XRobotsTagParser::cleanup() - Code Metrics - Inspection of "minor refactoring update" - VIPnytt/RobotsTagParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 26bda2...b142f3 )

by Jan-Petter

created 2016-02-08 23:47 UTC

XRobotsTagParser::cleanup() A

↳ Parent: XRobotsTagParser

Complexity

Conditions	1
Paths	1

Size

Total Lines	7
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Importance

Changes	4
Bugs	0	Features	0

Metric	Value
c	4
b	0
f	0
dl	0
loc	7
rs	9.4285
cc	1
eloc	5
nc	1
nop	0

<?php
/**
 * X-Robots-Tag HTTP header parser class
 *
 * @author VIP nytt ([email protected])
 * @author Jan-Petter Gundersen ([email protected])
 *
 * Project:
 * @link https://github.com/VIPnytt/X-Robots-Tag-parser
 * @license https://opensource.org/licenses/MIT MIT license
 *
 * Specification:
 * @link https://developers.google.com/webmasters/control-crawl-index/docs/robots_meta_tag#using-the-x-robots-tag-http-header
 */

namespace vipnytt;

use vipnytt\XRobotsTagParser\directive;
use vipnytt\XRobotsTagParser\URLParser;
use vipnytt\XRobotsTagParser\UserAgentParser;

class XRobotsTagParser
{
    const HEADER_RULE_IDENTIFIER = 'x-robots-tag';
    const USERAGENT_DEFAULT = '';

    const DIRECTIVE_ALL = 'all';
    const DIRECTIVE_NONE = 'none';
    const DIRECTIVE_NO_ARCHIVE = 'noarchive';
    const DIRECTIVE_NO_FOLLOW = 'nofollow';
    const DIRECTIVE_NO_IMAGE_INDEX = 'noimageindex';
    const DIRECTIVE_NO_INDEX = 'noindex';
    const DIRECTIVE_NO_ODP = 'noodp';
    const DIRECTIVE_NO_SNIPPET = 'nosnippet';
    const DIRECTIVE_NO_TRANSLATE = 'notranslate';
    const DIRECTIVE_UNAVAILABLE_AFTER = 'unavailable_after';

    private $strict = false;


    private $url = '';
    private $userAgent = self::USERAGENT_DEFAULT;

    private $headers = [];
    private $currentRule = '';
    private $currentUserAgent = self::USERAGENT_DEFAULT;
    private $currentDirective = '';
    private $currentValue = '';

    private $rules = [];

    /**
     * Constructor
     *
     * @param string $url
     * @param string $userAgent
     * @param array|null $headers
     */
    public function __construct($url, $userAgent = self::USERAGENT_DEFAULT, $headers = null)
    {
        // Parse URL
        $urlParser = new URLParser(trim($url));
        if (!$urlParser->isValid()) {
            trigger_error('Invalid URL', E_USER_WARNING);
        }
        $this->url = $urlParser->encode();
        // Get headers
        $this->useHeaders($headers);
        // Parse rules
        $this->parse();
        // Set User-Agent
        $parser = new UserAgentParser($userAgent);
        $this->userAgent = $parser->match(array_keys($this->rules), self::USERAGENT_DEFAULT);
    }

    /**
     * Request HTTP headers
     *
     * @param array|null|false $customHeaders - use these headers
     * @return bool
     */
    private function useHeaders($customHeaders = null)
    {
        if ($customHeaders === false) {
            trigger_error('Unable to fetch HTTP headers', E_USER_ERROR);
            return false;
        } elseif (!is_array($customHeaders) || empty($customHeaders)) {
            return $this->useHeaders(get_headers($this->url));
        }
        $this->headers = $customHeaders;
        return true;
    }

    /**
     * Parse HTTP headers
     *
     * @return void
     */
    private function parse()
    {
        foreach ($this->headers as $header) {
            $parts = explode(':', mb_strtolower($header), 2);
            if (count($parts) < 2 || $parts[0] != self::HEADER_RULE_IDENTIFIER) {
                // Header is not a rule
                continue;
            }
            $this->currentRule = trim($parts[1]);
            $this->detectDirectives();
        }

    }

    /**
     * Detect directives in rule
     *
     * @return void
     */
    private function detectDirectives()
    {
        $rules = explode(',', $this->currentRule);
        foreach ($rules as $rule) {
            $pair = array_map('trim', explode(':', $rule, 2));
            if ($rules[0] === $rule && count($pair) == 2 && !in_array($pair[0], $this->directiveArray())) {
                $this->currentUserAgent = $pair[0];
                $pair = array_map('trim', explode(':', $pair[1], 2));
            }
            if (in_array($pair[0], $this->directiveArray())) {
                $this->currentDirective = $pair[0];
                $this->currentValue = isset($pair[1]) ? $pair[1] : '';
                $this->addRule();
            }
        }
        $this->cleanup();
    }

    /**
     * Directives supported
     *
     * @return array
     */
    protected function directiveArray()
    {
        return [
            self::DIRECTIVE_ALL,
            self::DIRECTIVE_NONE,
            self::DIRECTIVE_NO_ARCHIVE,
            self::DIRECTIVE_NO_FOLLOW,
            self::DIRECTIVE_NO_IMAGE_INDEX,
            self::DIRECTIVE_NO_INDEX,
            self::DIRECTIVE_NO_ODP,
            self::DIRECTIVE_NO_SNIPPET,
            self::DIRECTIVE_NO_TRANSLATE,
            self::DIRECTIVE_UNAVAILABLE_AFTER
        ];
    }

    /**
     * Add rule
     *
     * @return void
     */
    private function addRule()
    {
        if (!isset($this->rules[$this->currentUserAgent])) {
            $this->rules[$this->currentUserAgent] = [];
        }
        $directive = new directive($this->currentDirective, $this->currentValue);
        $this->rules[$this->currentUserAgent] = array_merge($this->rules[$this->currentUserAgent], $directive->getArray());
    }

    /**
     * Cleanup before next rule is read
     *
     * @return void
     */
    private function cleanup()
    {
        $this->currentRule = '';
        $this->currentUserAgent = self::USERAGENT_DEFAULT;
        $this->currentDirective = '';
        $this->currentValue = '';
    }

    /**
     * Return all applicable rules
     *
     * @return array
     */
    public function getRules()
    {
        $rules = [];
        // Default UserAgent
        if (isset($this->rules[self::USERAGENT_DEFAULT])) {
            $rules = array_merge($rules, $this->rules[self::USERAGENT_DEFAULT]);
        }
        // Matching UserAgent
        if (isset($this->rules[$this->userAgent])) {
            $rules = array_merge($rules, $this->rules[$this->userAgent]);
        }
        // Result
        return $rules;
    }

    /**
     * Export all rules for all UserAgents
     *
     * @return array
     */
    public function export()
    {
        return $this->rules;
    }
}

1			<?php
2			/**
3			* X-Robots-Tag HTTP header parser class
4			*
5			* @author VIP nytt ([email protected])
6			* @author Jan-Petter Gundersen ([email protected])
7			*
8			* Project:
9			* @link https://github.com/VIPnytt/X-Robots-Tag-parser
10			* @license https://opensource.org/licenses/MIT MIT license
11			*
12			* Specification:
13			* @link https://developers.google.com/webmasters/control-crawl-index/docs/robots_meta_tag#using-the-x-robots-tag-http-header
14			*/
15
16			namespace vipnytt;
17
18			use vipnytt\XRobotsTagParser\directive;
19			use vipnytt\XRobotsTagParser\URLParser;
20			use vipnytt\XRobotsTagParser\UserAgentParser;
21
22			class XRobotsTagParser
23			{
24			const HEADER_RULE_IDENTIFIER = 'x-robots-tag';
25			const USERAGENT_DEFAULT = '';
26
27			const DIRECTIVE_ALL = 'all';
28			const DIRECTIVE_NONE = 'none';
29			const DIRECTIVE_NO_ARCHIVE = 'noarchive';
30			const DIRECTIVE_NO_FOLLOW = 'nofollow';
31			const DIRECTIVE_NO_IMAGE_INDEX = 'noimageindex';
32			const DIRECTIVE_NO_INDEX = 'noindex';
33			const DIRECTIVE_NO_ODP = 'noodp';
34			const DIRECTIVE_NO_SNIPPET = 'nosnippet';
35			const DIRECTIVE_NO_TRANSLATE = 'notranslate';
36			const DIRECTIVE_UNAVAILABLE_AFTER = 'unavailable_after';
37
38			private $strict = false;
			0 ignored issues – show Unused Code introduced 2016-02-08 23:49 UTC by Report Bug Copy Issue Report The property `$strict` is not used and could be removed. This check marks private properties in classes that are never used. Those properties can be removed. Loading history...
39
40			private $url = '';
41			private $userAgent = self::USERAGENT_DEFAULT;
42
43			private $headers = [];
44			private $currentRule = '';
45			private $currentUserAgent = self::USERAGENT_DEFAULT;
46			private $currentDirective = '';
47			private $currentValue = '';
48
49			private $rules = [];
50
51			/**
52			* Constructor
53			*
54			* @param string $url
55			* @param string $userAgent
56			* @param array\|null $headers
57			*/
58			public function __construct($url, $userAgent = self::USERAGENT_DEFAULT, $headers = null)
59			{
60			// Parse URL
61			$urlParser = new URLParser(trim($url));
62			if (!$urlParser->isValid()) {
63			trigger_error('Invalid URL', E_USER_WARNING);
64			}
65			$this->url = $urlParser->encode();
66			// Get headers
67			$this->useHeaders($headers);
68			// Parse rules
69			$this->parse();
70			// Set User-Agent
71			$parser = new UserAgentParser($userAgent);
72			$this->userAgent = $parser->match(array_keys($this->rules), self::USERAGENT_DEFAULT);
73			}
74
75			/**
76			* Request HTTP headers
77			*
78			* @param array\|null\|false $customHeaders - use these headers
79			* @return bool
80			*/
81			private function useHeaders($customHeaders = null)
82			{
83			if ($customHeaders === false) {
84			trigger_error('Unable to fetch HTTP headers', E_USER_ERROR);
85			return false;
86			} elseif (!is_array($customHeaders) \|\| empty($customHeaders)) {
87			return $this->useHeaders(get_headers($this->url));
88			}
89			$this->headers = $customHeaders;
90			return true;
91			}
92
93			/**
94			* Parse HTTP headers
95			*
96			* @return void
97			*/
98			private function parse()
99			{
100			foreach ($this->headers as $header) {
101			$parts = explode(':', mb_strtolower($header), 2);
102			if (count($parts) < 2 \|\| $parts[0] != self::HEADER_RULE_IDENTIFIER) {
103			// Header is not a rule
104			continue;
105			}
106			$this->currentRule = trim($parts[1]);
107			$this->detectDirectives();
108			}
109
110			}
111
112			/**
113			* Detect directives in rule
114			*
115			* @return void
116			*/
117			private function detectDirectives()
118			{
119			$rules = explode(',', $this->currentRule);
120			foreach ($rules as $rule) {
121			$pair = array_map('trim', explode(':', $rule, 2));
122			if ($rules[0] === $rule && count($pair) == 2 && !in_array($pair[0], $this->directiveArray())) {
123			$this->currentUserAgent = $pair[0];
124			$pair = array_map('trim', explode(':', $pair[1], 2));
125			}
126			if (in_array($pair[0], $this->directiveArray())) {
127			$this->currentDirective = $pair[0];
128			$this->currentValue = isset($pair[1]) ? $pair[1] : '';
129			$this->addRule();
130			}
131			}
132			$this->cleanup();
133			}
134
135			/**
136			* Directives supported
137			*
138			* @return array
139			*/
140			protected function directiveArray()
141			{
142			return [
143			self::DIRECTIVE_ALL,
144			self::DIRECTIVE_NONE,
145			self::DIRECTIVE_NO_ARCHIVE,
146			self::DIRECTIVE_NO_FOLLOW,
147			self::DIRECTIVE_NO_IMAGE_INDEX,
148			self::DIRECTIVE_NO_INDEX,
149			self::DIRECTIVE_NO_ODP,
150			self::DIRECTIVE_NO_SNIPPET,
151			self::DIRECTIVE_NO_TRANSLATE,
152			self::DIRECTIVE_UNAVAILABLE_AFTER
153			];
154			}
155
156			/**
157			* Add rule
158			*
159			* @return void
160			*/
161			private function addRule()
162			{
163			if (!isset($this->rules[$this->currentUserAgent])) {
164			$this->rules[$this->currentUserAgent] = [];
165			}
166			$directive = new directive($this->currentDirective, $this->currentValue);
167			$this->rules[$this->currentUserAgent] = array_merge($this->rules[$this->currentUserAgent], $directive->getArray());
168			}
169
170			/**
171			* Cleanup before next rule is read
172			*
173			* @return void
174			*/
175			private function cleanup()
176			{
177			$this->currentRule = '';
178			$this->currentUserAgent = self::USERAGENT_DEFAULT;
179			$this->currentDirective = '';
180			$this->currentValue = '';
181			}
182
183			/**
184			* Return all applicable rules
185			*
186			* @return array
187			*/
188			public function getRules()
189			{
190			$rules = [];
191			// Default UserAgent
192			if (isset($this->rules[self::USERAGENT_DEFAULT])) {
193			$rules = array_merge($rules, $this->rules[self::USERAGENT_DEFAULT]);
194			}
195			// Matching UserAgent
196			if (isset($this->rules[$this->userAgent])) {
197			$rules = array_merge($rules, $this->rules[$this->userAgent]);
198			}
199			// Result
200			return $rules;
201			}
202
203			/**
204			* Export all rules for all UserAgents
205			*
206			* @return array
207			*/
208			public function export()
209			{
210			return $this->rules;
211			}
212			}

VIPnytt / RobotsTagParser

Push — master ( 26bda2...b142f3 )

XRobotsTagParser::cleanup() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like