Harvest::getDomain() - Code Metrics - Inspection of "fix scrtnzr issues" - PiedWeb/UrlHarvester - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( f60a2c...2d7c5c )

by Dev

created 2019-01-08 18:09 UTC

Harvest::getDomain() A

↳ Parent: Harvest

Complexity

Conditions	2
Paths	2

Size

Total Lines	9
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	2

Importance

Changes

Metric	Value
cc	2
eloc	5
nc	2
nop	0
dl	0
loc	9
ccs	6
cts	6
cp	1
crap	2
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;

class Harvest
{
    use HarvestLinksTrait;

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

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

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

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

    public static function fromUrl(
        string $url,
        string $userAgent = 'Bot: Url Harvester',
        string $language = 'en,en-US;q=0.5',
        bool   $tryHttps = false
    ) {
        $request = Request::make($url, $userAgent, 'text/html', $language, $tryHttps);
        $response = $request->getResponse();

        if ($response instanceof Response) {

            return new self($response);
        }

        return $request->getError();
    }

    /**
     * @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) : '') : null;
    }

    /**
     * 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()
    {
        return $this->response->getEffectiveUrl() == $this->getCanonical();
    }

    public function getKws()
    {
        $kws = TextAnalyzer::get(
            $this->response->getContent(),
            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 or NULL if we didn't found breadcrumb
     */
    public function getBreadCrumb()
    {
        return ExtractBreadcrumb::get(
            $this->response->getContent(),
            $this->getBaseUrl(),
            $this->response->getEffectiveUrl()
        );
    }

    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;
    }
}


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
10		class Harvest
11		{
12		use HarvestLinksTrait;
13
14		/**
15		* @var Response
16		*/
17		protected $response;
18
19		/**
20		* @var simple_html_dom
21		*/
22		protected $dom;
23
24		/** @var string */
25		protected $baseUrl;
26
27		/** @var string */
28		protected $domain;
29
30	9	public static function fromUrl(
31		string $url,
32		string $userAgent = 'Bot: Url Harvester',
33		string $language = 'en,en-US;q=0.5',
34		bool $tryHttps = false
35		) {
36	9	$request = Request::make($url, $userAgent, 'text/html', $language, $tryHttps);
37	9	$response = $request->getResponse();
38
39	9	if ($response instanceof Response) {
		0 ignored issues – show introduced 2019-01-08 17:19 UTC by Report Bug Copy Issue Report `$response` is always a sub-type of `PiedWeb\Curl\Response`. Loading history...
40	9	return new self($response);
41		}
42
43		return $request->getError();
44		}
45
46		/**
47		* @param Response $response
48		*/
49	9	public function __construct(Response $response)
50		{
51	9	$this->response = $response;
52	9	}
53
54	3	public function getResponse()
55		{
56	3	return $this->response;
57		}
58
59		public function getRedirection()
60		{
61		$headers = $this->response->getHeaders();
62		$headers = array_change_key_case($headers ? $headers : []);
63		if (isset($headers['location'])) {
64		return phpUri::parse($this->response->getEffectiveUrl())->join($headers['location']);
65		}
66
67		return false;
68		}
69
70	6	public function getDom()
71		{
72	6	if (null === $this->dom) {
73	6	$this->dom = new simple_html_dom();
74	6	$this->dom->load($this->response->getContent());
75		}
76
77	6	return $this->dom;
78		}
79
80	3	private function find($selector, $number = null)
81		{
82	3	return $this->getDom()->find($selector, $number);
83		}
84
85	3	private function findOne($selector)
86		{
87	3	return $this->find($selector, 0);
88		}
89
90		/**
91		* Return content inside a selector.
92		*
93		* @return string
94		*/
95	3	public function getTag($selector)
96		{
97	3	$found = $this->findOne($selector);
98
99	3	return null !== $found ? Helper::clean($found->innertext) : null;
100		}
101
102		public function getUniqueTag($selector = 'title')
103		{
104		$found = $this->find($selector);
105		if ($found) {
106		if (count($found) > 1) {
107		return count($found).' `'.$selector.'` !!';
108		} else {
109		return Helper::clean($found[0]->innertext);
110		}
111		}
112		}
113
114		/**
115		* Return content inside a meta.
116		*
117		* @return string from content attribute
118		*/
119	3	public function getMeta(string $name)
120		{
121	3	$meta = $this->findOne('meta[name='.$name.']');
122
123	3	return null !== $meta ? (isset($meta->content) ? Helper::clean($meta->content) : '') : null;
124		}
125
126		/**
127		* Renvoie le contenu de l'attribut href de la balise link rel=canonical.
128		*
129		* @return string le contenu de l'attribute href sinon NULL si la balise n'existe pas
130		*/
131	3	public function getCanonical()
132		{
133	3	$canonical = $this->findOne('link[rel=canonical]');
134
135	3	return null !== $canonical ? (isset($canonical->href) ? $canonical->href : '') : null;
136		}
137
138		/*
139		* @return bool
140		*/
141	3	public function isCanonicalCorrect()
142		{
143	3	return $this->response->getEffectiveUrl() == $this->getCanonical();
144		}
145
146	3	public function getKws()
147		{
148	3	$kws = TextAnalyzer::get(
149	3	$this->response->getContent(),
150	3	true, // only sentences
151	3	1, // no expression, just words
152	3	0 // keep trail
153		);
154
155	3	return $kws->getExpressions(10);
156		}
157
158		/**
159		* @return int
160		*/
161	3	public function getRatioTxtCode(): int
162		{
163	3	$textLenght = strlen($this->getDom()->plaintext);
164	3	$htmlLenght = strlen(Helper::clean($this->response->getContent()));
165
166	3	return (int) ($htmlLenght > 0 ? round($textLenght / $htmlLenght * 100) : 0);
167		}
168
169		/**
170		* Return an array of object with two elements Link and anchor.
171		*
172		* @return array or NULL if we didn't found breadcrumb
173		*/
174	3	public function getBreadCrumb()
175		{
176	3	return ExtractBreadcrumb::get(
177	3	$this->response->getContent(),
178	3	$this->getBaseUrl(),
179	3	$this->response->getEffectiveUrl()
180		);
181		}
182
183	3	public function getBaseUrl()
184		{
185	3	if (!isset($this->baseUrl)) {
186	3	$base = $this->findOne('base');
187	3	if (null !== $base && isset($base->href) && filter_var($base->href, FILTER_VALIDATE_URL)) {
188		$this->baseUrl = $base->href;
189		} else {
190	3	$this->baseUrl = $this->response->getEffectiveUrl();
191		}
192		}
193
194	3	return $this->baseUrl;
195		}
196
197	9	public function getDomain()
198		{
199	9	if (!isset($this->domain)) {
200	9	$urlParsed = parse_url($this->response->getEffectiveUrl());
201	9	preg_match("/[^\.\/]+(\.com?)?\.[^\.\/]+$/", $urlParsed['host'], $match);
202	9	$this->domain = $match[0];
203		}
204
205	9	return $this->domain;
206		}
207		}
208

PiedWeb / UrlHarvester

Push — master ( f60a2c...2d7c5c )

Harvest::getDomain() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like