UriClient::getBaseUri() - Code Metrics - VIPnytt/RobotsTxtParser - Measure and Improve Code Quality continuously with Scrutinizer

UriClient::getBaseUri() A
last analyzed 2021-04-10 17:00 UTC

↳ Parent: UriClient

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	4
rs	10
c	0
b	0
f	0
cc	1
nc	1
nop	0

<?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;

use Composer\CaBundle\CaBundle;
use vipnytt\RobotsTxtParser\Parser\StatusCodeParser;
use vipnytt\RobotsTxtParser\Parser\UriParser;

/**
 * Class UriClient
 *
 * @see https://github.com/VIPnytt/RobotsTxtParser/blob/master/docs/methods/UriClient.md for documentation
 * @package vipnytt\RobotsTxtParser
 */
class UriClient extends TxtClient
{
    /**
     * User-agent
     */
    const CURL_USER_AGENT = 'RobotsTxtParser-VIPnytt/2.1 (+https://github.com/VIPnytt/RobotsTxtParser/blob/master/README.md)';

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

    /**
     * Header parser
     * @var Parser\HeaderParser
     */
    private $headerParser;

    /**
     * RequestClient timestamp
     * @var int
     */
    private $time;

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

    /**
     * Effective uri
     * @var string
     */
    private $effective;

    /**
     * Cache-Control max-age
     * @var int
     */
    private $rawMaxAge;

    /**
     * Robots.txt contents
     * @var string
     */
    private $rawContents;

    /**
     * Robots.txt character encoding
     * @var string
     */
    private $rawEncoding;

    /**
     * RequestClient constructor.
     *
     * @param string $baseUri
     * @param array $curlOptions
     * @param int|null $byteLimit
     */
    public function __construct($baseUri, array $curlOptions = [], $byteLimit = self::BYTE_LIMIT)
    {
        $uriParser = new UriParser($baseUri);
        $this->base = $uriParser->base();
        if ($this->request($curlOptions) === false) {
            $this->time = time();
            $this->effective = $this->base;
            $this->rawStatusCode = null;
            $this->rawContents = '';
            $this->rawEncoding = self::ENCODING;
            $this->rawMaxAge = 0;
        }
        parent::__construct($this->base, $this->rawStatusCode, $this->rawContents, $this->rawEncoding, $this->effective, $byteLimit);
    }

    /**
     * cURL request
     *
     * @param array $options
     * @return bool
     */
    private function request($options = [])
    {
        $curl = curl_init();
        $caPathOrFile = CaBundle::getSystemCaRootBundlePath();
        // Set default cURL options
        curl_setopt_array($curl, [
            CURLOPT_AUTOREFERER => true,
            CURLOPT_CONNECTTIMEOUT => 30,
            CURLOPT_ENCODING => 'identity',
            CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_NONE,
            CURLOPT_IPRESOLVE => CURL_IPRESOLVE_WHATEVER,
            CURLOPT_SSL_VERIFYHOST => 2,
            CURLOPT_SSL_VERIFYPEER => true,
            CURLOPT_TIMEOUT => 120,
            CURLOPT_USERAGENT => self::CURL_USER_AGENT,
            (is_dir($caPathOrFile) ||
                (
                    is_link($caPathOrFile) &&
                    is_dir(readlink($caPathOrFile))
                )
            ) ? CURLOPT_CAPATH : CURLOPT_CAINFO => $caPathOrFile
        ]);
        // Apply custom cURL options
        curl_setopt_array($curl, $options);
        // Initialize the header parser
        $this->headerParser = new Parser\HeaderParser($curl);
        // Make sure these cURL options stays untouched
        curl_setopt_array($curl, [
            CURLOPT_FAILONERROR => false,
            CURLOPT_FOLLOWLOCATION => true,
            CURLOPT_FTPSSLAUTH => CURLFTPAUTH_DEFAULT,
            CURLOPT_HEADER => false,
            CURLOPT_HEADERFUNCTION => [$this->headerParser, 'curlCallback'],
            CURLOPT_HTTPAUTH => CURLAUTH_BASIC,
            CURLOPT_MAXREDIRS => self::MAX_REDIRECTS,
            CURLOPT_NOBODY => false,
            CURLOPT_PROTOCOLS => CURLPROTO_FTP | CURLPROTO_FTPS | CURLPROTO_HTTP | CURLPROTO_HTTPS | CURLPROTO_SFTP,
            CURLOPT_REDIR_PROTOCOLS => CURLPROTO_FTP | CURLPROTO_FTPS | CURLPROTO_HTTP | CURLPROTO_HTTPS | CURLPROTO_SFTP,
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_URL => $this->base . self::PATH,
            CURLOPT_USERPWD => 'anonymous:anonymous@',
        ]);
        // Execute cURL request
        if (($this->rawContents = curl_exec($curl)) === false) {
            // Request failed
            return false;
        }
        $this->time = time();
        $this->rawStatusCode = curl_getinfo($curl, CURLINFO_HTTP_CODE); // also works with FTP status codes
        $uriParser = new UriParser(curl_getinfo($curl, CURLINFO_EFFECTIVE_URL));
        $this->effective = $uriParser->base();
        curl_close($curl);
        $this->rawEncoding = $this->headerParser->getCharset();
        $this->rawMaxAge = $this->headerParser->getMaxAge();
        return true;
    }

    /**
     * Base uri
     *
     * @return string
     */
    public function getBaseUri()
    {
        return $this->base;
    }

    /**
     * Effective uri
     *
     * @return string|null
     */
    public function getEffectiveUri()
    {
        return $this->effective;
    }

    /**
     * Status code
     *
     * @return int|null
     */
    public function getStatusCode()
    {
        $statusCodeParser = new StatusCodeParser($this->rawStatusCode, parse_url($this->effective, PHP_URL_SCHEME));
        return $statusCodeParser->isValid() ? $this->rawStatusCode : null;
    }

    /**
     * Body content
     *
     * @return string
     */
    public function getContents()
    {
        return $this->rawContents;
    }

    /**
     * Encoding
     *
     * @return string
     */
    public function getEncoding()
    {
        return $this->rawEncoding;
    }

    /**
     * Next update timestamp
     *
     * @return int
     */
    public function nextUpdate()
    {
        if ($this->rawStatusCode === 503 &&
            strpos($this->base, 'http') === 0
        ) {
            return $this->time + min(self::CACHE_TIME, $this->headerParser->getRetryAfter($this->time));
        }
        return $this->time + self::CACHE_TIME;
    }

    /**
     * Valid until timestamp
     *
     * @return int
     */
    public function validUntil()
    {
        return $this->time + max(self::CACHE_TIME, $this->rawMaxAge);
    }
}


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;
10
11			use Composer\CaBundle\CaBundle;
12			use vipnytt\RobotsTxtParser\Parser\StatusCodeParser;
13			use vipnytt\RobotsTxtParser\Parser\UriParser;
14
15			/**
16			* Class UriClient
17			*
18			* @see https://github.com/VIPnytt/RobotsTxtParser/blob/master/docs/methods/UriClient.md for documentation
19			* @package vipnytt\RobotsTxtParser
20			*/
21			class UriClient extends TxtClient
22			{
23			/**
24			* User-agent
25			*/
26			const CURL_USER_AGENT = 'RobotsTxtParser-VIPnytt/2.1 (+https://github.com/VIPnytt/RobotsTxtParser/blob/master/README.md)';
27
28			/**
29			* Base uri
30			* @var string
31			*/
32			private $base;
33
34			/**
35			* Header parser
36			* @var Parser\HeaderParser
37			*/
38			private $headerParser;
39
40			/**
41			* RequestClient timestamp
42			* @var int
43			*/
44			private $time;
45
46			/**
47			* Status code
48			* @var int\|null
49			*/
50			private $rawStatusCode;
51
52			/**
53			* Effective uri
54			* @var string
55			*/
56			private $effective;
57
58			/**
59			* Cache-Control max-age
60			* @var int
61			*/
62			private $rawMaxAge;
63
64			/**
65			* Robots.txt contents
66			* @var string
67			*/
68			private $rawContents;
69
70			/**
71			* Robots.txt character encoding
72			* @var string
73			*/
74			private $rawEncoding;
75
76			/**
77			* RequestClient constructor.
78			*
79			* @param string $baseUri
80			* @param array $curlOptions
81			* @param int\|null $byteLimit
82			*/
83			public function __construct($baseUri, array $curlOptions = [], $byteLimit = self::BYTE_LIMIT)
84			{
85			$uriParser = new UriParser($baseUri);
86			$this->base = $uriParser->base();
87			if ($this->request($curlOptions) === false) {
88			$this->time = time();
89			$this->effective = $this->base;
90			$this->rawStatusCode = null;
91			$this->rawContents = '';
92			$this->rawEncoding = self::ENCODING;
93			$this->rawMaxAge = 0;
94			}
95			parent::__construct($this->base, $this->rawStatusCode, $this->rawContents, $this->rawEncoding, $this->effective, $byteLimit);
96			}
97
98			/**
99			* cURL request
100			*
101			* @param array $options
102			* @return bool
103			*/
104			private function request($options = [])
105			{
106			$curl = curl_init();
107			$caPathOrFile = CaBundle::getSystemCaRootBundlePath();
108			// Set default cURL options
109			curl_setopt_array($curl, [
110			CURLOPT_AUTOREFERER => true,
111			CURLOPT_CONNECTTIMEOUT => 30,
112			CURLOPT_ENCODING => 'identity',
113			CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_NONE,
114			CURLOPT_IPRESOLVE => CURL_IPRESOLVE_WHATEVER,
115			CURLOPT_SSL_VERIFYHOST => 2,
116			CURLOPT_SSL_VERIFYPEER => true,
117			CURLOPT_TIMEOUT => 120,
118			CURLOPT_USERAGENT => self::CURL_USER_AGENT,
119			(is_dir($caPathOrFile) \|\|
120			(
121			is_link($caPathOrFile) &&
122			is_dir(readlink($caPathOrFile))
123			)
124			) ? CURLOPT_CAPATH : CURLOPT_CAINFO => $caPathOrFile
125			]);
126			// Apply custom cURL options
127			curl_setopt_array($curl, $options);
128			// Initialize the header parser
129			$this->headerParser = new Parser\HeaderParser($curl);
130			// Make sure these cURL options stays untouched
131			curl_setopt_array($curl, [
132			CURLOPT_FAILONERROR => false,
133			CURLOPT_FOLLOWLOCATION => true,
134			CURLOPT_FTPSSLAUTH => CURLFTPAUTH_DEFAULT,
135			CURLOPT_HEADER => false,
136			CURLOPT_HEADERFUNCTION => [$this->headerParser, 'curlCallback'],
137			CURLOPT_HTTPAUTH => CURLAUTH_BASIC,
138			CURLOPT_MAXREDIRS => self::MAX_REDIRECTS,
139			CURLOPT_NOBODY => false,
140			CURLOPT_PROTOCOLS => CURLPROTO_FTP \| CURLPROTO_FTPS \| CURLPROTO_HTTP \| CURLPROTO_HTTPS \| CURLPROTO_SFTP,
141			CURLOPT_REDIR_PROTOCOLS => CURLPROTO_FTP \| CURLPROTO_FTPS \| CURLPROTO_HTTP \| CURLPROTO_HTTPS \| CURLPROTO_SFTP,
142			CURLOPT_RETURNTRANSFER => true,
143			CURLOPT_URL => $this->base . self::PATH,
144			CURLOPT_USERPWD => 'anonymous:anonymous@',
145			]);
146			// Execute cURL request
147			if (($this->rawContents = curl_exec($curl)) === false) {
148			// Request failed
149			return false;
150			}
151			$this->time = time();
152			$this->rawStatusCode = curl_getinfo($curl, CURLINFO_HTTP_CODE); // also works with FTP status codes
153			$uriParser = new UriParser(curl_getinfo($curl, CURLINFO_EFFECTIVE_URL));
154			$this->effective = $uriParser->base();
155			curl_close($curl);
156			$this->rawEncoding = $this->headerParser->getCharset();
157			$this->rawMaxAge = $this->headerParser->getMaxAge();
158			return true;
159			}
160
161			/**
162			* Base uri
163			*
164			* @return string
165			*/
166			public function getBaseUri()
167			{
168			return $this->base;
169			}
170
171			/**
172			* Effective uri
173			*
174			* @return string\|null
175			*/
176			public function getEffectiveUri()
177			{
178			return $this->effective;
179			}
180
181			/**
182			* Status code
183			*
184			* @return int\|null
185			*/
186			public function getStatusCode()
187			{
188			$statusCodeParser = new StatusCodeParser($this->rawStatusCode, parse_url($this->effective, PHP_URL_SCHEME));
189			return $statusCodeParser->isValid() ? $this->rawStatusCode : null;
190			}
191
192			/**
193			* Body content
194			*
195			* @return string
196			*/
197			public function getContents()
198			{
199			return $this->rawContents;
200			}
201
202			/**
203			* Encoding
204			*
205			* @return string
206			*/
207			public function getEncoding()
208			{
209			return $this->rawEncoding;
210			}
211
212			/**
213			* Next update timestamp
214			*
215			* @return int
216			*/
217			public function nextUpdate()
218			{
219			if ($this->rawStatusCode === 503 &&
220			strpos($this->base, 'http') === 0
221			) {
222			return $this->time + min(self::CACHE_TIME, $this->headerParser->getRetryAfter($this->time));
223			}
224			return $this->time + self::CACHE_TIME;
225			}
226
227			/**
228			* Valid until timestamp
229			*
230			* @return int
231			*/
232			public function validUntil()
233			{
234			return $this->time + max(self::CACHE_TIME, $this->rawMaxAge);
235			}
236			}
237

VIPnytt / RobotsTxtParser

UriClient::getBaseUri() A last analyzed 2021-04-10 17:00 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

UriClient::getBaseUri() A
last analyzed 2021-04-10 17:00 UTC