|
1
|
|
|
<?php |
|
2
|
|
|
namespace CViniciusSDias\GoogleCrawler; |
|
3
|
|
|
|
|
4
|
|
|
use CViniciusSDias\GoogleCrawler\Exception\InvalidResultException; |
|
5
|
|
|
use CViniciusSDias\GoogleCrawler\Proxy\{ |
|
6
|
|
|
GoogleProxyInterface, NoProxy |
|
7
|
|
|
}; |
|
8
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
9
|
|
|
use Symfony\Component\DomCrawler\Crawler as DomCrawler; |
|
10
|
|
|
use Symfony\Component\DomCrawler\Link; |
|
11
|
|
|
use DOMElement; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* Google Crawler |
|
15
|
|
|
* |
|
16
|
|
|
* @package CViniciusSDias\GoogleCrawler |
|
17
|
|
|
* @author Vinicius Dias |
|
18
|
|
|
*/ |
|
19
|
|
|
class Crawler |
|
20
|
|
|
{ |
|
21
|
|
|
/** @var GoogleProxyInterface $proxy */ |
|
22
|
|
|
protected $proxy; |
|
23
|
|
|
/** @var SearchTermInterface $searchTerm */ |
|
24
|
|
|
private $searchTerm; |
|
25
|
|
|
/** @var string $countrySpecificSuffix */ |
|
26
|
|
|
private $googleDomain; |
|
27
|
|
|
/** @var string $countryCode */ |
|
28
|
|
|
private $countryCode; |
|
29
|
|
|
|
|
30
|
3 |
|
public function __construct( |
|
31
|
|
|
SearchTermInterface $searchTerm, GoogleProxyInterface $proxy = null, |
|
32
|
|
|
string $googleDomain = 'google.com', string $countryCode = '' |
|
33
|
|
|
) { |
|
34
|
3 |
|
$this->proxy = is_null($proxy) ? new NoProxy() : $proxy; |
|
|
|
|
|
|
35
|
3 |
|
$this->searchTerm = $searchTerm; |
|
36
|
|
|
|
|
37
|
3 |
|
if (mb_stripos($googleDomain, 'google.') === false || mb_stripos($googleDomain, 'http') === 0) { |
|
38
|
|
|
throw new \InvalidArgumentException('Invalid google domain'); |
|
39
|
|
|
} |
|
40
|
3 |
|
$this->googleDomain = $googleDomain; |
|
41
|
|
|
|
|
42
|
3 |
|
$this->countryCode = mb_strtoupper($countryCode); |
|
43
|
3 |
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Returns the 100 first found results for the specified search term |
|
47
|
|
|
* |
|
48
|
|
|
* @return ResultList |
|
49
|
|
|
* @throws \GuzzleHttp\Exception\ServerException If the proxy was overused |
|
50
|
|
|
* @throws \GuzzleHttp\Exception\ConnectException If the proxy is unavailable or $countrySpecificSuffix is invalid |
|
51
|
|
|
*/ |
|
52
|
3 |
|
public function getResults(): ResultList |
|
53
|
|
|
{ |
|
54
|
3 |
|
$googleUrl = $this->getGoogleUrl(); |
|
55
|
|
|
/** @var ResponseInterface $response */ |
|
56
|
3 |
|
$response = $this->proxy->getHttpResponse($googleUrl); |
|
57
|
3 |
|
$stringResponse = (string) $response->getBody(); |
|
58
|
3 |
|
$domCrawler = new DomCrawler($stringResponse); |
|
59
|
3 |
|
$googleResults = $domCrawler->filterXPath('//div[@class="g" and h3[@class="r" and a]]'); |
|
60
|
3 |
|
$resultList = new ResultList($googleResults->count()); |
|
61
|
|
|
|
|
62
|
3 |
|
foreach ($googleResults as $result) { |
|
63
|
2 |
|
$resultCrawler = new DomCrawler($result); |
|
64
|
2 |
|
$linkElement = $resultCrawler->filterXPath('//h3[@class="r"]/a')->getNode(0); |
|
65
|
2 |
|
$resultLink = new Link($linkElement, 'http://google.com/'); |
|
|
|
|
|
|
66
|
2 |
|
$descriptionElement = $resultCrawler->filterXPath('//span[@class="st"]')->getNode(0); |
|
67
|
|
|
try { |
|
68
|
2 |
|
$googleResult = $this->parseResult($resultLink, $descriptionElement); |
|
|
|
|
|
|
69
|
2 |
|
$resultList->addResult($googleResult); |
|
70
|
2 |
|
} catch (InvalidResultException $invalidResult) { |
|
71
|
|
|
// Maybe log this exception. Other than that, there's nothing to do, cause it isn't an error. |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
3 |
|
return $resultList; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* If $resultLink is a valid link, this method assembles the Result and adds it to $googleResults |
|
80
|
|
|
* |
|
81
|
|
|
* @param Link $resultLink |
|
82
|
|
|
* @param DOMElement $descriptionElement |
|
83
|
|
|
* @return Result |
|
84
|
|
|
* @throws InvalidResultException |
|
85
|
|
|
*/ |
|
86
|
2 |
|
private function parseResult(Link $resultLink, DOMElement $descriptionElement): Result |
|
87
|
|
|
{ |
|
88
|
2 |
|
$description = $descriptionElement->nodeValue |
|
89
|
2 |
|
?? 'A description for this result isn\'t available due to the robots.txt file.'; |
|
90
|
|
|
|
|
91
|
2 |
|
$googleResult = new Result(); |
|
92
|
|
|
$googleResult |
|
93
|
2 |
|
->setTitle($resultLink->getNode()->nodeValue) |
|
94
|
2 |
|
->setUrl($this->parseUrl($resultLink->getUri())) |
|
95
|
2 |
|
->setDescription($description); |
|
96
|
|
|
|
|
97
|
2 |
|
return $googleResult; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* Parses the URL using the parser provided by $proxy |
|
102
|
|
|
* |
|
103
|
|
|
* @param string $url |
|
104
|
|
|
* @return string |
|
105
|
|
|
* @throws InvalidResultException |
|
106
|
|
|
*/ |
|
107
|
2 |
|
private function parseUrl(string $url): string |
|
108
|
|
|
{ |
|
109
|
2 |
|
return $this->proxy->parseUrl($url); |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* Assembles the Google URL using the previously informed data |
|
114
|
|
|
*/ |
|
115
|
3 |
|
private function getGoogleUrl(): string |
|
116
|
|
|
{ |
|
117
|
3 |
|
$domain = $this->googleDomain; |
|
118
|
3 |
|
$url = "https://$domain/search?q={$this->searchTerm}&num=100"; |
|
119
|
3 |
|
if (!empty($this->countryCode)) { |
|
120
|
|
|
$url .= "&gl={$this->countryCode}"; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
3 |
|
return $url; |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
|
CViniciusSDias /
google-crawler
Loading history...
Our type inference engine has found an assignment to a property that is incompatible with the declared type of that property.
Either this assignment is in error or the assigned type should be added to the documentation/type hint for that property..