RequestRateParser::add() - Code Metrics - Inspection of "Merge remote-tracking branch 'origin/master'" - VIPnytt/RobotsTxtParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Branch — master (ed4a79)

by Jan-Petter

created 2018-02-14 21:37 UTC

RequestRateParser::add() A

↳ Parent: RequestRateParser

Complexity

Conditions	4
Paths	3

Size

Total Lines	18
Code Lines	13

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	18
rs	9.2
c	0
b	0
f	0
cc	4
eloc	13
nc	3
nop	1

<?php
/**
 * vipnytt/RobotsTxtParser
 *
 * @link https://github.com/VIPnytt/RobotsTxtParser
 * @license https://github.com/VIPnytt/RobotsTxtParser/blob/master/LICENSE The MIT License (MIT)
 */

namespace vipnytt\RobotsTxtParser\Parser\Directives;

use vipnytt\RobotsTxtParser\Client\Directives\RequestRateClient;
use vipnytt\RobotsTxtParser\Handler\RenderHandler;
use vipnytt\RobotsTxtParser\RobotsTxtInterface;

/**
 * Class RequestRateParser
 *
 * @package vipnytt\RobotsTxtParser\Parser\Directives
 */
class RequestRateParser implements ParserInterface, RobotsTxtInterface
{
    use DirectiveParserTrait;

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

    /**
     * RequestRate array
     * @var array
     */
    private $requestRates = [];

    /**
     * Sorted
     * @var bool
     */
    private $sorted = false;

    /**
     * Time units
     * @var int[]
     */
    private $units = [
        'w' => 604800,
        'd' => 86400,
        'h' => 3600,
        'm' => 60,
    ];

    /**
     * RequestRate constructor.
     *
     * @param string $base
     */
    public function __construct($base)
    {
        $this->base = $base;
    }

    /**
     * Add
     *
     * @param string $line
     * @return bool
     */
    public function add($line)
    {
        $array = preg_split('/\s+/', $line, 2);
        $result = [
            'rate' => $this->draftParseRate($array[0]),
            'from' => null,
            'to' => null,
        ];
        if ($result['rate'] === false) {
            return false;
        } elseif (!empty($array[1]) &&
            ($times = $this->draftParseTime($array[1])) !== false
        ) {
            $result = array_merge($result, $times);
        }
        $this->requestRates[] = $result;
        return true;
    }

    /**
     * Client rate as specified in the `Robot exclusion standard` version 2.0 draft
     * rate = numDocuments / timeUnit
     * @link http://www.conman.org/people/spc/robots2.html#format.directives.request-rate
     *
     * @param string $string
     * @return float|int|false
     */
    private function draftParseRate($string)
    {
        $parts = array_map('trim', explode('/', $string));
        if (count($parts) != 2) {
            return false;
        }
        $unit = strtolower(substr(preg_replace('/[^A-Za-z]/', '', filter_var($parts[1], FILTER_SANITIZE_STRING)), 0, 1));
        $multiplier = isset($this->units[$unit]) ? $this->units[$unit] : 1;
        $rate = abs(filter_var($parts[1], FILTER_SANITIZE_NUMBER_FLOAT, FILTER_FLAG_ALLOW_FRACTION)) * $multiplier / abs(filter_var($parts[0], FILTER_SANITIZE_NUMBER_INT));
        return $rate > 0 ? $rate : false;
    }

    /**
     * Client
     *
     * @param string $userAgent
     * @param float|int $fallbackValue
     * @return RequestRateClient
     */
    public function client($userAgent = self::USER_AGENT, $fallbackValue = 0)
    {
        $this->sort();
        return new RequestRateClient($this->base, $userAgent, $this->requestRates, $fallbackValue);
    }

    /**
     * Sort
     *
     * @return bool
     */
    private function sort()
    {
        if (!$this->sorted) {
            $this->sorted = true;
            return usort($this->requestRates, function (array $requestRateA, array $requestRateB) {
                // PHP 7: Switch to the <=> "Spaceship" operator
                return $requestRateB['rate'] > $requestRateA['rate'];
            });
        }
        return $this->sorted;
    }

    /**
     * Render
     *
     * @param RenderHandler $handler
     * @return bool
     */
    public function render(RenderHandler $handler)
    {
        $this->sort();
        foreach ($this->requestRates as $array) {
            $multiplyFactor = $this->decimalMultiplier($array['rate']);
            $multipliedRate = $array['rate'] * $multiplyFactor;
            $gcd = $this->getGCD($multiplyFactor, $multipliedRate);
            $requests = $multiplyFactor / $gcd;
            $time = $multipliedRate / $gcd;
            $suffix = 's';
            foreach ($this->units as $unit => $sec) {
                if ($time % $sec === 0) {
                    $suffix = $unit;
                    $time /= $sec;
                    break;
                }
            }
            if (isset($array['from']) &&
                isset($array['to'])
            ) {
                $suffix .= ' ' . $array['from'] . '-' . $array['to'];
            }
            $handler->add(self::DIRECTIVE_REQUEST_RATE, $requests . '/' . $time . $suffix);
        }
        return true;
    }

    /**
     * @param int|float $value
     * @return int
     */
    private function decimalMultiplier($value)
    {
        $multiplier = 1;
        while (fmod($value, 1) != 0) {
            $value *= 10;
            $multiplier *= 10;
        }
        return $multiplier;
    }

    /**
     * Returns the greatest common divisor of two integers using the Euclidean algorithm.
     *
     * @param int $a
     * @param int $b
     * @return int
     */
    private function getGCD($a, $b)
    {
        if (extension_loaded('gmp')) {
            return gmp_intval(gmp_gcd((string)$a, (string)$b));
        }
        $large = $a > $b ? $a : $b;
        $small = $a > $b ? $b : $a;
        $remainder = $large % $small;
        return 0 === $remainder ? $small : $this->getGCD($small, $remainder);
    }
}


1			<?php
2			/**
3			* vipnytt/RobotsTxtParser
4			*
5			* @link https://github.com/VIPnytt/RobotsTxtParser
6			* @license https://github.com/VIPnytt/RobotsTxtParser/blob/master/LICENSE The MIT License (MIT)
7			*/
8
9			namespace vipnytt\RobotsTxtParser\Parser\Directives;
10
11			use vipnytt\RobotsTxtParser\Client\Directives\RequestRateClient;
12			use vipnytt\RobotsTxtParser\Handler\RenderHandler;
13			use vipnytt\RobotsTxtParser\RobotsTxtInterface;
14
15			/**
16			* Class RequestRateParser
17			*
18			* @package vipnytt\RobotsTxtParser\Parser\Directives
19			*/
20			class RequestRateParser implements ParserInterface, RobotsTxtInterface
21			{
22			use DirectiveParserTrait;
23
24			/**
25			* Base uri
26			* @var string
27			*/
28			private $base;
29
30			/**
31			* RequestRate array
32			* @var array
33			*/
34			private $requestRates = [];
35
36			/**
37			* Sorted
38			* @var bool
39			*/
40			private $sorted = false;
41
42			/**
43			* Time units
44			* @var int[]
45			*/
46			private $units = [
47			'w' => 604800,
48			'd' => 86400,
49			'h' => 3600,
50			'm' => 60,
51			];
52
53			/**
54			* RequestRate constructor.
55			*
56			* @param string $base
57			*/
58			public function __construct($base)
59			{
60			$this->base = $base;
61			}
62
63			/**
64			* Add
65			*
66			* @param string $line
67			* @return bool
68			*/
69			public function add($line)
70			{
71			$array = preg_split('/\s+/', $line, 2);
72			$result = [
73			'rate' => $this->draftParseRate($array[0]),
74			'from' => null,
75			'to' => null,
76			];
77			if ($result['rate'] === false) {
78			return false;
79			} elseif (!empty($array[1]) &&
80			($times = $this->draftParseTime($array[1])) !== false
81			) {
82			$result = array_merge($result, $times);
83			}
84			$this->requestRates[] = $result;
85			return true;
86			}
87
88			/**
89			* Client rate as specified in the `Robot exclusion standard` version 2.0 draft
90			* rate = numDocuments / timeUnit
91			* @link http://www.conman.org/people/spc/robots2.html#format.directives.request-rate
92			*
93			* @param string $string
94			* @return float\|int\|false
95			*/
96			private function draftParseRate($string)
97			{
98			$parts = array_map('trim', explode('/', $string));
99			if (count($parts) != 2) {
100			return false;
101			}
102			$unit = strtolower(substr(preg_replace('/[^A-Za-z]/', '', filter_var($parts[1], FILTER_SANITIZE_STRING)), 0, 1));
103			$multiplier = isset($this->units[$unit]) ? $this->units[$unit] : 1;
104			$rate = abs(filter_var($parts[1], FILTER_SANITIZE_NUMBER_FLOAT, FILTER_FLAG_ALLOW_FRACTION)) * $multiplier / abs(filter_var($parts[0], FILTER_SANITIZE_NUMBER_INT));
105			return $rate > 0 ? $rate : false;
106			}
107
108			/**
109			* Client
110			*
111			* @param string $userAgent
112			* @param float\|int $fallbackValue
113			* @return RequestRateClient
114			*/
115			public function client($userAgent = self::USER_AGENT, $fallbackValue = 0)
116			{
117			$this->sort();
118			return new RequestRateClient($this->base, $userAgent, $this->requestRates, $fallbackValue);
119			}
120
121			/**
122			* Sort
123			*
124			* @return bool
125			*/
126			private function sort()
127			{
128			if (!$this->sorted) {
129			$this->sorted = true;
130			return usort($this->requestRates, function (array $requestRateA, array $requestRateB) {
131			// PHP 7: Switch to the <=> "Spaceship" operator
132			return $requestRateB['rate'] > $requestRateA['rate'];
133			});
134			}
135			return $this->sorted;
136			}
137
138			/**
139			* Render
140			*
141			* @param RenderHandler $handler
142			* @return bool
143			*/
144			public function render(RenderHandler $handler)
145			{
146			$this->sort();
147			foreach ($this->requestRates as $array) {
148			$multiplyFactor = $this->decimalMultiplier($array['rate']);
149			$multipliedRate = $array['rate'] * $multiplyFactor;
150			$gcd = $this->getGCD($multiplyFactor, $multipliedRate);
151			$requests = $multiplyFactor / $gcd;
152			$time = $multipliedRate / $gcd;
153			$suffix = 's';
154			foreach ($this->units as $unit => $sec) {
155			if ($time % $sec === 0) {
156			$suffix = $unit;
157			$time /= $sec;
158			break;
159			}
160			}
161			if (isset($array['from']) &&
162			isset($array['to'])
163			) {
164			$suffix .= ' ' . $array['from'] . '-' . $array['to'];
165			}
166			$handler->add(self::DIRECTIVE_REQUEST_RATE, $requests . '/' . $time . $suffix);
167			}
168			return true;
169			}
170
171			/**
172			* @param int\|float $value
173			* @return int
174			*/
175			private function decimalMultiplier($value)
176			{
177			$multiplier = 1;
178			while (fmod($value, 1) != 0) {
179			$value *= 10;
180			$multiplier *= 10;
181			}
182			return $multiplier;
183			}
184
185			/**
186			* Returns the greatest common divisor of two integers using the Euclidean algorithm.
187			*
188			* @param int $a
189			* @param int $b
190			* @return int
191			*/
192			private function getGCD($a, $b)
193			{
194			if (extension_loaded('gmp')) {
195			return gmp_intval(gmp_gcd((string)$a, (string)$b));
196			}
197			$large = $a > $b ? $a : $b;
198			$small = $a > $b ? $b : $a;
199			$remainder = $large % $small;
200			return 0 === $remainder ? $small : $this->getGCD($small, $remainder);
201			}
202			}
203

VIPnytt / RobotsTxtParser

Branch — master (ed4a79)

RequestRateParser::add() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like