|
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 Symfony\Component\DomCrawler\Crawler; |
|
13
|
|
|
|
|
14
|
|
|
final class EventListCrawler extends AbstractCrawler |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* @param string $username |
|
18
|
|
|
* |
|
19
|
|
|
* @return int[]|null |
|
20
|
|
|
*/ |
|
21
|
|
|
public function getUserYears(string $username): ? array |
|
22
|
|
|
{ |
|
23
|
|
|
$node = $this->crawlEventList($username); |
|
24
|
|
|
|
|
25
|
|
|
if (null === $node) { |
|
26
|
|
|
return null; |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
$years = $node->filter('.content-top .secondary-nav-item-link') |
|
30
|
|
|
->each(function (Crawler $node) { |
|
31
|
|
|
return (int) trim($node->text()); |
|
32
|
|
|
}); |
|
33
|
|
|
|
|
34
|
|
|
sort($years); |
|
35
|
|
|
array_shift($years); |
|
36
|
|
|
|
|
37
|
|
|
return $years; |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Get all events of a user. |
|
42
|
|
|
* |
|
43
|
|
|
* @param string $username |
|
44
|
|
|
* @param int|null $year |
|
45
|
|
|
* @param int $page |
|
46
|
|
|
* |
|
47
|
|
|
* @return array|null |
|
48
|
|
|
*/ |
|
49
|
|
|
public function getEvents(string $username, ? int $year, int $page = 1): ? array |
|
50
|
|
|
{ |
|
51
|
|
|
$node = $this->crawlEventList($username, $year, $page); |
|
52
|
|
|
|
|
53
|
|
|
if (null === $node) { |
|
54
|
|
|
return null; |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
return $node->filter('.events-list-item')->each(function (Crawler $node): array { |
|
58
|
|
|
$eventNode = $node->filter('.events-list-item-event--title a'); |
|
59
|
|
|
|
|
60
|
|
|
$url = $this->parseUrl($eventNode); |
|
61
|
|
|
$id = preg_replace('/.*\/(\d+)+.*/', '$1', $url); |
|
62
|
|
|
|
|
63
|
|
|
return array( |
|
64
|
|
|
'eventId' => (int) $id, |
|
65
|
|
|
'title' => $this->parseString($eventNode), |
|
66
|
|
|
'time' => new \DateTime($node->filter('time')->attr('datetime')), |
|
67
|
|
|
'url' => $url, |
|
68
|
|
|
); |
|
69
|
|
|
}); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* Gets the pages for a year. |
|
74
|
|
|
* |
|
75
|
|
|
* @param string $username |
|
76
|
|
|
* @param int|null $year |
|
77
|
|
|
* |
|
78
|
|
|
* @return int|null |
|
79
|
|
|
*/ |
|
80
|
|
|
public function getYearPages(string $username, ? int $year): ? int |
|
81
|
|
|
{ |
|
82
|
|
|
$node = $this->crawlEventList($username, $year); |
|
83
|
|
|
|
|
84
|
|
|
if (null === $node) { |
|
85
|
|
|
return null; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
return $this->countListPages($node); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Gets the event count for a year. |
|
93
|
|
|
* |
|
94
|
|
|
* @param string $username |
|
95
|
|
|
* @param int|null $year |
|
96
|
|
|
* @param int $page |
|
97
|
|
|
* |
|
98
|
|
|
* @return int|null |
|
99
|
|
|
*/ |
|
100
|
|
|
public function getYearCount(string $username, ? int $year, int $page = 1): ? int |
|
101
|
|
|
{ |
|
102
|
|
|
$node = $this->crawlEventList($username, $year, $page); |
|
103
|
|
|
|
|
104
|
|
|
if (null === $node) { |
|
105
|
|
|
return null; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
$perPage = $this->countListEvents($node); |
|
109
|
|
|
$pages = $this->countListPages($node); |
|
110
|
|
|
|
|
111
|
|
|
if ($pages) { |
|
112
|
|
|
$node = $this->crawlEventList($username, $year, $pages); |
|
113
|
|
|
|
|
114
|
|
|
if (null === $node) { |
|
115
|
|
|
return $perPage; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
$count = $this->countListEvents($node); |
|
119
|
|
|
|
|
120
|
|
|
return ($pages - 1) * $perPage + $count; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
return $perPage; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
/** |
|
127
|
|
|
* @param Crawler $node |
|
128
|
|
|
* |
|
129
|
|
|
* @return int |
|
130
|
|
|
*/ |
|
131
|
|
|
private function countListPages(Crawler $node): int |
|
132
|
|
|
{ |
|
133
|
|
|
$pagination = $this->parseString($node->filter('.pagination .pages')); |
|
134
|
|
|
|
|
135
|
|
|
return $pagination ? (int) preg_replace('/.* of /', '', $pagination) : 1; |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
/** |
|
139
|
|
|
* @param Crawler $node |
|
140
|
|
|
* |
|
141
|
|
|
* @return int |
|
142
|
|
|
*/ |
|
143
|
|
|
private function countListEvents(Crawler $node): int |
|
144
|
|
|
{ |
|
145
|
|
|
return $node->filter('.events-list-item')->count(); |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
/** |
|
149
|
|
|
* @param string $username |
|
150
|
|
|
* @param int|null $year |
|
151
|
|
|
* @param int $page |
|
152
|
|
|
* |
|
153
|
|
|
* @return Crawler|null |
|
154
|
|
|
*/ |
|
155
|
|
|
private function crawlEventList(string $username, ? int $year = null, int $page = 1): ? Crawler |
|
156
|
|
|
{ |
|
157
|
|
|
$url = 'http://www.last.fm/user/'.$username.'/events/'.($year ?: '').'?page='.$page; |
|
158
|
|
|
|
|
159
|
|
|
return $this->crawl($url); |
|
160
|
|
|
} |
|
161
|
|
|
} |
|
162
|
|
|
|
core23 /
lastfm-php-api
