CrawlerDetect::setUserAgent() - Code Metrics - Inspection of "Implement prefetch header checking" - JayBizzle/Crawler-Detect - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#218)

unknown

created 2017-07-18 19:43 UTC

CrawlerDetect::setUserAgent() A

↳ Parent: Project

Complexity

Conditions	4
Paths	2

Size

Total Lines	12
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	12
rs	9.2
c	0
b	0
f	0
cc	4
eloc	6
nc	2
nop	1

<?php

/*
 * This file is part of Crawler Detect - the web crawler detection library.
 *
 * (c) Mark Beech <[email protected]>
 *
 * 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\Headers;
use Jaybizzle\CrawlerDetect\Fixtures\Crawlers;
use Jaybizzle\CrawlerDetect\Fixtures\Exclusions;

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

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

    /**
     * Store regex matches.
     *
     * @var array
     */
    protected $matches = array();

    /**
     * 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;

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

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

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

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


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

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

    /**
     * Compile the regex patterns into one regex string.
     *
     * @param array
     *
     * @return string
     */
    public function compileRegex($patterns)
    {
        return '('.implode('|', $patterns).')';
    }

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

        // Clear existing headers.
        $this->httpHeaders = array();

        // 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 $userAgent
     */
    public function setUserAgent($userAgent)
    {
        if (is_null($userAgent)) {
            foreach ($this->getUaHttpHeaders() as $altHeader) {
                if (isset($this->httpHeaders[$altHeader])) {
                    $userAgent .= $this->httpHeaders[$altHeader].' ';
                }
            }
        }

        return $userAgent;
    }

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

        $agent = $userAgent ?: $this->userAgent;

        $agent = preg_replace('/'.$this->compiledExclusions.'/i', '', $agent);

        if (strlen(trim($agent)) == 0) {
            return false;
        }

        $result = preg_match('/'.$this->compiledRegex.'/i', trim($agent), $matches);

        if ($matches) {
            $this->matches = $matches;
        }

        return (bool) $result;
    }

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

    /**
     * Return true if headers contain a prefetch setting
     *
     * @return null
     */
    public function hasPrefetchHeaderSet()
    {
        foreach ($this->prefetchHttpHeaders as $header => $value) {
            if (isset($this->uaHttpHeaders[$header]) && $this->uaHttpHeaders[$header] == $value) {
                return [$header => $value];
            }
        }

        return false;
    }
}


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\Headers;
15			use Jaybizzle\CrawlerDetect\Fixtures\Crawlers;
16			use Jaybizzle\CrawlerDetect\Fixtures\Exclusions;
17
18			class CrawlerDetect
19			{
20			/**
21			* The user agent.
22			*
23			* @var null
24			*/
25			protected $userAgent = null;
26
27			/**
28			* Headers that contain a user agent.
29			*
30			* @var array
31			*/
32			protected $httpHeaders = array();
33
34			/**
35			* Store regex matches.
36			*
37			* @var array
38			*/
39			protected $matches = array();
40
41			/**
42			* Crawlers object.
43			*
44			* @var \Jaybizzle\CrawlerDetect\Fixtures\Crawlers
45			*/
46			protected $crawlers;
47
48			/**
49			* Exclusions object.
50			*
51			* @var \Jaybizzle\CrawlerDetect\Fixtures\Exclusions
52			*/
53			protected $exclusions;
54
55			/**
56			* Headers object.
57			*
58			* @var \Jaybizzle\CrawlerDetect\Fixtures\Headers
59			*/
60			protected $uaHttpHeaders;
61
62			/**
63			* Prefetch Headers object.
64			*
65			* @var \Jaybizzle\CrawlerDetect\Fixtures\Headers
66			*/
67			protected $prefetchHttpHeaders;
68
69			/**
70			* The compiled regex string.
71			*
72			* @var string
73			*/
74			protected $compiledRegex;
75
76			/**
77			* The compiled exclusions regex string.
78			*
79			* @var string
80			*/
81			protected $compiledExclusions;
82
83			/**
84			* Class constructor.
85			*/
86			public function __construct(array $headers = null, $userAgent = null)
87			{
88			$this->crawlers = new Crawlers();
89			$this->exclusions = new Exclusions();
90			$this->uaHttpHeaders = new Headers();
91			$this->prefetchHttpHeaders = new Headers()->getPrefetchHeaders();
			0 ignored issues – show Bug introduced 2017-07-18 19:45 UTC by Report Bug Copy Issue Report This code did not parse for me. Apparently, there is an error somewhere around this line: Syntax error, unexpected T_OBJECT_OPERATOR Loading history...
92
93			$this->compiledRegex = $this->compileRegex($this->crawlers->getAll());
94			$this->compiledExclusions = $this->compileRegex($this->exclusions->getAll());
95
96			$this->setHttpHeaders($headers);
97			$this->userAgent = $this->setUserAgent($userAgent);
98			}
99
100			/**
101			* Compile the regex patterns into one regex string.
102			*
103			* @param array
104			*
105			* @return string
106			*/
107			public function compileRegex($patterns)
108			{
109			return '('.implode('\|', $patterns).')';
110			}
111
112			/**
113			* Set HTTP headers.
114			*
115			* @param array\|null $httpHeaders
116			*/
117			public function setHttpHeaders($httpHeaders)
118			{
119			// Use global _SERVER if $httpHeaders aren't defined.
120			if (! is_array($httpHeaders) \|\| ! count($httpHeaders)) {
121			$httpHeaders = $_SERVER;
122			}
123
124			// Clear existing headers.
125			$this->httpHeaders = array();
126
127			// Only save HTTP headers. In PHP land, that means
128			// only _SERVER vars that start with HTTP_.
129			foreach ($httpHeaders as $key => $value) {
130			if (strpos($key, 'HTTP_') === 0) {
131			$this->httpHeaders[$key] = $value;
132			}
133			}
134			}
135
136			/**
137			* Return user agent headers.
138			*
139			* @return array
140			*/
141			public function getUaHttpHeaders()
142			{
143			return $this->uaHttpHeaders->getAll();
144			}
145
146			/**
147			* Set the user agent.
148			*
149			* @param string $userAgent
150			*/
151			public function setUserAgent($userAgent)
152			{
153			if (is_null($userAgent)) {
154			foreach ($this->getUaHttpHeaders() as $altHeader) {
155			if (isset($this->httpHeaders[$altHeader])) {
156			$userAgent .= $this->httpHeaders[$altHeader].' ';
157			}
158			}
159			}
160
161			return $userAgent;
162			}
163
164			/**
165			* Check user agent string against the regex.
166			*
167			* @param string\|null $userAgent
168			*
169			* @return bool
170			*/
171			public function isCrawler($userAgent = null)
172			{
173			$prefetchHeaders = $this->hasPrefetchHeaderSet;
174			if ($prefetchHeaders) {
175			$this->matches = $prefetchHeaders;
176			return true;
177			}
178
179			$agent = $userAgent ?: $this->userAgent;
180
181			$agent = preg_replace('/'.$this->compiledExclusions.'/i', '', $agent);
182
183			if (strlen(trim($agent)) == 0) {
184			return false;
185			}
186
187			$result = preg_match('/'.$this->compiledRegex.'/i', trim($agent), $matches);
188
189			if ($matches) {
190			$this->matches = $matches;
191			}
192
193			return (bool) $result;
194			}
195
196			/**
197			* Return the matches.
198			*
199			* @return string\|null
200			*/
201			public function getMatches()
202			{
203			return isset($this->matches[0]) ? $this->matches[0] : null;
204			}
205
206			/**
207			* Return true if headers contain a prefetch setting
208			*
209			* @return null
210			*/
211			public function hasPrefetchHeaderSet()
212			{
213			foreach ($this->prefetchHttpHeaders as $header => $value) {
214			if (isset($this->uaHttpHeaders[$header]) && $this->uaHttpHeaders[$header] == $value) {
215			return [$header => $value];
216			}
217			}
218
219			return false;
220			}
221			}
222

JayBizzle / Crawler-Detect

Pull Request — master (#218)

CrawlerDetect::setUserAgent() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like