|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Services; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\Client; |
|
6
|
|
|
use GuzzleHttp\Exception\GuzzleException; |
|
7
|
|
|
use Illuminate\Support\Facades\Cache; |
|
8
|
|
|
use Illuminate\Support\Facades\Log; |
|
9
|
|
|
use voku\helper\HtmlDomParser; |
|
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
class ImdbScraper |
|
12
|
|
|
{ |
|
13
|
|
|
protected Client $client; |
|
14
|
|
|
|
|
15
|
|
|
public function __construct() |
|
16
|
|
|
{ |
|
17
|
|
|
$this->client = new Client([ |
|
18
|
|
|
'timeout' => 10, |
|
19
|
|
|
'headers' => [ |
|
20
|
|
|
'User-Agent' => 'Mozilla/5.0 (compatible; NNTmuxBot/1.0; +https://github.com/NNTmux)', |
|
21
|
|
|
], |
|
22
|
|
|
]); |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Fetch a movie by IMDB numeric ID. |
|
27
|
|
|
* |
|
28
|
|
|
* @param string $id Numeric part without 'tt'. |
|
29
|
|
|
*/ |
|
30
|
|
|
public function fetchById(string $id): array|false |
|
31
|
|
|
{ |
|
32
|
|
|
$id = preg_replace('/[^0-9]/', '', $id); |
|
33
|
|
|
if ($id === '') { |
|
34
|
|
|
return false; |
|
35
|
|
|
} |
|
36
|
|
|
$cacheKey = 'imdb_scrape_id_'.$id; |
|
37
|
|
|
if (Cache::has($cacheKey)) { |
|
38
|
|
|
return Cache::get($cacheKey); |
|
39
|
|
|
} |
|
40
|
|
|
$url = 'https://www.imdb.com/title/tt'.$id.'/'; |
|
41
|
|
|
try { |
|
42
|
|
|
$res = $this->client->get($url); |
|
43
|
|
|
$html = (string) $res->getBody(); |
|
44
|
|
|
$dom = HtmlDomParser::str_get_html($html); |
|
45
|
|
|
if (! $dom) { |
|
46
|
|
|
Cache::put($cacheKey, false, now()->addHours(6)); |
|
47
|
|
|
|
|
48
|
|
|
return false; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
$titleNode = $dom->findOne('h1'); |
|
52
|
|
|
$title = trim($titleNode?->text() ?? ''); |
|
53
|
|
|
if ($title === '') { |
|
54
|
|
|
Cache::put($cacheKey, false, now()->addHours(6)); |
|
55
|
|
|
|
|
56
|
|
|
return false; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
// Year |
|
60
|
|
|
$year = ''; |
|
61
|
|
|
$yearNode = $dom->findOne("span[data-testid='title-details-releasedate'] a"); |
|
62
|
|
|
if ($yearNode) { |
|
63
|
|
|
if (preg_match('/(19|20)\d{2}/', $yearNode->text(), $m)) { |
|
64
|
|
|
$year = $m[0]; |
|
65
|
|
|
} |
|
66
|
|
|
} |
|
67
|
|
|
if ($year === '') { |
|
68
|
|
|
$altYearNode = $dom->findOne("span[data-testid='title-details-releasedate']"); |
|
69
|
|
|
if ($altYearNode && preg_match('/(19|20)\d{2}/', $altYearNode->text(), $m)) { |
|
70
|
|
|
$year = $m[0]; |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
// Plot |
|
75
|
|
|
$plot = ''; |
|
76
|
|
|
$plotNode = $dom->findOne("span[data-testid='plot-l']"); |
|
77
|
|
|
if (! $plotNode) { |
|
78
|
|
|
$plotNode = $dom->findOne("span[data-testid='plot-xl']"); |
|
79
|
|
|
} |
|
80
|
|
|
if ($plotNode) { |
|
81
|
|
|
$plot = trim($plotNode->text()); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
// Rating |
|
85
|
|
|
$rating = ''; |
|
86
|
|
|
$ratingNode = $dom->findOne("div[data-testid='hero-rating-bar__aggregate-rating__score'] span"); |
|
87
|
|
|
if ($ratingNode) { |
|
88
|
|
|
$rating = trim($ratingNode->text()); |
|
89
|
|
|
} |
|
90
|
|
|
$rating = preg_replace('/[^0-9\.]/', '', $rating); |
|
91
|
|
|
|
|
92
|
|
|
// Poster |
|
93
|
|
|
$cover = ''; |
|
94
|
|
|
$posterNode = $dom->findOne("div[data-testid='hero-media__poster'] img"); |
|
95
|
|
|
if ($posterNode) { |
|
96
|
|
|
$cover = $posterNode->getAttribute('src'); |
|
97
|
|
|
} |
|
98
|
|
|
if ($cover !== '' && str_contains($cover, '._V1_')) { |
|
99
|
|
|
// Attempt higher res by removing size suffix |
|
100
|
|
|
$cover = preg_replace('/\._V1_.*\./', '.', $cover); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
// Genres |
|
104
|
|
|
$genres = []; |
|
105
|
|
|
foreach ($dom->find("div[data-testid='genres'] a") as $g) { |
|
106
|
|
|
$gt = trim($g->text()); |
|
107
|
|
|
if ($gt !== '') { |
|
108
|
|
|
$genres[] = $gt; |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
// Directors |
|
113
|
|
|
$directors = []; |
|
114
|
|
|
foreach ($dom->find("li[data-testid='title-pc-principal-credit']:contains(Director) a") as $d) { |
|
115
|
|
|
$dt = trim($d->text()); |
|
116
|
|
|
if ($dt !== '') { |
|
117
|
|
|
$directors[] = $dt; |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
if (empty($directors)) { |
|
121
|
|
|
foreach ($dom->find("li[data-testid='title-pc-principal-credit']:contains(Creator) a") as $d) { |
|
122
|
|
|
$dt = trim($d->text()); |
|
123
|
|
|
if ($dt !== '') { |
|
124
|
|
|
$directors[] = $dt; |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
// Actors (top billed) |
|
130
|
|
|
$actors = []; |
|
131
|
|
|
foreach ($dom->find("div[data-testid='title-cast-item'] a[data-testid='title-cast-item__actor']") as $a) { |
|
132
|
|
|
$at = trim($a->text()); |
|
133
|
|
|
if ($at !== '') { |
|
134
|
|
|
$actors[] = $at; |
|
135
|
|
|
} |
|
136
|
|
|
if (count($actors) >= 10) { |
|
137
|
|
|
break; |
|
138
|
|
|
} |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
// Languages (from details section) |
|
142
|
|
|
$language = ''; |
|
143
|
|
|
foreach ($dom->find("li[data-testid='title-details-languages'] a") as $ln) { |
|
144
|
|
|
$lt = trim($ln->text()); |
|
145
|
|
|
if ($lt !== '') { |
|
146
|
|
|
$language .= $lt.', '; |
|
147
|
|
|
} |
|
148
|
|
|
} |
|
149
|
|
|
$language = rtrim($language, ', '); |
|
150
|
|
|
|
|
151
|
|
|
$data = [ |
|
152
|
|
|
'imdbid' => $id, |
|
153
|
|
|
'title' => $title, |
|
154
|
|
|
'year' => $year, |
|
155
|
|
|
'plot' => $plot, |
|
156
|
|
|
'rating' => $rating, |
|
157
|
|
|
'cover' => $cover, |
|
158
|
|
|
'genre' => $genres, |
|
159
|
|
|
'director' => $directors, |
|
160
|
|
|
'actors' => $actors, |
|
161
|
|
|
'language' => $language, |
|
162
|
|
|
'type' => 'movie', |
|
163
|
|
|
]; |
|
164
|
|
|
Cache::put($cacheKey, $data, now()->addDays(7)); |
|
165
|
|
|
|
|
166
|
|
|
return $data; |
|
167
|
|
|
} catch (GuzzleException|\Throwable $e) { |
|
168
|
|
|
Log::debug('IMDb fetch error tt'.$id.': '.$e->getMessage()); |
|
169
|
|
|
Cache::put($cacheKey, false, now()->addHours(6)); |
|
170
|
|
|
|
|
171
|
|
|
return false; |
|
172
|
|
|
} |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
/** |
|
176
|
|
|
* Search IMDb suggestion API for a title. |
|
177
|
|
|
*/ |
|
178
|
|
|
public function search(string $query): array |
|
179
|
|
|
{ |
|
180
|
|
|
$query = trim($query); |
|
181
|
|
|
if ($query === '') { |
|
182
|
|
|
return []; |
|
183
|
|
|
} |
|
184
|
|
|
$norm = strtolower(preg_replace('/[^a-z0-9 ]/i', '', $query)); |
|
185
|
|
|
$slug = str_replace(' ', '_', $norm); |
|
186
|
|
|
$prefix = substr($slug, 0, 1); |
|
187
|
|
|
$cacheKey = 'imdb_search_'.md5($slug); |
|
188
|
|
|
if (Cache::has($cacheKey)) { |
|
189
|
|
|
return Cache::get($cacheKey); |
|
190
|
|
|
} |
|
191
|
|
|
$url = 'https://v2.sg.media-imdb.com/suggestion/'.urlencode($prefix).'/'.urlencode($slug).'.json'; |
|
192
|
|
|
try { |
|
193
|
|
|
$res = $this->client->get($url); |
|
194
|
|
|
$json = json_decode((string) $res->getBody(), true); |
|
195
|
|
|
$results = []; |
|
196
|
|
|
foreach (($json['d'] ?? []) as $row) { |
|
197
|
|
|
if (! isset($row['id']) || ! str_starts_with($row['id'], 'tt')) { |
|
198
|
|
|
continue; |
|
199
|
|
|
} |
|
200
|
|
|
$id = substr($row['id'], 2); |
|
201
|
|
|
$title = $row['l'] ?? ''; |
|
202
|
|
|
$year = $row['y'] ?? ''; |
|
203
|
|
|
if ($title === '') { |
|
204
|
|
|
continue; |
|
205
|
|
|
} |
|
206
|
|
|
$results[] = [ |
|
207
|
|
|
'imdbid' => $id, |
|
208
|
|
|
'title' => $title, |
|
209
|
|
|
'year' => (string) $year, |
|
210
|
|
|
]; |
|
211
|
|
|
if (count($results) >= 25) { |
|
212
|
|
|
break; |
|
213
|
|
|
} |
|
214
|
|
|
} |
|
215
|
|
|
Cache::put($cacheKey, $results, now()->addHours(12)); |
|
216
|
|
|
|
|
217
|
|
|
return $results; |
|
218
|
|
|
} catch (GuzzleException|\Throwable $e) { |
|
219
|
|
|
Log::debug('IMDb search error '.$query.': '.$e->getMessage()); |
|
220
|
|
|
Cache::put($cacheKey, [], now()->addHours(6)); |
|
221
|
|
|
|
|
222
|
|
|
return []; |
|
223
|
|
|
} |
|
224
|
|
|
} |
|
225
|
|
|
} |
|
226
|
|
|
|
NNTmux /
newznab-tmux
The issue could also be caused by a filter entry in the build configuration. If the path has been excluded in your configuration, e.g.
excluded_paths: ["lib/*"], you can move it to the dependency path list as follows:For further information see https://scrutinizer-ci.com/docs/tools/php/php-scrutinizer/#list-dependency-paths