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

Completed

Push — master ( 2c8763...82b108 )

by Jan-Petter

created 2016-04-08 01:09 UTC

XRobotsTagParser::parse() A

↳ Parent: XRobotsTagParser

Complexity

Conditions	4
Paths	3

Size

Total Lines	13
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes	14
Bugs	4	Features	0

Metric	Value
c	14
b	4
f	0
dl	0
loc	13
rs	9.2
cc	4
eloc	8
nc	3
nop	1

<?php
namespace vipnytt;

/**
 * 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
 */

use vipnytt\XRobotsTagParser\Exceptions\XRobotsTagParserException;
use vipnytt\XRobotsTagParser\Rebuild;
use vipnytt\XRobotsTagParser\UserAgentParser;

class XRobotsTagParser
{
    const HEADER_RULE_IDENTIFIER = 'X-Robots-Tag';

    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';

    protected $userAgent = '';
    protected $userAgentMatch = '';

    protected $currentRule = '';
    protected $currentUserAgent;

    protected $rules = [];

    /**
     * Constructor
     *
     * @param string $userAgent
     * @param array $headers
     */
    public function __construct($userAgent = '', $headers = null)
    {
        $this->userAgent = $userAgent;
        if (isset($headers)) {
            $this->parse($headers);
        }
    }

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

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

    /**
     * Array of directives and their class names
     *
     * @return array
     */
    protected function directiveClasses()
    {
        return [
            self::DIRECTIVE_ALL => 'All',
            self::DIRECTIVE_NO_ARCHIVE => 'NoArchive',
            self::DIRECTIVE_NO_FOLLOW => 'NoFollow',
            self::DIRECTIVE_NO_IMAGE_INDEX => 'NoImageIndex',
            self::DIRECTIVE_NO_INDEX => 'NoIndex',
            self::DIRECTIVE_NONE => 'None',
            self::DIRECTIVE_NO_ODP => 'NoODP',
            self::DIRECTIVE_NO_SNIPPET => 'NoSnippet',
            self::DIRECTIVE_NO_TRANSLATE => 'NoTranslate',
            self::DIRECTIVE_UNAVAILABLE_AFTER => 'UnavailableAfter',
        ];
    }

    /**
     * Add rule
     *
     * @param string $directive
     * @return void
     * @throws XRobotsTagParserException
     */
    protected function addRule($directive)
    {
        if (!isset($this->rules[$this->currentUserAgent])) {
            $this->rules[$this->currentUserAgent] = [];
        }
        $class = "\\" . __CLASS__ . "\\directives\\" . $this->directiveClasses()[$directive];
        $object = new $class($this->currentRule);
        if (!$object instanceof XRobotsTagParser\directives\DirectiveInterface) {
            throw new XRobotsTagParserException('Unsupported directive class');
        }
        $this->rules[$this->currentUserAgent] = array_merge($this->rules[$this->currentUserAgent], [$object->getDirective() => $object->getValue()]);
    }

    /**
     * Cleanup before next rule is read
     *
     * @return void
     */
    protected function cleanup()
    {
        $this->currentRule = '';
        $this->currentUserAgent = '';
    }

    /**
     * Find the most rule-matching User-Agent
     *
     * @return string
     */
    protected function matchUserAgent()
    {
        $userAgentParser = new UserAgentParser($this->userAgent);
        $match = $userAgentParser->match(array_keys($this->rules), '');
        $this->userAgentMatch = (is_string($match)) ? $match : '';
        return $this->userAgentMatch;
    }

    /**
     * Return all applicable rules
     *
     * @param bool $raw
     * @return array
     */
    public function getRules($raw = false)
    {
        $rules = [];
        // Default UserAgent
        if (isset($this->rules[''])) {
            $rules = array_merge($rules, $this->rules['']);
        }
        // Matching UserAgent
        if (isset($this->rules[$this->userAgentMatch])) {
            $rules = array_merge($rules, $this->rules[$this->userAgentMatch]);
        }
        if (!$raw) {
            $rebuild = new Rebuild($rules);
            $rules = $rebuild->getResult();
        }
        // Result
        return $rules;
    }

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

    /**
     * Get the meaning of an Directive
     *
     * @param string $directive
     * @return string
     * @throws XRobotsTagParserException
     */
    public function getDirectiveMeaning($directive)
    {
        if (!in_array($directive, array_keys($this->directiveClasses()))) {
            throw new XRobotsTagParserException('Unknown directive');
        }
        $class = "\\" . __CLASS__ . "\\directives\\" . $this->directiveClasses()[$directive];
        $object = new $class($this->directiveClasses()[$directive]);
        if (!$object instanceof XRobotsTagParser\directives\DirectiveInterface) {
            throw new XRobotsTagParserException('Unsupported directive class');
        }
        return $object->getMeaning();
    }
}


1			<?php
2			namespace vipnytt;
3
4			/**
5			* X-Robots-Tag HTTP header parser class
6			*
7			* @author VIP nytt ([email protected])
8			* @author Jan-Petter Gundersen ([email protected])
9			*
10			* Project:
11			* @link https://github.com/VIPnytt/X-Robots-Tag-parser
12			* @license https://opensource.org/licenses/MIT MIT license
13			*
14			* Specification:
15			* @link https://developers.google.com/webmasters/control-crawl-index/docs/robots_meta_tag#using-the-x-robots-tag-http-header
16			*/
17
18			use vipnytt\XRobotsTagParser\Exceptions\XRobotsTagParserException;
19			use vipnytt\XRobotsTagParser\Rebuild;
20			use vipnytt\XRobotsTagParser\UserAgentParser;
21
22			class XRobotsTagParser
23			{
24			const HEADER_RULE_IDENTIFIER = 'X-Robots-Tag';
25
26			const DIRECTIVE_ALL = 'all';
27			const DIRECTIVE_NONE = 'none';
28			const DIRECTIVE_NO_ARCHIVE = 'noarchive';
29			const DIRECTIVE_NO_FOLLOW = 'nofollow';
30			const DIRECTIVE_NO_IMAGE_INDEX = 'noimageindex';
31			const DIRECTIVE_NO_INDEX = 'noindex';
32			const DIRECTIVE_NO_ODP = 'noodp';
33			const DIRECTIVE_NO_SNIPPET = 'nosnippet';
34			const DIRECTIVE_NO_TRANSLATE = 'notranslate';
35			const DIRECTIVE_UNAVAILABLE_AFTER = 'unavailable_after';
36
37			protected $userAgent = '';
38			protected $userAgentMatch = '';
39
40			protected $currentRule = '';
41			protected $currentUserAgent;
42
43			protected $rules = [];
44
45			/**
46			* Constructor
47			*
48			* @param string $userAgent
49			* @param array $headers
50			*/
51			public function __construct($userAgent = '', $headers = null)
52			{
53			$this->userAgent = $userAgent;
54			if (isset($headers)) {
55			$this->parse($headers);
56			}
57			}
58
59			/**
60			* Parse HTTP headers
61			*
62			* @param array $headers
63			* @return void
64			*/
65			public function parse(array $headers)
66			{
67			foreach ($headers as $header) {
68			$parts = array_map('trim', explode(':', mb_strtolower($header), 2));
69			if (count($parts) < 2 \|\| $parts[0] != mb_strtolower(self::HEADER_RULE_IDENTIFIER)) {
70			// Header is not a rule
71			continue;
72			}
73			$this->currentRule = $parts[1];
74			$this->detectDirectives();
75			}
76			$this->matchUserAgent();
77			}
78
79			/**
80			* Detect directives in rule
81			*
82			* @return void
83			*/
84			protected function detectDirectives()
85			{
86			$directives = array_map('trim', explode(',', $this->currentRule));
87			$pair = array_map('trim', explode(':', $directives[0], 2));
88			if (count($pair) == 2 && !in_array($pair[0], array_keys($this->directiveClasses()))) {
89			$this->currentUserAgent = $pair[0];
90			$directives[0] = $pair[1];
91			}
92			foreach ($directives as $rule) {
93			$directive = trim(explode(':', $rule, 2)[0]);
94			if (in_array($directive, array_keys($this->directiveClasses()))) {
95			$this->addRule($directive);
96			}
97			}
98			$this->cleanup();
99			}
100
101			/**
102			* Array of directives and their class names
103			*
104			* @return array
105			*/
106			protected function directiveClasses()
107			{
108			return [
109			self::DIRECTIVE_ALL => 'All',
110			self::DIRECTIVE_NO_ARCHIVE => 'NoArchive',
111			self::DIRECTIVE_NO_FOLLOW => 'NoFollow',
112			self::DIRECTIVE_NO_IMAGE_INDEX => 'NoImageIndex',
113			self::DIRECTIVE_NO_INDEX => 'NoIndex',
114			self::DIRECTIVE_NONE => 'None',
115			self::DIRECTIVE_NO_ODP => 'NoODP',
116			self::DIRECTIVE_NO_SNIPPET => 'NoSnippet',
117			self::DIRECTIVE_NO_TRANSLATE => 'NoTranslate',
118			self::DIRECTIVE_UNAVAILABLE_AFTER => 'UnavailableAfter',
119			];
120			}
121
122			/**
123			* Add rule
124			*
125			* @param string $directive
126			* @return void
127			* @throws XRobotsTagParserException
128			*/
129			protected function addRule($directive)
130			{
131			if (!isset($this->rules[$this->currentUserAgent])) {
132			$this->rules[$this->currentUserAgent] = [];
133			}
134			$class = "\\" . __CLASS__ . "\\directives\\" . $this->directiveClasses()[$directive];
135			$object = new $class($this->currentRule);
136			if (!$object instanceof XRobotsTagParser\directives\DirectiveInterface) {
137			throw new XRobotsTagParserException('Unsupported directive class');
138			}
139			$this->rules[$this->currentUserAgent] = array_merge($this->rules[$this->currentUserAgent], [$object->getDirective() => $object->getValue()]);
140			}
141
142			/**
143			* Cleanup before next rule is read
144			*
145			* @return void
146			*/
147			protected function cleanup()
148			{
149			$this->currentRule = '';
150			$this->currentUserAgent = '';
151			}
152
153			/**
154			* Find the most rule-matching User-Agent
155			*
156			* @return string
157			*/
158			protected function matchUserAgent()
159			{
160			$userAgentParser = new UserAgentParser($this->userAgent);
161			$match = $userAgentParser->match(array_keys($this->rules), '');
162			$this->userAgentMatch = (is_string($match)) ? $match : '';
163			return $this->userAgentMatch;
164			}
165
166			/**
167			* Return all applicable rules
168			*
169			* @param bool $raw
170			* @return array
171			*/
172			public function getRules($raw = false)
173			{
174			$rules = [];
175			// Default UserAgent
176			if (isset($this->rules[''])) {
177			$rules = array_merge($rules, $this->rules['']);
178			}
179			// Matching UserAgent
180			if (isset($this->rules[$this->userAgentMatch])) {
181			$rules = array_merge($rules, $this->rules[$this->userAgentMatch]);
182			}
183			if (!$raw) {
184			$rebuild = new Rebuild($rules);
185			$rules = $rebuild->getResult();
186			}
187			// Result
188			return $rules;
189			}
190
191			/**
192			* Export all rules for all UserAgents
193			*
194			* @return array
195			*/
196			public function export()
197			{
198			return $this->rules;
199			}
200
201			/**
202			* Get the meaning of an Directive
203			*
204			* @param string $directive
205			* @return string
206			* @throws XRobotsTagParserException
207			*/
208			public function getDirectiveMeaning($directive)
209			{
210			if (!in_array($directive, array_keys($this->directiveClasses()))) {
211			throw new XRobotsTagParserException('Unknown directive');
212			}
213			$class = "\\" . __CLASS__ . "\\directives\\" . $this->directiveClasses()[$directive];
214			$object = new $class($this->directiveClasses()[$directive]);
215			if (!$object instanceof XRobotsTagParser\directives\DirectiveInterface) {
216			throw new XRobotsTagParserException('Unsupported directive class');
217			}
218			return $object->getMeaning();
219			}
220			}
221

VIPnytt / RobotsTagParser

Push — master ( 2c8763...82b108 )

XRobotsTagParser::parse() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like