|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* vipnytt/UserAgentParser |
|
4
|
|
|
* |
|
5
|
|
|
* @link https://github.com/VIPnytt/UserAgentParser |
|
6
|
|
|
* @license https://github.com/VIPnytt/UserAgentParser/blob/master/LICENSE The MIT License (MIT) |
|
7
|
|
|
*/ |
|
8
|
|
|
|
|
9
|
|
|
namespace vipnytt; |
|
10
|
|
|
|
|
11
|
|
|
use vipnytt\UserAgentParser\Exceptions\FormatException; |
|
12
|
|
|
use vipnytt\UserAgentParser\Exceptions\ProductException; |
|
13
|
|
|
use vipnytt\UserAgentParser\Exceptions\VersionException; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Class UserAgentParser |
|
17
|
|
|
* |
|
18
|
|
|
* @link https://tools.ietf.org/html/rfc7231 |
|
19
|
|
|
* @link https://tools.ietf.org/html/rfc7230 |
|
20
|
|
|
* |
|
21
|
|
|
* @package vipnytt |
|
22
|
|
|
*/ |
|
23
|
|
|
class UserAgentParser |
|
24
|
|
|
{ |
|
25
|
|
|
/** |
|
26
|
|
|
* RFC 7231 - Section 5.5.3 - User-agent |
|
27
|
|
|
*/ |
|
28
|
|
|
const RFC_README = 'https://tools.ietf.org/html/rfc7231#section-5.5.3'; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* PREG pattern for valid characters |
|
32
|
|
|
*/ |
|
33
|
|
|
const PREG_PATTERN = '/[^\x21-\x7E]/'; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Origin product |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
private $originProduct; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Product |
|
43
|
|
|
* @var string |
|
44
|
|
|
*/ |
|
45
|
|
|
private $product; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Version |
|
49
|
|
|
* @var float|int|string|null |
|
50
|
|
|
*/ |
|
51
|
|
|
private $version; |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* UserAgentParser constructor |
|
55
|
|
|
* |
|
56
|
|
|
* @param string $product |
|
57
|
|
|
* @param float|int|string|null $version |
|
58
|
|
|
*/ |
|
59
|
|
|
public function __construct($product, $version = null) |
|
60
|
|
|
{ |
|
61
|
|
|
$this->product = $product; |
|
62
|
|
|
$this->version = $version; |
|
63
|
|
|
if (strpos($this->product, '/') !== false) { |
|
64
|
|
|
$this->split(); |
|
65
|
|
|
} |
|
66
|
|
|
$this->validateProduct(); |
|
67
|
|
|
$this->validateVersion(); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* Split Product and Version |
|
72
|
|
|
* |
|
73
|
|
|
* @return bool |
|
74
|
|
|
*/ |
|
75
|
|
|
private function split() |
|
76
|
|
|
{ |
|
77
|
|
|
if (count($parts = explode('/', trim($this->product . '/' . $this->version, '/'), 2)) === 2) { |
|
78
|
|
|
$this->product = $parts[0]; |
|
79
|
|
|
$this->version = $parts[1]; |
|
80
|
|
|
} |
|
81
|
|
|
return true; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* Validate the Product format |
|
86
|
|
|
* @link https://tools.ietf.org/html/rfc7231#section-5.5.3 |
|
87
|
|
|
* @link https://tools.ietf.org/html/rfc7230#section-3.2.4 |
|
88
|
|
|
* |
|
89
|
|
|
* @return bool |
|
90
|
|
|
* @throws ProductException |
|
91
|
|
|
*/ |
|
92
|
|
|
private function validateProduct() |
|
93
|
|
|
{ |
|
94
|
|
|
$this->blacklistCheck($this->product); |
|
95
|
|
|
$this->originProduct = $this->product; |
|
96
|
|
|
if ($this->originProduct !== ($this->product = preg_replace(self::PREG_PATTERN, '', $this->product))) { |
|
97
|
|
|
trigger_error("Product name contains invalid characters. Truncated to `$this->product`.", E_USER_NOTICE); |
|
98
|
|
|
} |
|
99
|
|
|
if (empty($this->product)) { |
|
100
|
|
|
throw new ProductException('Product string cannot be empty.'); |
|
101
|
|
|
} |
|
102
|
|
|
return true; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* Check for blacklisted strings or characters |
|
107
|
|
|
* |
|
108
|
|
|
* @param float|int|string|null $input |
|
109
|
|
|
* @return bool |
|
110
|
|
|
* @throws FormatException |
|
111
|
|
|
*/ |
|
112
|
|
|
private function blacklistCheck($input) |
|
113
|
|
|
{ |
|
114
|
|
|
foreach ([ |
|
115
|
|
|
'mozilla', |
|
116
|
|
|
'compatible', |
|
117
|
|
|
'(', |
|
118
|
|
|
')', |
|
119
|
|
|
] as $string) { |
|
120
|
|
|
if (stripos($input, $string) !== false) { |
|
121
|
|
|
throw new FormatException('Invalid User-agent format (`' . trim($this->product . '/' . $this->version, '/') . '`). Examples of valid User-agents: `MyCustomBot`, `MyFetcher-news`, `MyCrawler/2.1` and `MyBot-images/1.2`. See also ' . self::RFC_README); |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
return true; |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
/** |
|
128
|
|
|
* Validate the Version and it's format |
|
129
|
|
|
* @link https://tools.ietf.org/html/rfc7231#section-5.5.3 |
|
130
|
|
|
* |
|
131
|
|
|
* @return bool |
|
132
|
|
|
* @throws VersionException |
|
133
|
|
|
*/ |
|
134
|
|
|
private function validateVersion() |
|
135
|
|
|
{ |
|
136
|
|
|
$this->blacklistCheck($this->version); |
|
137
|
|
|
if ( |
|
138
|
|
|
!empty($this->version) && |
|
139
|
|
|
( |
|
140
|
|
|
str_replace('+', '', filter_var($this->version, FILTER_SANITIZE_NUMBER_FLOAT, FILTER_FLAG_ALLOW_FRACTION)) || |
|
141
|
|
|
version_compare($this->version, '0.0.1', '>=') === false |
|
142
|
|
|
) |
|
143
|
|
|
) { |
|
144
|
|
|
throw new VersionException("Invalid version format (` $this->version `). See http://semver.org/ for guidelines. In addition, dev/alpha/beta/rc tags is disallowed. See also " . self::RFC_README); |
|
145
|
|
|
} |
|
146
|
|
|
$new = trim($this->version, '.0'); |
|
147
|
|
|
$this->version = empty($new) ? null : $new; |
|
148
|
|
|
return true; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
/** |
|
152
|
|
|
* Get User-agent |
|
153
|
|
|
* |
|
154
|
|
|
* @return string |
|
155
|
|
|
*/ |
|
156
|
|
|
public function getUserAgent() |
|
157
|
|
|
{ |
|
158
|
|
|
$product = $this->getProduct(); |
|
159
|
|
|
$version = $this->getVersion(); |
|
160
|
|
|
return $version === null ? $product : $product . '/' . $version; |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
/** |
|
164
|
|
|
* Get product |
|
165
|
|
|
* |
|
166
|
|
|
* @return string |
|
167
|
|
|
*/ |
|
168
|
|
|
public function getProduct() |
|
169
|
|
|
{ |
|
170
|
|
|
return $this->product; |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
/** |
|
174
|
|
|
* Get version |
|
175
|
|
|
* |
|
176
|
|
|
* @return float|int|string|null |
|
177
|
|
|
*/ |
|
178
|
|
|
public function getVersion() |
|
179
|
|
|
{ |
|
180
|
|
|
return empty($this->version) ? null : $this->version; |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
|
|
/** |
|
184
|
|
|
* Find the best matching User-agent |
|
185
|
|
|
* |
|
186
|
|
|
* @param string[] $userAgents |
|
187
|
|
|
* @return string|false |
|
188
|
|
|
*/ |
|
189
|
|
|
public function getMostSpecific(array $userAgents) |
|
190
|
|
|
{ |
|
191
|
|
|
$array = []; |
|
192
|
|
|
foreach ($userAgents as $string) { |
|
193
|
|
|
// Strip non-US-ASCII characters |
|
194
|
|
|
$array[$string] = strtolower(preg_replace(self::PREG_PATTERN, '', $string)); |
|
195
|
|
|
} |
|
196
|
|
|
foreach (array_map('strtolower', $this->getUserAgents()) as $generated) { |
|
197
|
|
|
if (($result = array_search($generated, $array)) !== false) { |
|
198
|
|
|
// Match found |
|
199
|
|
|
return $result; |
|
200
|
|
|
} |
|
201
|
|
|
} |
|
202
|
|
|
return false; |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
|
|
/** |
|
206
|
|
|
* Get an array of all possible User-agent combinations |
|
207
|
|
|
* |
|
208
|
|
|
* @return string[] |
|
209
|
|
|
*/ |
|
210
|
|
|
public function getUserAgents() |
|
211
|
|
|
{ |
|
212
|
|
|
return array_merge( |
|
213
|
|
|
preg_filter('/^/', $this->product . '/', $this->getVersions()), |
|
214
|
|
|
$this->getProducts() |
|
215
|
|
|
); |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
/** |
|
219
|
|
|
* Get versions |
|
220
|
|
|
* |
|
221
|
|
|
* @return float[]|int[]|string[] |
|
222
|
|
|
*/ |
|
223
|
|
|
public function getVersions() |
|
224
|
|
|
{ |
|
225
|
|
|
while ( |
|
226
|
|
|
!empty($this->version) && |
|
227
|
|
|
substr_count($this->version, '.') < 2 |
|
228
|
|
|
) { |
|
229
|
|
|
$this->version .= '.0'; |
|
230
|
|
|
} |
|
231
|
|
|
// Remove part by part of the version. |
|
232
|
|
|
$result = array_merge( |
|
233
|
|
|
[$this->version], |
|
234
|
|
|
$this->explode($this->version, '.') |
|
235
|
|
|
); |
|
236
|
|
|
asort($result); |
|
237
|
|
|
usort($result, function ($a, $b) { |
|
238
|
|
|
// PHP 7: Switch to the <=> "Spaceship" operator |
|
239
|
|
|
return strlen($b) - strlen($a); |
|
240
|
|
|
}); |
|
241
|
|
|
return $this->filterDuplicates($result); |
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
/** |
|
245
|
|
|
* Explode |
|
246
|
|
|
* |
|
247
|
|
|
* @param string $string |
|
248
|
|
|
* @param string $delimiter |
|
249
|
|
|
* @return string[] |
|
250
|
|
|
*/ |
|
251
|
|
|
private function explode($string, $delimiter) |
|
252
|
|
|
{ |
|
253
|
|
|
$result = []; |
|
254
|
|
|
while (($pos = strrpos($string, $delimiter)) !== false) { |
|
255
|
|
|
$result[] = ($string = substr($string, 0, $pos)); |
|
256
|
|
|
} |
|
257
|
|
|
return $result; |
|
258
|
|
|
} |
|
259
|
|
|
|
|
260
|
|
|
/** |
|
261
|
|
|
* Filter duplicates from an array |
|
262
|
|
|
* |
|
263
|
|
|
* @param string[] $array |
|
264
|
|
|
* @return string[] |
|
265
|
|
|
*/ |
|
266
|
|
|
private function filterDuplicates($array) |
|
267
|
|
|
{ |
|
268
|
|
|
$result = []; |
|
269
|
|
|
foreach ($array as $value) { |
|
270
|
|
|
if (!in_array($value, $result)) { |
|
271
|
|
|
$result[] = $value; |
|
272
|
|
|
} |
|
273
|
|
|
} |
|
274
|
|
|
return array_filter($result); |
|
275
|
|
|
} |
|
276
|
|
|
|
|
277
|
|
|
/** |
|
278
|
|
|
* Get products |
|
279
|
|
|
* |
|
280
|
|
|
* @return string[] |
|
281
|
|
|
*/ |
|
282
|
|
|
public function getProducts() |
|
283
|
|
|
{ |
|
284
|
|
|
return $this->filterDuplicates(array_merge([ |
|
285
|
|
|
$this->originProduct, |
|
286
|
|
|
$this->product, |
|
287
|
|
|
preg_replace('/[^A-Za-z0-9]/', '', $this->product), // in case of special characters |
|
288
|
|
|
], |
|
289
|
|
|
$this->explode($this->product, '-') |
|
290
|
|
|
)); |
|
291
|
|
|
} |
|
292
|
|
|
} |
|
293
|
|
|
|