Harvest::getRedirection() - Code Metrics - Inspection of "update piedweb/curl" - PiedWeb/UrlHarvester - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( acb1e6...d7f53c )

by Dev

created 2019-01-18 18:13 UTC

Harvest::getRedirection() A

↳ Parent: Harvest

Complexity

Conditions	3
Paths	2

Size

Total Lines	9
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	5
CRAP Score	3.0416

Importance

Changes

Metric	Value
cc	3
eloc	5
nc	2
nop	0
dl	0
loc	9
ccs	5
cts	6
cp	0.8333
crap	3.0416
rs	10
c	0
b	0
f	0

<?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;

    const DEFAULT_USER_AGENT = 'SeoPocketCrawler - Open Source Bot for SEO Metrics';

    /**
     * @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 = self::DEFAULT_USER_AGENT,
        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		const DEFAULT_USER_AGENT = 'SeoPocketCrawler - Open Source Bot for SEO Metrics';
21
22		/**
23		* @var Response
24		*/
25		protected $response;
26
27		/**
28		* @var simple_html_dom
29		*/
30		protected $dom;
31
32		/** @var string */
33		protected $baseUrl;
34
35		/** @var string */
36		protected $domain;
37
38		/** @var string */
39		private $domainWithScheme;
40
41		/**
42		* @return self\|int
43		*/
44	15	public static function fromUrl(
45		string $url,
46		string $userAgent = self::DEFAULT_USER_AGENT,
47		string $language = 'en,en-US;q=0.5',
48		?CurlRequest $previousRequest = null
49		) {
50	15	$response = Request::makeFromRequest($previousRequest, $url, $userAgent, $language);
51
52	15	if ($response instanceof Response) {
53	15	return new self($response);
54		}
55
56		return $response;
57		}
58
59		/**
60		* @param Response $response
61		*/
62	18	public function __construct(Response $response)
63		{
64	18	$this->response = $response;
65	18	}
66
67	15	public function getResponse()
68		{
69	15	return $this->response;
70		}
71
72	3	public function getRedirection()
73		{
74	3	$headers = $this->response->getHeaders();
75	3	$headers = array_change_key_case($headers ? $headers : []);
76	3	if (isset($headers['location'])) {
77	3	return phpUri::parse($this->response->getEffectiveUrl())->join($headers['location']);
78		}
79
80		return false;
81		}
82
83	18	public function getDom()
84		{
85	18	if (null === $this->dom) {
86	9	$this->dom = new simple_html_dom();
87	9	$this->dom->load($this->response->getContent());
88		}
89
90	18	return $this->dom;
91		}
92
93	15	private function find($selector, $number = null)
94		{
95	15	return $this->getDom()->find($selector, $number);
96		}
97
98	12	private function findOne($selector)
99		{
100	12	return $this->find($selector, 0);
101		}
102
103		/**
104		* Return content inside a selector.
105		*
106		* @return string
107		*/
108	3	public function getTag($selector)
109		{
110	3	$found = $this->findOne($selector);
111
112	3	return null !== $found ? Helper::clean($found->innertext) : null;
113		}
114
115	6	public function getUniqueTag($selector = 'title')
116		{
117	6	$found = $this->find($selector);
118	6	if ($found) {
119	6	if (count($found) > 1) {
120		return count($found).' `'.$selector.'` !!';
121		} else {
122	6	return Helper::clean($found[0]->innertext);
123		}
124		}
125		}
126
127		/**
128		* Return content inside a meta.
129		*
130		* @return string from content attribute
131		*/
132	12	public function getMeta(string $name)
133		{
134	12	$meta = $this->findOne('meta[name='.$name.']');
135
136	12	return null !== $meta ? (isset($meta->content) ? Helper::clean($meta->content) : '') : '';
137		}
138
139		/**
140		* Renvoie le contenu de l'attribut href de la balise link rel=canonical.
141		*
142		* @return string le contenu de l'attribute href sinon NULL si la balise n'existe pas
143		*/
144	9	public function getCanonical()
145		{
146	9	$canonical = $this->findOne('link[rel=canonical]');
147
148	9	return null !== $canonical ? (isset($canonical->href) ? $canonical->href : '') : null;
149		}
150
151		/*
152		* @return bool
153		*/
154	9	public function isCanonicalCorrect()
155		{
156	9	$canonical = $this->getCanonical();
157
158	9	return $canonical ? $this->response->getEffectiveUrl() == $canonical : true;
159		}
160
161	3	public function getKws()
162		{
163	3	$kws = TextAnalyzer::get(
164	3	$this->getDom(),
165	3	true, // only sentences
166	3	1, // no expression, just words
167	3	0 // keep trail
168		);
169
170	3	return $kws->getExpressions(10);
171		}
172
173		/**
174		* @return int
175		*/
176	3	public function getRatioTxtCode(): int
177		{
178	3	$textLenght = strlen($this->getDom()->plaintext);
179	3	$htmlLenght = strlen(Helper::clean($this->response->getContent()));
180
181	3	return (int) ($htmlLenght > 0 ? round($textLenght / $htmlLenght * 100) : 0);
182		}
183
184		/**
185		* Return an array of object with two elements Link and anchor.
186		*
187		* @return array\|null if we didn't found breadcrumb
188		*/
189	3	public function getBreadCrumb(?string $separator = null)
190		{
191	3	$breadcrumb = ExtractBreadcrumb::get(
192	3	$this->response->getContent(),
193	3	$this->getBaseUrl(),
194	3	$this->response->getEffectiveUrl()
195		);
196
197	3	if (null !== $separator && is_array($breadcrumb)) {
198		$breadcrumb = array_map(function ($item) {
199		return $item->getCleanName();
200		}, $breadcrumb);
201		$breadcrumb = implode($separator, $breadcrumb);
202		}
203
204	3	return $breadcrumb;
205		}
206
207		/**
208		* @return string\|false
209		*/
210		public function amIRedirectToHttps()
211		{
212		$headers = $this->response->getHeaders();
213		$headers = array_change_key_case(null !== $headers ? $headers : []);
214		$redirUrl = isset($headers['location']) ? $headers['location'] : null;
215		$url = $this->response->getUrl();
216		if (null !== $redirUrl && ($httpsUrl = preg_replace('#^http:#', 'https:', $url, 1)) == $redirUrl) {
217		return $httpsUrl;
218		}
219
220		return false;
221		}
222
223	3	public function getBaseUrl()
224		{
225	3	if (!isset($this->baseUrl)) {
226	3	$base = $this->findOne('base');
227	3	if (null !== $base && isset($base->href) && filter_var($base->href, FILTER_VALIDATE_URL)) {
228		$this->baseUrl = $base->href;
229		} else {
230	3	$this->baseUrl = $this->response->getEffectiveUrl();
231		}
232		}
233
234	3	return $this->baseUrl;
235		}
236
237	6	public function getDomain()
238		{
239	6	if (!isset($this->domain)) {
240	6	$urlParsed = parse_url($this->response->getEffectiveUrl());
241	6	preg_match("/[^\.\/]+(\.com?)?\.[^\.\/]+$/", $urlParsed['host'], $match);
242	6	$this->domain = $match[0];
243		}
244
245	6	return $this->domain;
246		}
247
248		/**
249		* @return int correspond to a const from Indexable
250		*/
251	9	public function isIndexable(string $userAgent = 'googlebot')
252		{
253	9	return Indexable::isIndexable($this, $userAgent);
254		}
255
256	12	public function getDomainAndScheme()
257		{
258	12	if (null === $this->domainWithScheme) {
259	12	$this->domainWithScheme = self::getDomainAndSchemeFrom($this->response->getEffectiveUrl());
260		}
261
262	12	return $this->domainWithScheme;
263		}
264
265	12	public static function getDomainAndSchemeFrom(string $url)
266		{
267	12	$url = parse_url($url);
268
269	12	return $url['scheme'].'://'.$url['host'];
270		}
271		}
272

PiedWeb / UrlHarvester

Push — master ( acb1e6...d7f53c )

Harvest::getRedirection() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like