|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace App\Movies\Service; |
|
6
|
|
|
|
|
7
|
|
|
use App\Movies\Entity\Movie; |
|
8
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
9
|
|
|
|
|
10
|
|
|
class ImdbReleaseDateParserService |
|
11
|
|
|
{ |
|
12
|
|
|
private const IMDB_RELEASE_DATES_URL = 'https://www.imdb.com/title/{id}/releaseinfo?ref_=tt_dt_dt'; |
|
13
|
|
|
|
|
14
|
|
|
private $imdbMapper; |
|
15
|
|
|
|
|
16
|
|
|
public function __construct(ImdbDataMapper $mapper) |
|
17
|
|
|
{ |
|
18
|
|
|
$this->imdbMapper = $mapper; |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
public function getReleaseDates(Movie $movie): array |
|
22
|
|
|
{ |
|
23
|
|
|
if ($movie->getImdbId() === null) { |
|
24
|
|
|
return []; |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
$html = $this->loadImdbReleaseDatesPageHtml($movie->getImdbId()); |
|
28
|
|
|
|
|
29
|
|
|
// Because filterXPath('//*[@id="release_dates"]//td') don't work correctly due errors in html from imdb page |
|
30
|
|
|
$releasesPosition = mb_strpos($html, 'id="releases"'); |
|
31
|
|
|
if ($releasesPosition === false) { |
|
32
|
|
|
// no releases yet on imdb |
|
33
|
|
|
return []; |
|
34
|
|
|
} |
|
35
|
|
|
$html = mb_substr($html, $releasesPosition); |
|
36
|
|
|
|
|
37
|
|
|
$akasPosition = mb_strpos($html, 'id="akas"'); |
|
38
|
|
|
if ($akasPosition === false) { |
|
39
|
|
|
$html = mb_substr($html, 0); |
|
40
|
|
|
} else { |
|
41
|
|
|
$html = mb_substr($html, 0, $akasPosition); |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
$crawler = new Crawler($html, $this->getEndpoint($movie->getImdbId())); |
|
45
|
|
|
|
|
46
|
|
|
$tds = $crawler->filterXPath('//td')->getIterator(); |
|
47
|
|
|
|
|
48
|
|
|
$result = []; |
|
49
|
|
|
$country = ''; |
|
50
|
|
|
foreach ($tds as $td) { |
|
51
|
|
|
if ($td->getAttribute('class') === 'release-date-item__date') { |
|
52
|
|
|
$result[$country] = $this->imdbMapper->dateToObject($td->textContent); |
|
53
|
|
|
continue; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
if ($td->hasChildNodes() && $td->getElementsByTagName('a')->item(0) !== null) { |
|
57
|
|
|
$country = $td->getElementsByTagName('a')->item(0)->textContent; |
|
58
|
|
|
$country = trim($country); |
|
59
|
|
|
$country = $this->imdbMapper->countryToCode($country); |
|
60
|
|
|
continue; |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
return $result; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
private function getEndpoint(string $imdbId): string |
|
68
|
|
|
{ |
|
69
|
|
|
return str_replace('{id}', $imdbId, self::IMDB_RELEASE_DATES_URL); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
private function loadImdbReleaseDatesPageHtml(string $imdbId): string |
|
73
|
|
|
{ |
|
74
|
|
|
$endpoint = $this->getEndpoint($imdbId); |
|
75
|
|
|
|
|
76
|
|
|
$c = curl_init($endpoint); |
|
77
|
|
|
curl_setopt($c, CURLOPT_RETURNTRANSFER, true); |
|
78
|
|
|
|
|
79
|
|
|
$html = curl_exec($c); |
|
80
|
|
|
|
|
81
|
|
|
if ($html === false) { |
|
82
|
|
|
//todo write to log curl_error() |
|
83
|
|
|
return ''; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
curl_close($c); |
|
87
|
|
|
|
|
88
|
|
|
return $html; |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|
svbackend /
my-art-lib
