|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
use Carbon\Carbon; |
|
4
|
|
|
use Goutte\Client; |
|
5
|
|
|
use Illuminate\Database\Eloquent\Collection; |
|
6
|
|
|
use Illuminate\Queue\Jobs\Job; |
|
7
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
8
|
|
|
|
|
9
|
|
|
class EarningsScraperWorker { |
|
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* @type Client |
|
13
|
|
|
*/ |
|
14
|
|
|
private $client; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @param Client $client |
|
18
|
|
|
*/ |
|
19
|
|
|
function __construct(Client $client) { |
|
|
|
|
|
|
20
|
|
|
$this->client = $client; |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* Scrape earnings |
|
25
|
|
|
* |
|
26
|
|
|
* @param Job $job |
|
27
|
|
|
* @param $data |
|
28
|
|
|
*/ |
|
29
|
|
|
public function fire(Job $job, $data) { |
|
30
|
|
|
if ($job->attempts() > 3) { |
|
31
|
|
|
$job->delete(); |
|
32
|
|
|
|
|
33
|
|
|
return; |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** @type Carbon $date */ |
|
37
|
|
|
$date = $data['day'] instanceof Carbon ? $data['day'] : new Carbon($data['day']); |
|
38
|
|
|
list($dateFor, $earnings) = $this->getEarnings($date); |
|
39
|
|
|
|
|
40
|
|
|
$this->storeInteresting($dateFor, $earnings); |
|
41
|
|
|
|
|
42
|
|
|
$job->delete(); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Get earnings data |
|
47
|
|
|
* |
|
48
|
|
|
* @param Carbon $date |
|
49
|
|
|
* |
|
50
|
|
|
* @return array |
|
51
|
|
|
*/ |
|
52
|
|
|
private function getEarnings(Carbon $date) { |
|
53
|
|
|
$crawler = $this->client->request('GET', $this->getUrl($date), [], [], [ |
|
54
|
|
|
'HTTP_USER_AGENT' => "Mozilla/5.0 (Windows NT 10.0; WOW64; rv:45.0) Gecko/20100101 Firefox/45.0" |
|
55
|
|
|
]); |
|
56
|
|
|
|
|
57
|
|
|
$result_date = new Carbon($crawler->filter('font[face="Verdana"][size="4"] > b')->text()); |
|
58
|
|
|
$movies = $crawler->filter('#body table[border="0"][cellpadding="5"] tr:not([bgcolor="#dcdcdc"])')->each(function (Crawler $node) { |
|
59
|
|
|
$cols = $node->children(); |
|
60
|
|
|
|
|
61
|
|
|
$url = $cols->eq(2)->children()->eq(0)->children()->eq(0)->attr('href'); |
|
62
|
|
|
|
|
63
|
|
|
$info = [ |
|
64
|
|
|
'boxmojo_id' => str_replace('/movies/?page=daily&id=', '', str_replace('.htm', '', $url)), |
|
65
|
|
|
'domestic_total' => intval(str_replace(['$', ','], '', $cols->eq(9)->text())), |
|
66
|
|
|
]; |
|
67
|
|
|
|
|
68
|
|
|
return $info; |
|
69
|
|
|
}); |
|
70
|
|
|
|
|
71
|
|
|
return [$result_date, $movies]; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* Get the URL to scrape |
|
76
|
|
|
* |
|
77
|
|
|
* @param Carbon $date |
|
78
|
|
|
* |
|
79
|
|
|
* @return string |
|
80
|
|
|
*/ |
|
81
|
|
|
private function getUrl(Carbon $date) { |
|
82
|
|
|
return 'http://www.boxofficemojo.com/daily/chart/?view=1day&sortdate=' . $date->toDateString(); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* Store |
|
87
|
|
|
* |
|
88
|
|
|
* @param Carbon $dateFor |
|
89
|
|
|
* @param $earnings |
|
90
|
|
|
*/ |
|
91
|
|
|
private function storeInteresting(Carbon $dateFor, $earnings) { |
|
92
|
|
|
$movies = $this->getActiveMovies($dateFor)->keyBy('boxmojo_id'); |
|
93
|
|
|
|
|
94
|
|
|
foreach ($earnings as $info) { |
|
95
|
|
|
if (! $movies->has($info['boxmojo_id'])) { |
|
96
|
|
|
continue; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** @type Movie $movie */ |
|
100
|
|
|
$movie = $movies->get($info['boxmojo_id']); |
|
101
|
|
|
$earnings = $movie->earnings()->where('date', $dateFor)->first(); |
|
102
|
|
|
if (! $earnings) { |
|
103
|
|
|
$earnings = new MovieEarning([ |
|
104
|
|
|
'movie_id' => $movie->id, |
|
105
|
|
|
'date' => $dateFor, |
|
106
|
|
|
]); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
$earnings->domestic = $info['domestic_total']; |
|
110
|
|
|
|
|
111
|
|
|
if ($earnings->isDirty()) { |
|
112
|
|
|
$earnings->save(); |
|
113
|
|
|
|
|
114
|
|
|
if (! $movie->latestEarnings || $movie->latestEarnings->date < $earnings->date) { |
|
115
|
|
|
$movie->latest_earnings_id = $earnings->id; |
|
116
|
|
|
$movie->save(); |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
// Find any LeagueMovies that might should be updated |
|
120
|
|
|
$leagueMovies = DB::table('league_movies') |
|
121
|
|
|
->where('league_movies.movie_id', $movie->id) |
|
122
|
|
|
->join('leagues', 'league_movies.league_id', '=', 'leagues.id') |
|
123
|
|
|
->where('leagues.end_date', '>', $dateFor) |
|
124
|
|
|
->get(['league_movies.id']); // . infront bypasses clearing of dots, which is actually needed here |
|
125
|
|
|
|
|
126
|
|
|
foreach ($leagueMovies as $leagueMovie) { |
|
127
|
|
|
Queue::push('UpdateLeagueMovieEarnings', [ |
|
128
|
|
|
'league_movie_id' => $leagueMovie->id, 'earnings_id' => $earnings->id |
|
129
|
|
|
]); |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* Get movies that are being used in an active League |
|
139
|
|
|
* |
|
140
|
|
|
* @param Carbon $dateFor |
|
141
|
|
|
* |
|
142
|
|
|
* @return Collection |
|
143
|
|
|
*/ |
|
144
|
|
|
private function getActiveMovies(Carbon $dateFor) { |
|
145
|
|
|
$movies = Movie::query() |
|
146
|
|
|
->where('release', '<', $dateFor) |
|
147
|
|
|
->whereExists(function (\Illuminate\Database\Query\Builder $query) use ($dateFor) { |
|
148
|
|
|
$query->select(DB::raw(1)) |
|
149
|
|
|
->from('league_movies') |
|
150
|
|
|
->join('leagues', 'league_movies.league_id', '=', 'leagues.id') |
|
151
|
|
|
->whereRaw('league_movies.movie_id = movies.id') |
|
152
|
|
|
->where('leagues.end_date', '>', $dateFor); |
|
153
|
|
|
})->get(); |
|
154
|
|
|
|
|
155
|
|
|
return $movies; |
|
156
|
|
|
} |
|
157
|
|
|
} |
t2t2 /
draft
Loading history...
You can fix this by adding a namespace to your class:
When choosing a vendor namespace, try to pick something that is not too generic to avoid conflicts with other libraries.