UserAgentClient::getRules() - Code Metrics - Inspection of "Merge branch '2.0-dev' of https://github.com/VIPny..." - VIPnytt/RobotsTxtParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Branch — 2.0-dev (4f313a)

by Jan-Petter

created 2016-06-03 21:10 UTC

UserAgentClient::getRules() A

↳ Parent: UserAgentClient

Complexity

Conditions	1
Paths	1

Size

Total Lines	13
Code Lines	10

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	13
rs	9.4285
cc	1
eloc	10
nc	1
nop	0

<?php
namespace vipnytt\RobotsTxtParser\Client\Directives;

use vipnytt\RobotsTxtParser\Exceptions\ClientException;
use vipnytt\RobotsTxtParser\Parser\Directives\SubDirectiveHandler;
use vipnytt\RobotsTxtParser\Parser\StatusCodeParser;
use vipnytt\RobotsTxtParser\Parser\UrlParser;
use vipnytt\RobotsTxtParser\RobotsTxtInterface;

/**
 * Class UserAgentClient
 *
 * @package vipnytt\RobotsTxtParser\Client\Directives
 */
class UserAgentClient implements RobotsTxtInterface
{
    use UrlParser;

    /**
     * Rules
     * @var SubDirectiveHandler
     */
    private $handler;

    /**
     * Base Uri
     * @var string
     */
    private $base;

    /**
     * Status code
     * @var int|null
     */
    private $statusCode;

    /**
     * UserAgentClient constructor.
     *
     * @param SubDirectiveHandler $handler
     * @param string $baseUri
     * @param int|null $statusCode
     */
    public function __construct(SubDirectiveHandler $handler, $baseUri, $statusCode)
    {
        $this->handler = $handler;
        $this->base = $baseUri;
        $this->statusCode = $statusCode;
    }

    /**
     * UserAgentClient destructor.
     */
    public function __destruct()
    {
        $this->comment();
    }

    /**
     * Comment
     *
     * @return CommentClient
     */
    public function comment()
    {
        return $this->handler->comment()->client();
    }

    /**
     * Check if URL is allowed to crawl
     *
     * @param string $url
     * @return bool
     */
    public function isAllowed($url)
    {
        return $this->check(self::DIRECTIVE_ALLOW, $url);
    }

    /**
     * Check
     *
     * @param string $directive
     * @param string $url - URL to check
     * @return bool
     * @throws ClientException
     */
    private function check($directive, $url)
    {
        $url = $this->urlConvertToFull($url, $this->base);
        if (!$this->isUrlApplicable([$url, $this->base])) {
            throw new ClientException('URL belongs to a different robots.txt');
        }
        $statusCodeParser = new StatusCodeParser($this->statusCode, parse_url($this->base, PHP_URL_SCHEME));
        $statusCodeParser->replaceUnofficial();
        if (($result = $statusCodeParser->check()) !== null) {
            return $directive === $result;
        }
        $result = self::DIRECTIVE_ALLOW;
        foreach (
            [
                self::DIRECTIVE_DISALLOW => $this->handler->disallow(),
                self::DIRECTIVE_ALLOW => $this->handler->allow()
            ] as $currentDirective => $currentRules
        ) {
            if ($currentRules->check($url)) {
                $result = $currentDirective;
            }
        }
        return $directive === $result;
    }

    /**
     * Check if the URL belongs to current robots.txt
     *
     * @param string[] $urls
     * @return bool
     */
    private function isUrlApplicable($urls)
    {
        foreach ($urls as $url) {
            $parsed = parse_url($url);
            $parsed['port'] = is_int($port = parse_url($url, PHP_URL_PORT)) ? $port : getservbyname($parsed['scheme'], 'tcp');
            $assembled = $parsed['scheme'] . '://' . $parsed['host'] . ':' . $parsed['port'];
            if (!isset($result)) {
                $result = $assembled;
            } elseif ($result !== $assembled) {
                return false;
            }
        }
        return true;
    }

    /**
     * Check if URL is disallowed to crawl
     *
     * @param string $url
     * @return bool
     */
    public function isDisallowed($url)
    {
        return $this->check(self::DIRECTIVE_DISALLOW, $url);
    }

    /**
     * Cache-delay
     *
     * @return DelayClient
     */
    public function cacheDelay()
    {
        return $this->handler->cacheDelay()->client($this->crawlDelay()->get());
    }

    /**
     * Crawl-delay
     *
     * @return DelayClient
     */
    public function crawlDelay()
    {
        return $this->handler->crawlDelay()->client($this->requestRate()->get());
    }

    /**
     * RequestClient-rate
     *
     * @return RequestRateClient
     */
    public function requestRate()
    {
        return $this->handler->requestRate()->client();
    }

    /**
     * Robot-version
     *
     * @return RobotVersionClient
     */
    public function robotVersion()
    {
        return $this->handler->robotVersion()->client();
    }

    /**
     * Rule export
     *
     * @return array
     */
    public function getRules()
    {
        return array_merge(
            $this->handler->allow()->getRules(),
            $this->handler->comment()->getRules(),
            $this->handler->cacheDelay()->getRules(),
            $this->handler->crawlDelay()->getRules(),
            $this->handler->disallow()->getRules(),
            $this->handler->requestRate()->getRules(),
            $this->handler->robotVersion()->getRules(),
            $this->handler->visitTime()->getRules()
        );
    }

    /**
     * Visit-time
     *
     * @return VisitTimeClient
     */
    public function visitTime()
    {
        return $this->handler->visitTime()->client();
    }
}


1			<?php
2			namespace vipnytt\RobotsTxtParser\Client\Directives;
3
4			use vipnytt\RobotsTxtParser\Exceptions\ClientException;
5			use vipnytt\RobotsTxtParser\Parser\Directives\SubDirectiveHandler;
6			use vipnytt\RobotsTxtParser\Parser\StatusCodeParser;
7			use vipnytt\RobotsTxtParser\Parser\UrlParser;
8			use vipnytt\RobotsTxtParser\RobotsTxtInterface;
9
10			/**
11			* Class UserAgentClient
12			*
13			* @package vipnytt\RobotsTxtParser\Client\Directives
14			*/
15			class UserAgentClient implements RobotsTxtInterface
16			{
17			use UrlParser;
18
19			/**
20			* Rules
21			* @var SubDirectiveHandler
22			*/
23			private $handler;
24
25			/**
26			* Base Uri
27			* @var string
28			*/
29			private $base;
30
31			/**
32			* Status code
33			* @var int\|null
34			*/
35			private $statusCode;
36
37			/**
38			* UserAgentClient constructor.
39			*
40			* @param SubDirectiveHandler $handler
41			* @param string $baseUri
42			* @param int\|null $statusCode
43			*/
44			public function __construct(SubDirectiveHandler $handler, $baseUri, $statusCode)
45			{
46			$this->handler = $handler;
47			$this->base = $baseUri;
48			$this->statusCode = $statusCode;
49			}
50
51			/**
52			* UserAgentClient destructor.
53			*/
54			public function __destruct()
55			{
56			$this->comment();
57			}
58
59			/**
60			* Comment
61			*
62			* @return CommentClient
63			*/
64			public function comment()
65			{
66			return $this->handler->comment()->client();
67			}
68
69			/**
70			* Check if URL is allowed to crawl
71			*
72			* @param string $url
73			* @return bool
74			*/
75			public function isAllowed($url)
76			{
77			return $this->check(self::DIRECTIVE_ALLOW, $url);
78			}
79
80			/**
81			* Check
82			*
83			* @param string $directive
84			* @param string $url - URL to check
85			* @return bool
86			* @throws ClientException
87			*/
88			private function check($directive, $url)
89			{
90			$url = $this->urlConvertToFull($url, $this->base);
91			if (!$this->isUrlApplicable([$url, $this->base])) {
92			throw new ClientException('URL belongs to a different robots.txt');
93			}
94			$statusCodeParser = new StatusCodeParser($this->statusCode, parse_url($this->base, PHP_URL_SCHEME));
95			$statusCodeParser->replaceUnofficial();
96			if (($result = $statusCodeParser->check()) !== null) {
97			return $directive === $result;
98			}
99			$result = self::DIRECTIVE_ALLOW;
100			foreach (
101			[
102			self::DIRECTIVE_DISALLOW => $this->handler->disallow(),
103			self::DIRECTIVE_ALLOW => $this->handler->allow()
104			] as $currentDirective => $currentRules
105			) {
106			if ($currentRules->check($url)) {
107			$result = $currentDirective;
108			}
109			}
110			return $directive === $result;
111			}
112
113			/**
114			* Check if the URL belongs to current robots.txt
115			*
116			* @param string[] $urls
117			* @return bool
118			*/
119			private function isUrlApplicable($urls)
120			{
121			foreach ($urls as $url) {
122			$parsed = parse_url($url);
123			$parsed['port'] = is_int($port = parse_url($url, PHP_URL_PORT)) ? $port : getservbyname($parsed['scheme'], 'tcp');
124			$assembled = $parsed['scheme'] . '://' . $parsed['host'] . ':' . $parsed['port'];
125			if (!isset($result)) {
126			$result = $assembled;
127			} elseif ($result !== $assembled) {
128			return false;
129			}
130			}
131			return true;
132			}
133
134			/**
135			* Check if URL is disallowed to crawl
136			*
137			* @param string $url
138			* @return bool
139			*/
140			public function isDisallowed($url)
141			{
142			return $this->check(self::DIRECTIVE_DISALLOW, $url);
143			}
144
145			/**
146			* Cache-delay
147			*
148			* @return DelayClient
149			*/
150			public function cacheDelay()
151			{
152			return $this->handler->cacheDelay()->client($this->crawlDelay()->get());
153			}
154
155			/**
156			* Crawl-delay
157			*
158			* @return DelayClient
159			*/
160			public function crawlDelay()
161			{
162			return $this->handler->crawlDelay()->client($this->requestRate()->get());
163			}
164
165			/**
166			* RequestClient-rate
167			*
168			* @return RequestRateClient
169			*/
170			public function requestRate()
171			{
172			return $this->handler->requestRate()->client();
173			}
174
175			/**
176			* Robot-version
177			*
178			* @return RobotVersionClient
179			*/
180			public function robotVersion()
181			{
182			return $this->handler->robotVersion()->client();
183			}
184
185			/**
186			* Rule export
187			*
188			* @return array
189			*/
190			public function getRules()
191			{
192			return array_merge(
193			$this->handler->allow()->getRules(),
194			$this->handler->comment()->getRules(),
195			$this->handler->cacheDelay()->getRules(),
196			$this->handler->crawlDelay()->getRules(),
197			$this->handler->disallow()->getRules(),
198			$this->handler->requestRate()->getRules(),
199			$this->handler->robotVersion()->getRules(),
200			$this->handler->visitTime()->getRules()
201			);
202			}
203
204			/**
205			* Visit-time
206			*
207			* @return VisitTimeClient
208			*/
209			public function visitTime()
210			{
211			return $this->handler->visitTime()->client();
212			}
213			}
214

VIPnytt / RobotsTxtParser

Branch — 2.0-dev (4f313a)

UserAgentClient::getRules() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like