|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Lbc\Crawler; |
|
4
|
|
|
|
|
5
|
|
|
use Lbc\Filter\PrixSanitizer; |
|
6
|
|
|
use Lbc\Parser\AdUrlParser; |
|
7
|
|
|
use Lbc\Parser\SearchResultUrlParser; |
|
8
|
|
|
use League\Uri\Schemes\Http; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Class SearchResultAdCrawler |
|
12
|
|
|
* @package Lbc\Crawler |
|
13
|
|
|
*/ |
|
14
|
|
|
class SearchResultAdCrawler extends CrawlerAbstract |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* @var AdUrlParser |
|
18
|
|
|
*/ |
|
19
|
|
|
protected $url; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @param $url |
|
23
|
|
|
* @return SearchResultUrlParser |
|
24
|
|
|
*/ |
|
25
|
14 |
|
protected function setUrlParser($url) |
|
26
|
|
|
{ |
|
27
|
14 |
|
$this->url = new AdUrlParser($url); |
|
28
|
14 |
|
} |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* Return the Ad's ID |
|
32
|
|
|
* |
|
33
|
|
|
* @return string |
|
34
|
|
|
*/ |
|
35
|
14 |
|
public function getId() |
|
36
|
14 |
|
{ |
|
37
|
10 |
|
return $this->url->getId(); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Return the title |
|
43
|
|
|
* |
|
44
|
|
|
* @return string |
|
45
|
|
|
*/ |
|
46
|
10 |
|
public function getTitle() |
|
47
|
|
|
{ |
|
48
|
10 |
|
return $this->getFieldValue($this->node->filter('h2'), ''); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Return the price |
|
53
|
|
|
* |
|
54
|
|
|
* @return int |
|
55
|
|
|
*/ |
|
56
|
10 |
|
public function getPrice() |
|
57
|
|
|
{ |
|
58
|
10 |
|
return $this->getFieldValue( |
|
59
|
10 |
|
$this->node->filter('*[itemprop=price]'), |
|
60
|
10 |
|
0, |
|
61
|
|
|
function ($value) { |
|
62
|
10 |
|
return PrixSanitizer::clean($value); |
|
63
|
|
|
} |
|
64
|
10 |
|
); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Return the Ad's URL |
|
69
|
|
|
* |
|
70
|
|
|
* @return string |
|
71
|
|
|
*/ |
|
72
|
10 |
|
public function getUrl() |
|
73
|
|
|
{ |
|
74
|
10 |
|
return (string)Http::createFromString($this->url)->withScheme('https'); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
/** |
|
78
|
|
|
* Return the data and time the ad was created |
|
79
|
|
|
* |
|
80
|
|
|
* @return string |
|
81
|
|
|
*/ |
|
82
|
10 |
|
public function getCreatedAt() |
|
83
|
|
|
{ |
|
84
|
10 |
|
return $this->node |
|
85
|
10 |
|
->filter('*[itemprop=availabilityStarts]') |
|
86
|
10 |
|
->first() |
|
87
|
10 |
|
->attr('content'); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* Return the thumb picture url |
|
92
|
|
|
* |
|
93
|
|
|
* @return null|string |
|
94
|
|
|
*/ |
|
95
|
12 |
|
public function getThumb() |
|
96
|
|
|
{ |
|
97
|
12 |
|
$image = $this->node |
|
98
|
12 |
|
->filter('.item_imagePic .lazyload[data-imgsrc]') |
|
99
|
12 |
|
->first(); |
|
100
|
|
|
|
|
101
|
12 |
|
if (0 === $image->count()) { |
|
102
|
2 |
|
return null; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
$src = $image |
|
106
|
10 |
|
->attr('data-imgsrc'); |
|
107
|
|
|
|
|
108
|
10 |
|
return (string)Http::createFromString($src)->withScheme('https'); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* Return the number of picture of the ad |
|
113
|
|
|
* |
|
114
|
|
|
* @return int |
|
115
|
|
|
*/ |
|
116
|
12 |
|
public function getNbImage() |
|
117
|
|
|
{ |
|
118
|
12 |
|
$node = $this->node->filter('.item_imageNumber'); |
|
119
|
|
|
|
|
120
|
12 |
|
return $this->getFieldValue($node, 0); |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
/** |
|
124
|
|
|
* @return mixed |
|
125
|
|
|
*/ |
|
126
|
10 |
|
public function getPlacement() |
|
127
|
|
|
{ |
|
128
|
10 |
|
$node = $this->node->filter('*[itemprop=availableAtOrFrom]'); |
|
129
|
|
|
|
|
130
|
|
|
return $this->getFieldValue($node, '', function ($value) { |
|
131
|
10 |
|
return preg_replace('/\s+/', ' ', trim($value)); |
|
132
|
10 |
|
}); |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
/** |
|
136
|
|
|
* @return mixed |
|
137
|
|
|
*/ |
|
138
|
10 |
|
public function getIsPro() |
|
139
|
|
|
{ |
|
140
|
10 |
|
return $this->getFieldValue( |
|
141
|
10 |
|
$this->node->filter('.ispro'), |
|
142
|
10 |
|
false, |
|
143
|
8 |
|
function ($value) { |
|
144
|
8 |
|
return true || $value; |
|
145
|
|
|
} |
|
146
|
10 |
|
); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* @return array |
|
151
|
|
|
*/ |
|
152
|
10 |
|
public function getAll() |
|
153
|
|
|
{ |
|
154
|
|
|
return [ |
|
155
|
10 |
|
'id' => $this->getId(), |
|
156
|
10 |
|
'titre' => $this->getTitle(), |
|
157
|
10 |
|
'is_pro' => $this->getIsPro(), |
|
158
|
10 |
|
'prix' => $this->getPrice(), |
|
159
|
10 |
|
'url' => $this->getUrl(), |
|
160
|
10 |
|
'created_at' => $this->getCreatedAt(), |
|
161
|
10 |
|
'images_thumbs' => $this->getThumb(), |
|
162
|
10 |
|
'nb_image' => $this->getNbImage(), |
|
163
|
10 |
|
'placement' => $this->getPlacement(), |
|
164
|
10 |
|
]; |
|
165
|
|
|
} |
|
166
|
|
|
} |
|
167
|
|
|
|
rfussien /
leboncoin-crawler
