|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Services\Categorization\Categorizers; |
|
4
|
|
|
|
|
5
|
|
|
use App\Models\Category; |
|
6
|
|
|
use App\Services\Categorization\CategorizationResult; |
|
7
|
|
|
use App\Services\Categorization\ReleaseContext; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Categorizer for Movie content including HD, SD, UHD, 3D, Blu-ray, DVD, etc. |
|
11
|
|
|
*/ |
|
12
|
|
|
class MovieCategorizer extends AbstractCategorizer |
|
13
|
|
|
{ |
|
14
|
|
|
protected int $priority = 25; |
|
15
|
|
|
|
|
16
|
|
|
public function getName(): string |
|
17
|
|
|
{ |
|
18
|
|
|
return 'Movie'; |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
public function shouldSkip(ReleaseContext $context): bool |
|
22
|
|
|
{ |
|
23
|
|
|
// Skip if this looks like adult content |
|
24
|
|
|
if ($context->hasAdultMarkers()) { |
|
25
|
|
|
return true; |
|
26
|
|
|
} |
|
27
|
|
|
|
|
28
|
|
|
// Skip if it looks like a TV episode (S01E01) or season pack (S01.1080p) |
|
29
|
|
|
if (preg_match('/[._ -]S\d{1,3}[._ -]?(E\d|D\d|Complete|Full|1080|720|480|2160|WEB|HDTV|BluRay|NF|AMZN)/i', $context->releaseName)) { |
|
30
|
|
|
return true; |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
// Skip episode-only patterns (E01, E02) - common in anime |
|
34
|
|
|
if (preg_match('/[._ -]E\d{1,4}[._ -]/i', $context->releaseName)) { |
|
35
|
|
|
return true; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
// Skip known anime release groups |
|
39
|
|
|
if (preg_match('/[.\-_ ](URANiME|ANiHLS|HaiKU|ANiURL|SkyAnime|Erai-raws|LostYears|Vodes|SubsPlease|Judas|Ember|YuiSubs|ASW|Tsundere-Raws|Anime-Raws)[.\-_ ]?/i', $context->releaseName)) { |
|
40
|
|
|
return true; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
return false; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
public function categorize(ReleaseContext $context): CategorizationResult |
|
47
|
|
|
{ |
|
48
|
|
|
$name = $context->releaseName; |
|
49
|
|
|
|
|
50
|
|
|
// Check if it looks like movie content |
|
51
|
|
|
if (!$this->looksLikeMovie($name)) { |
|
52
|
|
|
return $this->noMatch(); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
// Try specific movie subcategories in order of specificity |
|
56
|
|
|
if ($context->categorizeForeign && ($result = $this->checkForeign($name))) { |
|
57
|
|
|
return $result; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
if ($result = $this->checkX265($name)) { |
|
61
|
|
|
return $result; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
if ($result = $this->checkUHD($name)) { |
|
65
|
|
|
return $result; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
if ($result = $this->check3D($name)) { |
|
69
|
|
|
return $result; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
if ($result = $this->checkBluRay($name)) { |
|
73
|
|
|
return $result; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
if ($result = $this->checkDVD($name)) { |
|
77
|
|
|
return $result; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
if ($context->catWebDL && ($result = $this->checkWebDL($name))) { |
|
81
|
|
|
return $result; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
if ($result = $this->checkHD($name, $context->catWebDL)) { |
|
85
|
|
|
return $result; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
if ($result = $this->checkSD($name)) { |
|
89
|
|
|
return $result; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
if ($result = $this->checkOther($name)) { |
|
93
|
|
|
return $result; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
return $this->noMatch(); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* Check if release name looks like movie content. |
|
101
|
|
|
*/ |
|
102
|
|
|
protected function looksLikeMovie(string $name): bool |
|
103
|
|
|
{ |
|
104
|
|
|
return (bool) preg_match('/[._ -]AVC|[BH][DR]RIP|(Bluray|Blu-Ray)|BD[._ -]?(25|50)?|\bBR\b|Camrip|[._ -]\d{4}[._ -].+(720p|1080p|Cam|HDTS|2160p)|DIVX|[._ -]DVD[._ -]|DVD-?(5|9|R|Rip)|Untouched|VHSRip|XVID|[._ -](DTS|TVrip|webrip|WEBDL|WEB-DL)[._ -]|\b(2160)p\b.*\b(Netflix|Amazon|NF|AMZN|Disney)\b/i', $name); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
protected function checkForeign(string $name): ?CategorizationResult |
|
108
|
|
|
{ |
|
109
|
|
|
if (preg_match('/(danish|flemish|Deutsch|dutch|french|german|heb|hebrew|nl[._ -]?sub|dub(bed|s)?|\.NL|norwegian|swedish|swesub|spanish|Staffel)[._ -]|\(german\)|Multisub/i', $name)) { |
|
110
|
|
|
return $this->matched(Category::MOVIE_FOREIGN, 0.8, 'foreign_language'); |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
if (stripos($name, 'Castellano') !== false) { |
|
114
|
|
|
return $this->matched(Category::MOVIE_FOREIGN, 0.8, 'foreign_castellano'); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
if (preg_match('/(720p|1080p|AC3|AVC|DIVX|DVD(5|9|RIP|R)|XVID)[._ -](Dutch|French|German|ITA)|\(?(Dutch|French|German|ITA)\)?[._ -](720P|1080p|AC3|AVC|DIVX|DVD(5|9|RIP|R)|WEB(-DL|-?RIP)|HD[._ -]|XVID)/i', $name)) { |
|
118
|
|
|
return $this->matched(Category::MOVIE_FOREIGN, 0.85, 'foreign_pattern'); |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
return null; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
protected function checkX265(string $name): ?CategorizationResult |
|
125
|
|
|
{ |
|
126
|
|
|
if (preg_match('/(\w+[\.-_\s]+).*(x265).*(Tigole|SESKAPiLE|CHD|IAMABLE|THREESOME|OohLaLa|DEFLATE|NCmt)/i', $name)) { |
|
127
|
|
|
return $this->matched(Category::MOVIE_X265, 0.9, 'x265_group'); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
return null; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
protected function checkUHD(string $name): ?CategorizationResult |
|
134
|
|
|
{ |
|
135
|
|
|
// Skip TV shows |
|
136
|
|
|
if (preg_match('/(S\d+).*(2160p).*(Netflix|Amazon|NF|AMZN).*(TrollUHD|NTb|VLAD|DEFLATE|CMRG)/i', $name)) { |
|
137
|
|
|
return null; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
// Check for UHD indicators |
|
141
|
|
|
if (stripos($name, '2160p') !== false || |
|
142
|
|
|
preg_match('/\b(UHD|Ultra[._ -]HD|4K)\b/i', $name) || |
|
143
|
|
|
(preg_match('/\b(HDR|HDR10|HDR10\+|Dolby[._ -]?Vision)\b/i', $name) && |
|
144
|
|
|
preg_match('/\b(HEVC|H\.?265|x265)\b/i', $name)) || |
|
145
|
|
|
(stripos($name, 'UHD') !== false && |
|
146
|
|
|
preg_match('/\b(BR|BluRay|Blu[._ -]?Ray)\b/i', $name))) { |
|
147
|
|
|
return $this->matched(Category::MOVIE_UHD, 0.9, 'uhd'); |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
return null; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
protected function check3D(string $name): ?CategorizationResult |
|
154
|
|
|
{ |
|
155
|
|
|
if (preg_match('/[._ -]3D\s?[\.\-_\[ ](1080p|(19|20)\d\d|AVC|BD(25|50)|Blu[._ -]?ray|CEE|Complete|GER|MVC|MULTi|SBS|H(-)?SBS)[._ -]/i', $name)) { |
|
156
|
|
|
return $this->matched(Category::MOVIE_3D, 0.9, '3d'); |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
return null; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
protected function checkBluRay(string $name): ?CategorizationResult |
|
163
|
|
|
{ |
|
164
|
|
|
if (preg_match('/bluray-|[._ -]bd?[._ -]?(25|50)|blu-ray|Bluray\s-\sUntouched|[._ -]untouched[._ -]/i', $name) && |
|
165
|
|
|
!preg_match('/SecretUsenet\.com$/i', $name)) { |
|
166
|
|
|
return $this->matched(Category::MOVIE_BLURAY, 0.9, 'bluray'); |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
return null; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
protected function checkDVD(string $name): ?CategorizationResult |
|
173
|
|
|
{ |
|
174
|
|
|
if (preg_match('/(dvd\-?r|[._ -]dvd|dvd9|dvd5|[._ -]r5)[._ -]/i', $name)) { |
|
175
|
|
|
return $this->matched(Category::MOVIE_DVD, 0.85, 'dvd'); |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
return null; |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
protected function checkWebDL(string $name): ?CategorizationResult |
|
182
|
|
|
{ |
|
183
|
|
|
if (preg_match('/web[._ -]dl|web-?rip/i', $name)) { |
|
184
|
|
|
return $this->matched(Category::MOVIE_WEBDL, 0.85, 'webdl'); |
|
185
|
|
|
} |
|
186
|
|
|
|
|
187
|
|
|
return null; |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
|
|
protected function checkHD(string $name, bool $catWebDL): ?CategorizationResult |
|
191
|
|
|
{ |
|
192
|
|
|
if (preg_match('/720p|1080p|AVC|VC1|VC-1|web-dl|wmvhd|x264|XvidHD|bdrip/i', $name)) { |
|
193
|
|
|
return $this->matched(Category::MOVIE_HD, 0.85, 'hd'); |
|
194
|
|
|
} |
|
195
|
|
|
|
|
196
|
|
|
if (!$catWebDL && preg_match('/web[._ -]dl|web-?rip/i', $name)) { |
|
197
|
|
|
return $this->matched(Category::MOVIE_HD, 0.8, 'hd_webdl_fallback'); |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
return null; |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
protected function checkSD(string $name): ?CategorizationResult |
|
204
|
|
|
{ |
|
205
|
|
|
if (preg_match('/(divx|dvdscr|extrascene|dvdrip|\.CAM|HDTS(-LINE)?|vhsrip|xvid(vd)?)[._ -]/i', $name)) { |
|
206
|
|
|
return $this->matched(Category::MOVIE_SD, 0.8, 'sd'); |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
return null; |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
|
|
protected function checkOther(string $name): ?CategorizationResult |
|
213
|
|
|
{ |
|
214
|
|
|
if (preg_match('/[._ -]cam[._ -]/i', $name)) { |
|
215
|
|
|
return $this->matched(Category::MOVIE_OTHER, 0.6, 'cam'); |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
return null; |
|
219
|
|
|
} |
|
220
|
|
|
} |
|
221
|
|
|
|
|
222
|
|
|
|
NNTmux /
newznab-tmux
