XRobotsTagParser::detectDirectives() - Code Metrics - Inspection of "User-Agent and GuzzleHttp bug fixes" - VIPnytt/RobotsTagParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 1084b1...9b56e4 )

by Jan-Petter

created 2016-04-06 16:55 UTC

XRobotsTagParser::detectDirectives() B

↳ Parent: XRobotsTagParser

Complexity

Conditions	5
Paths	6

Size

Total Lines	16
Code Lines	11

Duplication

Lines	0
Ratio	0 %

Importance

Changes	10
Bugs	2	Features	0

Metric	Value
c	10
b	2
f	0
dl	0
loc	16
rs	8.8571
cc	5
eloc	11
nc	6
nop	0

<?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 GuzzleHttp;
use vipnytt\XRobotsTagParser\Exceptions\XRobotsTagParserException;
use vipnytt\XRobotsTagParser\Rebuild;
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';

    protected $url = '';
    protected $userAgent = self::USERAGENT_DEFAULT;
    protected $userAgentMatch = self::USERAGENT_DEFAULT;
    protected $config = [];

    protected $headers = [];
    protected $currentRule = '';
    protected $currentUserAgent = self::USERAGENT_DEFAULT;

    protected $rules = [];

    /**
     * Constructor
     *
     * @param string $url
     * @param string $userAgent
     * @param array $config
     * @throws XRobotsTagParserException
     */
    public function __construct($url, $userAgent = self::USERAGENT_DEFAULT, array $config = [])
    {
        $this->url = $url;
        if (!filter_var($this->url, FILTER_VALIDATE_URL)) {
            throw new XRobotsTagParserException('Invalid URL provided');
        }
        // User-Agent for HTTP request
        $this->userAgent = $userAgent;
        // Set any optional configuration options
        $this->config = $config;
        // Parse rules
        $this->parse();
        // User-Agent matching rules
        $parser = new UserAgentParser($this->userAgent);
        $this->userAgentMatch = $parser->match(array_keys($this->rules), self::USERAGENT_DEFAULT);
    }

    /**
     * Parse HTTP headers
     *
     * @return void
     */
    protected function parse()
    {
        $this->headers = $this->selectHeaderSource();
        foreach ($this->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();
        }
    }

    /**
     * Select HTTP header source
     *
     * @return array
     */
    protected function selectHeaderSource()
    {
        if (isset($this->config['headers']) && is_array($this->config['headers'])) {
            return $this->config['headers'];
        }
        // No provided HTTP headers
        return $this->getHeaders();
    }

    /**
     * Request the HTTP headers from an URL
     *
     * @return array Raw HTTP headers
     * @throws XRobotsTagParserException
     */
    protected function getHeaders()
    {
        try {
            if (!isset($this->config['guzzle']['headers']['User-Agent'])) {
                $this->config['guzzle']['headers']['User-Agent'] = $this->userAgent;
            }
            $client = new GuzzleHttp\Client();
            $res = $client->head($this->url, $this->config['guzzle']);
            $headers = [];
            foreach ($res->getHeader(self::HEADER_RULE_IDENTIFIER) as $name => $values) {
                $headers[] = $name . ': ' . implode(' ', $values) . "\r\n";
            }
            return $headers;
        } catch (GuzzleHttp\Exception\TransferException $e) {
            throw new XRobotsTagParserException($e->getMessage());
        }
    }

    /**
     * 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 = self::USERAGENT_DEFAULT;
    }

    /**
     * Return all applicable rules
     *
     * @param bool $raw
     * @return array
     */
    public function getRules($raw = false)
    {
        $rules = [];
        // Default UserAgent
        if (isset($this->rules[self::USERAGENT_DEFAULT])) {
            $rules = array_merge($rules, $this->rules[self::USERAGENT_DEFAULT]);
        }
        // Matching UserAgent
        if (isset($this->rules[$this->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 GuzzleHttp;
19			use vipnytt\XRobotsTagParser\Exceptions\XRobotsTagParserException;
20			use vipnytt\XRobotsTagParser\Rebuild;
21			use vipnytt\XRobotsTagParser\UserAgentParser;
22
23			class XRobotsTagParser
24			{
25			const HEADER_RULE_IDENTIFIER = 'X-Robots-Tag';
26			const USERAGENT_DEFAULT = '';
27
28			const DIRECTIVE_ALL = 'all';
29			const DIRECTIVE_NONE = 'none';
30			const DIRECTIVE_NO_ARCHIVE = 'noarchive';
31			const DIRECTIVE_NO_FOLLOW = 'nofollow';
32			const DIRECTIVE_NO_IMAGE_INDEX = 'noimageindex';
33			const DIRECTIVE_NO_INDEX = 'noindex';
34			const DIRECTIVE_NO_ODP = 'noodp';
35			const DIRECTIVE_NO_SNIPPET = 'nosnippet';
36			const DIRECTIVE_NO_TRANSLATE = 'notranslate';
37			const DIRECTIVE_UNAVAILABLE_AFTER = 'unavailable_after';
38
39			protected $url = '';
40			protected $userAgent = self::USERAGENT_DEFAULT;
41			protected $userAgentMatch = self::USERAGENT_DEFAULT;
42			protected $config = [];
43
44			protected $headers = [];
45			protected $currentRule = '';
46			protected $currentUserAgent = self::USERAGENT_DEFAULT;
47
48			protected $rules = [];
49
50			/**
51			* Constructor
52			*
53			* @param string $url
54			* @param string $userAgent
55			* @param array $config
56			* @throws XRobotsTagParserException
57			*/
58			public function __construct($url, $userAgent = self::USERAGENT_DEFAULT, array $config = [])
59			{
60			$this->url = $url;
61			if (!filter_var($this->url, FILTER_VALIDATE_URL)) {
62			throw new XRobotsTagParserException('Invalid URL provided');
63			}
64			// User-Agent for HTTP request
65			$this->userAgent = $userAgent;
66			// Set any optional configuration options
67			$this->config = $config;
68			// Parse rules
69			$this->parse();
70			// User-Agent matching rules
71			$parser = new UserAgentParser($this->userAgent);
72			$this->userAgentMatch = $parser->match(array_keys($this->rules), self::USERAGENT_DEFAULT);
73			}
74
75			/**
76			* Parse HTTP headers
77			*
78			* @return void
79			*/
80			protected function parse()
81			{
82			$this->headers = $this->selectHeaderSource();
83			foreach ($this->headers as $header) {
84			$parts = array_map('trim', explode(':', mb_strtolower($header), 2));
85			if (count($parts) < 2 \|\| $parts[0] != mb_strtolower(self::HEADER_RULE_IDENTIFIER)) {
86			// Header is not a rule
87			continue;
88			}
89			$this->currentRule = $parts[1];
90			$this->detectDirectives();
91			}
92			}
93
94			/**
95			* Select HTTP header source
96			*
97			* @return array
98			*/
99			protected function selectHeaderSource()
100			{
101			if (isset($this->config['headers']) && is_array($this->config['headers'])) {
102			return $this->config['headers'];
103			}
104			// No provided HTTP headers
105			return $this->getHeaders();
106			}
107
108			/**
109			* Request the HTTP headers from an URL
110			*
111			* @return array Raw HTTP headers
112			* @throws XRobotsTagParserException
113			*/
114			protected function getHeaders()
115			{
116			try {
117			if (!isset($this->config['guzzle']['headers']['User-Agent'])) {
118			$this->config['guzzle']['headers']['User-Agent'] = $this->userAgent;
119			}
120			$client = new GuzzleHttp\Client();
121			$res = $client->head($this->url, $this->config['guzzle']);
122			$headers = [];
123			foreach ($res->getHeader(self::HEADER_RULE_IDENTIFIER) as $name => $values) {
124			$headers[] = $name . ': ' . implode(' ', $values) . "\r\n";
125			}
126			return $headers;
127			} catch (GuzzleHttp\Exception\TransferException $e) {
128			throw new XRobotsTagParserException($e->getMessage());
129			}
130			}
131
132			/**
133			* Detect directives in rule
134			*
135			* @return void
136			*/
137			protected function detectDirectives()
138			{
139			$directives = array_map('trim', explode(',', $this->currentRule));
140			$pair = array_map('trim', explode(':', $directives[0], 2));
141			if (count($pair) == 2 && !in_array($pair[0], array_keys($this->directiveClasses()))) {
142			$this->currentUserAgent = $pair[0];
143			$directives[0] = $pair[1];
144			}
145			foreach ($directives as $rule) {
146			$directive = trim(explode(':', $rule, 2)[0]);
147			if (in_array($directive, array_keys($this->directiveClasses()))) {
148			$this->addRule($directive);
149			}
150			}
151			$this->cleanup();
152			}
153
154			/**
155			* Array of directives and their class names
156			*
157			* @return array
158			*/
159			protected function directiveClasses()
160			{
161			return [
162			self::DIRECTIVE_ALL => 'All',
163			self::DIRECTIVE_NO_ARCHIVE => 'NoArchive',
164			self::DIRECTIVE_NO_FOLLOW => 'NoFollow',
165			self::DIRECTIVE_NO_IMAGE_INDEX => 'NoImageIndex',
166			self::DIRECTIVE_NO_INDEX => 'NoIndex',
167			self::DIRECTIVE_NONE => 'None',
168			self::DIRECTIVE_NO_ODP => 'NoODP',
169			self::DIRECTIVE_NO_SNIPPET => 'NoSnippet',
170			self::DIRECTIVE_NO_TRANSLATE => 'NoTranslate',
171			self::DIRECTIVE_UNAVAILABLE_AFTER => 'UnavailableAfter',
172			];
173			}
174
175			/**
176			* Add rule
177			*
178			* @param string $directive
179			* @return void
180			* @throws XRobotsTagParserException
181			*/
182			protected function addRule($directive)
183			{
184			if (!isset($this->rules[$this->currentUserAgent])) {
185			$this->rules[$this->currentUserAgent] = [];
186			}
187			$class = "\\" . __CLASS__ . "\\directives\\" . $this->directiveClasses()[$directive];
188			$object = new $class($this->currentRule);
189			if (!$object instanceof XRobotsTagParser\directives\directiveInterface) {
190			throw new XRobotsTagParserException('Unsupported directive class');
191			}
192			$this->rules[$this->currentUserAgent] = array_merge($this->rules[$this->currentUserAgent], [$object->getDirective() => $object->getValue()]);
193			}
194
195			/**
196			* Cleanup before next rule is read
197			*
198			* @return void
199			*/
200			protected function cleanup()
201			{
202			$this->currentRule = '';
203			$this->currentUserAgent = self::USERAGENT_DEFAULT;
204			}
205
206			/**
207			* Return all applicable rules
208			*
209			* @param bool $raw
210			* @return array
211			*/
212			public function getRules($raw = false)
213			{
214			$rules = [];
215			// Default UserAgent
216			if (isset($this->rules[self::USERAGENT_DEFAULT])) {
217			$rules = array_merge($rules, $this->rules[self::USERAGENT_DEFAULT]);
218			}
219			// Matching UserAgent
220			if (isset($this->rules[$this->userAgentMatch])) {
221			$rules = array_merge($rules, $this->rules[$this->userAgentMatch]);
222			}
223			if (!$raw) {
224			$rebuild = new Rebuild($rules);
225			$rules = $rebuild->getResult();
226			}
227			// Result
228			return $rules;
229			}
230
231			/**
232			* Export all rules for all UserAgents
233			*
234			* @return array
235			*/
236			public function export()
237			{
238			return $this->rules;
239			}
240
241			/**
242			* Get the meaning of an Directive
243			*
244			* @param string $directive
245			* @return string
246			* @throws XRobotsTagParserException
247			*/
248			public function getDirectiveMeaning($directive)
249			{
250			if (!in_array($directive, array_keys($this->directiveClasses()))) {
251			throw new XRobotsTagParserException('Unknown directive');
252			}
253			$class = "\\" . __CLASS__ . "\\directives\\" . $this->directiveClasses()[$directive];
254			$object = new $class($this->directiveClasses()[$directive]);
255			if (!$object instanceof XRobotsTagParser\directives\directiveInterface) {
256			throw new XRobotsTagParserException('Unsupported directive class');
257			}
258			return $object->getMeaning();
259			}
260			}
261

VIPnytt / RobotsTagParser

Push — master ( 1084b1...9b56e4 )

XRobotsTagParser::detectDirectives() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like