|
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\HttpClient; |
|
15
|
|
|
|
|
16
|
|
|
use GuzzleHttp\Client as GuzzleClient; |
|
17
|
|
|
use GuzzleHttp\Exception\ConnectException; |
|
18
|
|
|
use GuzzleHttp\Exception\TransferException; |
|
19
|
|
|
use GuzzleHttp\HandlerStack; |
|
20
|
|
|
use GuzzleHttp\MessageFormatter; |
|
21
|
|
|
use GuzzleHttp\Middleware; |
|
22
|
|
|
use GuzzleHttp\Pool; |
|
23
|
|
|
use GuzzleHttp\Promise\FulfilledPromise; |
|
24
|
|
|
use GuzzleHttp\Promise\PromiseInterface; |
|
25
|
|
|
use GuzzleHttp\RequestOptions; |
|
26
|
|
|
use Psr\Http\Message\RequestInterface; |
|
27
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
28
|
|
|
use Psr\SimpleCache\CacheInterface; |
|
29
|
|
|
use Psr\SimpleCache\InvalidArgumentException; |
|
30
|
|
|
|
|
31
|
|
|
class HttpClient |
|
32
|
|
|
{ |
|
33
|
|
|
public const DEFAULT_CONCURRENCY = 4; |
|
34
|
|
|
|
|
35
|
|
|
/** @var \Psr\SimpleCache\CacheInterface|null */ |
|
36
|
|
|
private $cache; |
|
37
|
|
|
|
|
38
|
|
|
/** @var \GuzzleHttp\Client */ |
|
39
|
|
|
private $client; |
|
40
|
|
|
|
|
41
|
|
|
/** @var array */ |
|
42
|
|
|
private $options = []; |
|
43
|
|
|
|
|
44
|
60 |
|
public function __construct(?GuzzleClient $client = null, ?CacheInterface $cache = null) |
|
45
|
|
|
{ |
|
46
|
2 |
|
if ($client === null) { |
|
47
|
2 |
|
$proxy = getenv('HTTP_PROXY'); |
|
48
|
|
|
|
|
49
|
|
|
$defaultOptions = [ |
|
50
|
|
|
RequestOptions::HEADERS => [ |
|
51
|
|
|
'User-Agent' => 'Mozilla/5.0 (X11; Linux x86_64; rv:97.0) Gecko/20100101 Firefox/97.0', |
|
52
|
|
|
], |
|
53
|
|
|
]; |
|
54
|
|
|
|
|
55
|
2 |
|
if ($proxy !== false) { |
|
56
|
|
|
$defaultOptions[RequestOptions::PROXY] = $proxy; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
2 |
|
$stack = HandlerStack::create(); |
|
60
|
2 |
|
$logTemplate = $config['logTemplate'] |
|
|
|
|
|
|
61
|
|
|
?? '🌎 [{ts}] "{method} {url} HTTP/{version}" {code} "{phrase}" - "{req_headers}"' |
|
62
|
|
|
. ' - {res_header_Content-Length}'; |
|
63
|
2 |
|
$stack->push(Middleware::log(new ConsoleLog(), new MessageFormatter($logTemplate)), 'logger'); |
|
64
|
2 |
|
$stack->push( |
|
65
|
2 |
|
Middleware::retry( |
|
66
|
2 |
|
static function ( |
|
67
|
|
|
int $retries, |
|
68
|
|
|
RequestInterface $request, |
|
69
|
|
|
?ResponseInterface $response = null, |
|
70
|
|
|
?TransferException $exception = null |
|
71
|
|
|
) { |
|
72
|
60 |
|
return $retries < 3 && ( |
|
73
|
60 |
|
$exception instanceof ConnectException |
|
74
|
|
|
|| ( |
|
75
|
60 |
|
$response !== null |
|
76
|
60 |
|
&& \in_array($response->getStatusCode(), [408, 429, 500, 502, 503, 522], true) |
|
77
|
|
|
) |
|
78
|
|
|
); |
|
79
|
|
|
}, |
|
80
|
2 |
|
static function (int $retries) { |
|
81
|
|
|
return 2 ** $retries * 1000; |
|
82
|
|
|
} |
|
83
|
|
|
), |
|
84
|
|
|
'retry' |
|
85
|
|
|
); |
|
86
|
2 |
|
$defaultOptions['handler'] = $stack; |
|
87
|
|
|
|
|
88
|
2 |
|
$client = new GuzzleClient($defaultOptions); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
2 |
|
$this->client = $client; |
|
92
|
2 |
|
$this->cache = $cache; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* @return \Psr\SimpleCache\CacheInterface|null |
|
97
|
|
|
*/ |
|
98
|
|
|
public function getCache(): ?CacheInterface |
|
99
|
|
|
{ |
|
100
|
|
|
return $this->cache; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* @return \GuzzleHttp\Client |
|
105
|
|
|
*/ |
|
106
|
3 |
|
public function getClient(): GuzzleClient |
|
107
|
|
|
{ |
|
108
|
3 |
|
return $this->client; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* @param \Nelexa\GPlay\HttpClient\Request $request |
|
113
|
|
|
* @param \Closure|null $onRejected |
|
114
|
|
|
* |
|
115
|
|
|
* @return mixed |
|
116
|
|
|
*/ |
|
117
|
36 |
|
public function request(Request $request, ?\Closure $onRejected = null) |
|
118
|
|
|
{ |
|
119
|
36 |
|
$promise = $this->getRequestPromise($request); |
|
120
|
36 |
|
$promise->otherwise( |
|
121
|
36 |
|
$onRejected ?? static function (\Throwable $throwable) { |
|
122
|
1 |
|
return $throwable; |
|
123
|
|
|
} |
|
124
|
|
|
); |
|
125
|
|
|
|
|
126
|
36 |
|
return $promise->wait(); |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* @param \Nelexa\GPlay\HttpClient\Request $request |
|
131
|
|
|
* |
|
132
|
|
|
* @return \GuzzleHttp\Promise\PromiseInterface |
|
133
|
|
|
* |
|
134
|
|
|
* @internal |
|
135
|
|
|
*/ |
|
136
|
63 |
|
public function getRequestPromise(Request $request): PromiseInterface |
|
137
|
|
|
{ |
|
138
|
63 |
|
$options = array_merge($this->options, $request->getOptions()); |
|
139
|
63 |
|
$cacheKey = null; |
|
140
|
|
|
|
|
141
|
|
|
if ( |
|
142
|
63 |
|
$this->cache !== null |
|
143
|
63 |
|
&& !\array_key_exists('no_cache', $options) |
|
144
|
63 |
|
&& \array_key_exists('cache_ttl', $options) |
|
145
|
|
|
) { |
|
146
|
63 |
|
$cacheKey = $options['cache_key'] ?? sprintf( |
|
147
|
|
|
'http_client_gplay.v1.%s.%s', |
|
148
|
63 |
|
HashUtil::hashCallable($request->getParseHandler()), |
|
149
|
63 |
|
HashUtil::getRequestHash($request->getPsrRequest()) |
|
150
|
|
|
); |
|
151
|
|
|
try { |
|
152
|
63 |
|
$cachedValue = $this->cache->get($cacheKey); |
|
153
|
|
|
} catch (InvalidArgumentException $e) { |
|
154
|
|
|
throw new \RuntimeException('Error fetch cache'); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
63 |
|
if ($cachedValue !== null) { |
|
158
|
15 |
|
return new FulfilledPromise($cachedValue); |
|
159
|
|
|
} |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
57 |
|
return $this->client |
|
163
|
57 |
|
->sendAsync($request->getPsrRequest(), $request->getOptions()) |
|
164
|
57 |
|
->then(function (ResponseInterface $response) use ($request, $cacheKey, $options) { |
|
165
|
55 |
|
$parseResult = $request->getParseHandler()($request->getPsrRequest(), $response, $options); |
|
166
|
55 |
|
if ($cacheKey !== null && $parseResult !== null) { |
|
167
|
55 |
|
$this->cache->set($cacheKey, $parseResult, $options['cache_ttl']); |
|
|
|
|
|
|
168
|
|
|
} |
|
169
|
|
|
|
|
170
|
55 |
|
return $parseResult; |
|
171
|
|
|
}) |
|
172
|
|
|
; |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
/** |
|
176
|
|
|
* @param array<Request> $requests |
|
177
|
|
|
* @param \Closure|null $onRejected |
|
178
|
|
|
* |
|
179
|
|
|
* @return array |
|
180
|
|
|
*/ |
|
181
|
27 |
|
public function requestPool(array $requests, ?\Closure $onRejected = null): array |
|
182
|
|
|
{ |
|
183
|
27 |
|
$makeRequests = function () use ($requests): \Generator { |
|
184
|
27 |
|
foreach ($requests as $key => $request) { |
|
185
|
27 |
|
yield $key => function () use ($request): PromiseInterface { |
|
186
|
27 |
|
return $this->getRequestPromise($request); |
|
187
|
|
|
}; |
|
188
|
|
|
} |
|
189
|
|
|
}; |
|
190
|
|
|
|
|
191
|
27 |
|
$results = []; |
|
192
|
27 |
|
$pool = new Pool($this->client, $makeRequests(), [ |
|
193
|
27 |
|
'concurrency' => $options['concurrency'] ?? self::DEFAULT_CONCURRENCY, |
|
|
|
|
|
|
194
|
27 |
|
'fulfilled' => static function ($result, $key) use (&$results): void { |
|
195
|
26 |
|
$results[$key] = $result; |
|
196
|
|
|
}, |
|
197
|
27 |
|
'rejected' => $onRejected ?? static function (\Throwable $throwable, $key): void { |
|
|
|
|
|
|
198
|
2 |
|
throw $throwable; |
|
199
|
|
|
}, |
|
200
|
|
|
]); |
|
201
|
|
|
|
|
202
|
27 |
|
$pool->promise()->wait(); |
|
203
|
|
|
|
|
204
|
25 |
|
return $results; |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
/** |
|
208
|
|
|
* @param \Psr\SimpleCache\CacheInterface|null $cache |
|
209
|
|
|
* |
|
210
|
|
|
* @return HttpClient |
|
211
|
|
|
*/ |
|
212
|
12 |
|
public function setCache(?CacheInterface $cache): self |
|
213
|
|
|
{ |
|
214
|
12 |
|
$this->cache = $cache; |
|
215
|
|
|
|
|
216
|
12 |
|
return $this; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
/** |
|
220
|
|
|
* @param \GuzzleHttp\Client $client |
|
221
|
|
|
* |
|
222
|
|
|
* @return HttpClient |
|
223
|
|
|
*/ |
|
224
|
|
|
public function setClient(GuzzleClient $client): self |
|
225
|
|
|
{ |
|
226
|
|
|
$this->client = $client; |
|
227
|
|
|
|
|
228
|
|
|
return $this; |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
12 |
|
public function setOption(string $key, $value): self |
|
232
|
|
|
{ |
|
233
|
12 |
|
$this->options[$key] = $value; |
|
234
|
|
|
|
|
235
|
12 |
|
return $this; |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
4 |
|
public function setConcurrency(int $concurrency): self |
|
239
|
|
|
{ |
|
240
|
4 |
|
$this->options['concurrency'] = max(1, $concurrency); |
|
241
|
|
|
|
|
242
|
4 |
|
return $this; |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
2 |
|
public function getConcurrency(): int |
|
246
|
|
|
{ |
|
247
|
2 |
|
return $this->options['concurrency'] ?? self::DEFAULT_CONCURRENCY; |
|
248
|
|
|
} |
|
249
|
|
|
|
|
250
|
|
|
public function setConnectTimeout(float $connectTimeout): self |
|
251
|
|
|
{ |
|
252
|
|
|
$this->options[RequestOptions::CONNECT_TIMEOUT] = max(0, $connectTimeout); |
|
253
|
|
|
|
|
254
|
|
|
return $this; |
|
255
|
|
|
} |
|
256
|
|
|
|
|
257
|
|
|
public function setTimeout(float $timeout): self |
|
258
|
|
|
{ |
|
259
|
|
|
$this->options[RequestOptions::TIMEOUT] = max(0, $timeout); |
|
260
|
|
|
|
|
261
|
|
|
return $this; |
|
262
|
|
|
} |
|
263
|
|
|
} |
|
264
|
|
|
|