|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpCfdi\SatCatalogosPopulate\Origins; |
|
6
|
|
|
|
|
7
|
|
|
use LogicException; |
|
8
|
|
|
use RuntimeException; |
|
9
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
10
|
|
|
|
|
11
|
|
|
class ScrapingReviewer implements ReviewerInterface |
|
12
|
|
|
{ |
|
13
|
|
|
/** @var ResourcesGatewayInterface */ |
|
14
|
|
|
private $gateway; |
|
15
|
|
|
|
|
16
|
|
|
public function __construct(ResourcesGatewayInterface $gateway) |
|
17
|
|
|
{ |
|
18
|
|
|
$this->gateway = $gateway; |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
public function gateway(): ResourcesGatewayInterface |
|
22
|
|
|
{ |
|
23
|
|
|
return $this->gateway; |
|
24
|
|
|
} |
|
25
|
|
|
|
|
26
|
|
|
public function accepts(OriginInterface $origin): bool |
|
27
|
|
|
{ |
|
28
|
|
|
return ($origin instanceof ScrapingOrigin); |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
public function review(OriginInterface $origin): Review |
|
32
|
|
|
{ |
|
33
|
|
|
if (! $origin instanceof ScrapingOrigin) { |
|
34
|
|
|
throw new LogicException('This reviewer can only handle ScrapingOrigin objects'); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
if (! $origin->hasDownloadUrl()) { |
|
38
|
|
|
try { |
|
39
|
|
|
$origin = $this->resolveOrigin($origin); |
|
40
|
|
|
} catch (RuntimeException $exception) { |
|
41
|
|
|
return new Review($origin, ReviewStatus::notFound()); |
|
42
|
|
|
} |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
$response = $this->gateway->headers($origin->downloadUrl()); |
|
46
|
|
|
|
|
47
|
|
|
// si no se pudo obtener el recurso |
|
48
|
|
|
if (! $response->isSuccess()) { |
|
49
|
|
|
return new Review($origin, ReviewStatus::notFound()); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
// si el recurso no coincide con la última versión |
|
53
|
|
|
if (! $origin->hasLastVersion() || ! $response->dateMatch($origin->lastVersion())) { |
|
54
|
|
|
return new Review($origin, ReviewStatus::notUpdated()); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
// entonces el recurso coincide |
|
58
|
|
|
return new Review($origin, ReviewStatus::uptodate()); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
public function resolveOrigin(ScrapingOrigin $origin): ScrapingOrigin |
|
62
|
|
|
{ |
|
63
|
|
|
$baseResource = $this->gateway->get($origin->url(), ''); |
|
64
|
|
|
$downloadUrl = $this->resolveHtmlToLink($baseResource, $origin->linkText()); |
|
65
|
|
|
return $origin->withDownloadUrl($downloadUrl); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
public function resolveHtmlToLink(UrlResponse $response, string $linkText): string |
|
69
|
|
|
{ |
|
70
|
|
|
if (empty($response->body())) { |
|
71
|
|
|
throw new RuntimeException('Content is empty'); |
|
72
|
|
|
} |
|
73
|
|
|
$crawler = new Crawler($response->body(), $response->url()); |
|
74
|
|
|
$link = $crawler->selectLink($linkText)->link(); |
|
75
|
|
|
$downloadUrl = $link->getUri(); |
|
76
|
|
|
if (empty($downloadUrl)) { |
|
77
|
|
|
throw new RuntimeException('The link was found but it does not contains the url to download'); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
return $downloadUrl; |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
phpcfdi /
sat-catalogos-populate
