|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* (c) Christian Gripp <[email protected]> |
|
5
|
|
|
* |
|
6
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
7
|
|
|
* file that was distributed with this source code. |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace Core23\LastFm\Crawler; |
|
11
|
|
|
|
|
12
|
|
|
use Core23\LastFm\Model\Event; |
|
13
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
14
|
|
|
|
|
15
|
|
|
final class ArtistEventCrawler extends AbstractCrawler implements ArtistEventCrawlerInterface |
|
16
|
|
|
{ |
|
17
|
|
|
public function getArtistYears(string $artist): array |
|
18
|
|
|
{ |
|
19
|
|
|
$node = $this->crawlUrl($artist); |
|
20
|
|
|
|
|
21
|
|
|
if (null === $node) { |
|
22
|
|
|
return []; |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
$years = $node->filter('.content-top .secondary-nav-item-link') |
|
26
|
|
|
->each(static function (Crawler $node) { |
|
27
|
|
|
return (int) trim($node->text()); |
|
28
|
|
|
}) |
|
29
|
|
|
; |
|
30
|
|
|
|
|
31
|
|
|
sort($years); |
|
32
|
|
|
array_shift($years); |
|
33
|
|
|
|
|
34
|
|
|
return $years; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function getEvents(string $artist, ?int $year, int $page = 1): array |
|
38
|
|
|
{ |
|
39
|
|
|
$node = $this->crawlUrl($artist, $year, $page); |
|
40
|
|
|
|
|
41
|
|
|
if (null === $node) { |
|
42
|
|
|
return []; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
return $node->filter('.events-list-item')->each(function (Crawler $node): Event { |
|
46
|
|
|
return $this->parseEvent($node); |
|
47
|
|
|
}); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
public function getYearPages(string $artist, ?int $year): int |
|
51
|
|
|
{ |
|
52
|
|
|
$node = $this->crawlUrl($artist, $year); |
|
53
|
|
|
|
|
54
|
|
|
if (null === $node) { |
|
55
|
|
|
return 0; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
return $this->countListPages($node); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
public function getYearCount(string $artist, ?int $year, int $page = 1): int |
|
62
|
|
|
{ |
|
63
|
|
|
$node = $this->crawlUrl($artist, $year, $page); |
|
64
|
|
|
|
|
65
|
|
|
if (null === $node) { |
|
66
|
|
|
return 0; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
$perPage = $this->countListEvents($node); |
|
70
|
|
|
$pages = $this->countListPages($node); |
|
71
|
|
|
|
|
72
|
|
|
$node = $this->crawlUrl($artist, $year, $pages); |
|
73
|
|
|
|
|
74
|
|
|
if (null === $node) { |
|
75
|
|
|
return $perPage; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
$count = $this->countListEvents($node); |
|
79
|
|
|
|
|
80
|
|
|
return ($pages - 1) * $perPage + $count; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
private function countListPages(Crawler $node): int |
|
84
|
|
|
{ |
|
85
|
|
|
$pagination = $this->parseString($node->filter('.pagination .pagination-page')->last()); |
|
86
|
|
|
|
|
87
|
|
|
return null !== $pagination ? (int) preg_replace('/.* of /', '', $pagination) : 1; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
private function countListEvents(Crawler $node): int |
|
91
|
|
|
{ |
|
92
|
|
|
return $node->filter('.events-list-item')->count(); |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
private function crawlUrl(string $artist, ?int $year = null, int $page = 1): ?Crawler |
|
96
|
|
|
{ |
|
97
|
|
|
$url = 'https://www.last.fm/artist/'.$artist.'/+events/'.($year ?: ''); |
|
98
|
|
|
|
|
99
|
|
|
return $this->crawl($url, [ |
|
100
|
|
|
'page' => $page, |
|
101
|
|
|
]); |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
core23 /
lastfm-php-api
