|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Platform\YouTube; |
|
4
|
|
|
|
|
5
|
|
|
use App\Domain\Collection\Contents; |
|
6
|
|
|
use App\Domain\Content; |
|
7
|
|
|
use App\Domain\Collection\Path; |
|
8
|
|
|
use App\Domain\PathPart; |
|
9
|
|
|
use App\Platform\Platform; |
|
10
|
|
|
use App\Platform\YouTube\Exception as YouTubeException; |
|
11
|
|
|
use Symfony\Component\Filesystem\Filesystem; |
|
12
|
|
|
use Symfony\Component\Finder\Finder; |
|
13
|
|
|
|
|
14
|
|
|
final class YouTube implements Platform |
|
15
|
|
|
{ |
|
16
|
|
|
use FilesystemManager; |
|
17
|
|
|
|
|
18
|
|
|
// See https://stackoverflow.com/a/37704433/389519 |
|
19
|
|
|
private const YOUTUBE_URL_REGEX = <<<REGEX |
|
20
|
|
|
/\b((?:https?:)?\/\/)?((?:www|m)\.)?((?:youtube\.com|youtu.be))(\/(?:[\w\-]+\?v=|embed\/|v\/)?)([\w\-]+)(\S+)?/i |
|
21
|
|
|
REGEX; |
|
22
|
|
|
private const YOUTUBE_URL_REGEX_MATCHES_ID_INDEX = 5; |
|
23
|
|
|
private const YOUTUBE_URL_PREFIX = 'https://www.youtube.com/watch?v='; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @param \App\Domain\Collection\Contents $contents |
|
27
|
|
|
* @param \App\Domain\PathPart $rootPathPart |
|
28
|
|
|
* |
|
29
|
|
|
* @throws \RuntimeException |
|
30
|
|
|
*/ |
|
31
|
|
|
public function synchronizeContents(Contents $contents, PathPart $rootPathPart): void |
|
32
|
|
|
{ |
|
33
|
|
|
if ($contents->isEmpty()) { |
|
34
|
|
|
return; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
$platformPathPart = new PathPart($this->options['path_part']); |
|
38
|
|
|
$downloadPath = new Path([$rootPathPart, $platformPathPart]); |
|
39
|
|
|
|
|
40
|
|
|
// Try to create the downloads directory... 'cause if it fails, nothing will work. |
|
41
|
|
|
(new Filesystem())->mkdir((string) $downloadPath); |
|
42
|
|
|
|
|
43
|
|
|
// Add the platform path part and get a collection of downloads |
|
44
|
|
|
$downloads = new Downloads(); |
|
|
|
|
|
|
45
|
|
|
foreach ($contents as $content) { |
|
46
|
|
|
$content->getPath()->add($platformPathPart); |
|
47
|
|
|
|
|
48
|
|
|
foreach ($this->extractDownloads($content) as $download) { |
|
49
|
|
|
$downloads->add($download); |
|
50
|
|
|
} |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
$this->cleanFilesystem($downloads, $downloadPath); |
|
54
|
|
|
|
|
55
|
|
|
$downloads = $this->filterAlreadyDownloaded($downloads); |
|
56
|
|
|
|
|
57
|
|
|
if ($this->shouldDownload($downloads, $downloadPath)) { |
|
58
|
|
|
$this->download($downloads, $downloadPath); |
|
59
|
|
|
} |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** @noinspection LowerAccessLevelInspection */ |
|
63
|
|
|
/** |
|
64
|
|
|
* @param \App\Domain\Collection\Path $downloadPath |
|
65
|
|
|
* |
|
66
|
|
|
* @return \Symfony\Component\Finder\Finder |
|
67
|
|
|
*/ |
|
68
|
|
|
protected function getAllDownloadsFolderFinder(Path $downloadPath): Finder |
|
69
|
|
|
{ |
|
70
|
|
|
/** @noinspection ExceptionsAnnotatingAndHandlingInspection */ |
|
71
|
|
|
return (new Finder()) |
|
72
|
|
|
->directories() |
|
73
|
|
|
->in((string) $downloadPath) |
|
74
|
|
|
->sort(function (\SplFileInfo $fileInfoA, \SplFileInfo $fileInfoB) { |
|
75
|
|
|
// Sort the result by folder depth |
|
76
|
|
|
$a = substr_count($fileInfoA->getRealPath(), DIRECTORY_SEPARATOR); |
|
77
|
|
|
$b = substr_count($fileInfoB->getRealPath(), DIRECTORY_SEPARATOR); |
|
78
|
|
|
|
|
79
|
|
|
return $a <=> $b; |
|
80
|
|
|
}); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** @noinspection LowerAccessLevelInspection */ |
|
84
|
|
|
/** |
|
85
|
|
|
* @param \App\Platform\YouTube\Download $download |
|
86
|
|
|
* |
|
87
|
|
|
* @return \Symfony\Component\Finder\Finder|\SplFileInfo[] |
|
88
|
|
|
* @throws \InvalidArgumentException |
|
89
|
|
|
*/ |
|
90
|
|
|
protected function getDownloadFolderFinder(Download $download): Finder |
|
91
|
|
|
{ |
|
92
|
|
|
$placeholders = [ |
|
93
|
|
|
'%video_id%' => $download->getVideoId(), |
|
94
|
|
|
'%file_extension%' => $download->getFileExtension(), |
|
95
|
|
|
]; |
|
96
|
|
|
|
|
97
|
|
|
$downloadPathPart = new PathPart([ |
|
98
|
|
|
'path' => (string) $download->getPath(), |
|
99
|
|
|
'priority' => 0, |
|
100
|
|
|
]); |
|
101
|
|
|
$folderPathPart = new PathPart([ |
|
102
|
|
|
'path' => $this->options['patterns']['folder'], |
|
103
|
|
|
'priority' => 1, |
|
104
|
|
|
'substitutions' => $placeholders, |
|
105
|
|
|
]); |
|
106
|
|
|
|
|
107
|
|
|
return (new Finder()) |
|
108
|
|
|
->files() |
|
109
|
|
|
->depth('== 0') |
|
110
|
|
|
->in((string) new Path([$downloadPathPart, $folderPathPart])) |
|
111
|
|
|
->name( |
|
112
|
|
|
str_replace( |
|
113
|
|
|
array_keys($placeholders), |
|
114
|
|
|
array_values($placeholders), |
|
115
|
|
|
$this->options['patterns']['filename'] |
|
116
|
|
|
) |
|
117
|
|
|
); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** @noinspection LowerAccessLevelInspection */ |
|
121
|
|
|
/** |
|
122
|
|
|
* @return bool |
|
123
|
|
|
*/ |
|
124
|
|
|
protected function skip(): bool |
|
125
|
|
|
{ |
|
126
|
|
|
if ($this->ui->isDryRun()) { |
|
127
|
|
|
$this->ui->writeln('<info>[DRY-RUN]</info> Not doing anything...'.PHP_EOL); |
|
128
|
|
|
|
|
129
|
|
|
return true; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
return false; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
/** |
|
136
|
|
|
* @param \App\Domain\Content $content |
|
137
|
|
|
* |
|
138
|
|
|
* @return \App\Platform\YouTube\Downloads |
|
139
|
|
|
*/ |
|
140
|
|
|
private function extractDownloads(Content $content): Downloads |
|
141
|
|
|
{ |
|
142
|
|
|
$downloads = new Downloads(); |
|
|
|
|
|
|
143
|
|
|
|
|
144
|
|
|
if (preg_match_all(static::YOUTUBE_URL_REGEX, $content->getData(), $youtubeUrls)) { |
|
145
|
|
|
foreach ((array) $youtubeUrls[static::YOUTUBE_URL_REGEX_MATCHES_ID_INDEX] as $youtubeId) { |
|
146
|
|
|
foreach (array_keys($this->options['youtube_dl']['options']) as $videoFileType) { |
|
147
|
|
|
$downloads->add( |
|
148
|
|
|
new Download( |
|
149
|
|
|
$content->getPath(), |
|
150
|
|
|
$youtubeId, |
|
151
|
|
|
$videoFileType, |
|
152
|
|
|
$this->options['patterns']['extensions'][$videoFileType] |
|
153
|
|
|
) |
|
154
|
|
|
); |
|
155
|
|
|
} |
|
156
|
|
|
} |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
return $downloads; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* @param \App\Platform\YouTube\Downloads $downloads |
|
164
|
|
|
* |
|
165
|
|
|
* @return \App\Platform\YouTube\Downloads |
|
166
|
|
|
*/ |
|
167
|
|
|
private function filterAlreadyDownloaded(Downloads $downloads): Downloads |
|
168
|
|
|
{ |
|
169
|
|
|
return $downloads->filter( |
|
170
|
|
|
function (Download $download) { |
|
171
|
|
|
$shouldBeDownloaded = true; |
|
172
|
|
|
try { |
|
173
|
|
|
if ($this->getDownloadFolderFinder($download)->hasResults()) { |
|
174
|
|
|
$shouldBeDownloaded = false; |
|
175
|
|
|
} |
|
176
|
|
|
} catch (\InvalidArgumentException $e) { |
|
177
|
|
|
// Here we know that the download folder will exist. |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
return $shouldBeDownloaded; |
|
181
|
|
|
} |
|
182
|
|
|
); |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
/** |
|
186
|
|
|
* @param \App\Platform\YouTube\Downloads $downloads |
|
187
|
|
|
* @param \App\Domain\Collection\Path $downloadPath |
|
188
|
|
|
* |
|
189
|
|
|
* @return bool |
|
190
|
|
|
*/ |
|
191
|
|
|
private function shouldDownload(Downloads $downloads, Path $downloadPath): bool |
|
192
|
|
|
{ |
|
193
|
|
|
$this->ui->writeln('Download files from YouTube... '.PHP_EOL); |
|
194
|
|
|
|
|
195
|
|
|
if ($downloads->isEmpty()) { |
|
196
|
|
|
$this->ui->writeln($this->ui->indent().'<comment>Nothing to download.</comment>'.PHP_EOL); |
|
197
|
|
|
|
|
198
|
|
|
return false; |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
$this->ui->writeln( |
|
202
|
|
|
sprintf( |
|
203
|
|
|
'%sThe script is about to download <question> %s </question> files into <info>%s</info>. '.PHP_EOL, |
|
204
|
|
|
$this->ui->indent(), |
|
205
|
|
|
$downloads->count(), |
|
206
|
|
|
(string) $downloadPath |
|
207
|
|
|
) |
|
208
|
|
|
); |
|
209
|
|
|
|
|
210
|
|
|
$this->ui->write($this->ui->indent()); |
|
211
|
|
|
if ($this->skip() || !$this->ui->confirm()) { |
|
212
|
|
|
$this->ui->writeln(($this->ui->isDryRun() ? '' : PHP_EOL).'<info>Done.</info>'.PHP_EOL); |
|
213
|
|
|
|
|
214
|
|
|
return false; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
return true; |
|
218
|
|
|
} |
|
219
|
|
|
|
|
220
|
|
|
/** |
|
221
|
|
|
* @param \App\Platform\YouTube\Downloads $downloads |
|
222
|
|
|
* @param \App\Domain\Collection\Path $downloadPath |
|
223
|
|
|
* |
|
224
|
|
|
* @throws \RuntimeException |
|
225
|
|
|
*/ |
|
226
|
|
|
private function download(Downloads $downloads, Path $downloadPath) |
|
227
|
|
|
{ |
|
228
|
|
|
$errors = []; |
|
229
|
|
|
foreach ($downloads as $download) { |
|
230
|
|
|
$this->ui->write( |
|
231
|
|
|
sprintf( |
|
232
|
|
|
'%s* [<comment>%s</comment>][<comment>%s</comment>] Download the %s file in <info>%s</info>... ', |
|
233
|
|
|
$this->ui->indent(2), |
|
234
|
|
|
$download->getVideoId(), |
|
235
|
|
|
$download->getFileType(), |
|
236
|
|
|
$download->getFileExtension(), |
|
237
|
|
|
str_replace((string) $downloadPath.DIRECTORY_SEPARATOR, '', $download->getPath()) |
|
238
|
|
|
) |
|
239
|
|
|
); |
|
240
|
|
|
|
|
241
|
|
|
$options = $this->options['youtube_dl']['options'][$download->getFileType()]; |
|
242
|
|
|
$nbAttempts = \count($options); |
|
243
|
|
|
|
|
244
|
|
|
$attempt = 0; |
|
245
|
|
|
while (true) { |
|
246
|
|
|
try { |
|
247
|
|
|
$this->doDownload($download, $options[$attempt]); |
|
248
|
|
|
|
|
249
|
|
|
$this->ui->writeln('<info>Done.</info>'); |
|
250
|
|
|
break; |
|
251
|
|
|
|
|
252
|
|
|
} catch (YouTubeException\ChannelRemovedByUserException |
|
253
|
|
|
| YouTubeException\VideoBlockedByCopyrightException |
|
254
|
|
|
| YouTubeException\VideoRemovedByUserException |
|
255
|
|
|
| YouTubeException\VideoUnavailableException |
|
256
|
|
|
$e) { |
|
257
|
|
|
$this->ui->logError($e->getMessage(), $errors); |
|
258
|
|
|
break; |
|
259
|
|
|
|
|
260
|
|
|
// These are (supposedly) connection/download errors, so we try again |
|
261
|
|
|
} catch (\Exception $e) { |
|
262
|
|
|
$attempt++; |
|
263
|
|
|
|
|
264
|
|
|
// Maximum number of attempts reached, move along... |
|
265
|
|
|
if ($attempt === $nbAttempts) { |
|
266
|
|
|
$this->ui->logError($e->getMessage(), $errors); |
|
267
|
|
|
break; |
|
268
|
|
|
} |
|
269
|
|
|
} |
|
270
|
|
|
} |
|
271
|
|
|
} |
|
272
|
|
|
$this->ui->displayErrors($errors, 'download of files', 'error', 1); |
|
273
|
|
|
|
|
274
|
|
|
$this->ui->writeln(PHP_EOL.'<info>Done.</info>'.PHP_EOL); |
|
275
|
|
|
} |
|
276
|
|
|
|
|
277
|
|
|
/** |
|
278
|
|
|
* @param \App\Platform\YouTube\Download $download |
|
279
|
|
|
* @param array $downloadOptions |
|
280
|
|
|
* |
|
281
|
|
|
* @throws \Exception |
|
282
|
|
|
*/ |
|
283
|
|
|
private function doDownload(Download $download, array $downloadOptions) |
|
284
|
|
|
{ |
|
285
|
|
|
$youtubeDlOptions = $this->options['youtube_dl']; |
|
286
|
|
|
|
|
287
|
|
|
/** @var \YouTubeDl\YouTubeDl $dl */ |
|
288
|
|
|
$dl = new $youtubeDlOptions['class_name']($downloadOptions); |
|
289
|
|
|
$dl->setDownloadPath((string) $download->getPath()); |
|
290
|
|
|
|
|
291
|
|
|
try { |
|
292
|
|
|
(new Filesystem())->mkdir((string) $download->getPath()); |
|
293
|
|
|
$dl->download(static::YOUTUBE_URL_PREFIX.$download->getVideoId()); |
|
294
|
|
|
} catch (\Exception $e) { |
|
295
|
|
|
|
|
296
|
|
|
// Add more custom exceptions than those already provided by YoutubeDl |
|
297
|
|
|
if (preg_match('/this video is unavailable/i', $e->getMessage())) { |
|
298
|
|
|
throw new YouTubeException\VideoUnavailableException( |
|
299
|
|
|
sprintf('The video %s is unavailable.', $download->getVideoId()), 0, $e |
|
300
|
|
|
); |
|
301
|
|
|
} |
|
302
|
|
|
if (preg_match('/this video has been removed by the user/i', $e->getMessage())) { |
|
303
|
|
|
throw new YouTubeException\VideoRemovedByUserException( |
|
304
|
|
|
sprintf('The video %s has been removed by its user.', $download->getVideoId()), 0, $e |
|
305
|
|
|
); |
|
306
|
|
|
} |
|
307
|
|
|
if (preg_match('/the uploader has closed their YouTube account/i', $e->getMessage())) { |
|
308
|
|
|
throw new YouTubeException\ChannelRemovedByUserException( |
|
309
|
|
|
sprintf('The channel that published the video %s has been removed.', $download->getVideoId()), 0, $e |
|
310
|
|
|
); |
|
311
|
|
|
} |
|
312
|
|
|
if (preg_match('/who has blocked it on copyright grounds/i', $e->getMessage())) { |
|
313
|
|
|
throw new YouTubeException\VideoBlockedByCopyrightException( |
|
314
|
|
|
sprintf('The video %s has been block for copyright infringement.', $download->getVideoId()), 0, $e |
|
315
|
|
|
); |
|
316
|
|
|
} |
|
317
|
|
|
|
|
318
|
|
|
throw $e; |
|
319
|
|
|
} |
|
320
|
|
|
} |
|
321
|
|
|
} |
|
322
|
|
|
|
This check looks for function calls that miss required arguments.