|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Lbc\Parser; |
|
4
|
|
|
|
|
5
|
|
|
use League\Uri\Modifiers\MergeQuery; |
|
6
|
|
|
use League\Uri\Modifiers\RemoveQueryKeys; |
|
7
|
|
|
use League\Uri\Schemes\Http; |
|
8
|
|
|
use Psr\Http\Message\UriInterface; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Class SearchResultUrlParser |
|
12
|
|
|
* @package Lbc\Parser |
|
13
|
|
|
*/ |
|
14
|
|
|
class SearchResultUrlParser |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* @var Http |
|
18
|
|
|
*/ |
|
19
|
|
|
protected $url; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @var int |
|
23
|
|
|
*/ |
|
24
|
|
|
protected $nbPages; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @param string $url |
|
28
|
|
|
* @param int $nbPages |
|
29
|
|
|
*/ |
|
30
|
44 |
|
public function __construct($url, $nbPages = 1) |
|
31
|
|
|
{ |
|
32
|
44 |
|
$this->url = Http::createFromString($url); |
|
33
|
|
|
|
|
34
|
44 |
|
$this->nbPages = $nbPages; |
|
35
|
44 |
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @return Http |
|
39
|
|
|
*/ |
|
40
|
36 |
|
public function current() |
|
41
|
|
|
{ |
|
42
|
|
|
// set the default page to 1 unless it is set |
|
43
|
36 |
|
if (!$this->url->query->hasKey('o')) { |
|
44
|
24 |
|
$newUrl = new MergeQuery('o=1'); |
|
45
|
24 |
|
$this->url = $newUrl($this->url); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
// remove th (thumb image) |
|
49
|
36 |
|
$newUrl = new RemoveQueryKeys(['th']); |
|
50
|
36 |
|
$this->url = $newUrl($this->url); |
|
51
|
|
|
|
|
52
|
36 |
|
return $this->url; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Return the next page URL or null if non. |
|
57
|
|
|
* |
|
58
|
|
|
* @return UriInterface |
|
59
|
|
|
*/ |
|
60
|
10 |
View Code Duplication |
public function next() |
|
61
|
|
|
{ |
|
62
|
10 |
|
if ((int) $this->current()->query->getValue('o') >= $this->nbPages) { |
|
63
|
2 |
|
return null; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
8 |
|
return $this->getIndexUrl(+1); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Return the previous page URL or null if none |
|
71
|
|
|
* |
|
72
|
|
|
* @return UriInterface |
|
73
|
|
|
*/ |
|
74
|
10 |
View Code Duplication |
public function previous() |
|
75
|
|
|
{ |
|
76
|
10 |
|
if ((int) $this->current()->query->getValue('o') === 1) { |
|
77
|
4 |
|
return null; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
8 |
|
return $this->getIndexUrl(-1); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* @param int $index |
|
85
|
|
|
* |
|
86
|
|
|
* @return UriInterface |
|
87
|
|
|
*/ |
|
88
|
10 |
|
public function getIndexUrl($index) |
|
89
|
|
|
{ |
|
90
|
10 |
|
$oParam = (int) $this->current()->query->getValue('o') + $index; |
|
91
|
10 |
|
$newQuery = new MergeQuery('o='.$oParam); |
|
92
|
|
|
|
|
93
|
10 |
|
return $newQuery($this->url); |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* Return a meta array containing the nav links and the page |
|
98
|
|
|
* |
|
99
|
|
|
* @return array |
|
100
|
|
|
*/ |
|
101
|
6 |
|
public function getNav() |
|
102
|
|
|
{ |
|
103
|
|
|
return [ |
|
104
|
6 |
|
'page' => (int) $this->current()->query->getValue('o'), |
|
105
|
|
|
'links' => [ |
|
106
|
6 |
|
'current' => (string) $this->current(), |
|
107
|
6 |
|
'previous' => (string) $this->previous(), |
|
108
|
6 |
|
'next' => (string) $this->next(), |
|
109
|
|
|
] |
|
110
|
|
|
]; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
/** |
|
114
|
|
|
* Return the category |
|
115
|
|
|
* |
|
116
|
|
|
* @return string |
|
117
|
|
|
*/ |
|
118
|
10 |
|
public function getCategory() |
|
119
|
|
|
{ |
|
120
|
10 |
|
$category = $this->current()->path->getSegment(0); |
|
121
|
|
|
|
|
122
|
10 |
|
if ($category === 'annonces') { |
|
123
|
4 |
|
return null; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
6 |
|
return $category; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* Return the search area |
|
131
|
|
|
* |
|
132
|
|
|
* @return string |
|
133
|
|
|
*/ |
|
134
|
8 |
|
public function getSearchArea() |
|
135
|
|
|
{ |
|
136
|
8 |
|
if ($this->current()->path->getSegment(3) === 'occasions') { |
|
137
|
2 |
|
return 'toute la france'; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
8 |
|
if ($this->current()->path->getSegment(3) === 'bonnes_affaires') { |
|
141
|
2 |
|
return 'regions voisines'; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
8 |
|
return $this->current()->path->getSegment(2); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Return one or more location (separated by a comma) or null if none |
|
149
|
|
|
* |
|
150
|
|
|
* @return null |
|
151
|
|
|
*/ |
|
152
|
10 |
|
public function getLocation() |
|
153
|
|
|
{ |
|
154
|
10 |
|
return $this->current()->query->getValue('location', null); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
/** |
|
158
|
|
|
* Return the type of the ads |
|
159
|
|
|
* |
|
160
|
|
|
* @return string all|part|pro |
|
161
|
|
|
*/ |
|
162
|
8 |
|
public function getType() |
|
163
|
|
|
{ |
|
164
|
8 |
|
switch ($this->current()->query->getValue('f')) { |
|
165
|
8 |
|
case 'p': |
|
166
|
2 |
|
return 'part'; |
|
167
|
8 |
|
case 'c': |
|
168
|
2 |
|
return 'pro'; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
8 |
|
return 'all'; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
/** |
|
175
|
|
|
* Return the sorting type |
|
176
|
|
|
* |
|
177
|
|
|
* @return string price|date |
|
178
|
|
|
*/ |
|
179
|
8 |
|
public function getSortType() |
|
180
|
|
|
{ |
|
181
|
8 |
|
if ((int) $this->current()->query->getValue('sp') === 1) { |
|
182
|
2 |
|
return 'price'; |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
8 |
|
return 'date'; |
|
186
|
|
|
} |
|
187
|
|
|
} |
|
188
|
|
|
|
rfussien /
leboncoin-crawler
