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
|
51 |
|
public function __construct(?GuzzleClient $client = null, ?CacheInterface $cache = null) |
45
|
|
|
{ |
46
|
2 |
|
if ($client === null) { |
47
|
2 |
|
$proxy = getenv('HTTP_PROXY'); |
48
|
|
|
|
49
|
|
|
$defaultOptions = [ |
50
|
|
|
RequestOptions::HEADERS => [ |
51
|
2 |
|
'User-Agent' => 'Mozilla/5.0 (X11; Linux x86_64; rv:103.0) Gecko/20100101 Firefox/103.0', |
52
|
|
|
], |
53
|
|
|
]; |
54
|
|
|
|
55
|
2 |
|
if ($proxy !== false) { |
56
|
|
|
$defaultOptions[RequestOptions::PROXY] = $proxy; |
57
|
|
|
} |
58
|
|
|
|
59
|
2 |
|
$stack = HandlerStack::create(); |
60
|
2 |
|
if (\PHP_SAPI === 'cli') { |
61
|
2 |
|
$logTemplate = $config['logTemplate'] |
|
|
|
|
62
|
|
|
?? '🌎 [{ts}] "{method} {url} HTTP/{version}" {code} "{phrase}" - {res_header_Content-Length}'; |
63
|
2 |
|
$stack->push(Middleware::log(new ConsoleLog(), new MessageFormatter($logTemplate)), 'logger'); |
64
|
|
|
} |
65
|
2 |
|
$stack->push( |
66
|
2 |
|
Middleware::retry( |
67
|
2 |
|
static function ( |
68
|
|
|
int $retries, |
69
|
|
|
RequestInterface $request, |
70
|
|
|
?ResponseInterface $response = null, |
71
|
|
|
?TransferException $exception = null |
72
|
|
|
) { |
73
|
51 |
|
return $retries < 3 && ( |
74
|
51 |
|
$exception instanceof ConnectException |
75
|
|
|
|| ( |
76
|
51 |
|
$response !== null |
77
|
51 |
|
&& \in_array($response->getStatusCode(), [408, 429, 500, 502, 503, 522], true) |
78
|
|
|
) |
79
|
|
|
); |
80
|
|
|
}, |
81
|
2 |
|
static function (int $retries) { |
82
|
|
|
return 2 ** $retries * 1000; |
83
|
|
|
} |
84
|
|
|
), |
85
|
|
|
'retry' |
86
|
|
|
); |
87
|
2 |
|
$defaultOptions['handler'] = $stack; |
88
|
|
|
|
89
|
2 |
|
$client = new GuzzleClient($defaultOptions); |
90
|
|
|
} |
91
|
|
|
|
92
|
2 |
|
$this->client = $client; |
93
|
2 |
|
$this->cache = $cache; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* @return \Psr\SimpleCache\CacheInterface|null |
98
|
|
|
*/ |
99
|
|
|
public function getCache(): ?CacheInterface |
100
|
|
|
{ |
101
|
|
|
return $this->cache; |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @return \GuzzleHttp\Client |
106
|
|
|
*/ |
107
|
3 |
|
public function getClient(): GuzzleClient |
108
|
|
|
{ |
109
|
3 |
|
return $this->client; |
110
|
|
|
} |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* @param \Nelexa\GPlay\HttpClient\Request $request |
114
|
|
|
* @param \Closure|null $onRejected |
115
|
|
|
* |
116
|
|
|
* @return mixed |
117
|
|
|
*/ |
118
|
36 |
|
public function request(Request $request, ?\Closure $onRejected = null) |
119
|
|
|
{ |
120
|
36 |
|
$promise = $this->getRequestPromise($request); |
121
|
36 |
|
$promise->otherwise( |
122
|
36 |
|
$onRejected ?? static function (\Throwable $throwable) { |
123
|
1 |
|
return $throwable; |
124
|
|
|
} |
125
|
|
|
); |
126
|
|
|
|
127
|
36 |
|
return $promise->wait(); |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
/** |
131
|
|
|
* @param \Nelexa\GPlay\HttpClient\Request $request |
132
|
|
|
* |
133
|
|
|
* @return \GuzzleHttp\Promise\PromiseInterface |
134
|
|
|
* |
135
|
|
|
* @internal |
136
|
|
|
*/ |
137
|
49 |
|
public function getRequestPromise(Request $request): PromiseInterface |
138
|
|
|
{ |
139
|
49 |
|
$options = array_merge($this->options, $request->getOptions()); |
140
|
49 |
|
$cacheKey = null; |
141
|
|
|
|
142
|
|
|
if ( |
143
|
49 |
|
$this->cache !== null |
144
|
49 |
|
&& !\array_key_exists('no_cache', $options) |
145
|
49 |
|
&& \array_key_exists('cache_ttl', $options) |
146
|
|
|
) { |
147
|
|
|
$cacheKey = $options['cache_key'] ?? sprintf( |
148
|
|
|
'http_client_gplay.v1.%s.%s', |
149
|
|
|
HashUtil::hashCallable($request->getParseHandler()), |
150
|
|
|
HashUtil::getRequestHash($request->getPsrRequest()) |
151
|
|
|
); |
152
|
|
|
try { |
153
|
|
|
$cachedValue = $this->cache->get($cacheKey); |
154
|
|
|
} catch (InvalidArgumentException $e) { |
155
|
|
|
throw new \RuntimeException('Error fetch cache'); |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
if ($cachedValue !== null) { |
159
|
|
|
return new FulfilledPromise($cachedValue); |
160
|
|
|
} |
161
|
|
|
} |
162
|
|
|
|
163
|
49 |
|
return $this->client |
164
|
49 |
|
->sendAsync($request->getPsrRequest(), $request->getOptions()) |
165
|
49 |
|
->then(function (ResponseInterface $response) use ($request, $cacheKey, $options) { |
166
|
47 |
|
$parseResult = $request->getParseHandler()($request->getPsrRequest(), $response, $options); |
167
|
47 |
|
if ($cacheKey !== null && $parseResult !== null) { |
168
|
|
|
$this->cache->set($cacheKey, $parseResult, $options['cache_ttl']); |
|
|
|
|
169
|
|
|
} |
170
|
|
|
|
171
|
47 |
|
return $parseResult; |
172
|
|
|
}) |
173
|
|
|
; |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
/** |
177
|
|
|
* @param array<Request> $requests |
178
|
|
|
* @param \Closure|null $onRejected |
179
|
|
|
* |
180
|
|
|
* @return array |
181
|
|
|
*/ |
182
|
13 |
|
public function requestPool(array $requests, ?\Closure $onRejected = null): array |
183
|
|
|
{ |
184
|
13 |
|
$makeRequests = function () use ($requests): \Generator { |
185
|
13 |
|
foreach ($requests as $key => $request) { |
186
|
13 |
|
yield $key => function () use ($request): PromiseInterface { |
187
|
13 |
|
return $this->getRequestPromise($request); |
188
|
|
|
}; |
189
|
|
|
} |
190
|
|
|
}; |
191
|
|
|
|
192
|
13 |
|
$results = []; |
193
|
13 |
|
$pool = new Pool($this->client, $makeRequests(), [ |
194
|
13 |
|
'concurrency' => $options['concurrency'] ?? self::DEFAULT_CONCURRENCY, |
|
|
|
|
195
|
13 |
|
'fulfilled' => static function ($result, $key) use (&$results): void { |
196
|
12 |
|
$results[$key] = $result; |
197
|
|
|
}, |
198
|
13 |
|
'rejected' => $onRejected ?? static function (\Throwable $throwable, $key): void { |
|
|
|
|
199
|
2 |
|
throw $throwable; |
200
|
|
|
}, |
201
|
|
|
]); |
202
|
|
|
|
203
|
13 |
|
$pool->promise()->wait(); |
204
|
|
|
|
205
|
11 |
|
return $results; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
/** |
209
|
|
|
* @param \Psr\SimpleCache\CacheInterface|null $cache |
210
|
|
|
* |
211
|
|
|
* @return HttpClient |
212
|
|
|
*/ |
213
|
|
|
public function setCache(?CacheInterface $cache): self |
214
|
|
|
{ |
215
|
|
|
$this->cache = $cache; |
216
|
|
|
|
217
|
|
|
return $this; |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
/** |
221
|
|
|
* @param \GuzzleHttp\Client $client |
222
|
|
|
* |
223
|
|
|
* @return HttpClient |
224
|
|
|
*/ |
225
|
|
|
public function setClient(GuzzleClient $client): self |
226
|
|
|
{ |
227
|
|
|
$this->client = $client; |
228
|
|
|
|
229
|
|
|
return $this; |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
public function setOption(string $key, $value): self |
233
|
|
|
{ |
234
|
|
|
$this->options[$key] = $value; |
235
|
|
|
|
236
|
|
|
return $this; |
237
|
|
|
} |
238
|
|
|
|
239
|
3 |
|
public function setConcurrency(int $concurrency): self |
240
|
|
|
{ |
241
|
3 |
|
$this->options['concurrency'] = max(1, $concurrency); |
242
|
|
|
|
243
|
3 |
|
return $this; |
244
|
|
|
} |
245
|
|
|
|
246
|
2 |
|
public function getConcurrency(): int |
247
|
|
|
{ |
248
|
2 |
|
return $this->options['concurrency'] ?? self::DEFAULT_CONCURRENCY; |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
public function setConnectTimeout(float $connectTimeout): self |
252
|
|
|
{ |
253
|
|
|
$this->options[RequestOptions::CONNECT_TIMEOUT] = max(0, $connectTimeout); |
254
|
|
|
|
255
|
|
|
return $this; |
256
|
|
|
} |
257
|
|
|
|
258
|
|
|
public function setTimeout(float $timeout): self |
259
|
|
|
{ |
260
|
|
|
$this->options[RequestOptions::TIMEOUT] = max(0, $timeout); |
261
|
|
|
|
262
|
|
|
return $this; |
263
|
|
|
} |
264
|
|
|
} |
265
|
|
|
|