|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* File: ModelOffersHtmlProvider.php |
|
4
|
|
|
* |
|
5
|
|
|
* @author Maciej Sławik <[email protected]> |
|
6
|
|
|
* Github: https://github.com/maciejslawik |
|
7
|
|
|
*/ |
|
8
|
|
|
|
|
9
|
|
|
namespace MSlwk\Otomoto\App\Stats; |
|
10
|
|
|
|
|
11
|
|
|
use MSlwk\Otomoto\App\Base\UrlPoviderInterface; |
|
12
|
|
|
use MSlwk\Otomoto\App\Manufacturer\Data\ManufacturerDTOInterface; |
|
13
|
|
|
use MSlwk\Otomoto\App\Manufacturer\Url\ManufacturerUrlSuffixProviderInterface; |
|
14
|
|
|
use MSlwk\Otomoto\App\Model\Data\ModelDTOInterface; |
|
15
|
|
|
use MSlwk\Otomoto\App\Model\Url\ModelUrlSuffixProviderInterface; |
|
16
|
|
|
use MSlwk\Otomoto\App\Stats\Filter\FilterArray; |
|
17
|
|
|
use MSlwk\Otomoto\App\Stats\Filter\FilterInterface; |
|
18
|
|
|
use MSlwk\Otomoto\App\Stats\Pager\PagerInterface; |
|
19
|
|
|
use MSlwk\Otomoto\Middleware\Webpage\Adapter\GETHandlerInterface; |
|
20
|
|
|
use MSlwk\Otomoto\Middleware\Webpage\Data\UrlDTO; |
|
21
|
|
|
use MSlwk\Otomoto\Middleware\Webpage\Data\UrlDTOArray; |
|
22
|
|
|
use MSlwk\Otomoto\Middleware\Webpage\Data\WebpageDTOArray; |
|
23
|
|
|
use MSlwk\Otomoto\Middleware\Webpage\Data\WebpageDTOInterface; |
|
24
|
|
|
use MSlwk\Otomoto\Middleware\Webpage\Factory\GETHandlerAdapterFactoryInterface; |
|
25
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Class ModelOffersHtmlProvider |
|
29
|
|
|
* @package MSlwk\Otomoto\App\Stats |
|
30
|
|
|
*/ |
|
31
|
|
|
class ModelOffersHtmlProvider |
|
32
|
|
|
{ |
|
33
|
|
|
const HTML_PAGER_SELECTOR = 'ul.om-pager'; |
|
34
|
|
|
const HTML_PAGER_ELEMENT_SELECTOR = 'span.page'; |
|
35
|
|
|
const HTML_OFFER_LIST_SELECTOR = 'div.offers.list'; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var GETHandlerAdapterFactoryInterface |
|
39
|
|
|
*/ |
|
40
|
|
|
private $adapterFactory; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @var UrlPoviderInterface |
|
44
|
|
|
*/ |
|
45
|
|
|
private $urlPovider; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* @var ManufacturerUrlSuffixProviderInterface |
|
49
|
|
|
*/ |
|
50
|
|
|
private $manufacturerUrlSuffixProvider; |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* @var ModelUrlSuffixProviderInterface |
|
54
|
|
|
*/ |
|
55
|
|
|
private $modelUrlSuffixProvider; |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* @var PagerInterface |
|
59
|
|
|
*/ |
|
60
|
|
|
private $pager; |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* @var GETHandlerInterface |
|
64
|
|
|
*/ |
|
65
|
|
|
private $GETAdapter; |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* ModelOffersHtmlProvider constructor. |
|
69
|
|
|
* @param GETHandlerAdapterFactoryInterface $adapterFactory |
|
70
|
|
|
* @param UrlPoviderInterface $urlPovider |
|
71
|
|
|
* @param ManufacturerUrlSuffixProviderInterface $manufacturerUrlSuffixProvider |
|
72
|
|
|
* @param ModelUrlSuffixProviderInterface $modelUrlSuffixProvider |
|
73
|
|
|
* @param PagerInterface $pager |
|
74
|
|
|
*/ |
|
75
|
2 |
|
public function __construct( |
|
76
|
|
|
GETHandlerAdapterFactoryInterface $adapterFactory, |
|
77
|
|
|
UrlPoviderInterface $urlPovider, |
|
78
|
|
|
ManufacturerUrlSuffixProviderInterface $manufacturerUrlSuffixProvider, |
|
79
|
|
|
ModelUrlSuffixProviderInterface $modelUrlSuffixProvider, |
|
80
|
|
|
PagerInterface $pager |
|
81
|
|
|
) { |
|
82
|
2 |
|
$this->adapterFactory = $adapterFactory; |
|
83
|
2 |
|
$this->urlPovider = $urlPovider; |
|
84
|
2 |
|
$this->manufacturerUrlSuffixProvider = $manufacturerUrlSuffixProvider; |
|
85
|
2 |
|
$this->modelUrlSuffixProvider = $modelUrlSuffixProvider; |
|
86
|
2 |
|
$this->pager = $pager; |
|
87
|
2 |
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* @param ManufacturerDTOInterface $manufacturer |
|
91
|
|
|
* @param ModelDTOInterface $model |
|
92
|
|
|
* @param FilterArray $filters |
|
93
|
|
|
* @return string |
|
94
|
|
|
*/ |
|
95
|
1 |
|
public function getAllPagesHtml( |
|
96
|
|
|
ManufacturerDTOInterface $manufacturer, |
|
97
|
|
|
ModelDTOInterface $model, |
|
98
|
|
|
FilterArray $filters |
|
99
|
|
|
): string { |
|
100
|
1 |
|
$baseUrl = $this->buildBaseUrl($manufacturer, $model, $filters); |
|
101
|
1 |
|
$this->GETAdapter = $this->adapterFactory->create(); |
|
102
|
|
|
|
|
103
|
1 |
|
$baseUrlDTO = new UrlDTOArray(new UrlDTO($baseUrl)); |
|
104
|
1 |
|
$baseWebpageDTO = $this->GETAdapter->getWebpages($baseUrlDTO)->current(); |
|
105
|
|
|
|
|
106
|
1 |
|
$lastPage = $this->getLastPageNumber($baseWebpageDTO); |
|
107
|
1 |
|
$allPagesDTOs = new WebpageDTOArray(); |
|
108
|
1 |
|
if ($lastPage >= 2) { |
|
109
|
1 |
|
$allPagesDTOs = $this->getNextPages($lastPage, $baseUrl); |
|
110
|
|
|
} |
|
111
|
1 |
|
$allPagesDTOs->add($baseWebpageDTO); |
|
112
|
|
|
|
|
113
|
1 |
|
$html = $this->extractOffersHtml($allPagesDTOs); |
|
114
|
|
|
|
|
115
|
1 |
|
return $html; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* @param ManufacturerDTOInterface $manufacturer |
|
120
|
|
|
* @param ModelDTOInterface $model |
|
121
|
|
|
* @param FilterArray $filters |
|
122
|
|
|
* @return string |
|
123
|
|
|
*/ |
|
124
|
1 |
|
private function buildBaseUrl( |
|
125
|
|
|
ManufacturerDTOInterface $manufacturer, |
|
126
|
|
|
ModelDTOInterface $model, |
|
127
|
|
|
FilterArray $filters |
|
128
|
|
|
): string { |
|
129
|
1 |
|
$baseUrl = $this->urlPovider->getBaseUrl(); |
|
130
|
1 |
|
$baseUrl .= $this->manufacturerUrlSuffixProvider->getUrlSuffix($manufacturer); |
|
131
|
1 |
|
$baseUrl .= $this->modelUrlSuffixProvider->getUrlSuffix($model); |
|
132
|
1 |
|
$baseUrl .= '?'; |
|
133
|
|
|
/** @var FilterInterface $filter */ |
|
134
|
1 |
|
foreach ($filters as $filter) { |
|
135
|
1 |
|
$baseUrl .= "{$filter->getName()}={$filter->getValue()}&"; |
|
136
|
|
|
} |
|
137
|
1 |
|
return $baseUrl; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* @param WebpageDTOInterface $baseWebpageDTO |
|
142
|
|
|
* @return int |
|
143
|
|
|
*/ |
|
144
|
1 |
|
private function getLastPageNumber(WebpageDTOInterface $baseWebpageDTO): int |
|
145
|
|
|
{ |
|
146
|
1 |
|
$crawler = new Crawler($baseWebpageDTO->getHtml()); |
|
147
|
|
|
$pagerListItems = $crawler |
|
148
|
1 |
|
->filter(self::HTML_PAGER_SELECTOR) |
|
149
|
1 |
|
->filter(self::HTML_PAGER_ELEMENT_SELECTOR); |
|
150
|
1 |
|
$lastPagerItem = $pagerListItems->getNode($pagerListItems->count() - 2); |
|
151
|
1 |
|
return $lastPagerItem ? (int)$lastPagerItem->textContent : 0; |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
|
/** |
|
155
|
|
|
* @param int $lastPage |
|
156
|
|
|
* @param string $baseUrl |
|
157
|
|
|
* @return WebpageDTOArray |
|
158
|
|
|
*/ |
|
159
|
1 |
|
private function getNextPages(int $lastPage, string $baseUrl): WebpageDTOArray |
|
160
|
|
|
{ |
|
161
|
1 |
|
$urlDTOs = new UrlDTOArray(); |
|
162
|
1 |
|
for ($i = 2; $i <= $lastPage; $i++) { |
|
163
|
1 |
|
$urlDTO = new UrlDTO($baseUrl . $this->pager->getPagerParameter($i)); |
|
164
|
1 |
|
$urlDTOs->add($urlDTO); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
1 |
|
$allPagesDTOs = $this->GETAdapter->getWebpages($urlDTOs); |
|
168
|
1 |
|
return $allPagesDTOs; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
/** |
|
172
|
|
|
* @param WebpageDTOArray $allPagesDTOs |
|
173
|
|
|
* @return string |
|
174
|
|
|
*/ |
|
175
|
1 |
|
private function extractOffersHtml(WebpageDTOArray $allPagesDTOs): string |
|
176
|
|
|
{ |
|
177
|
1 |
|
$html = ''; |
|
178
|
1 |
|
foreach ($allPagesDTOs as $pageDTO) { |
|
179
|
1 |
|
$crawler = new Crawler($pageDTO->getHtml()); |
|
180
|
1 |
|
$offersList = $crawler->filter(self::HTML_OFFER_LIST_SELECTOR); |
|
181
|
1 |
|
if ($offersList->count()) { |
|
182
|
1 |
|
$html .= $offersList->html(); |
|
183
|
|
|
} |
|
184
|
|
|
} |
|
185
|
1 |
|
return $html; |
|
186
|
|
|
} |
|
187
|
|
|
} |
|
188
|
|
|
|