Parser::convertEncoding() - Code Metrics - Inspection of "robots.txt render feature" - VIPnytt/RobotsTxtParser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( a6d8dc...680f8e )

by Jan-Petter

created 2016-05-17 15:03 UTC

Parser::convertEncoding() A

↳ Parent: Parser

Complexity

Conditions	3
Paths	3

Size

Total Lines	12
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	1	Features	0

Metric	Value
c	1
b	1
f	0
dl	0
loc	12
rs	9.4285
cc	3
eloc	8
nc	3
nop	2

<?php
namespace vipnytt\RobotsTxtParser;

use vipnytt\RobotsTxtParser\Parser\CharacterEncodingConvert;
use vipnytt\RobotsTxtParser\Parser\Directives\CleanParam;
use vipnytt\RobotsTxtParser\Parser\Directives\Host;
use vipnytt\RobotsTxtParser\Parser\Directives\Sitemap;
use vipnytt\RobotsTxtParser\Parser\Directives\UserAgent;
use vipnytt\RobotsTxtParser\Parser\RobotsTxtInterface;
use vipnytt\RobotsTxtParser\Parser\Toolbox;

/**
 * Class Parser
 *
 * @package vipnytt\RobotsTxtParser
 */
abstract class Parser implements RobotsTxtInterface
{
    use Toolbox;

    /**
     * Directive white list
     */
    const TOP_LEVEL_DIRECTIVES = [
        self::DIRECTIVE_CLEAN_PARAM,
        self::DIRECTIVE_HOST,
        self::DIRECTIVE_SITEMAP,
        self::DIRECTIVE_USER_AGENT,
    ];

    /**
     * Previous directive
     * @var string
     */
    protected $previousDirective;

    /**
     * Current user-agent(s)
     * @var array
     */
    protected $userAgentValues;

    /**
     * Clean-param class
     * @var CleanParam
     */
    protected $cleanParam;

    /**
     * Host class
     * @var Host
     */
    protected $host;

    /**
     * Sitemap class
     * @var Sitemap
     */
    protected $sitemap;

    /**
     * User-agent class
     * @var UserAgent
     */
    protected $userAgent;

    /**
     * Core constructor.
     *
     * @param string $content - file content
     * @param string $encoding - character encoding
     * @param int|null $byteLimit - maximum of bytes to parse
     */
    public function __construct($content, $encoding = self::ENCODING, $byteLimit = self::BYTE_LIMIT)
    {
        $this->cleanParam = new CleanParam();
        $this->host = new Host();
        $this->sitemap = new Sitemap();
        $this->userAgent = new UserAgent();
        $content = $this->convertEncoding($encoding, $content);
        if (is_int($byteLimit) && $byteLimit > 0) {
            $content = mb_strcut($content, 0, $byteLimit);
        }
        $this->parseTxt($content);
    }

    /**
     * Convert character encoding
     *
     * @param string $encoding
     * @param string $content
     * @return string
     */
    protected function convertEncoding($encoding, $content)
    {
        mb_internal_encoding(self::ENCODING);
        if ($encoding == self::ENCODING) {
            return $content;
        }
        $convert = new CharacterEncodingConvert($content, $encoding, self::ENCODING);
        if (($result = $convert->auto()) !== false) {
            return $result;
        }
        return $content;
    }

    /**
     * Parse robots.txt
     *
     * @param string $txt
     * @return void
     */
    private function parseTxt($txt)
    {
        $lines = array_filter(array_map('trim', mb_split('\r\n|\n|\r', $txt)));
        // Parse each line individually
        foreach ($lines as $line) {
            // Limit rule length
            $line = mb_substr($line, 0, self::MAX_LENGTH_RULE);
            // Remove comments
            $line = mb_split('#', $line, 2)[0];
            // Parse line
            $this->add($line);
        }
    }

    /**
     * Add line
     *
     * @param string $line
     * @return bool
     */
    public function add($line)
    {
        $previousDirective = $this->previousDirective;
        $pair = $this->generateRulePair($line, self::TOP_LEVEL_DIRECTIVES);
        if ($pair['directive'] === self::DIRECTIVE_USER_AGENT) {
            if ($previousDirective !== self::DIRECTIVE_USER_AGENT) {
                $this->userAgentValues = [];
            }
            $this->userAgentValues[] = $pair['value'];
        }
        $this->previousDirective = $pair['directive'];
        switch ($pair['directive']) {
            case self::DIRECTIVE_CLEAN_PARAM:
                return $this->cleanParam->add($pair['value']);
            case self::DIRECTIVE_HOST:
                return $this->host->add($pair['value']);
            case self::DIRECTIVE_SITEMAP:
                return $this->sitemap->add($pair['value']);
            case self::DIRECTIVE_USER_AGENT:
                return $this->userAgent->set($this->userAgentValues);
        }
        return $this->userAgent->add($line);
    }

    /**
     * Render
     *
     * @return string
     */
    public function render()
    {
        return implode("\r\n", array_merge(
            $this->cleanParam->render(),
            $this->host->render(),
            $this->sitemap->render(),
            $this->userAgent->render()
        ));
    }

    /**
     * Export rules
     *
     * @return array
     */
    public function export()
    {
        return array_merge(
            $this->cleanParam->export(),
            $this->host->export(),
            $this->sitemap->export(),
            $this->userAgent->export()
        );
    }
}


1			<?php
2			namespace vipnytt\RobotsTxtParser;
3
4			use vipnytt\RobotsTxtParser\Parser\CharacterEncodingConvert;
5			use vipnytt\RobotsTxtParser\Parser\Directives\CleanParam;
6			use vipnytt\RobotsTxtParser\Parser\Directives\Host;
7			use vipnytt\RobotsTxtParser\Parser\Directives\Sitemap;
8			use vipnytt\RobotsTxtParser\Parser\Directives\UserAgent;
9			use vipnytt\RobotsTxtParser\Parser\RobotsTxtInterface;
10			use vipnytt\RobotsTxtParser\Parser\Toolbox;
11
12			/**
13			* Class Parser
14			*
15			* @package vipnytt\RobotsTxtParser
16			*/
17			abstract class Parser implements RobotsTxtInterface
18			{
19			use Toolbox;
20
21			/**
22			* Directive white list
23			*/
24			const TOP_LEVEL_DIRECTIVES = [
25			self::DIRECTIVE_CLEAN_PARAM,
26			self::DIRECTIVE_HOST,
27			self::DIRECTIVE_SITEMAP,
28			self::DIRECTIVE_USER_AGENT,
29			];
30
31			/**
32			* Previous directive
33			* @var string
34			*/
35			protected $previousDirective;
36
37			/**
38			* Current user-agent(s)
39			* @var array
40			*/
41			protected $userAgentValues;
42
43			/**
44			* Clean-param class
45			* @var CleanParam
46			*/
47			protected $cleanParam;
48
49			/**
50			* Host class
51			* @var Host
52			*/
53			protected $host;
54
55			/**
56			* Sitemap class
57			* @var Sitemap
58			*/
59			protected $sitemap;
60
61			/**
62			* User-agent class
63			* @var UserAgent
64			*/
65			protected $userAgent;
66
67			/**
68			* Core constructor.
69			*
70			* @param string $content - file content
71			* @param string $encoding - character encoding
72			* @param int\|null $byteLimit - maximum of bytes to parse
73			*/
74			public function __construct($content, $encoding = self::ENCODING, $byteLimit = self::BYTE_LIMIT)
75			{
76			$this->cleanParam = new CleanParam();
77			$this->host = new Host();
78			$this->sitemap = new Sitemap();
79			$this->userAgent = new UserAgent();
80			$content = $this->convertEncoding($encoding, $content);
81			if (is_int($byteLimit) && $byteLimit > 0) {
82			$content = mb_strcut($content, 0, $byteLimit);
83			}
84			$this->parseTxt($content);
85			}
86
87			/**
88			* Convert character encoding
89			*
90			* @param string $encoding
91			* @param string $content
92			* @return string
93			*/
94			protected function convertEncoding($encoding, $content)
95			{
96			mb_internal_encoding(self::ENCODING);
97			if ($encoding == self::ENCODING) {
98			return $content;
99			}
100			$convert = new CharacterEncodingConvert($content, $encoding, self::ENCODING);
101			if (($result = $convert->auto()) !== false) {
102			return $result;
103			}
104			return $content;
105			}
106
107			/**
108			* Parse robots.txt
109			*
110			* @param string $txt
111			* @return void
112			*/
113			private function parseTxt($txt)
114			{
115			$lines = array_filter(array_map('trim', mb_split('\r\n\|\n\|\r', $txt)));
116			// Parse each line individually
117			foreach ($lines as $line) {
118			// Limit rule length
119			$line = mb_substr($line, 0, self::MAX_LENGTH_RULE);
120			// Remove comments
121			$line = mb_split('#', $line, 2)[0];
122			// Parse line
123			$this->add($line);
124			}
125			}
126
127			/**
128			* Add line
129			*
130			* @param string $line
131			* @return bool
132			*/
133			public function add($line)
134			{
135			$previousDirective = $this->previousDirective;
136			$pair = $this->generateRulePair($line, self::TOP_LEVEL_DIRECTIVES);
137			if ($pair['directive'] === self::DIRECTIVE_USER_AGENT) {
138			if ($previousDirective !== self::DIRECTIVE_USER_AGENT) {
139			$this->userAgentValues = [];
140			}
141			$this->userAgentValues[] = $pair['value'];
142			}
143			$this->previousDirective = $pair['directive'];
144			switch ($pair['directive']) {
145			case self::DIRECTIVE_CLEAN_PARAM:
146			return $this->cleanParam->add($pair['value']);
147			case self::DIRECTIVE_HOST:
148			return $this->host->add($pair['value']);
149			case self::DIRECTIVE_SITEMAP:
150			return $this->sitemap->add($pair['value']);
151			case self::DIRECTIVE_USER_AGENT:
152			return $this->userAgent->set($this->userAgentValues);
153			}
154			return $this->userAgent->add($line);
155			}
156
157			/**
158			* Render
159			*
160			* @return string
161			*/
162			public function render()
163			{
164			return implode("\r\n", array_merge(
165			$this->cleanParam->render(),
166			$this->host->render(),
167			$this->sitemap->render(),
168			$this->userAgent->render()
169			));
170			}
171
172			/**
173			* Export rules
174			*
175			* @return array
176			*/
177			public function export()
178			{
179			return array_merge(
180			$this->cleanParam->export(),
181			$this->host->export(),
182			$this->sitemap->export(),
183			$this->userAgent->export()
184			);
185			}
186			}
187

VIPnytt / RobotsTxtParser

Push — master ( a6d8dc...680f8e )

Parser::convertEncoding() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like