top-10 / vendor / jaybizzle / crawler-detect / src / CrawlerDetect.php

CrawlerDetect.php in WebberZone Top 10 — Popular Posts 4.4.1, at vendor/jaybizzle/crawler-detect/src/CrawlerDetect.php

245 lines 5.6 KB

Raw Download Zip

1	<?php
2
3	/*
4	* This file is part of Crawler Detect - the web crawler detection library.
5	*
6	* (c) Mark Beech <[email protected]>
7	*
8	* This source file is subject to the MIT license that is bundled
9	* with this source code in the file LICENSE.
10	*/
11
12	namespace Jaybizzle\CrawlerDetect;
13
14	use Jaybizzle\CrawlerDetect\Fixtures\AbstractProvider;
15	use Jaybizzle\CrawlerDetect\Fixtures\Crawlers;
16	use Jaybizzle\CrawlerDetect\Fixtures\Exclusions;
17	use Jaybizzle\CrawlerDetect\Fixtures\Headers;
18
19	class CrawlerDetect
20	{
21	/**
22	* The user agent.
23	*
24	* @var string\|null
25	*/
26	protected $userAgent;
27
28	/**
29	* Headers that contain a user agent.
30	*
31	* @var array
32	*/
33	protected $httpHeaders = [];
34
35	/**
36	* Store regex matches.
37	*
38	* @var array
39	*/
40	protected $matches = [];
41
42	/**
43	* Crawlers object.
44	*
45	* @var \Jaybizzle\CrawlerDetect\Fixtures\Crawlers
46	*/
47	protected $crawlers;
48
49	/**
50	* Exclusions object.
51	*
52	* @var \Jaybizzle\CrawlerDetect\Fixtures\Exclusions
53	*/
54	protected $exclusions;
55
56	/**
57	* Headers object.
58	*
59	* @var \Jaybizzle\CrawlerDetect\Fixtures\Headers
60	*/
61	protected $uaHttpHeaders;
62
63	/**
64	* The compiled regex string.
65	*
66	* @var string
67	*/
68	protected $compiledRegex;
69
70	/**
71	* The compiled exclusions regex string.
72	*
73	* @var string
74	*/
75	protected $compiledExclusions;
76
77	/**
78	* Cache of compiled regex strings keyed by fixture class name, shared
79	* across instances so per-request `new CrawlerDetect` calls don't
80	* re-implode the (~1500-entry) pattern list each time.
81	*
82	* @var array<string, string>
83	*/
84	protected static $compileCache = [];
85
86	/**
87	* Class constructor.
88	*/
89	public function __construct(?array $headers = null, $userAgent = null)
90	{
91	$this->crawlers = new Crawlers;
92	$this->exclusions = new Exclusions;
93	$this->uaHttpHeaders = new Headers;
94
95	$this->compiledRegex = $this->compileFixtureRegex($this->crawlers);
96	$this->compiledExclusions = $this->compileFixtureRegex($this->exclusions);
97
98	$this->setHttpHeaders($headers);
99	$this->setUserAgent($userAgent);
100	}
101
102	/**
103	* Compile and memoize the regex for a fixture provider.
104	*
105	* The pattern list is a fixed class property, so the class name is a
106	* sufficient cache key — cheaper than hashing the patterns themselves.
107	*
108	* @return string
109	*/
110	protected function compileFixtureRegex(AbstractProvider $fixture)
111	{
112	$class = get_class($fixture);
113
114	if (! isset(self::$compileCache[$class])) {
115	self::$compileCache[$class] = $this->compileRegex($fixture->getAll());
116	}
117
118	return self::$compileCache[$class];
119	}
120
121	/**
122	* Compile the regex patterns into one regex string.
123	*
124	* A non-capturing group is used because callers only need the full
125	* match (preg_match's $matches[0]), not a back-reference.
126	*
127	* @param array $patterns
128	* @return string
129	*/
130	public function compileRegex($patterns)
131	{
132	return '(?:'.implode('\|', $patterns).')';
133	}
134
135	/**
136	* Set HTTP headers.
137	*
138	* @param array\|null $httpHeaders
139	*/
140	public function setHttpHeaders($httpHeaders = null)
141	{
142	// Use global _SERVER if $httpHeaders aren't defined.
143	if (! is_array($httpHeaders) \|\| ! count($httpHeaders)) {
144	$httpHeaders = $_SERVER;
145	}
146
147	// Clear existing headers.
148	$this->httpHeaders = [];
149
150	// Only save HTTP headers. In PHP land, that means
151	// only _SERVER vars that start with HTTP_.
152	foreach ($httpHeaders as $key => $value) {
153	if (strpos($key, 'HTTP_') === 0) {
154	$this->httpHeaders[$key] = $value;
155	}
156	}
157	}
158
159	/**
160	* Return user agent headers.
161	*
162	* @return array
163	*/
164	public function getUaHttpHeaders()
165	{
166	return $this->uaHttpHeaders->getAll();
167	}
168
169	/**
170	* Set the user agent.
171	*
172	* @param string\|null $userAgent
173	*/
174	public function setUserAgent($userAgent = null)
175	{
176	if (is_null($userAgent)) {
177	$userAgent = '';
178
179	foreach ($this->getUaHttpHeaders() as $altHeader) {
180	if (isset($this->httpHeaders[$altHeader])) {
181	$userAgent .= $this->httpHeaders[$altHeader].' ';
182	}
183	}
184
185	// If no headers were found, keep it as null.
186	if ($userAgent === '') {
187	$userAgent = null;
188	}
189	}
190
191	return $this->userAgent = $userAgent;
192	}
193
194	/**
195	* Check user agent string against the regex.
196	*
197	* @param string\|null $userAgent
198	* @return bool
199	*/
200	public function isCrawler($userAgent = null)
201	{
202	$this->matches = [];
203
204	$agent = preg_replace(
205	"/{$this->compiledExclusions}/i",
206	'',
207	$userAgent ?: $this->userAgent ?: ''
208	);
209
210	if ($agent === null \|\| trim($agent) === '') {
211	return false;
212	}
213
214	$agent = trim($agent);
215
216	$result = preg_match("/{$this->compiledRegex}/i", $agent, $this->matches);
217
218	if ($result === false) {
219	$this->matches = [];
220
221	return false;
222	}
223
224	return (bool) $result;
225	}
226
227	/**
228	* Return the matches.
229	*
230	* @return string\|null
231	*/
232	public function getMatches()
233	{
234	return isset($this->matches[0]) ? $this->matches[0] : null;
235	}
236
237	/**
238	* @return string\|null
239	*/
240	public function getUserAgent()
241	{
242	return $this->userAgent;
243	}
244	}
245