Harvest::getBaseUrl() - Code Metrics - Inspection of "move some code to trait" - PiedWeb/UrlHarvester - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 7e536e...5c7ae7 )

by Dev

created 2019-01-16 12:15 UTC

Harvest::getBaseUrl() A

↳ Parent: Harvest

Complexity

Conditions	5
Paths	3

Size

Total Lines	12
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	5
CRAP Score	5.583

Importance

Changes

Metric	Value
eloc	7
dl	0
loc	12
rs	9.6111
c	0
b	0
f	0
ccs	5
cts	7
cp	0.7143
cc	5
nc	3
nop	0
crap	5.583

<?php

namespace PiedWeb\UrlHarvester;

use PiedWeb\Curl\Response;
use PiedWeb\TextAnalyzer\Analyzer as TextAnalyzer;
use phpuri;
use simple_html_dom;
use PiedWeb\Curl\Request as CurlRequest;

class Harvest
{
    use HarvestLinksTrait, RobotsTxtTrait;

    const LINK_SELF = 1;
    const LINK_INTERNAL = 2;
    const LINK_SUB = 3;
    const LINK_EXTERNAL = 4;

    /**
     * @var Response
     */
    protected $response;

    /**
     * @var simple_html_dom
     */
    protected $dom;

    /** @var string */
    protected $baseUrl;

    /** @var string */
    protected $domain;

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

    /**
     * @return self|int
     */
    public static function fromUrl(
        string $url,
        string $userAgent = 'SeoPocketCrawler - Open Source Bot for SEO Metrics',
        string $language = 'en,en-US;q=0.5',
        ?CurlRequest $previousRequest = null
    ) {
        $response = Request::makeFromRequest($previousRequest, $url, $userAgent, $language);

        if ($response instanceof Response) {
            return new self($response);
        }

        return $response;
    }

    /**
     * @param Response $response
     */
    public function __construct(Response $response)
    {
        $this->response = $response;
    }

    public function getResponse()
    {
        return $this->response;
    }

    public function getRedirection()
    {
        $headers = $this->response->getHeaders();
        $headers = array_change_key_case($headers ? $headers : []);
        if (isset($headers['location'])) {
            return phpUri::parse($this->response->getEffectiveUrl())->join($headers['location']);
        }

        return false;
    }

    public function getDom()
    {
        if (null === $this->dom) {
            $this->dom = new simple_html_dom();
            $this->dom->load($this->response->getContent());
        }

        return $this->dom;
    }

    private function find($selector, $number = null)
    {
        return $this->getDom()->find($selector, $number);
    }

    private function findOne($selector)
    {
        return $this->find($selector, 0);
    }

    /**
     * Return content inside a selector.
     *
     * @return string
     */
    public function getTag($selector)
    {
        $found = $this->findOne($selector);

        return null !== $found ? Helper::clean($found->innertext) : null;
    }

    public function getUniqueTag($selector = 'title')
    {
        $found = $this->find($selector);
        if ($found) {
            if (count($found) > 1) {
                return count($found).' `'.$selector.'` !!';
            } else {
                return Helper::clean($found[0]->innertext);
            }
        }
    }

    /**
     * Return content inside a meta.
     *
     * @return string from content attribute
     */
    public function getMeta(string $name)
    {
        $meta = $this->findOne('meta[name='.$name.']');

        return null !== $meta ? (isset($meta->content) ? Helper::clean($meta->content) : '') : '';
    }

    /**
     * Renvoie le contenu de l'attribut href de la balise link rel=canonical.
     *
     * @return string le contenu de l'attribute href sinon NULL si la balise n'existe pas
     */
    public function getCanonical()
    {
        $canonical = $this->findOne('link[rel=canonical]');

        return null !== $canonical ? (isset($canonical->href) ? $canonical->href : '') : null;
    }

    /*
     * @return bool
     */
    public function isCanonicalCorrect()
    {
        $canonical = $this->getCanonical();

        return $canonical ? $this->response->getEffectiveUrl() == $canonical : true;
    }

    public function getKws()
    {
        $kws = TextAnalyzer::get(
            $this->getDom(),
            true,   // only sentences
            1,      // no expression, just words
            0      // keep trail
        );

        return $kws->getExpressions(10);
    }

    /**
     * @return int
     */
    public function getRatioTxtCode(): int
    {
        $textLenght = strlen($this->getDom()->plaintext);
        $htmlLenght = strlen(Helper::clean($this->response->getContent()));

        return (int) ($htmlLenght > 0 ? round($textLenght / $htmlLenght * 100) : 0);
    }

    /**
     * Return an array of object with two elements Link and anchor.
     *
     * @return array|null if we didn't found breadcrumb
     */
    public function getBreadCrumb(?string $separator = null)
    {
        $breadcrumb = ExtractBreadcrumb::get(
            $this->response->getContent(),
            $this->getBaseUrl(),
            $this->response->getEffectiveUrl()
        );

        if (null !== $separator && is_array($breadcrumb)) {
            $breadcrumb = array_map(function ($item) {
                return $item->getCleanName();
            }, $breadcrumb);
            $breadcrumb = implode($separator, $breadcrumb);
        }

        return $breadcrumb;
    }

    /**
     * @return string|false
     */
    public function amIRedirectToHttps()
    {
        $headers = $this->response->getHeaders();
        $headers = array_change_key_case(null !== $headers ? $headers : []);
        $redirUrl = isset($headers['location']) ? $headers['location'] : null;
        $url = $this->response->getUrl();
        if (null !== $redirUrl && ($httpsUrl = preg_replace('#^http:#', 'https:', $url, 1)) == $redirUrl) {
            return $httpsUrl;
        }

        return false;
    }

    public function getBaseUrl()
    {
        if (!isset($this->baseUrl)) {
            $base = $this->findOne('base');
            if (null !== $base && isset($base->href) && filter_var($base->href, FILTER_VALIDATE_URL)) {
                $this->baseUrl = $base->href;
            } else {
                $this->baseUrl = $this->response->getEffectiveUrl();
            }
        }

        return $this->baseUrl;
    }

    public function getDomain()
    {
        if (!isset($this->domain)) {
            $urlParsed = parse_url($this->response->getEffectiveUrl());
            preg_match("/[^\.\/]+(\.com?)?\.[^\.\/]+$/", $urlParsed['host'], $match);
            $this->domain = $match[0];
        }

        return $this->domain;
    }

    /**
     * @return int correspond to a const from Indexable
     */
    public function isIndexable(string $userAgent = 'googlebot')
    {
        return Indexable::isIndexable($this, $userAgent);
    }

    public function getDomainAndScheme()
    {
        if (null === $this->domainWithScheme) {
            $this->domainWithScheme = self::getDomainAndSchemeFrom($this->response->getEffectiveUrl());
        }

        return $this->domainWithScheme;
    }

    public static function getDomainAndSchemeFrom(string $url)
    {
        $url = parse_url($url);

        return $url['scheme'].'://'.$url['host'];
    }
}


1		<?php
2
3		namespace PiedWeb\UrlHarvester;
4
5		use PiedWeb\Curl\Response;
6		use PiedWeb\TextAnalyzer\Analyzer as TextAnalyzer;
7		use phpuri;
8		use simple_html_dom;
9		use PiedWeb\Curl\Request as CurlRequest;
10
11		class Harvest
12		{
13		use HarvestLinksTrait, RobotsTxtTrait;
14
15		const LINK_SELF = 1;
16		const LINK_INTERNAL = 2;
17		const LINK_SUB = 3;
18		const LINK_EXTERNAL = 4;
19
20		/**
21		* @var Response
22		*/
23		protected $response;
24
25		/**
26		* @var simple_html_dom
27		*/
28		protected $dom;
29
30		/** @var string */
31		protected $baseUrl;
32
33		/** @var string */
34		protected $domain;
35
36		/** @var string */
37		private $domainWithScheme;
38
39		/**
40		* @return self\|int
41		*/
42		public static function fromUrl(
43		string $url,
44		string $userAgent = 'SeoPocketCrawler - Open Source Bot for SEO Metrics',
45		string $language = 'en,en-US;q=0.5',
46	15	?CurlRequest $previousRequest = null
47		) {
48		$response = Request::makeFromRequest($previousRequest, $url, $userAgent, $language);
49
50		if ($response instanceof Response) {
51		return new self($response);
52	15	}
53
54	15	return $response;
55	15	}
56
57		/**
58		* @param Response $response
59		*/
60		public function __construct(Response $response)
61		{
62		$this->response = $response;
63		}
64	15
65		public function getResponse()
66	15	{
67	15	return $this->response;
68		}
69	12
70		public function getRedirection()
71	12	{
72		$headers = $this->response->getHeaders();
73		$headers = array_change_key_case($headers ? $headers : []);
74	3	if (isset($headers['location'])) {
75		return phpUri::parse($this->response->getEffectiveUrl())->join($headers['location']);
76	3	}
77	3
78	3	return false;
79	3	}
80
81		public function getDom()
82		{
83		if (null === $this->dom) {
84		$this->dom = new simple_html_dom();
85	15	$this->dom->load($this->response->getContent());
86		}
87	15
88	6	return $this->dom;
89	6	}
90
91		private function find($selector, $number = null)
92	15	{
93		return $this->getDom()->find($selector, $number);
94		}
95	12
96		private function findOne($selector)
97	12	{
98		return $this->find($selector, 0);
99		}
100	9
101		/**
102	9	* Return content inside a selector.
103		*
104		* @return string
105		*/
106		public function getTag($selector)
107		{
108		$found = $this->findOne($selector);
109
110	3	return null !== $found ? Helper::clean($found->innertext) : null;
111		}
112	3
113		public function getUniqueTag($selector = 'title')
114	3	{
115		$found = $this->find($selector);
116		if ($found) {
117	6	if (count($found) > 1) {
118		return count($found).' `'.$selector.'` !!';
119	6	} else {
120	6	return Helper::clean($found[0]->innertext);
121	6	}
122		}
123		}
124	6
125		/**
126		* Return content inside a meta.
127		*
128		* @return string from content attribute
129		*/
130		public function getMeta(string $name)
131		{
132		$meta = $this->findOne('meta[name='.$name.']');
133
134	9	return null !== $meta ? (isset($meta->content) ? Helper::clean($meta->content) : '') : '';
135		}
136	9
137		/**
138	9	* Renvoie le contenu de l'attribut href de la balise link rel=canonical.
139		*
140		* @return string le contenu de l'attribute href sinon NULL si la balise n'existe pas
141		*/
142		public function getCanonical()
143		{
144		$canonical = $this->findOne('link[rel=canonical]');
145
146	6	return null !== $canonical ? (isset($canonical->href) ? $canonical->href : '') : null;
147		}
148	6
149		/*
150	6	* @return bool
151		*/
152		public function isCanonicalCorrect()
153		{
154		$canonical = $this->getCanonical();
155
156	6	return $canonical ? $this->response->getEffectiveUrl() == $canonical : true;
157		}
158	6
159		public function getKws()
160	6	{
161		$kws = TextAnalyzer::get(
162		$this->getDom(),
163	3	true, // only sentences
164		1, // no expression, just words
165	3	0 // keep trail
166	3	);
167	3
168	3	return $kws->getExpressions(10);
169	3	}
170
171		/**
172	3	* @return int
173		*/
174		public function getRatioTxtCode(): int
175		{
176		$textLenght = strlen($this->getDom()->plaintext);
177		$htmlLenght = strlen(Helper::clean($this->response->getContent()));
178	3
179		return (int) ($htmlLenght > 0 ? round($textLenght / $htmlLenght * 100) : 0);
180	3	}
181	3
182		/**
183	3	* Return an array of object with two elements Link and anchor.
184		*
185		* @return array\|null if we didn't found breadcrumb
186		*/
187		public function getBreadCrumb(?string $separator = null)
188		{
189		$breadcrumb = ExtractBreadcrumb::get(
190		$this->response->getContent(),
191	3	$this->getBaseUrl(),
192		$this->response->getEffectiveUrl()
193	3	);
194	3
195	3	if (null !== $separator && is_array($breadcrumb)) {
196	3	$breadcrumb = array_map(function ($item) {
197		return $item->getCleanName();
198		}, $breadcrumb);
199	3	$breadcrumb = implode($separator, $breadcrumb);
200		}
201
202		return $breadcrumb;
203		}
204
205		/**
206	3	* @return string\|false
207		*/
208		public function amIRedirectToHttps()
209		{
210		$headers = $this->response->getHeaders();
211		$headers = array_change_key_case(null !== $headers ? $headers : []);
212		$redirUrl = isset($headers['location']) ? $headers['location'] : null;
213		$url = $this->response->getUrl();
214		if (null !== $redirUrl && ($httpsUrl = preg_replace('#^http:#', 'https:', $url, 1)) == $redirUrl) {
215		return $httpsUrl;
216		}
217
218		return false;
219		}
220
221		public function getBaseUrl()
222		{
223		if (!isset($this->baseUrl)) {
224		$base = $this->findOne('base');
225	3	if (null !== $base && isset($base->href) && filter_var($base->href, FILTER_VALIDATE_URL)) {
226		$this->baseUrl = $base->href;
227	3	} else {
228	3	$this->baseUrl = $this->response->getEffectiveUrl();
229	3	}
230		}
231
232	3	return $this->baseUrl;
233		}
234
235		public function getDomain()
236	3	{
237		if (!isset($this->domain)) {
238		$urlParsed = parse_url($this->response->getEffectiveUrl());
239	6	preg_match("/[^\.\/]+(\.com?)?\.[^\.\/]+$/", $urlParsed['host'], $match);
240		$this->domain = $match[0];
241	6	}
242	6
243	6	return $this->domain;
244	6	}
245
246		/**
247	6	* @return int correspond to a const from Indexable
248		*/
249		public function isIndexable(string $userAgent = 'googlebot')
250		{
251		return Indexable::isIndexable($this, $userAgent);
252		}
253	6
254		public function getDomainAndScheme()
255	6	{
256		if (null === $this->domainWithScheme) {
257		$this->domainWithScheme = self::getDomainAndSchemeFrom($this->response->getEffectiveUrl());
258		}
259
260		return $this->domainWithScheme;
261		}
262	6
263		public static function getDomainAndSchemeFrom(string $url)
264	6	{
265	6	$url = parse_url($url);
266
267	6	return $url['scheme'].'://'.$url['host'];
268		}
269		}
270

PiedWeb / UrlHarvester

Push — master ( 7e536e...5c7ae7 )

Harvest::getBaseUrl() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like