|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Lbc; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\Client; |
|
6
|
|
|
use Lbc\Crawler\AdCrawler; |
|
7
|
|
|
use Lbc\Crawler\SearchResultCrawler; |
|
8
|
|
|
use Lbc\Parser\SearchResultUrlParser; |
|
9
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Class GetFrom |
|
13
|
|
|
* @package Lbc |
|
14
|
|
|
*/ |
|
15
|
|
|
class GetFrom |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* @var Client |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $client; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* GetFrom constructor. |
|
24
|
|
|
* @param Client|null $client |
|
25
|
|
|
*/ |
|
26
|
12 |
|
public function __construct(Client $client = null) |
|
27
|
|
|
{ |
|
28
|
12 |
|
$this->client = $client ?: new Client(); |
|
29
|
12 |
|
} |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* Return the http client |
|
33
|
|
|
* (useful to mock the response for unit testing) |
|
34
|
|
|
* |
|
35
|
|
|
* @return Client |
|
36
|
|
|
*/ |
|
37
|
2 |
|
public function getHttpClient() |
|
38
|
|
|
{ |
|
39
|
2 |
|
return $this->client; |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* retrieve the search result data from the given url |
|
44
|
|
|
* |
|
45
|
|
|
* @param $url |
|
46
|
|
|
* @param bool $detailedAd |
|
47
|
|
|
* |
|
48
|
|
|
* @return array |
|
49
|
|
|
*/ |
|
50
|
4 |
|
public function search($url, $detailedAd = false) |
|
51
|
|
|
{ |
|
52
|
4 |
|
$searchData = new SearchResultCrawler( |
|
53
|
4 |
|
new Crawler((string) $this->client->get($url)->getBody()), |
|
54
|
4 |
|
$url |
|
55
|
|
|
); |
|
56
|
|
|
|
|
57
|
4 |
|
$url = new SearchResultUrlParser($url, $searchData->getNbPages()); |
|
58
|
|
|
|
|
59
|
4 |
|
$ads = ($detailedAd) ? $searchData->getAds() : $searchData->getAdsId(); |
|
60
|
|
|
|
|
61
|
|
|
$sumarize = [ |
|
62
|
4 |
|
'total_ads' => $searchData->getNbAds(), |
|
63
|
4 |
|
'total_page' => $searchData->getNbPages(), |
|
64
|
4 |
|
'ads_per_page' => $searchData->getNbAdsPerPage(), |
|
65
|
4 |
|
'category' => $searchData->getUrlParser()->getCategory(), |
|
66
|
4 |
|
'location' => $searchData->getUrlParser()->getLocation(), |
|
67
|
4 |
|
'search_area' => $searchData->getUrlParser()->getSearchArea(), |
|
68
|
4 |
|
'sort_by' => $searchData->getUrlParser()->getSortType(), |
|
69
|
4 |
|
'type' => $searchData->getUrlParser()->getType(), |
|
70
|
4 |
|
'ads' => $ads, |
|
71
|
|
|
]; |
|
72
|
|
|
|
|
73
|
4 |
|
return array_merge($url->getNav(), $sumarize); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Retrieve the ad's data from an ad's ID and its category |
|
78
|
|
|
* |
|
79
|
|
|
* @param $id |
|
80
|
|
|
* @param $category |
|
81
|
|
|
* |
|
82
|
|
|
* @return array |
|
83
|
|
|
*/ |
|
84
|
2 |
|
private function adById($id, $category) |
|
85
|
|
|
{ |
|
86
|
2 |
|
return $this->ad("https://www.leboncoin.fr/{$category}/{$id}.htm"); |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* Retrieve the ad's data from the given url |
|
91
|
|
|
* |
|
92
|
|
|
* @param $url |
|
93
|
|
|
* @return array |
|
94
|
|
|
*/ |
|
95
|
4 |
|
private function adByUrl($url) |
|
96
|
|
|
{ |
|
97
|
4 |
|
$content = $this->client->get($url)->getBody()->getContents(); |
|
98
|
|
|
|
|
99
|
4 |
|
$adData = new AdCrawler(new Crawler(utf8_encode($content)), $url); |
|
100
|
|
|
|
|
101
|
4 |
|
return $adData->getAll(); |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* Dynamique method to retrive the data by url OR id and category |
|
106
|
|
|
* |
|
107
|
|
|
* @return bool|mixed |
|
108
|
|
|
*/ |
|
109
|
6 |
|
public function ad() |
|
110
|
|
|
{ |
|
111
|
6 |
|
if (func_num_args() === 1) { |
|
112
|
4 |
|
return call_user_func_array([$this, 'adByUrl'], func_get_args()); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
4 |
|
if (func_num_args() === 2) { |
|
116
|
2 |
|
return call_user_func_array([$this, 'adById'], func_get_args()); |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
2 |
|
throw new \InvalidArgumentException('Bad number of argument'); |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
|
rfussien /
leboncoin-crawler
