|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SomeWork\Minjust\Parser; |
|
4
|
|
|
|
|
5
|
|
|
use PHPHtmlParser\Dom; |
|
6
|
|
|
use PHPHtmlParser\Dom\Collection; |
|
7
|
|
|
use PHPHtmlParser\Dom\HtmlNode; |
|
8
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
|
9
|
|
|
use PHPHtmlParser\Exceptions\NotLoadedException; |
|
10
|
|
|
use SomeWork\Minjust\Entity\DetailLawyer; |
|
11
|
|
|
use SomeWork\Minjust\Entity\LawFormation; |
|
12
|
|
|
use SomeWork\Minjust\Entity\Lawyer; |
|
13
|
|
|
use SomeWork\Minjust\FindResponse; |
|
14
|
|
|
|
|
15
|
|
|
class DomParser implements ParserInterface |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* @var string |
|
19
|
|
|
*/ |
|
20
|
|
|
protected const CURRENT_PAGE_SELECTOR = 'span.currentStep'; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var string |
|
24
|
|
|
*/ |
|
25
|
|
|
protected const PAGINATION_BLOCK_SELECTOR = 'div.pagination'; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var string |
|
29
|
|
|
*/ |
|
30
|
|
|
protected const PAGINATION_STEP_SELECTOR = 'a.step'; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var string |
|
34
|
|
|
*/ |
|
35
|
|
|
protected const LAWYERS_LIST_BLOCK_SELECTOR = 'table.persons > tbody > tr'; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var string |
|
39
|
|
|
*/ |
|
40
|
|
|
protected const LAWYER_DETAIL_SELECTOR = '.floating > p.row'; |
|
41
|
|
|
|
|
42
|
5 |
|
public function list(string $body): FindResponse |
|
43
|
|
|
{ |
|
44
|
5 |
|
$dom = (new Dom())->loadStr($body); |
|
45
|
|
|
|
|
46
|
5 |
|
return (new FindResponse()) |
|
47
|
5 |
|
->setPage($this->getCurrentPage($dom)) |
|
48
|
5 |
|
->setTotalPage($this->getTotalPage($dom)) |
|
49
|
5 |
|
->setLawyers($this->getListLawyers($dom)); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
5 |
|
protected function getCurrentPage(Dom $dom): int |
|
53
|
|
|
{ |
|
54
|
5 |
|
if ($span = $dom->find(static::CURRENT_PAGE_SELECTOR, 0)) { |
|
55
|
4 |
|
return (int) $span->text(); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
1 |
|
return 1; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
4 |
|
protected function getTotalPage(Dom $dom): int |
|
62
|
|
|
{ |
|
63
|
|
|
/** |
|
64
|
|
|
* @var HtmlNode[] $collection |
|
65
|
|
|
*/ |
|
66
|
|
|
$collection = $dom |
|
67
|
4 |
|
->find(static::PAGINATION_BLOCK_SELECTOR, 0) |
|
68
|
4 |
|
->find(static::PAGINATION_STEP_SELECTOR) |
|
69
|
4 |
|
->toArray(); |
|
70
|
4 |
|
if (0 === count($collection)) { |
|
71
|
1 |
|
return 1; |
|
72
|
|
|
} |
|
73
|
3 |
|
$lastStep = (int) end($collection)->text(); |
|
74
|
3 |
|
$currentPage = $this->getCurrentPage($dom); |
|
75
|
|
|
|
|
76
|
3 |
|
return $lastStep > $currentPage ? $lastStep : $currentPage; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param \PHPHtmlParser\Dom $dom |
|
81
|
|
|
* |
|
82
|
|
|
* @return Lawyer[] |
|
83
|
|
|
* @throws ChildNotFoundException |
|
84
|
|
|
* @throws NotLoadedException |
|
85
|
|
|
*/ |
|
86
|
5 |
|
protected function getListLawyers(Dom $dom): array |
|
87
|
|
|
{ |
|
88
|
5 |
|
$data = []; |
|
89
|
|
|
/** |
|
90
|
|
|
* @var Dom\HtmlNode[]|Collection $nodes |
|
91
|
|
|
*/ |
|
92
|
5 |
|
$nodes = $dom->find(static::LAWYERS_LIST_BLOCK_SELECTOR); |
|
93
|
5 |
|
foreach ($nodes as $node) { |
|
94
|
|
|
/** |
|
95
|
|
|
* @var Dom\HtmlNode[]|Collection $tds |
|
96
|
|
|
*/ |
|
97
|
5 |
|
$tds = $node->find('td'); |
|
98
|
|
|
$tds = array_filter($tds->toArray(), static function (HtmlNode $node) { |
|
99
|
5 |
|
return $node->outerHtml() !== '' && $node->getAttribute('class') !== 'empty'; |
|
100
|
5 |
|
}); |
|
101
|
5 |
|
$data[] = (new Lawyer()) |
|
102
|
5 |
|
->setRegisterNumber($tds[3]->text()) |
|
103
|
5 |
|
->setFullName($tds[4]->text(true)) |
|
104
|
5 |
|
->setUrl($tds[4]->firstChild()->getAttribute('href')) |
|
105
|
5 |
|
->setTerritorialSubject($tds[5]->text()) |
|
106
|
5 |
|
->setCertificateNumber($tds[6]->text()) |
|
107
|
5 |
|
->setStatus($tds[7]->text()); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
5 |
|
return $data; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
2 |
|
public function detail(string $body): DetailLawyer |
|
114
|
|
|
{ |
|
115
|
2 |
|
$dom = (new Dom())->loadStr($body); |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @var Dom\HtmlNode[] $nodes |
|
119
|
|
|
*/ |
|
120
|
2 |
|
$nodes = $dom->find(static::LAWYER_DETAIL_SELECTOR)->toArray(); |
|
121
|
|
|
|
|
122
|
2 |
|
$lawyer = (new DetailLawyer()) |
|
123
|
2 |
|
->setChamberOfLaw(trim($nodes[5]->text())); |
|
124
|
2 |
|
if (($organizationForm = trim($nodes[7]->text())) !== '') { |
|
125
|
1 |
|
$lawyer->setLawFormation( |
|
126
|
1 |
|
(new LawFormation()) |
|
127
|
1 |
|
->setOrganizationalForm($organizationForm) |
|
128
|
1 |
|
->setName(trim($nodes[9]->text())) |
|
129
|
1 |
|
->setAddress(trim($nodes[11]->text())) |
|
130
|
1 |
|
->setPhone(trim($nodes[13]->text())) |
|
131
|
1 |
|
->setEmail(trim($nodes[15]->text())) |
|
132
|
|
|
); |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
2 |
|
return $lawyer; |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
somework /
minjust
