UserAgentClient::determineRequestRates() - Code Metrics - Inspection of "Refactoring + readme" - VIPnytt/RobotsTxtParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( e5e536...a6d8dc )

by Jan-Petter

created 2016-05-14 23:06 UTC

UserAgentClient::determineRequestRates() C

↳ Parent: UserAgentClient

Complexity

Conditions	7
Paths	6

Size

Total Lines	26
Code Lines	18

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	26
rs	6.7272
cc	7
eloc	18
nc	6
nop	1

<?php
namespace vipnytt\RobotsTxtParser\Client;

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

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

    /**
     * Rules
     * @var array
     */
    protected $rules;

    /**
     * User-agent
     * @var string
     */
    protected $userAgent;

    /**
     * Robots.txt base URL
     * @var string
     */
    protected $base;

    /**
     * Status code parser
     * @var StatusCodeParser
     */
    protected $statusCodeParser;

    /**
     * Comment export status
     * @var bool
     */
    protected $commentsExported = false;

    /**
     * UserAgentClient constructor.
     *
     * @param array $rules
     * @param string $userAgent
     * @param string $baseUrl
     * @param int|null $statusCode
     */
    public function __construct(array $rules, $userAgent, $baseUrl, $statusCode)
    {
        $this->statusCodeParser = new StatusCodeParser($statusCode, parse_url($baseUrl, PHP_URL_SCHEME));
        $this->userAgent = $userAgent;
        $this->rules = $rules;
        $this->base = $baseUrl;
    }

    /**
     * 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 $url - URL to check
     * @return bool
     * @throws ClientException
     */
    protected function check($directive, $url)
    {
        $url = $this->urlConvertToFull($url, $this->base);
        if (!$this->isUrlApplicable([$url, $this->base])) {
            throw new ClientException('URL belongs to a different robots.txt, please check it against that one instead');
        }
        $this->statusCodeParser->replaceUnofficial();
        if (($result = $this->statusCodeParser->check()) !== null) {
            return $directive === $result;
        }
        $result = self::DIRECTIVE_ALLOW;
        foreach ([self::DIRECTIVE_DISALLOW, self::DIRECTIVE_ALLOW] as $currentDirective) {
            if ($this->rules[$currentDirective]->check($url)) {
                $result = $currentDirective;
            }
        }
        return $directive === $result;
    }

    /**
     * Check if the URL belongs to current robots.txt
     *
     * @param $urls
     * @return bool
     */
    protected function isUrlApplicable($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);
    }

    /**
     * Get Cache-delay
     *
     * @return float|int
     */
    public function getCacheDelay()
    {
        $delay = $this->rules[self::DIRECTIVE_CACHE_DELAY]->export();
        return isset($delay[self::DIRECTIVE_CACHE_DELAY]) ? $delay[self::DIRECTIVE_CACHE_DELAY] : $this->getCrawlDelay();
    }

    /**
     * Get Crawl-delay
     *
     * @return float|int
     */
    public function getCrawlDelay()
    {
        $delay = $this->rules[self::DIRECTIVE_CRAWL_DELAY]->export();
        return isset($delay[self::DIRECTIVE_CRAWL_DELAY]) ? $delay[self::DIRECTIVE_CRAWL_DELAY] : $this->getRequestRate();
    }

    /**
     * Get Request-rate for current timestamp
     *
     * @param int|null $timestamp
     * @return float|int
     */
    protected function getRequestRate($timestamp = null)
    {
        if (!is_int($timestamp)) {
            $timestamp = time();
        }
        $values = $this->determineRequestRates($timestamp);
        if (
            count($values) > 0 &&
            ($rate = min($values)) > 0
        ) {
            return $rate;
        }
        return 0;
    }

    /**
     * Determine Request rates
     *
     * @param $timestamp
     * @return array
     */
    protected function determineRequestRates($timestamp)
    {
        $rates = $this->getRequestRates();
        $values = [];
        foreach ($rates as $array) {
            if (
                !isset($array['from']) ||
                !isset($array['to'])
            ) {
                $values[] = $array['rate'];
                continue;
            }
            $fromTime = gmmktime(mb_substr($array['from'], 0, mb_strlen($array['from']) - 2), mb_substr($array['from'], -2, 2), 0);
            $toTime = gmmktime(mb_substr($array['to'], 0, mb_strlen($array['to']) - 2), mb_substr($array['to'], -2, 2), 59);
            if ($fromTime > $toTime) {
                $toTime = gmmktime(mb_substr($array['to'] + 24, 0, mb_strlen($array['to']) - 2), mb_substr($array['to'], -2, 2), 59);
            }
            if (
                $timestamp >= $fromTime &&
                $timestamp <= $toTime
            ) {
                $values[] = $array['rate'];
            }
        }
        return $values;
    }

    /**
     * Get Request-rates
     *
     * @return array
     */
    public function getRequestRates()
    {
        $array = $this->rules[self::DIRECTIVE_REQUEST_RATE]->export();
        return isset($array[self::DIRECTIVE_REQUEST_RATE]) ? $array[self::DIRECTIVE_REQUEST_RATE] : [];
    }

    /**
     * Rule export
     *
     * @return array
     */
    public function export()
    {
        $result = [];
        foreach ($this->rules as $directive => $object) {
            if (!empty($export = $object->export())) {
                $result[$directive] = $export[$directive];
            }
        }
        return $result;
    }

    /**
     * UserAgentClient destructor.
     */
    public function __destruct()
    {
        if (!$this->commentsExported) {
            // Comment from the `Comments` directive exists, but has not been exported.
            foreach ($this->getComments() as $message) {
                trigger_error('Comment for `' . $this->userAgent . '` at `' . $this->base . '/robots.txt`: ' . $message, E_USER_NOTICE);
            }
        }
    }

    /**
     * Get Comments
     *
     * @return array
     */
    public function getComments()
    {
        $this->commentsExported = true;
        $comments = $this->rules[self::DIRECTIVE_COMMENT]->export();
        return isset($comments[self::DIRECTIVE_COMMENT]) ? $comments[self::DIRECTIVE_COMMENT] : [];
    }

    /**
     * Get Visit-time
     *
     * @return array|false
     */
    public function getVisitTime()
    {
        $times = $this->rules[self::DIRECTIVE_VISIT_TIME]->export();
        return isset($times[self::DIRECTIVE_VISIT_TIME]) ? $times[self::DIRECTIVE_VISIT_TIME] : [];
    }
}


1			<?php
2			namespace vipnytt\RobotsTxtParser\Client;
3
4			use vipnytt\RobotsTxtParser\Exceptions\ClientException;
5			use vipnytt\RobotsTxtParser\Parser\RobotsTxtInterface;
6			use vipnytt\RobotsTxtParser\Parser\StatusCodeParser;
7			use vipnytt\RobotsTxtParser\Parser\UrlParser;
8
9			/**
10			* Class UserAgentClient
11			*
12			* @package vipnytt\RobotsTxtParser\Client
13			*/
14			class UserAgentClient implements RobotsTxtInterface
15			{
16			use UrlParser;
17
18			/**
19			* Rules
20			* @var array
21			*/
22			protected $rules;
23
24			/**
25			* User-agent
26			* @var string
27			*/
28			protected $userAgent;
29
30			/**
31			* Robots.txt base URL
32			* @var string
33			*/
34			protected $base;
35
36			/**
37			* Status code parser
38			* @var StatusCodeParser
39			*/
40			protected $statusCodeParser;
41
42			/**
43			* Comment export status
44			* @var bool
45			*/
46			protected $commentsExported = false;
47
48			/**
49			* UserAgentClient constructor.
50			*
51			* @param array $rules
52			* @param string $userAgent
53			* @param string $baseUrl
54			* @param int\|null $statusCode
55			*/
56			public function __construct(array $rules, $userAgent, $baseUrl, $statusCode)
57			{
58			$this->statusCodeParser = new StatusCodeParser($statusCode, parse_url($baseUrl, PHP_URL_SCHEME));
59			$this->userAgent = $userAgent;
60			$this->rules = $rules;
61			$this->base = $baseUrl;
62			}
63
64			/**
65			* Check if URL is allowed to crawl
66			*
67			* @param string $url
68			* @return bool
69			*/
70			public function isAllowed($url)
71			{
72			return $this->check(self::DIRECTIVE_ALLOW, $url);
73			}
74
75			/**
76			* Check
77			*
78			* @param string $directive
79			* @param string $url - URL to check
80			* @return bool
81			* @throws ClientException
82			*/
83			protected function check($directive, $url)
84			{
85			$url = $this->urlConvertToFull($url, $this->base);
86			if (!$this->isUrlApplicable([$url, $this->base])) {
87			throw new ClientException('URL belongs to a different robots.txt, please check it against that one instead');
88			}
89			$this->statusCodeParser->replaceUnofficial();
90			if (($result = $this->statusCodeParser->check()) !== null) {
91			return $directive === $result;
92			}
93			$result = self::DIRECTIVE_ALLOW;
94			foreach ([self::DIRECTIVE_DISALLOW, self::DIRECTIVE_ALLOW] as $currentDirective) {
95			if ($this->rules[$currentDirective]->check($url)) {
96			$result = $currentDirective;
97			}
98			}
99			return $directive === $result;
100			}
101
102			/**
103			* Check if the URL belongs to current robots.txt
104			*
105			* @param $urls
106			* @return bool
107			*/
108			protected function isUrlApplicable($urls)
109			{
110			foreach ($urls as $url) {
111			$parsed = parse_url($url);
112			$parsed['port'] = is_int($port = parse_url($url, PHP_URL_PORT)) ? $port : getservbyname($parsed['scheme'], 'tcp');
113			$assembled = $parsed['scheme'] . '://' . $parsed['host'] . ':' . $parsed['port'];
114			if (!isset($result)) {
115			$result = $assembled;
116			} elseif ($result !== $assembled) {
117			return false;
118			}
119			}
120			return true;
121			}
122
123			/**
124			* Check if URL is disallowed to crawl
125			*
126			* @param string $url
127			* @return bool
128			*/
129			public function isDisallowed($url)
130			{
131			return $this->check(self::DIRECTIVE_DISALLOW, $url);
132			}
133
134			/**
135			* Get Cache-delay
136			*
137			* @return float\|int
138			*/
139			public function getCacheDelay()
140			{
141			$delay = $this->rules[self::DIRECTIVE_CACHE_DELAY]->export();
142			return isset($delay[self::DIRECTIVE_CACHE_DELAY]) ? $delay[self::DIRECTIVE_CACHE_DELAY] : $this->getCrawlDelay();
143			}
144
145			/**
146			* Get Crawl-delay
147			*
148			* @return float\|int
149			*/
150			public function getCrawlDelay()
151			{
152			$delay = $this->rules[self::DIRECTIVE_CRAWL_DELAY]->export();
153			return isset($delay[self::DIRECTIVE_CRAWL_DELAY]) ? $delay[self::DIRECTIVE_CRAWL_DELAY] : $this->getRequestRate();
154			}
155
156			/**
157			* Get Request-rate for current timestamp
158			*
159			* @param int\|null $timestamp
160			* @return float\|int
161			*/
162			protected function getRequestRate($timestamp = null)
163			{
164			if (!is_int($timestamp)) {
165			$timestamp = time();
166			}
167			$values = $this->determineRequestRates($timestamp);
168			if (
169			count($values) > 0 &&
170			($rate = min($values)) > 0
171			) {
172			return $rate;
173			}
174			return 0;
175			}
176
177			/**
178			* Determine Request rates
179			*
180			* @param $timestamp
181			* @return array
182			*/
183			protected function determineRequestRates($timestamp)
184			{
185			$rates = $this->getRequestRates();
186			$values = [];
187			foreach ($rates as $array) {
188			if (
189			!isset($array['from']) \|\|
190			!isset($array['to'])
191			) {
192			$values[] = $array['rate'];
193			continue;
194			}
195			$fromTime = gmmktime(mb_substr($array['from'], 0, mb_strlen($array['from']) - 2), mb_substr($array['from'], -2, 2), 0);
196			$toTime = gmmktime(mb_substr($array['to'], 0, mb_strlen($array['to']) - 2), mb_substr($array['to'], -2, 2), 59);
197			if ($fromTime > $toTime) {
198			$toTime = gmmktime(mb_substr($array['to'] + 24, 0, mb_strlen($array['to']) - 2), mb_substr($array['to'], -2, 2), 59);
199			}
200			if (
201			$timestamp >= $fromTime &&
202			$timestamp <= $toTime
203			) {
204			$values[] = $array['rate'];
205			}
206			}
207			return $values;
208			}
209
210			/**
211			* Get Request-rates
212			*
213			* @return array
214			*/
215			public function getRequestRates()
216			{
217			$array = $this->rules[self::DIRECTIVE_REQUEST_RATE]->export();
218			return isset($array[self::DIRECTIVE_REQUEST_RATE]) ? $array[self::DIRECTIVE_REQUEST_RATE] : [];
219			}
220
221			/**
222			* Rule export
223			*
224			* @return array
225			*/
226			public function export()
227			{
228			$result = [];
229			foreach ($this->rules as $directive => $object) {
230			if (!empty($export = $object->export())) {
231			$result[$directive] = $export[$directive];
232			}
233			}
234			return $result;
235			}
236
237			/**
238			* UserAgentClient destructor.
239			*/
240			public function __destruct()
241			{
242			if (!$this->commentsExported) {
243			// Comment from the `Comments` directive exists, but has not been exported.
244			foreach ($this->getComments() as $message) {
245			trigger_error('Comment for `' . $this->userAgent . '` at `' . $this->base . '/robots.txt`: ' . $message, E_USER_NOTICE);
246			}
247			}
248			}
249
250			/**
251			* Get Comments
252			*
253			* @return array
254			*/
255			public function getComments()
256			{
257			$this->commentsExported = true;
258			$comments = $this->rules[self::DIRECTIVE_COMMENT]->export();
259			return isset($comments[self::DIRECTIVE_COMMENT]) ? $comments[self::DIRECTIVE_COMMENT] : [];
260			}
261
262			/**
263			* Get Visit-time
264			*
265			* @return array\|false
266			*/
267			public function getVisitTime()
268			{
269			$times = $this->rules[self::DIRECTIVE_VISIT_TIME]->export();
270			return isset($times[self::DIRECTIVE_VISIT_TIME]) ? $times[self::DIRECTIVE_VISIT_TIME] : [];
271			}
272			}
273

VIPnytt / RobotsTxtParser

Push — master ( e5e536...a6d8dc )

UserAgentClient::determineRequestRates() C

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like