|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace rOpenDev\Google; |
|
4
|
|
|
|
|
5
|
|
|
abstract class Search |
|
6
|
|
|
{ |
|
7
|
|
|
use CacheTrait; |
|
8
|
|
|
use ConfSearchTrait; |
|
9
|
|
|
use SleepTrait; |
|
10
|
|
|
|
|
11
|
|
|
// ======= |
|
12
|
|
|
// ------- |
|
13
|
|
|
// ======= |
|
14
|
|
|
|
|
15
|
|
|
/** @var int Current page * */ |
|
16
|
|
|
protected $page = 1; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* @var string contain the current error |
|
20
|
|
|
*/ |
|
21
|
|
|
protected $error; |
|
22
|
|
|
|
|
23
|
|
|
protected $errors = [ |
|
24
|
|
|
1 => 'Google Captcha', |
|
25
|
|
|
2 => 'Google `We\'re sorry` (flagged as automated request)', |
|
26
|
|
|
3 => 'Erreurs cURL', |
|
27
|
|
|
4 => 'Google is redirecting us indefinitely via meta refresh.', |
|
28
|
|
|
]; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* Contient les erreurs provenant du cURL. |
|
32
|
|
|
*/ |
|
33
|
|
|
public $cErrors; |
|
34
|
|
|
|
|
35
|
|
|
public function generateGoogleSearchUrl() |
|
36
|
|
|
{ |
|
37
|
|
|
$this->setParameter('q', $this->q); |
|
38
|
|
|
$url = 'https://www.google.'.$this->tld.'/search?'.$this->generateParameters(); |
|
39
|
|
|
|
|
40
|
|
|
return $url; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
protected function generateParameters() |
|
44
|
|
|
{ |
|
45
|
|
|
return http_build_query($this->parameters, '', '&'); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @return string|false Contenu html de la page |
|
50
|
|
|
*/ |
|
51
|
|
|
abstract protected function requestGoogle(string $url); |
|
52
|
|
|
|
|
53
|
|
|
/* |
|
54
|
|
|
* Am I Kicked By Google ? Did you reach the google limits ?! |
|
55
|
|
|
* |
|
56
|
|
|
* @param string $output Html source |
|
57
|
|
|
* |
|
58
|
|
|
* @return int|false |
|
59
|
|
|
*/ |
|
60
|
|
|
protected function amIKickedByGoogleThePowerful($output) |
|
61
|
|
|
{ |
|
62
|
|
|
/* Google respond : |
|
63
|
|
|
* We're sorry...... but your computer or network may be sending automated queries. |
|
64
|
|
|
* To protect our users, we can't process your request right now.' |
|
65
|
|
|
*/ |
|
66
|
|
|
if (false !== strpos($output, '<title>Sorry...</title>')) { |
|
67
|
|
|
return 2; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/* Captcha Google */ |
|
71
|
|
|
elseif (false !== strpos($output, 'document.getElementById(\'captcha')) { |
|
72
|
|
|
return 1; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
/* RAS */ |
|
76
|
|
|
return false; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @return string explaining the error |
|
81
|
|
|
*/ |
|
82
|
|
|
public function getError() |
|
83
|
|
|
{ |
|
84
|
|
|
if (null !== $this->error) { |
|
85
|
|
|
return $this->errors[$this->error]; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
return false; |
|
|
|
|
|
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @return array|false containing the results with column type, link, title |
|
93
|
|
|
*/ |
|
94
|
|
|
public function extractResults() |
|
95
|
|
|
{ |
|
96
|
|
|
for ($this->page = 1; $this->page <= $this->nbrPage; ++$this->page) { |
|
97
|
|
|
if (! isset($url)) {// On génère l'url pour la première requète... Ensuite, on utilisera le lien Suivant. |
|
98
|
|
|
$url = $this->generateGoogleSearchUrl(); |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
$output = $this->requestGoogle($url); |
|
102
|
|
|
//file_put_contents('debug.html', $output); |
|
103
|
|
|
if (false === $output) { |
|
104
|
|
|
return false; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
$extract = new ExtractResults($output); |
|
108
|
|
|
|
|
109
|
|
|
$this->result = array_merge($this->result, $extract->getOrganicResults()); |
|
110
|
|
|
|
|
111
|
|
|
//h3 > a[href] |
|
112
|
|
|
if ($this->nbrPage > 1 && $extract->getNextPageLink()) { |
|
113
|
|
|
$this->execHalfSleep(); |
|
114
|
|
|
$url = 'https://www.google.'.$this->tld.str_replace('&', '&', $extract->getNextPageLink()); |
|
115
|
|
|
} else { |
|
116
|
|
|
break; |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
return $this->result; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
/** |
|
124
|
|
|
* getNbrResults va chercher le nombre de résulats que Google affiche proposer. |
|
125
|
|
|
* |
|
126
|
|
|
* @return int |
|
127
|
|
|
*/ |
|
128
|
|
|
public function getNbrResults() |
|
129
|
|
|
{ |
|
130
|
|
|
$url = $this->generateGoogleSearchUrl(); |
|
131
|
|
|
$output = $this->requestGoogle($url); |
|
132
|
|
|
if (false !== $output) { |
|
133
|
|
|
$html = new \simple_html_dom(); |
|
134
|
|
|
$html->load($output); |
|
135
|
|
|
|
|
136
|
|
|
$rS = $html->find('#resultStats'); |
|
137
|
|
|
if (isset($rS[0]->plaintext)) { |
|
138
|
|
|
$s = (string) $this->normalizeTextFromGoogle($rS[0]->plaintext); |
|
|
|
|
|
|
139
|
|
|
|
|
140
|
|
|
return (int) (preg_replace('/[^0-9]/', '', $s)); |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
RobinDev /
Google
Loading history...