UriClient::headerCharset() - Code Metrics - Inspection of "Bug fixes and other minor improvements" - VIPnytt/RobotsTxtParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( b23c5f...0489e6 )

by Jan-Petter

created 2016-06-25 18:00 UTC

UriClient::headerCharset() A

↳ Parent: UriClient

Complexity

Conditions	2
Paths	2

Size

Total Lines	7
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	1	Features	0

Metric	Value
c	1
b	1
f	0
dl	0
loc	7
rs	9.4285
cc	2
eloc	4
nc	2
nop	0

<?php
namespace vipnytt\RobotsTxtParser;

use GuzzleHttp;

/**
 * Class UriClient
 *
 * @package vipnytt\RobotsTxtParser
 */
class UriClient extends TxtClient
{
    /**
     * ANSI C's asctime() format
     */
    const DATE_ASCTIME = 'D M j h:i:s Y';

    /**
     * HTTP date formats
     */
    const DATE_HTTP = [
        DATE_RFC1123,
        DATE_RFC850,
        self::DATE_ASCTIME,
    ];

    /**
     * GuzzleHttp config
     */
    const GUZZLE_HTTP_CONFIG = [
        'allow_redirects' => [
            'max' => self::MAX_REDIRECTS,
            'referer' => true,
            'strict' => true,
        ],
        'decode_content' => false,
        'headers' => [
            'accept' => 'text/plain;q=1.0, text/*;q=0.8, */*;q=0.1',
            'accept-charset' => 'utf-8;q=1.0, *;q=0.1',
            'accept-encoding' => 'identity;q=1.0, *;q=0.1',
            'user-agent' => 'RobotsTxtParser-VIPnytt/2.0 (+https://github.com/VIPnytt/RobotsTxtParser/blob/master/README.md)',
        ],
        'http_errors' => false,
        'verify' => true,
    ];

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

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

    /**
     * @var \Psr\Http\Message\ResponseInterface
     */
    private $response;

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

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

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

    /**
     * RequestClient constructor.
     *
     * @param string $baseUri
     * @param array $guzzleConfig
     * @param int|null $byteLimit
     */
    public function __construct($baseUri, array $guzzleConfig = [], $byteLimit = self::BYTE_LIMIT)
    {
        $this->base = $this->urlBase($this->urlEncode($baseUri));
        $this->time = time();
        try {
            $client = new GuzzleHttp\Client(
                array_merge_recursive(
                    self::GUZZLE_HTTP_CONFIG,
                    $guzzleConfig,
                    [
                        'base_uri' => $this->base,
                    ]
                )
            );
            $this->response = $client->request('GET', self::PATH);
            $this->time = time();
            $this->statusCode = $this->response->getStatusCode();
            $this->contents = $this->response->getBody()->getContents();
            $this->encoding = $this->headerCharset();
            $this->maxAge = $this->headerMaxAge();
        } catch (GuzzleHttp\Exception\TransferException $e) {
            $this->statusCode = null;
            $this->contents = '';
            $this->encoding = self::ENCODING;
            $this->maxAge = 0;
        }
        parent::__construct($this->base, $this->statusCode, $this->contents, $this->encoding, $byteLimit);
    }

    /**
     * Content-Type encoding HTTP header
     *
     * @link https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.17
     *
     * @return string
     */
    private function headerCharset()
    {
        if (($value = $this->parseHeader($this->response->getHeader('content-type'), 'charset', ';')) !== false) {
            return $value;
        }
        return self::ENCODING;
    }

    /**
     * Client header
     *
     * @param string[] $headers
     * @param string $part
     * @param string $delimiter
     * @return string|false
     */
    private function parseHeader(array $headers, $part, $delimiter = ";")
    {
        foreach ($headers as $header) {
            foreach (array_map('trim', mb_split($delimiter, $header)) as $string) {
                if (mb_stripos($string, $part . '=') === 0) {
                    return mb_split('=', $string, 2)[1];
                }
            }
        }
        return false;
    }

    /**
     * Cache-Control max-age HTTP header
     *
     * @link https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.9.3
     *
     * @return int
     */
    private function headerMaxAge()
    {
        if (($value = $this->parseHeader($this->response->getHeader('cache-control'), 'max-age', ',')) !== false) {
            return intval($value);
        }
        return 0;
    }

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

    /**
     * Status code
     *
     * @return int|null
     */
    public function getStatusCode()
    {
        return $this->statusCode;
    }

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

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

    /**
     * Next update timestamp
     *
     * @return int
     */
    public function nextUpdate()
    {
        if (
            $this->statusCode === 503 &&
            ($retryTime = $this->headerRetryAfter()) !== false
        ) {
            return min($this->time + self::CACHE_TIME, $retryTime);
        }
        return $this->time + self::CACHE_TIME;
    }

    /**
     * Cache-Control Retry-After HTTP header
     *
     * @link https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.37
     *
     * @return int|false
     */
    private function headerRetryAfter()
    {
        foreach ($this->response->getHeader('retry-after') as $parts) {
            $value = implode(', ', $parts);
            if (is_numeric($value)) {
                return $this->time + $value;
            } elseif (($time = $this->parseHttpDate($value)) !== false) {
                return $time;
            }
        }
        return false;
    }

    /**
     * Parse HTTP-date
     *
     * @param string $string
     * @return int|false
     */
    private function parseHttpDate($string)
    {
        foreach (self::DATE_HTTP as $format) {
            $dateTime = date_create_from_format($format, $string, new \DateTimeZone('GMT'));
            if ($dateTime !== false) {
                return (int)date_format($dateTime, 'U');
            }
        }
        return false;
    }

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


1			<?php
2			namespace vipnytt\RobotsTxtParser;
3
4			use GuzzleHttp;
5
6			/**
7			* Class UriClient
8			*
9			* @package vipnytt\RobotsTxtParser
10			*/
11			class UriClient extends TxtClient
12			{
13			/**
14			* ANSI C's asctime() format
15			*/
16			const DATE_ASCTIME = 'D M j h:i:s Y';
17
18			/**
19			* HTTP date formats
20			*/
21			const DATE_HTTP = [
22			DATE_RFC1123,
23			DATE_RFC850,
24			self::DATE_ASCTIME,
25			];
26
27			/**
28			* GuzzleHttp config
29			*/
30			const GUZZLE_HTTP_CONFIG = [
31			'allow_redirects' => [
32			'max' => self::MAX_REDIRECTS,
33			'referer' => true,
34			'strict' => true,
35			],
36			'decode_content' => false,
37			'headers' => [
38			'accept' => 'text/plain;q=1.0, text/;q=0.8, /*;q=0.1',
39			'accept-charset' => 'utf-8;q=1.0, *;q=0.1',
40			'accept-encoding' => 'identity;q=1.0, *;q=0.1',
41			'user-agent' => 'RobotsTxtParser-VIPnytt/2.0 (+https://github.com/VIPnytt/RobotsTxtParser/blob/master/README.md)',
42			],
43			'http_errors' => false,
44			'verify' => true,
45			];
46
47			/**
48			* Base uri
49			* @var string
50			*/
51			private $base;
52
53			/**
54			* RequestClient timestamp
55			* @var int
56			*/
57			private $time;
58
59			/**
60			* @var \Psr\Http\Message\ResponseInterface
61			*/
62			private $response;
63
64			/**
65			* Cache-Control max-age
66			* @var int
67			*/
68			private $maxAge;
69
70			/**
71			* Robots.txt contents
72			* @var string
73			*/
74			private $contents;
75
76			/**
77			* Robots.txt character encoding
78			* @var string
79			*/
80			private $encoding;
81
82			/**
83			* RequestClient constructor.
84			*
85			* @param string $baseUri
86			* @param array $guzzleConfig
87			* @param int\|null $byteLimit
88			*/
89			public function __construct($baseUri, array $guzzleConfig = [], $byteLimit = self::BYTE_LIMIT)
90			{
91			$this->base = $this->urlBase($this->urlEncode($baseUri));
92			$this->time = time();
93			try {
94			$client = new GuzzleHttp\Client(
95			array_merge_recursive(
96			self::GUZZLE_HTTP_CONFIG,
97			$guzzleConfig,
98			[
99			'base_uri' => $this->base,
100			]
101			)
102			);
103			$this->response = $client->request('GET', self::PATH);
104			$this->time = time();
105			$this->statusCode = $this->response->getStatusCode();
106			$this->contents = $this->response->getBody()->getContents();
107			$this->encoding = $this->headerCharset();
108			$this->maxAge = $this->headerMaxAge();
109			} catch (GuzzleHttp\Exception\TransferException $e) {
110			$this->statusCode = null;
111			$this->contents = '';
112			$this->encoding = self::ENCODING;
113			$this->maxAge = 0;
114			}
115			parent::__construct($this->base, $this->statusCode, $this->contents, $this->encoding, $byteLimit);
116			}
117
118			/**
119			* Content-Type encoding HTTP header
120			*
121			* @link https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.17
122			*
123			* @return string
124			*/
125			private function headerCharset()
126			{
127			if (($value = $this->parseHeader($this->response->getHeader('content-type'), 'charset', ';')) !== false) {
128			return $value;
129			}
130			return self::ENCODING;
131			}
132
133			/**
134			* Client header
135			*
136			* @param string[] $headers
137			* @param string $part
138			* @param string $delimiter
139			* @return string\|false
140			*/
141			private function parseHeader(array $headers, $part, $delimiter = ";")
142			{
143			foreach ($headers as $header) {
144			foreach (array_map('trim', mb_split($delimiter, $header)) as $string) {
145			if (mb_stripos($string, $part . '=') === 0) {
146			return mb_split('=', $string, 2)[1];
147			}
148			}
149			}
150			return false;
151			}
152
153			/**
154			* Cache-Control max-age HTTP header
155			*
156			* @link https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.9.3
157			*
158			* @return int
159			*/
160			private function headerMaxAge()
161			{
162			if (($value = $this->parseHeader($this->response->getHeader('cache-control'), 'max-age', ',')) !== false) {
163			return intval($value);
164			}
165			return 0;
166			}
167
168			/**
169			* Base UriClient
170			*
171			* @return string
172			*/
173			public function getBaseUri()
174			{
175			return $this->base;
176			}
177
178			/**
179			* Status code
180			*
181			* @return int\|null
182			*/
183			public function getStatusCode()
184			{
185			return $this->statusCode;
186			}
187
188			/**
189			* URL content
190			*
191			* @return string
192			*/
193			public function getContents()
194			{
195			return $this->contents;
196			}
197
198			/**
199			* Encoding
200			*
201			* @return string
202			*/
203			public function getEncoding()
204			{
205			return $this->encoding;
206			}
207
208			/**
209			* Next update timestamp
210			*
211			* @return int
212			*/
213			public function nextUpdate()
214			{
215			if (
216			$this->statusCode === 503 &&
217			($retryTime = $this->headerRetryAfter()) !== false
218			) {
219			return min($this->time + self::CACHE_TIME, $retryTime);
220			}
221			return $this->time + self::CACHE_TIME;
222			}
223
224			/**
225			* Cache-Control Retry-After HTTP header
226			*
227			* @link https://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.37
228			*
229			* @return int\|false
230			*/
231			private function headerRetryAfter()
232			{
233			foreach ($this->response->getHeader('retry-after') as $parts) {
234			$value = implode(', ', $parts);
235			if (is_numeric($value)) {
236			return $this->time + $value;
237			} elseif (($time = $this->parseHttpDate($value)) !== false) {
238			return $time;
239			}
240			}
241			return false;
242			}
243
244			/**
245			* Parse HTTP-date
246			*
247			* @param string $string
248			* @return int\|false
249			*/
250			private function parseHttpDate($string)
251			{
252			foreach (self::DATE_HTTP as $format) {
253			$dateTime = date_create_from_format($format, $string, new \DateTimeZone('GMT'));
254			if ($dateTime !== false) {
255			return (int)date_format($dateTime, 'U');
256			}
257			}
258			return false;
259			}
260
261			/**
262			* Valid until timestamp
263			*
264			* @return int
265			*/
266			public function validUntil()
267			{
268			return $this->time + max(self::CACHE_TIME, $this->maxAge);
269			}
270			}
271

VIPnytt / RobotsTxtParser

Push — master ( b23c5f...0489e6 )

UriClient::headerCharset() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like