Search::setCache() - Code Metrics - Inspection of "Big Update" - RobinDev/Google - Measure and Improve Code Quality continuously with Scrutinizer

Test Failed

Push — master ( 138499...684014 )

by Dev

created 2019-01-03 15:41 UTC

Search::setCache() A

↳ Parent: Search::amIKickedByGoogleThePowerful()

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	2
nc	1
nop	1
dl	0
loc	5
rs	10
c	0
b	0
f	0

<?php

namespace rOpenDev\Google;

use rOpenDev\curl\CurlRequest;

abstract class Search
{
    use CacheTrait, ConfSearchTrait, SleepTrait;

    // =======
    // -------
    // =======

    /**
     * @var array from \rOpenDev\Google\ResultTypes.php
     */
    protected $types;

    /** @var int Current page * */
    protected $page = 1;

    /**
     * @var object dom
     */
    protected $html;

    /**
     * @var string contain the current error
     */
    protected $error;

    protected $errors = [
        1 => 'Google Captcha',
        2 => 'Google `We\'re sorry` (flagged as automated request)',
        3 => 'Erreurs cURL',
    ];

    /**
     * Contient les erreurs provenant du cURL.
     */
    public $cErrors;

    public function generateGoogleSearchUrl()
    {
        $this->setParameter('q', $this->q);
        $url = 'https://www.google.'.$this->tld.'/search?'.$this->generateParameters(); //$url = 'https://www.google.'.$this->tld.'/search?q='.self::encodeLikeGoogle($kw)
        return $url;
    }

    protected function generateParameters()
    {
        return http_build_query($this->parameters, '', '&');
    }

    /**
     * @return string|false Contenu html de la page
     */
    abstract protected function requestGoogle(string $url);

    /*
     * Am I Kicked By Google ? Did you reach the google limits ?!
     *
     * @param string $output Html source
     *
     * @return int|false
     */
    protected function amIKickedByGoogleThePowerful($output)
    {
        /* Google respond :
         * We're sorry...... but your computer or network may be sending automated queries.
         * To protect our users, we can't process your request right now.'
         */
        if (false !== strpos($output, '<title>Sorry...</title>')) {
            return 2;
        }

        /* Captcha Google */
        elseif (false !== strpos($output, 'e=document.getElementById(\'captcha\');if(e){e.focus();}')) {
            return 1;
        }

        /* RAS */
        return false;
    }

    /**
     * @return string explaining the error
     */
    public function getError()
    {
        if (null !== $this->error) {
            return $this->errors[$this->error];
        }

        return false;

    }

    /**
     * @return array|false containing the results with column type, link, title, page, page_pos
     */
    public function extractResults()
    {
        for ($this->page = 1; $this->page <= $this->nbrPage; ++$this->page) {
            if (!isset($url)) {// On génère l'url pour la première requète... Ensuite, on utilisera le lien Suivant.
                $url = $this->generateGoogleSearchUrl();
            }

            $output = $this->requestGoogle($url);
            if (false === $output) {
                return false;
            }

            $this->html = new \simple_html_dom();
            $this->html->load($output);

            $this->exctractOrganicResults();
            // todo : extract other results

            if ($this->nbrPage > 1 && isset($this->html->find('#pnnext')[0])) { // #pnnext // td.b a
                $url = 'https://www.google.'.$this->tld.str_replace('&amp;', '&', $this->html->find('#pnnext')[0]->href);
            } else {
                break;
            }
        }

        return $this->result;
    }

    protected function exctractOrganicResults()
    {
        $results = $this->html->find(str_replace(';', ', ',$this->getSelector('organic')));
        $page_pos = 1;
        foreach ($results as $r) {
            $title = $r->find('h3, [role=heading]', 0);
            if ($title) {
                $this->result[] = [
                    'type' => 'organic',
                    'title' => $this->normalizeTextFromGoogle($title->innertext),
                    'link' => $this->getUrlFromGoogleSerpFromat($r->find('a', 0)->href),
                    'page' => $this->page,
                    'page_pos' => $page_pos,
                ];
                ++$page_pos;
            }
        }
    }

    protected function getSelector(string $type)
    {
        if (null === $this->types) {
            $this->types = ResultsTypes::get();
        }

        return $this->types[$type];
    }

    /**
     * getNbrResults va chercher le nombre de résulats que Google affiche proposer.
     *
     * @return int
     */
    public function getNbrResults()
    {
        $url = $this->generateGoogleSearchUrl();
        $output = $this->requestGoogle($url);
        if (false !== $output) {
            $html = new \simple_html_dom();
            $html->load($output);

            $rS = $html->find('#resultStats');
            if (isset($rS[0]->plaintext)) {
                $s = (string) $this->normalizeTextFromGoogle($rS[0]->plaintext);

                return intval(preg_replace('/[^0-9]/', '', $s));
            }
        }
    }

    protected static function getUrlFromGoogleSerpFromat($str)
    {
        preg_match('/\/url\?.*(q|url)=(http.+)&/SiU', $str, $m1);
        $str = isset($m1[2]) ? $m1[2] : $str;

        return $str;
    }

    protected function normalizeTextFromGoogle($text)
    {
        return htmlspecialchars_decode(html_entity_decode(strip_tags($text)));
    }
}


1			<?php
2
3			namespace rOpenDev\Google;
4
5			use rOpenDev\curl\CurlRequest;
6
7			abstract class Search
8			{
9			use CacheTrait, ConfSearchTrait, SleepTrait;
10
11			// =======
12			// -------
13			// =======
14
15			/**
16			* @var array from \rOpenDev\Google\ResultTypes.php
17			*/
18			protected $types;
19
20			/** @var int Current page * */
21			protected $page = 1;
22
23			/**
24			* @var object dom
25			*/
26			protected $html;
27
28			/**
29			* @var string contain the current error
30			*/
31			protected $error;
32
33			protected $errors = [
34			1 => 'Google Captcha',
35			2 => 'Google `We\'re sorry` (flagged as automated request)',
36			3 => 'Erreurs cURL',
37			];
38
39			/**
40			* Contient les erreurs provenant du cURL.
41			*/
42			public $cErrors;
43
44			public function generateGoogleSearchUrl()
45			{
46			$this->setParameter('q', $this->q);
47			$url = 'https://www.google.'.$this->tld.'/search?'.$this->generateParameters(); //$url = 'https://www.google.'.$this->tld.'/search?q='.self::encodeLikeGoogle($kw)
48			return $url;
49			}
50
51			protected function generateParameters()
52			{
53			return http_build_query($this->parameters, '', '&');
54			}
55
56			/**
57			* @return string\|false Contenu html de la page
58			*/
59			abstract protected function requestGoogle(string $url);
60
61			/*
62			* Am I Kicked By Google ? Did you reach the google limits ?!
63			*
64			* @param string $output Html source
65			*
66			* @return int\|false
67			*/
68			protected function amIKickedByGoogleThePowerful($output)
69			{
70			/* Google respond :
71			* We're sorry...... but your computer or network may be sending automated queries.
72			* To protect our users, we can't process your request right now.'
73			*/
74			if (false !== strpos($output, '<title>Sorry...</title>')) {
75			return 2;
76			}
77
78			/* Captcha Google */
79			elseif (false !== strpos($output, 'e=document.getElementById(\'captcha\');if(e){e.focus();}')) {
80			return 1;
81			}
82
83			/* RAS */
84			return false;
85			}
86
87			/**
88			* @return string explaining the error
89			*/
90			public function getError()
91			{
92			if (null !== $this->error) {
93			return $this->errors[$this->error];
94			}
95
96			return false;
			0 ignored issues – show Bug Best Practice introduced 2019-01-03 07:40 UTC by Report Bug Copy Issue Report The expression `return false` returns the type `false` which is incompatible with the documented return type `string`. Loading history...
97			}
98
99			/**
100			* @return array\|false containing the results with column type, link, title, page, page_pos
101			*/
102			public function extractResults()
103			{
104			for ($this->page = 1; $this->page <= $this->nbrPage; ++$this->page) {
105			if (!isset($url)) {// On génère l'url pour la première requète... Ensuite, on utilisera le lien Suivant.
106			$url = $this->generateGoogleSearchUrl();
107			}
108
109			$output = $this->requestGoogle($url);
110			if (false === $output) {
111			return false;
112			}
113
114			$this->html = new \simple_html_dom();
115			$this->html->load($output);
116
117			$this->exctractOrganicResults();
118			// todo : extract other results
119
120			if ($this->nbrPage > 1 && isset($this->html->find('#pnnext')[0])) { // #pnnext // td.b a
121			$url = 'https://www.google.'.$this->tld.str_replace('&', '&', $this->html->find('#pnnext')[0]->href);
122			} else {
123			break;
124			}
125			}
126
127			return $this->result;
128			}
129
130			protected function exctractOrganicResults()
131			{
132			$results = $this->html->find(str_replace(';', ', ',$this->getSelector('organic')));
133			$page_pos = 1;
134			foreach ($results as $r) {
135			$title = $r->find('h3, [role=heading]', 0);
136			if ($title) {
137			$this->result[] = [
138			'type' => 'organic',
139			'title' => $this->normalizeTextFromGoogle($title->innertext),
140			'link' => $this->getUrlFromGoogleSerpFromat($r->find('a', 0)->href),
141			'page' => $this->page,
142			'page_pos' => $page_pos,
143			];
144			++$page_pos;
145			}
146			}
147			}
148
149			protected function getSelector(string $type)
150			{
151			if (null === $this->types) {
152			$this->types = ResultsTypes::get();
153			}
154
155			return $this->types[$type];
156			}
157
158			/**
159			* getNbrResults va chercher le nombre de résulats que Google affiche proposer.
160			*
161			* @return int
162			*/
163			public function getNbrResults()
164			{
165			$url = $this->generateGoogleSearchUrl();
166			$output = $this->requestGoogle($url);
167			if (false !== $output) {
168			$html = new \simple_html_dom();
169			$html->load($output);
170
171			$rS = $html->find('#resultStats');
172			if (isset($rS[0]->plaintext)) {
173			$s = (string) $this->normalizeTextFromGoogle($rS[0]->plaintext);
174
175			return intval(preg_replace('/[^0-9]/', '', $s));
176			}
177			}
178			}
179
180			protected static function getUrlFromGoogleSerpFromat($str)
181			{
182			preg_match('/\/url\?.*(q\|url)=(http.+)&/SiU', $str, $m1);
183			$str = isset($m1[2]) ? $m1[2] : $str;
184
185			return $str;
186			}
187
188			protected function normalizeTextFromGoogle($text)
189			{
190			return htmlspecialchars_decode(html_entity_decode(strip_tags($text)));
191			}
192			}
193

RobinDev / Google

Push — master ( 138499...684014 )

Search::setCache() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like