29077650452

Committed 10 Jul 2026 07:45AM UTC coverage: 95.455% (-4.5%) from 100.0%

Build # 29077650452

Build Type

Pull #616

github

Committed by

web-flow

Commit Message

Merge 65957a0bf into ed842b98a

Pull Request Pull Request #616: Add signatures for Palo Alto Cortex Xpanse, Infrawatch, ModatScanner,…

Coverage Stats

42 of 44 relevant lines covered (95.45%)

14.32 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

95.35

/src/CrawlerDetect.php

<?php

/*
 * This file is part of Crawler Detect - the web crawler detection library.
 *
 * (c) Mark Beech <m@rkbee.ch>
 *
 * This source file is subject to the MIT license that is bundled
 * with this source code in the file LICENSE.
 */

namespace Jaybizzle\CrawlerDetect;

use Jaybizzle\CrawlerDetect\Fixtures\AbstractProvider;
use Jaybizzle\CrawlerDetect\Fixtures\Crawlers;
use Jaybizzle\CrawlerDetect\Fixtures\Exclusions;
use Jaybizzle\CrawlerDetect\Fixtures\Headers;

class CrawlerDetect
{
    /**
     * The user agent.
     *
     * @var string|null
     */
    protected $userAgent;

    /**
     * Headers that contain a user agent.
     *
     * @var array
     */
    protected $httpHeaders = [];

    /**
     * Store regex matches.
     *
     * @var array
     */
    protected $matches = [];

    /**
     * Crawlers object.
     *
     * @var \Jaybizzle\CrawlerDetect\Fixtures\Crawlers
     */
    protected $crawlers;

    /**
     * Exclusions object.
     *
     * @var \Jaybizzle\CrawlerDetect\Fixtures\Exclusions
     */
    protected $exclusions;

    /**
     * Headers object.
     *
     * @var \Jaybizzle\CrawlerDetect\Fixtures\Headers
     */
    protected $uaHttpHeaders;

    /**
     * The compiled regex string.
     *
     * @var string
     */
    protected $compiledRegex;

    /**
     * The compiled exclusions regex string.
     *
     * @var string
     */
    protected $compiledExclusions;

    /**
     * Cache of compiled regex strings keyed by fixture class name, shared
     * across instances so per-request `new CrawlerDetect` calls don't
     * re-implode the (~1500-entry) pattern list each time.
     *
     * @var array<string, string>
     */
    protected static $compileCache = [];

    /**
     * Class constructor.
     */
    public function __construct(?array $headers = null, $userAgent = null)
    {
        $this->crawlers = new Crawlers;
        $this->exclusions = new Exclusions;
        $this->uaHttpHeaders = new Headers;

        $this->compiledRegex = $this->compileFixtureRegex($this->crawlers);
        $this->compiledExclusions = $this->compileFixtureRegex($this->exclusions);

        $this->setHttpHeaders($headers);
        $this->setUserAgent($userAgent);
    }

    /**
     * Compile and memoize the regex for a fixture provider.
     *
     * The pattern list is a fixed class property, so the class name is a
     * sufficient cache key — cheaper than hashing the patterns themselves.
     *
     * @return string
     */
    protected function compileFixtureRegex(AbstractProvider $fixture)
    {
        $class = get_class($fixture);

        if (! isset(self::$compileCache[$class])) {
            self::$compileCache[$class] = $this->compileRegex($fixture->getAll());
        }

        return self::$compileCache[$class];
    }

    /**
     * Compile the regex patterns into one regex string.
     *
     * A non-capturing group is used because callers only need the full
     * match (preg_match's $matches[0]), not a back-reference.
     *
     * @param  array  $patterns
     * @return string
     */
    public function compileRegex($patterns)
    {
        return '(?:'.implode('|', $patterns).')';
    }

    /**
     * Set HTTP headers.
     *
     * @param  array|null  $httpHeaders
     */
    public function setHttpHeaders($httpHeaders = null)
    {
        // Use global _SERVER if $httpHeaders aren't defined.
        if (! is_array($httpHeaders) || ! count($httpHeaders)) {
            $httpHeaders = $_SERVER;
        }

        // Clear existing headers.
        $this->httpHeaders = [];

        // Only save HTTP headers. In PHP land, that means
        // only _SERVER vars that start with HTTP_.
        foreach ($httpHeaders as $key => $value) {
            if (strpos($key, 'HTTP_') === 0) {
                $this->httpHeaders[$key] = $value;
            }
        }
    }

    /**
     * Return user agent headers.
     *
     * @return array
     */
    public function getUaHttpHeaders()
    {
        return $this->uaHttpHeaders->getAll();
    }

    /**
     * Set the user agent.
     *
     * @param  string|null  $userAgent
     */
    public function setUserAgent($userAgent = null)
    {
        if (is_null($userAgent)) {
            $userAgent = '';

            foreach ($this->getUaHttpHeaders() as $altHeader) {
                if (isset($this->httpHeaders[$altHeader])) {
                    $userAgent .= $this->httpHeaders[$altHeader].' ';
                }
            }

            // If no headers were found, keep it as null.
            if ($userAgent === '') {
                $userAgent = null;
            }
        }

        return $this->userAgent = $userAgent;
    }

    /**
     * Check user agent string against the regex.
     *
     * @param  string|null  $userAgent
     * @return bool
     */
    public function isCrawler($userAgent = null)
    {
        $this->matches = [];

        $agent = preg_replace(
            "/{$this->compiledExclusions}/i",
            '',
            $userAgent ?: $this->userAgent ?: ''
        );

        if ($agent === null || trim($agent) === '') {
            return false;
        }

        $agent = trim($agent);

        $result = preg_match("/{$this->compiledRegex}/i", $agent, $this->matches);

        if ($result === false) {
            $this->matches = [];

            return false;
        }

        return (bool) $result;
    }

    /**
     * Return the matches.
     *
     * @return string|null
     */
    public function getMatches()
    {
        return isset($this->matches[0]) ? $this->matches[0] : null;
    }

    /**
     * @return string|null
     */
    public function getUserAgent()
    {
        return $this->userAgent;
    }
}

1	<?php
2
3	/*
4	* This file is part of Crawler Detect - the web crawler detection library.
5	*
6	* (c) Mark Beech <m@rkbee.ch>
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;	19✔
92	$this->exclusions = new Exclusions;	19✔
93	$this->uaHttpHeaders = new Headers;	19✔
94
95	$this->compiledRegex = $this->compileFixtureRegex($this->crawlers);	19✔
96	$this->compiledExclusions = $this->compileFixtureRegex($this->exclusions);	19✔
97
98	$this->setHttpHeaders($headers);	19✔
99	$this->setUserAgent($userAgent);	19✔
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);	19✔
113
114	if (! isset(self::$compileCache[$class])) {	19✔
115	self::$compileCache[$class] = $this->compileRegex($fixture->getAll());	1✔
116	}
117
118	return self::$compileCache[$class];	19✔
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).')';	1✔
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)) {	19✔
144	$httpHeaders = $_SERVER;	19✔
145	}
146
147	// Clear existing headers.
148	$this->httpHeaders = [];	19✔
149
150	// Only save HTTP headers. In PHP land, that means
151	// only _SERVER vars that start with HTTP_.
152	foreach ($httpHeaders as $key => $value) {	19✔
153	if (strpos($key, 'HTTP_') === 0) {	19✔
154	$this->httpHeaders[$key] = $value;	2✔
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();	19✔
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)) {	19✔
177	$userAgent = '';	19✔
178
179	foreach ($this->getUaHttpHeaders() as $altHeader) {	19✔
180	if (isset($this->httpHeaders[$altHeader])) {	19✔
181	$userAgent .= $this->httpHeaders[$altHeader].' ';	2✔
182	}
183	}
184
185	// If no headers were found, keep it as null.
186	if ($userAgent === '') {	19✔
187	$userAgent = null;	19✔
188	}
189	}
190
191	return $this->userAgent = $userAgent;	19✔
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 = [];	15✔
203
204	$agent = preg_replace(	15✔
205	"/{$this->compiledExclusions}/i",	15✔
206	'',	15✔
207	$userAgent ?: $this->userAgent ?: ''	15✔
208	);	15✔
209
210	if ($agent === null \|\| trim($agent) === '') {	15✔
211	return false;	4✔
212	}
213
214	$agent = trim($agent);	13✔
215
216	$result = preg_match("/{$this->compiledRegex}/i", $agent, $this->matches);	13✔
217
218	if ($result === false) {	13✔
219	$this->matches = [];	×
220
221	return false;	×
222	}
223
224	return (bool) $result;	13✔
225	}
226
227	/**
228	* Return the matches.
229	*
230	* @return string\|null
231	*/
232	public function getMatches()
233	{
234	return isset($this->matches[0]) ? $this->matches[0] : null;	5✔
235	}
236
237	/**
238	* @return string\|null
239	*/
240	public function getUserAgent()
241	{
242	return $this->userAgent;	2✔
243	}
244	}

JayBizzle / Crawler-Detect / 29077650452

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous