1
|
|
|
<?php |
2
|
|
|
/** @noinspection MultiAssignmentUsageInspection */ |
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* @author Ne-Lexa |
7
|
|
|
* @license MIT |
8
|
|
|
* @link https://github.com/Ne-Lexa/google-play-scraper |
9
|
|
|
*/ |
10
|
|
|
|
11
|
|
|
namespace Nelexa\GPlay\Scraper; |
12
|
|
|
|
13
|
|
|
use Nelexa\GPlay\Exception\GooglePlayException; |
14
|
|
|
use Nelexa\GPlay\GPlayApps; |
15
|
|
|
use Nelexa\GPlay\Http\ResponseHandlerInterface; |
16
|
|
|
use Nelexa\GPlay\Model\AppDetail; |
17
|
|
|
use Nelexa\GPlay\Model\AppId; |
18
|
|
|
use Nelexa\GPlay\Model\Category; |
19
|
|
|
use Nelexa\GPlay\Model\Developer; |
20
|
|
|
use Nelexa\GPlay\Model\GoogleImage; |
21
|
|
|
use Nelexa\GPlay\Model\HistogramRating; |
22
|
|
|
use Nelexa\GPlay\Model\Review; |
23
|
|
|
use Nelexa\GPlay\Model\Video; |
24
|
|
|
use Nelexa\GPlay\Scraper\Extractor\ReviewsExtractor; |
25
|
|
|
use Nelexa\GPlay\Util\DateStringFormatter; |
26
|
|
|
use Nelexa\GPlay\Util\LocaleHelper; |
27
|
|
|
use Nelexa\GPlay\Util\ScraperUtil; |
28
|
|
|
use Psr\Http\Message\RequestInterface; |
29
|
|
|
use Psr\Http\Message\ResponseInterface; |
30
|
|
|
use function GuzzleHttp\Psr7\parse_query; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @internal |
34
|
|
|
*/ |
35
|
|
|
class AppDetailScraper implements ResponseHandlerInterface |
36
|
|
|
{ |
37
|
|
|
/** |
38
|
|
|
* @param RequestInterface $request |
39
|
|
|
* @param ResponseInterface $response |
40
|
|
|
* @return AppDetail |
41
|
|
|
* @throws GooglePlayException |
42
|
|
|
*/ |
43
|
7 |
|
public function __invoke(RequestInterface $request, ResponseInterface $response): AppDetail |
44
|
|
|
{ |
45
|
7 |
|
$query = parse_query($request->getUri()->getQuery()); |
46
|
|
|
|
47
|
7 |
|
$id = $query[GPlayApps::REQ_PARAM_ID]; |
48
|
7 |
|
$locale = $query[GPlayApps::REQ_PARAM_LOCALE] ?? GPlayApps::DEFAULT_LOCALE; |
49
|
7 |
|
$country = $query[GPlayApps::REQ_PARAM_COUNTRY] ?? GPlayApps::DEFAULT_COUNTRY; |
50
|
|
|
|
51
|
|
|
[ |
52
|
|
|
$scriptDataInfo, |
53
|
|
|
$scriptDataRating, |
54
|
|
|
$scriptDataPrice, |
55
|
|
|
$scriptDataVersion, |
56
|
|
|
$scriptDataReviews, |
57
|
7 |
|
] = $this->getScriptData($request, $response); |
58
|
|
|
|
59
|
7 |
|
$name = $scriptDataInfo[0][0][0]; |
60
|
7 |
|
$description = $this->extractDescription($scriptDataInfo); |
61
|
7 |
|
$translatedFromLocale = $this->extractTranslatedFromLocale($scriptDataInfo, $locale); |
62
|
7 |
|
$developer = $this->extractDeveloper($scriptDataInfo); |
63
|
7 |
|
$category = $this->extractCategory($scriptDataInfo[0][12][13][0]); |
64
|
7 |
|
$summary = $this->extractSummary($scriptDataInfo); |
65
|
7 |
|
$installs = $scriptDataInfo[0][12][9][2] ?? 0; |
66
|
7 |
|
$score = (float)($scriptDataRating[0][6][0][1] ?? 0); |
67
|
7 |
|
$numberVoters = (int)($scriptDataRating[0][6][2][1] ?? 0); |
68
|
7 |
|
$numberReviews = (int)($scriptDataRating[0][6][3][1] ?? 0); |
69
|
7 |
|
$histogramRating = $this->extractHistogramRating($scriptDataRating); |
70
|
7 |
|
$price = $this->extractPrice($scriptDataPrice); |
71
|
7 |
|
$currency = $scriptDataPrice[0][2][0][0][0][1][0][1]; |
72
|
7 |
|
$priceText = $scriptDataPrice[0][2][0][0][0][1][0][2] ?: null; |
73
|
7 |
|
$offersIAPCost = $scriptDataInfo[0][12][12][0] ?? null; |
74
|
7 |
|
$containsAds = (bool)$scriptDataInfo[0][12][14][0]; |
75
|
|
|
|
76
|
7 |
|
[$size, $appVersion, $androidVersion] = $scriptDataVersion; |
77
|
7 |
|
if (LocaleHelper::isDependOnDevice($locale, $size)) { |
78
|
5 |
|
$size = null; |
79
|
|
|
} |
80
|
7 |
|
if (LocaleHelper::isDependOnDevice($locale, $appVersion)) { |
81
|
3 |
|
$appVersion = null; |
82
|
|
|
} |
83
|
7 |
|
if (LocaleHelper::isDependOnDevice($locale, $androidVersion)) { |
84
|
3 |
|
$androidVersion = null; |
85
|
3 |
|
$minAndroidVersion = null; |
86
|
|
|
} else { |
87
|
4 |
|
$minAndroidVersion = preg_replace('~.*?(\d+(\.\d+)*).*~', '$1', $androidVersion); |
88
|
|
|
} |
89
|
|
|
|
90
|
7 |
|
$editorsChoice = !empty($scriptDataInfo[0][12][15][1][1]); |
91
|
7 |
|
$privacyPoliceUrl = $scriptDataInfo[0][12][7][2] ?? ''; |
92
|
7 |
|
$categoryFamily = $this->extractCategory($scriptDataInfo[0][12][13][1] ?? []); |
93
|
7 |
|
$icon = $this->extractIcon($scriptDataInfo); |
94
|
7 |
|
$cover = $this->extractCover($scriptDataInfo); |
95
|
7 |
|
$screenshots = $this->extractScreenshots($scriptDataInfo); |
96
|
7 |
|
$video = $this->extractVideo($scriptDataInfo); |
97
|
7 |
|
$contentRating = $scriptDataInfo[0][12][4][0] ?? ''; |
98
|
7 |
|
$released = $this->extractReleaseDate($scriptDataInfo, $locale); |
99
|
7 |
|
$updated = $this->extractUpdatedDate($scriptDataInfo); |
100
|
7 |
|
$recentChanges = $this->extractRecentChanges($scriptDataInfo); |
101
|
7 |
|
$reviews = $this->extractReviews(new AppId($id, $locale, $country), $scriptDataReviews); |
102
|
|
|
|
103
|
7 |
|
return new AppDetail( |
104
|
7 |
|
AppDetail::newBuilder() |
105
|
7 |
|
->setId($id) |
106
|
7 |
|
->setLocale($locale) |
107
|
7 |
|
->setCountry($country) |
108
|
7 |
|
->setName($name) |
109
|
7 |
|
->setDescription($description) |
110
|
7 |
|
->setTranslatedFromLocale($translatedFromLocale) |
111
|
7 |
|
->setSummary($summary) |
112
|
7 |
|
->setIcon($icon) |
113
|
7 |
|
->setCover($cover) |
114
|
7 |
|
->setScreenshots($screenshots) |
115
|
7 |
|
->setDeveloper($developer) |
116
|
7 |
|
->setCategory($category) |
117
|
7 |
|
->setCategoryFamily($categoryFamily) |
118
|
7 |
|
->setVideo($video) |
119
|
7 |
|
->setRecentChanges($recentChanges) |
120
|
7 |
|
->setEditorsChoice($editorsChoice) |
121
|
7 |
|
->setPrivacyPoliceUrl($privacyPoliceUrl) |
122
|
7 |
|
->setInstalls($installs) |
123
|
7 |
|
->setScore($score) |
124
|
7 |
|
->setRecentChanges($recentChanges) |
125
|
7 |
|
->setEditorsChoice($editorsChoice) |
126
|
7 |
|
->setPrivacyPoliceUrl($privacyPoliceUrl) |
127
|
7 |
|
->setInstalls($installs) |
128
|
7 |
|
->setScore($score) |
129
|
7 |
|
->setNumberVoters($numberVoters) |
130
|
7 |
|
->setHistogramRating($histogramRating) |
131
|
7 |
|
->setPrice($price) |
132
|
7 |
|
->setCurrency($currency) |
133
|
7 |
|
->setPriceText($priceText) |
134
|
7 |
|
->setOffersIAPCost($offersIAPCost) |
135
|
7 |
|
->setContainsAds($containsAds) |
136
|
7 |
|
->setSize($size) |
137
|
7 |
|
->setAppVersion($appVersion) |
138
|
7 |
|
->setAndroidVersion($androidVersion) |
139
|
7 |
|
->setMinAndroidVersion($minAndroidVersion) |
140
|
7 |
|
->setContentRating($contentRating) |
141
|
7 |
|
->setReleased($released) |
142
|
7 |
|
->setUpdated($updated) |
143
|
7 |
|
->setNumberReviews($numberReviews) |
144
|
7 |
|
->setReviews($reviews) |
145
|
|
|
); |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* @param RequestInterface $request |
150
|
|
|
* @param ResponseInterface $response |
151
|
|
|
* @return array |
152
|
|
|
* @throws GooglePlayException |
153
|
|
|
*/ |
154
|
7 |
|
private function getScriptData(RequestInterface $request, ResponseInterface $response): array |
155
|
|
|
{ |
156
|
7 |
|
$scriptData = ScraperUtil::extractScriptData($response->getBody()->getContents()); |
157
|
|
|
|
158
|
7 |
|
$scriptDataInfo = null; |
159
|
7 |
|
$scriptDataRating = null; |
160
|
7 |
|
$scriptDataPrice = null; |
161
|
7 |
|
$scriptDataVersion = null; |
162
|
7 |
|
$scriptDataReviews = []; |
163
|
|
|
|
164
|
7 |
|
foreach ($scriptData as $key => $scriptValue) { |
165
|
7 |
|
if (isset($scriptValue[0][12][5][5][4][2])) { // ds:5 |
166
|
7 |
|
$scriptDataInfo = $scriptValue; |
167
|
7 |
|
} elseif (isset($scriptValue[0][2][0][0][0][1][0][0])) { // ds:3 |
168
|
7 |
|
$scriptDataPrice = $scriptValue; |
169
|
7 |
|
} elseif (isset($scriptValue[0][0][0]) |
170
|
7 |
|
&& is_string($scriptValue[0][0][0]) |
171
|
7 |
|
&& strpos($scriptValue[0][0][0], 'gp:') === 0) { // ds:15 |
172
|
7 |
|
$scriptDataReviews = $scriptValue; |
173
|
7 |
|
} elseif (isset($scriptValue[0][6][3][1])) { // ds:7 |
174
|
7 |
|
$scriptDataRating = $scriptValue; |
175
|
7 |
|
} elseif (isset($scriptValue[0]) |
176
|
7 |
|
&& is_string($scriptValue[0]) |
177
|
7 |
|
&& count($scriptValue) === 3) { // ds:8 |
178
|
7 |
|
$scriptDataVersion = $scriptValue; |
179
|
|
|
} |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
if ( |
183
|
7 |
|
$scriptDataInfo === null || |
184
|
7 |
|
$scriptDataRating === null || |
185
|
7 |
|
$scriptDataPrice === null || |
186
|
7 |
|
$scriptDataVersion === null |
|
|
|
|
187
|
|
|
) { |
188
|
|
|
throw (new GooglePlayException('Unable to get data for this application.'))->setUrl($request->getUri()->__toString()); |
189
|
|
|
} |
190
|
7 |
|
return [$scriptDataInfo, $scriptDataRating, $scriptDataPrice, $scriptDataVersion, $scriptDataReviews]; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* @param $scriptDataInfo |
195
|
|
|
* @param string $locale |
196
|
|
|
* @return string|null |
197
|
|
|
*/ |
198
|
7 |
|
private function extractTranslatedFromLocale(array $scriptDataInfo, string $locale): ?string |
199
|
|
|
{ |
200
|
7 |
|
return isset($scriptDataInfo[0][19][1]) ? |
201
|
2 |
|
LocaleHelper::findPreferredLanguage( |
202
|
2 |
|
$locale, |
203
|
2 |
|
$scriptDataInfo[0][19][1] |
204
|
|
|
) : |
205
|
7 |
|
null; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
/** |
209
|
|
|
* @param array $scriptDataInfo |
210
|
|
|
* @return string |
211
|
|
|
*/ |
212
|
7 |
|
private function extractDescription(array $scriptDataInfo): string |
213
|
|
|
{ |
214
|
7 |
|
if (isset($scriptDataInfo[0][19][0][0][1])) { |
215
|
2 |
|
return ScraperUtil::html2text($scriptDataInfo[0][19][0][0][1]); |
216
|
|
|
} |
217
|
|
|
|
218
|
7 |
|
return ScraperUtil::html2text($scriptDataInfo[0][10][0][1]); |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
/** |
222
|
|
|
* @param $scriptDataInfo |
223
|
|
|
* @return string|null |
224
|
|
|
*/ |
225
|
7 |
|
private function extractSummary(array $scriptDataInfo): ?string |
226
|
|
|
{ |
227
|
7 |
|
return empty($scriptDataInfo[0][10][1][1]) ? |
228
|
|
|
null : |
229
|
7 |
|
ScraperUtil::html2text($scriptDataInfo[0][10][1][1]); |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
/** |
233
|
|
|
* @param array $scriptDataInfo |
234
|
|
|
* @return Developer |
235
|
|
|
*/ |
236
|
7 |
|
private function extractDeveloper(array $scriptDataInfo): Developer |
237
|
|
|
{ |
238
|
7 |
|
$developerPage = GPlayApps::GOOGLE_PLAY_URL . $scriptDataInfo[0][12][5][5][4][2]; |
239
|
7 |
|
$developerId = parse_query(parse_url($developerPage, PHP_URL_QUERY))[GPlayApps::REQ_PARAM_ID]; |
240
|
7 |
|
$developerName = $scriptDataInfo[0][12][5][1]; |
241
|
7 |
|
$developerEmail = $scriptDataInfo[0][12][5][2][0]; |
242
|
7 |
|
$developerWebsite = $scriptDataInfo[0][12][5][3][5][2]; |
243
|
7 |
|
$developerAddress = $scriptDataInfo[0][12][5][4][0]; |
244
|
|
|
// $developerInternalID = (int)$scriptDataInfo[0][12][5][0][0]; |
245
|
|
|
|
246
|
7 |
|
return new Developer( |
247
|
7 |
|
Developer::newBuilder() |
248
|
7 |
|
->setId($developerId) |
249
|
7 |
|
->setUrl($developerPage) |
250
|
7 |
|
->setName($developerName) |
251
|
7 |
|
->setEmail($developerEmail) |
252
|
7 |
|
->setAddress($developerAddress) |
253
|
7 |
|
->setWebsite($developerWebsite) |
254
|
|
|
); |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
/** |
258
|
|
|
* @param array $data |
259
|
|
|
* @return Category|null |
260
|
|
|
*/ |
261
|
7 |
|
private function extractCategory(array $data): ?Category |
262
|
|
|
{ |
263
|
7 |
|
if (isset($data[0]) && $data[0] !== null && $data[2] !== null) { |
264
|
7 |
|
$genreId = (string)$data[2]; |
265
|
7 |
|
$genreName = (string)$data[0]; |
266
|
7 |
|
return new Category($genreId, $genreName); |
267
|
|
|
} |
268
|
7 |
|
return null; |
269
|
|
|
} |
270
|
|
|
|
271
|
|
|
/** |
272
|
|
|
* @param array $scriptDataRating |
273
|
|
|
* @return HistogramRating |
274
|
|
|
*/ |
275
|
7 |
|
private function extractHistogramRating(array $scriptDataRating): HistogramRating |
276
|
|
|
{ |
277
|
7 |
|
return new HistogramRating( |
278
|
7 |
|
$scriptDataRating[0][6][1][5][1] ?? 0, |
279
|
7 |
|
$scriptDataRating[0][6][1][4][1] ?? 0, |
280
|
7 |
|
$scriptDataRating[0][6][1][3][1] ?? 0, |
281
|
7 |
|
$scriptDataRating[0][6][1][2][1] ?? 0, |
282
|
7 |
|
$scriptDataRating[0][6][1][1][1] ?? 0 |
283
|
|
|
); |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* @param $scriptDataPrice |
288
|
|
|
* @return float |
289
|
|
|
*/ |
290
|
7 |
|
protected function extractPrice(array $scriptDataPrice): ?float |
291
|
|
|
{ |
292
|
7 |
|
return isset($scriptDataPrice[0][2][0][0][0][1][0][0]) ? |
293
|
7 |
|
(float)($scriptDataPrice[0][2][0][0][0][1][0][0] / 1000000) : |
294
|
7 |
|
0.0; |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* @param array $scriptDataInfo |
299
|
|
|
* @return GoogleImage|null |
300
|
|
|
*/ |
301
|
7 |
|
protected function extractIcon(array $scriptDataInfo): ?GoogleImage |
302
|
|
|
{ |
303
|
7 |
|
return empty($scriptDataInfo[0][12][1][3][2]) ? |
304
|
|
|
null : |
305
|
7 |
|
new GoogleImage($scriptDataInfo[0][12][1][3][2]); |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
/** |
309
|
|
|
* @param array $scriptDataInfo |
310
|
|
|
* @return GoogleImage|null |
311
|
|
|
*/ |
312
|
7 |
|
protected function extractCover(array $scriptDataInfo): ?GoogleImage |
313
|
|
|
{ |
314
|
7 |
|
return empty($scriptDataInfo[0][12][2][3][2]) ? |
315
|
|
|
null : |
316
|
7 |
|
new GoogleImage($scriptDataInfo[0][12][2][3][2]); |
317
|
|
|
} |
318
|
|
|
|
319
|
|
|
/** |
320
|
|
|
* @param array $scriptDataInfo |
321
|
|
|
* @return GoogleImage[] |
322
|
|
|
*/ |
323
|
7 |
|
private function extractScreenshots(array $scriptDataInfo): array |
324
|
|
|
{ |
325
|
|
|
return !empty($scriptDataInfo[0][12][0]) ? array_map(static function (array $v) { |
326
|
7 |
|
return new GoogleImage($v[3][2]); |
327
|
7 |
|
}, $scriptDataInfo[0][12][0]) : []; |
328
|
|
|
} |
329
|
|
|
|
330
|
|
|
/** |
331
|
|
|
* @param array $scriptDataInfo |
332
|
|
|
* @return Video|null |
333
|
|
|
*/ |
334
|
7 |
|
private function extractVideo(array $scriptDataInfo): ?Video |
335
|
|
|
{ |
336
|
|
|
if ( |
337
|
7 |
|
isset($scriptDataInfo[0][12][3][0][3][2]) && |
338
|
7 |
|
$scriptDataInfo[0][12][3][0][3][2] !== null && |
339
|
7 |
|
$scriptDataInfo[0][12][3][1][3][2] !== null |
340
|
|
|
) { |
341
|
2 |
|
$videoThumb = (string)$scriptDataInfo[0][12][3][1][3][2]; |
342
|
2 |
|
$videoUrl = (string)$scriptDataInfo[0][12][3][0][3][2]; |
343
|
|
|
|
344
|
2 |
|
return new Video($videoThumb, $videoUrl); |
345
|
|
|
} |
346
|
6 |
|
return null; |
347
|
|
|
} |
348
|
|
|
|
349
|
|
|
/** |
350
|
|
|
* @param array $scriptDataInfo |
351
|
|
|
* @param string $locale |
352
|
|
|
* @return \DateTimeInterface|null |
353
|
|
|
*/ |
354
|
7 |
|
private function extractReleaseDate(array $scriptDataInfo, string $locale): ?\DateTimeInterface |
355
|
|
|
{ |
356
|
7 |
|
if (isset($scriptDataInfo[0][12][36])) { |
357
|
6 |
|
return DateStringFormatter::formatted($locale, $scriptDataInfo[0][12][36]); |
358
|
|
|
} |
359
|
1 |
|
return null; |
360
|
|
|
} |
361
|
|
|
|
362
|
|
|
/** |
363
|
|
|
* @param array $scriptDataInfo |
364
|
|
|
* @return \DateTimeInterface|null |
365
|
|
|
*/ |
366
|
7 |
|
private function extractUpdatedDate(array $scriptDataInfo): ?\DateTimeInterface |
367
|
|
|
{ |
368
|
7 |
|
if (isset($scriptDataInfo[0][12][8][0])) { |
369
|
7 |
|
return DateStringFormatter::unixTimeToDateTime($scriptDataInfo[0][12][8][0]); |
370
|
|
|
} |
371
|
|
|
return null; |
372
|
|
|
} |
373
|
|
|
|
374
|
|
|
/** |
375
|
|
|
* @param $scriptDataInfo |
376
|
|
|
* @return string|null |
377
|
|
|
*/ |
378
|
7 |
|
protected function extractRecentChanges($scriptDataInfo): ?string |
379
|
|
|
{ |
380
|
7 |
|
return empty($scriptDataInfo[0][12][6][1]) ? |
381
|
|
|
null : |
382
|
7 |
|
ScraperUtil::html2text($scriptDataInfo[0][12][6][1]); |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
/** |
386
|
|
|
* @param AppId $appId |
387
|
|
|
* @param array $scriptDataReviews |
388
|
|
|
* @param int $limit |
389
|
|
|
* @return Review[] |
390
|
|
|
*/ |
391
|
7 |
|
private function extractReviews(AppId $appId, array $scriptDataReviews, int $limit = 4): array |
392
|
|
|
{ |
393
|
7 |
|
if (empty($scriptDataReviews[0])) { |
394
|
1 |
|
return []; |
395
|
|
|
} |
396
|
|
|
|
397
|
7 |
|
return ReviewsExtractor::extractReviews( |
398
|
7 |
|
$appId, |
399
|
7 |
|
array_slice($scriptDataReviews[0], 0, $limit) |
400
|
|
|
); |
401
|
|
|
} |
402
|
|
|
} |
403
|
|
|
|