|null |
|
205
|
|
|
*/ |
|
206
|
10 |
|
private function extractCategory(array $data): ?Category |
|
207
|
|
|
{ |
|
208
|
10 |
|
if (isset($data[1][4][2], $data[0], $data[2])) { |
|
209
|
10 |
|
$categorySlug = (string) $data[2]; |
|
210
|
10 |
|
$categoryName = (string) $data[0]; |
|
211
|
|
|
|
|
212
|
10 |
|
return new Category($categorySlug, $categoryName); |
|
213
|
|
|
} |
|
214
|
|
|
|
|
215
|
9 |
|
return null; |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
/** |
|
219
|
|
|
* @param array|null $data |
|
220
|
|
|
* |
|
221
|
|
|
* @return HistogramRating |
|
222
|
|
|
*/ |
|
223
|
10 |
|
private function extractHistogramRating(array $data): HistogramRating |
|
224
|
|
|
{ |
|
225
|
10 |
|
return new HistogramRating( |
|
226
|
10 |
|
$data[1][1] ?? 0, |
|
227
|
10 |
|
$data[2][1] ?? 0, |
|
228
|
10 |
|
$data[3][1] ?? 0, |
|
229
|
10 |
|
$data[4][1] ?? 0, |
|
230
|
10 |
|
$data[5][1] ?? 0 |
|
231
|
|
|
); |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* @param array $scriptDataPrice |
|
236
|
|
|
* |
|
237
|
|
|
* @return float |
|
238
|
|
|
*/ |
|
239
|
10 |
|
protected function extractPrice(array $scriptDataPrice): ?float |
|
240
|
|
|
{ |
|
241
|
10 |
|
return isset($scriptDataPrice[0][0]) |
|
242
|
10 |
|
? (float) ($scriptDataPrice[0][0] / 1000000) |
|
243
|
10 |
|
: 0.0; |
|
244
|
|
|
} |
|
245
|
|
|
|
|
246
|
|
|
/** |
|
247
|
|
|
* @param array $data |
|
248
|
|
|
* |
|
249
|
|
|
* @return GoogleImage|null |
|
250
|
|
|
*/ |
|
251
|
10 |
|
protected function extractIcon(array $data): ?GoogleImage |
|
252
|
|
|
{ |
|
253
|
10 |
|
return empty($data[95][0][3][2]) |
|
254
|
|
|
? null |
|
255
|
10 |
|
: new GoogleImage($data[95][0][3][2]); |
|
256
|
|
|
} |
|
257
|
|
|
|
|
258
|
|
|
/** |
|
259
|
|
|
* @param array $data |
|
260
|
|
|
* |
|
261
|
|
|
* @return GoogleImage|null |
|
262
|
|
|
*/ |
|
263
|
10 |
|
protected function extractCover(array $data): ?GoogleImage |
|
264
|
|
|
{ |
|
265
|
10 |
|
return empty($data[96][0][3][2]) |
|
266
|
|
|
? null |
|
267
|
10 |
|
: new GoogleImage($data[96][0][3][2]); |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
|
|
/** |
|
271
|
|
|
* @param array $data |
|
272
|
|
|
* |
|
273
|
|
|
* @return GoogleImage[] |
|
274
|
|
|
*/ |
|
275
|
10 |
|
private function extractScreenshots(array $data): array |
|
276
|
|
|
{ |
|
277
|
10 |
|
return !empty($data[78][0][0][3][2]) ? array_map( |
|
278
|
10 |
|
static function (array $v) { |
|
279
|
10 |
|
return new GoogleImage($v[3][2]); |
|
280
|
|
|
}, |
|
281
|
10 |
|
$data[78][0] |
|
282
|
10 |
|
) : []; |
|
283
|
|
|
} |
|
284
|
|
|
|
|
285
|
|
|
/** |
|
286
|
|
|
* @param array $data |
|
287
|
|
|
* |
|
288
|
|
|
* @return Video|null |
|
289
|
|
|
*/ |
|
290
|
10 |
|
private function extractVideo(array $data): ?Video |
|
291
|
|
|
{ |
|
292
|
10 |
|
if (isset($data[100][0][0][4], $data[100][0][1][3][3])) { |
|
293
|
|
|
$videoThumb = (string) $data[100][0][1][3][2]; |
|
294
|
|
|
$youtubeId = (string) $data[100][0][4]; |
|
295
|
|
|
$youtubeId = str_replace('yt:', '', strtok($youtubeId, '?')); |
|
296
|
|
|
$videoUrl = 'https://www.youtube.com/embed/' . $youtubeId . '?ps=play&vq=large&rel=0&autohide=1&showinfo=0'; |
|
297
|
|
|
|
|
298
|
|
|
return new Video($videoThumb, $videoUrl); |
|
299
|
|
|
} |
|
300
|
|
|
|
|
301
|
10 |
|
return null; |
|
302
|
|
|
} |
|
303
|
|
|
|
|
304
|
|
|
/** |
|
305
|
|
|
* @param ?int $timestamp |
|
306
|
|
|
* |
|
307
|
|
|
* @return \DateTimeInterface|null |
|
308
|
|
|
*/ |
|
309
|
10 |
|
private function convertDate(?int $timestamp): ?\DateTimeInterface |
|
310
|
|
|
{ |
|
311
|
10 |
|
if ($timestamp !== null) { |
|
312
|
10 |
|
return DateStringFormatter::unixTimeToDateTime($timestamp); |
|
313
|
|
|
} |
|
314
|
|
|
|
|
315
|
2 |
|
return null; |
|
316
|
|
|
} |
|
317
|
|
|
|
|
318
|
|
|
/** |
|
319
|
|
|
* @param array $data |
|
320
|
|
|
* |
|
321
|
|
|
* @return string|null |
|
322
|
|
|
*/ |
|
323
|
10 |
|
protected function extractRecentChanges(array $data): ?string |
|
324
|
|
|
{ |
|
325
|
10 |
|
return empty($data[144][1][1]) |
|
326
|
2 |
|
? null |
|
327
|
10 |
|
: ScraperUtil::html2text($data[144][1][1]); |
|
328
|
|
|
} |
|
329
|
|
|
|
|
330
|
|
|
/** |
|
331
|
|
|
* @param AppId $appId |
|
332
|
|
|
* @param array $data |
|
333
|
|
|
* @param array $scripData |
|
334
|
|
|
* |
|
335
|
|
|
* @return Review[] |
|
336
|
|
|
*/ |
|
337
|
10 |
|
private function extractReviews(AppId $appId, array $scripData): array |
|
338
|
|
|
{ |
|
339
|
10 |
|
$data = null; |
|
340
|
10 |
|
foreach ($scripData as $value) { |
|
341
|
|
|
if ( |
|
342
|
10 |
|
isset($value[0][0][0]) |
|
343
|
10 |
|
&& \is_string($value[0][0][0]) |
|
344
|
10 |
|
&& preg_match('~^[0-9a-f]{8}-[0-9a-f]{4}-~', $value[0][0][0]) |
|
345
|
|
|
) { |
|
346
|
10 |
|
$data = $value[0]; |
|
347
|
10 |
|
break; |
|
348
|
|
|
} |
|
349
|
|
|
} |
|
350
|
|
|
|
|
351
|
10 |
|
if ($data === null) { |
|
352
|
1 |
|
return []; |
|
353
|
|
|
} |
|
354
|
|
|
|
|
355
|
10 |
|
return ReviewsExtractor::extractReviews($appId, $data); |
|
356
|
|
|
} |
|
357
|
|
|
} |
|
358
|
|
|
|
Ne-Lexa /
google-play-scraper
