ExternPage::getPrettyDomainName() - Code Metrics - Inspection of "Style : unused" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Branch — dev (6bb8f6)

by Dispositif

created 2020-04-13 22:33 UTC

ExternPage::getPrettyDomainName() A

↳ Parent: ExternPage

Complexity

Conditions	5
Paths	3

Size

Total Lines	13
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	1

Metric	Value
cc	5
eloc	6
nc	3
nop	0
dl	0
loc	13
rs	9.6111
c	1
b	0
f	1

<?php
/**
 * This file is part of dispositif/wikibot application (@github)
 * 2019/2020 © Philippe M. <[email protected]>
 * For the full copyright and MIT license information, please view the license file.
 */

declare(strict_types=1);


namespace App\Domain;

use App\Application\Http\ExternHttpClient;
use App\Infrastructure\TagParser;
use Exception;
use Psr\Log\LoggerInterface;

/**
 * Représente une page web d'un Lien Externe (hors wiki)
 * Class ExternPage
 *
 * @package App\Domain
 */
class ExternPage
{
    /**
     * @var string
     */
    private $url;

    /**
     * @var string
     */
    private $html;

    /**
     * @var LoggerInterface|null
     */
    private $log;

    /**
     * ExternPage constructor.
     *
     * @param string               $url
     * @param string               $html
     * @param LoggerInterface|null $log
     *
     * @throws Exception
     */
    public function __construct(string $url, string $html, ?LoggerInterface $log = null)
    {
        if (!ExternHttpClient::isWebURL($url)) {
            throw new Exception('string is not an URL');
        }
        $this->url = $url;
        $this->html = $html;
        $this->log = $log;
    }

    /**
     * @return string
     */
    public function getUrl(): string
    {
        return $this->url;
    }

    /**
     * @return array
     * @throws Exception
     */
    public function getData(): array
    {
        $ld = $this->parseLdJson($this->html);
        $meta = $this->parseMetaTags($this->html);

        return ['JSON-LD' => $ld, 'meta' => $meta];
    }

    /**
     * extract LD-JSON metadata from <script type="application/ld+json">.
     *
     * @param string $html
     *
     * @return array
     * @throws Exception
     * @throws Exception
     */
    private function parseLdJson(string $html): array
    {
        $parser = new TagParser();
        $results = $parser->importHtml($html)->xpathResults(
            '//script[@type="application/ld+json"]'
        );

        foreach ($results as $result) {
            $json = trim($result);
            // filtrage empty value (todo?)
            if (0 === strlen($json)) {
                continue;
            }
            $data = json_decode($json, true);
            if (!is_array($data)) {
                return [];
            }

            return $data;
        }

        return [];
    }

    /**
     * todo move? /refac/delete?
     *
     * @param string $str
     *
     * @return array
     */
    private function parseMetaTags(string $str): array
    {
        $pattern = '
              ~<\s*meta\s
              # using lookahead to capture type to $1
                (?=[^>]*?
                \b(?:name|property|http-equiv)\s*=\s*
                (?|"\s*([^"]*?)\s*"|\'\s*([^\']*?)\s*\'|
                ([^"\'>]*?)(?=\s*/?\s*>|\s\w+\s*=))
              )
              # capture content to $2
              [^>]*?\bcontent\s*=\s*
                (?|"\s*([^"]*?)\s*"|\'\s*([^\']*?)\s*\'|
                ([^"\'>]*?)(?=\s*/?\s*>|\s\w+\s*=))
              [^>]*>
              ~ix';

        if (preg_match_all($pattern, $str, $out)) {
            $combine = array_combine($out[1], $out[2]);

            return $combine ? $combine : [];
        }

        return [];
    }

    /**
     * test.com => Test.com
     * bla.test.com => Test.com
     * test.co.uk => google.co.uk (national commercial subdomain)
     * site.google.com => site.google.com ;)
     *
     * @return string
     */
    public function getPrettyDomainName(): string
    {
        $subDomain = $this->getSubDomain();
        if (!strpos($subDomain, '.co.uk') && !strpos($subDomain, '.co.ma')
            && !strpos($subDomain, 'site.google.')
        ) {
            // bla.test.com => Test.com
            if (preg_match('#\w+\.\w+$#', $subDomain, $matches)) {
                return ucfirst($matches[0]);
            }
        }

        return $subDomain;
    }

    public function getSubDomain(): string
    {
        try {
            return ExternDomains::extractSubDomain($this->url);
        } catch (Exception $e) {
            echo "erreur QDF";
interface ReturnsInt {
    public function returnsIntHinted(): int;
}

class MyClass implements ReturnsInt {
    public function returnsIntHinted(): int
    {
        if (foo()) {
            return 123;
        }
        // here: null is implicitly returned
    }
}
        }
    }

}


1			<?php
2			/**
3			* This file is part of dispositif/wikibot application (@github)
4			* 2019/2020 © Philippe M. <[email protected]>
5			* For the full copyright and MIT license information, please view the license file.
6			*/
7
8			declare(strict_types=1);
9
10
11			namespace App\Domain;
12
13			use App\Application\Http\ExternHttpClient;
14			use App\Infrastructure\TagParser;
15			use Exception;
16			use Psr\Log\LoggerInterface;
17
18			/**
19			* Représente une page web d'un Lien Externe (hors wiki)
20			* Class ExternPage
21			*
22			* @package App\Domain
23			*/
24			class ExternPage
25			{
26			/**
27			* @var string
28			*/
29			private $url;
30
31			/**
32			* @var string
33			*/
34			private $html;
35
36			/**
37			* @var LoggerInterface\|null
38			*/
39			private $log;
40
41			/**
42			* ExternPage constructor.
43			*
44			* @param string $url
45			* @param string $html
46			* @param LoggerInterface\|null $log
47			*
48			* @throws Exception
49			*/
50			public function __construct(string $url, string $html, ?LoggerInterface $log = null)
51			{
52			if (!ExternHttpClient::isWebURL($url)) {
53			throw new Exception('string is not an URL');
54			}
55			$this->url = $url;
56			$this->html = $html;
57			$this->log = $log;
58			}
59
60			/**
61			* @return string
62			*/
63			public function getUrl(): string
64			{
65			return $this->url;
66			}
67
68			/**
69			* @return array
70			* @throws Exception
71			*/
72			public function getData(): array
73			{
74			$ld = $this->parseLdJson($this->html);
75			$meta = $this->parseMetaTags($this->html);
76
77			return ['JSON-LD' => $ld, 'meta' => $meta];
78			}
79
80			/**
81			* extract LD-JSON metadata from <script type="application/ld+json">.
82			*
83			* @param string $html
84			*
85			* @return array
86			* @throws Exception
87			* @throws Exception
88			*/
89			private function parseLdJson(string $html): array
90			{
91			$parser = new TagParser();
92			$results = $parser->importHtml($html)->xpathResults(
93			'//script[@type="application/ld+json"]'
94			);
95
96			foreach ($results as $result) {
97			$json = trim($result);
98			// filtrage empty value (todo?)
99			if (0 === strlen($json)) {
100			continue;
101			}
102			$data = json_decode($json, true);
103			if (!is_array($data)) {
104			return [];
105			}
106
107			return $data;
108			}
109
110			return [];
111			}
112
113			/**
114			* todo move? /refac/delete?
115			*
116			* @param string $str
117			*
118			* @return array
119			*/
120			private function parseMetaTags(string $str): array
121			{
122			$pattern = '
123			~<\s*meta\s
124			# using lookahead to capture type to $1
125			(?=[^>]*?
126			\b(?:name\|property\|http-equiv)\s=\s
127			(?\|"\s([^"]?)\s"\|\'\s([^\']?)\s\'\|
128			([^"\'>]?)(?=\s/?\s>\|\s\w+\s=))
129			)
130			# capture content to $2
131			[^>]?\bcontent\s=\s*
132			(?\|"\s([^"]?)\s"\|\'\s([^\']?)\s\'\|
133			([^"\'>]?)(?=\s/?\s>\|\s\w+\s=))
134			[^>]*>
135			~ix';
136
137			if (preg_match_all($pattern, $str, $out)) {
138			$combine = array_combine($out[1], $out[2]);
139
140			return $combine ? $combine : [];
141			}
142
143			return [];
144			}
145
146			/**
147			* test.com => Test.com
148			* bla.test.com => Test.com
149			* test.co.uk => google.co.uk (national commercial subdomain)
150			* site.google.com => site.google.com ;)
151			*
152			* @return string
153			*/
154			public function getPrettyDomainName(): string
155			{
156			$subDomain = $this->getSubDomain();
157			if (!strpos($subDomain, '.co.uk') && !strpos($subDomain, '.co.ma')
158			&& !strpos($subDomain, 'site.google.')
159			) {
160			// bla.test.com => Test.com
161			if (preg_match('#\w+\.\w+$#', $subDomain, $matches)) {
162			return ucfirst($matches[0]);
163			}
164			}
165
166			return $subDomain;
167			}
168
169			public function getSubDomain(): string
170			{
171			try {
172			return ExternDomains::extractSubDomain($this->url);
173			} catch (Exception $e) {
174			echo "erreur QDF";
			0 ignored issues – show Bug Best Practice introduced 2020-04-13 20:21 UTC by Report Bug Copy Issue Report In this branch, the function will implicitly return `null` which is incompatible with the type-hinted return `string`. Consider adding a return statement or allowing `null` as return value. For hinted functions/methods where all return statements with the correct type are only reachable via conditions, ?null? gets implicitly returned which may be incompatible with the hinted type. Let?s take a look at an example: interface ReturnsInt { public function returnsIntHinted(): int; } class MyClass implements ReturnsInt { public function returnsIntHinted(): int { if (foo()) { return 123; } // here: null is implicitly returned } } Loading history...
175			}
176			}
177
178			}
179

Dispositif / Wikibot

Branch — dev (6bb8f6)

ExternPage::getPrettyDomainName() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like