|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Lbc\Crawler; |
|
4
|
|
|
|
|
5
|
|
|
use Lbc\Parser\SearchResultUrlParser; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* Class SearchResultCrawler |
|
9
|
|
|
* @package Lbc\node |
|
10
|
|
|
*/ |
|
11
|
|
|
class SearchResultCrawler extends CrawlerAbstract |
|
12
|
|
|
{ |
|
13
|
|
|
/** |
|
14
|
|
|
* @var SearchResultUrlParser |
|
15
|
|
|
*/ |
|
16
|
|
|
protected $url; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* @param $url |
|
20
|
|
|
* @return SearchResultUrlParser |
|
21
|
|
|
*/ |
|
22
|
14 |
|
protected function setUrlParser($url) |
|
23
|
|
|
{ |
|
24
|
14 |
|
$this->url = new SearchResultUrlParser($url); |
|
25
|
14 |
|
} |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Return the total number of ads of the search |
|
29
|
|
|
* |
|
30
|
|
|
* @return int |
|
31
|
|
|
*/ |
|
32
|
8 |
|
public function getNbAds() |
|
33
|
|
|
{ |
|
34
|
8 |
|
$nbAds = $this->node |
|
35
|
8 |
|
->filter('a.tabsSwitch span.tabsSwitchNumbers') |
|
36
|
8 |
|
->first(); |
|
37
|
|
|
|
|
38
|
8 |
|
if ($nbAds->count()) { |
|
39
|
6 |
|
$nbAds = preg_replace('/\s+/', '', $nbAds->text()); |
|
40
|
6 |
|
return (int) $nbAds; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
2 |
|
return 0; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Return the number of ads per page. |
|
48
|
|
|
* |
|
49
|
|
|
* Could be dynamically guessed in future, if Leboncoin change it frequently |
|
50
|
|
|
* Or if they add the ability for user to change it on result pages. |
|
51
|
|
|
* |
|
52
|
|
|
* |
|
53
|
|
|
* @return int |
|
54
|
|
|
*/ |
|
55
|
8 |
|
public function getNbAdsPerPage() |
|
56
|
|
|
{ |
|
57
|
8 |
|
return 35; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* Return the number of page |
|
62
|
|
|
* |
|
63
|
|
|
* @return int |
|
64
|
|
|
*/ |
|
65
|
8 |
|
public function getNbPages() |
|
66
|
|
|
{ |
|
67
|
8 |
|
return (int) ceil($this->getNbAds() / $this->getNbAdsPerPage()); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* Get an array containing the ads of the current result page |
|
72
|
|
|
* |
|
73
|
|
|
* @return array |
|
74
|
|
|
*/ |
|
75
|
10 |
|
public function getAds() |
|
76
|
|
|
{ |
|
77
|
10 |
|
$ads = array(); |
|
78
|
|
|
|
|
79
|
10 |
|
$this->node->filter('[itemtype="http://schema.org/Offer"]') |
|
80
|
10 |
|
->each(function ($node) use (&$ads) { |
|
81
|
8 |
|
$ad = (new SearchResultAdCrawler( |
|
82
|
|
|
$node, |
|
83
|
8 |
|
$node->filter('a')->attr('href') |
|
84
|
8 |
|
))->getAll(); |
|
85
|
|
|
|
|
86
|
8 |
|
$ads [$ad['id']] = $ad; |
|
87
|
10 |
|
}); |
|
88
|
|
|
|
|
89
|
10 |
|
return $ads; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* Return the Ads's ID only |
|
94
|
|
|
* |
|
95
|
|
|
* @return array |
|
96
|
|
|
*/ |
|
97
|
6 |
|
public function getAdsId() |
|
98
|
|
|
{ |
|
99
|
6 |
|
return array_keys($this->getAds()); |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
rfussien /
leboncoin-crawler
