1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* Nextcloud - OCR |
5
|
|
|
* |
6
|
|
|
* This file is licensed under the Affero General Public License version 3 or |
7
|
|
|
* later. See the COPYING file. |
8
|
|
|
* |
9
|
|
|
* @author Janis Koehr <[email protected]> |
10
|
|
|
* @copyright Janis Koehr 2017 |
11
|
|
|
*/ |
12
|
|
|
namespace OCA\Ocr\Service; |
13
|
|
|
|
14
|
|
|
use OCA\Ocr\Db\OcrJobMapper; |
15
|
|
|
use OC\Files\View; |
16
|
|
|
use OCP\ILogger; |
17
|
|
|
use OCP\IL10N; |
18
|
|
|
use OCP\ITempManager; |
19
|
|
|
use OCA\Ocr\Db\OcrJob; |
20
|
|
|
use OCA\Ocr\Db\File; |
21
|
|
|
use OCA\Ocr\Constants\OcrConstants; |
22
|
|
|
use OCA\Ocr\Util\PHPUtil; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Class JobService |
26
|
|
|
* |
27
|
|
|
* @package OCA\Ocr\Service |
28
|
|
|
*/ |
29
|
|
|
class JobService { |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* |
33
|
|
|
* @var ILogger |
34
|
|
|
*/ |
35
|
|
|
private $logger; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* |
39
|
|
|
* @var RedisService |
40
|
|
|
*/ |
41
|
|
|
private $redisService; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* |
45
|
|
|
* @var OcrJobMapper |
46
|
|
|
*/ |
47
|
|
|
private $jobMapper; |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* |
51
|
|
|
* @var View |
52
|
|
|
*/ |
53
|
|
|
private $view; |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* |
57
|
|
|
* @var String |
58
|
|
|
*/ |
59
|
|
|
private $userId; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* |
63
|
|
|
* @var IL10N |
64
|
|
|
*/ |
65
|
|
|
private $l10n; |
66
|
|
|
|
67
|
|
|
/** |
68
|
|
|
* |
69
|
|
|
* @var FileService |
70
|
|
|
*/ |
71
|
|
|
private $fileService; |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* |
75
|
|
|
* @var ITempManager |
76
|
|
|
*/ |
77
|
|
|
private $tempM; |
78
|
|
|
|
79
|
|
|
/** |
80
|
|
|
* |
81
|
|
|
* @var AppConfigService |
82
|
|
|
*/ |
83
|
|
|
private $appConfigService; |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* |
87
|
|
|
* @var PHPUtil |
88
|
|
|
*/ |
89
|
|
|
private $phpUtil; |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* JobService constructor. |
93
|
|
|
* |
94
|
|
|
* @param IL10N $l10n |
95
|
|
|
* @param ILogger $logger |
96
|
|
|
* @param string $userId |
97
|
|
|
* @param View $view |
98
|
|
|
* @param RedisService $queueService |
99
|
|
|
* @param OcrJobMapper $mapper |
100
|
|
|
* @param FileService $fileService |
101
|
|
|
* @param AppConfigService $appConfigService |
102
|
|
|
* @param PHPUtil $phpUtil |
103
|
|
|
*/ |
104
|
4 |
|
public function __construct(IL10N $l10n, ILogger $logger, $userId, View $view, ITempManager $tempManager, RedisService $queueService, OcrJobMapper $mapper, FileService $fileService, AppConfigService $appConfigService, PHPUtil $phpUtil) { |
105
|
4 |
|
$this->logger = $logger; |
106
|
4 |
|
$this->redisService = $queueService; |
107
|
4 |
|
$this->jobMapper = $mapper; |
108
|
4 |
|
$this->view = $view; |
109
|
4 |
|
$this->userId = $userId; |
110
|
4 |
|
$this->l10n = $l10n; |
111
|
4 |
|
$this->fileService = $fileService; |
112
|
4 |
|
$this->tempM = $tempManager; |
113
|
4 |
|
$this->appConfigService = $appConfigService; |
114
|
4 |
|
$this->phpUtil = $phpUtil; |
115
|
4 |
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Processes and prepares the files for OCR. |
119
|
|
|
* Sends the stuff to the client in order to OCR async. |
120
|
|
|
* |
121
|
|
|
* @param string[] $languages |
122
|
|
|
* @param array $files |
123
|
|
|
* @return string |
124
|
|
|
*/ |
125
|
|
|
public function process($languages, $files) { |
126
|
|
|
try { |
127
|
|
|
$this->logger->debug ( 'Will now process files: {files} with languages: {languages}', [ |
128
|
|
|
'files' => json_encode ( $files ), |
129
|
|
|
'languages' => json_encode ( $languages ) |
130
|
|
|
] ); |
131
|
|
|
// Check if files and language not empty |
132
|
|
|
$noLang = $this->noLanguage ( $languages ); |
133
|
|
|
if (! empty ( $files ) && ($this->checkForAcceptedLanguages ( $languages ) || $noLang)) { |
134
|
|
|
// language part: |
135
|
|
|
if ($noLang) { |
136
|
|
|
$languages = [ ]; |
137
|
|
|
} |
138
|
|
|
// file part: |
139
|
|
|
$fileInfo = $this->fileService->buildFileInfo ( $files ); |
140
|
|
|
foreach ( $fileInfo as $fInfo ) { |
141
|
|
|
// Check Shared |
142
|
|
|
$shared = $this->fileService->checkSharedWithInitiator ( $fInfo ); |
143
|
|
|
$target = $this->fileService->buildTarget ( $fInfo, $shared ); |
|
|
|
|
144
|
|
|
$source = $this->fileService->buildSource ( $fInfo, $shared ); |
|
|
|
|
145
|
|
|
|
146
|
|
|
// set the running type |
147
|
|
|
$fType = $this->fileService->getCorrectType ( $fInfo ); |
148
|
|
|
|
149
|
|
|
// create a temp file for ocr processing purposes |
150
|
|
|
$tempFile = $this->getTempFile ( $fType ); |
151
|
|
|
|
152
|
|
|
// Create job object |
153
|
|
|
$job = new OcrJob ( OcrConstants::STATUS_PENDING, $source, $target, $tempFile, $fType, $this->userId, false, $fInfo->getName (), null ); |
|
|
|
|
154
|
|
|
|
155
|
|
|
// Init client and send task / job |
156
|
|
|
// Feed the worker |
157
|
|
|
$this->redisService->sendJob ( $job, $languages, \OC::$SERVERROOT ); |
158
|
|
|
} |
159
|
|
|
return 'PROCESSING'; |
160
|
|
|
} else { |
161
|
|
|
throw new NotFoundException ( $this->l10n->t ( 'Empty parameters passed.' ) ); |
162
|
|
|
} |
163
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
164
|
|
|
$this->handleException ( $e ); |
165
|
|
|
} |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
/** |
169
|
|
|
* Delete an ocr job for a given id and userId. |
170
|
|
|
* |
171
|
|
|
* @param |
172
|
|
|
* $jobId |
173
|
|
|
* @param string $userId |
174
|
|
|
* @return OcrJob |
175
|
|
|
*/ |
176
|
|
|
public function deleteJob($jobId, $userId) { |
177
|
|
|
try { |
178
|
|
|
$job = $this->jobMapper->find ( $jobId ); |
179
|
|
|
if ($job->getUserId () !== $userId) { |
180
|
|
|
throw new NotFoundException ( $this->l10n->t ( 'Cannot delete. Wrong owner.' ) ); |
181
|
|
|
} else { |
182
|
|
|
$job = $this->jobMapper->delete ( $job ); |
183
|
|
|
} |
184
|
|
|
$job->setTarget ( null ); |
185
|
|
|
$job->setSource ( null ); |
186
|
|
|
$job->setStatus ( null ); |
187
|
|
|
$job->setTempFile ( null ); |
188
|
|
|
$job->setType ( null ); |
189
|
|
|
$job->setUserId ( null ); |
190
|
|
|
$job->setErrorDisplayed ( null ); |
191
|
|
|
return $job; |
192
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
193
|
|
|
if ($e instanceof DoesNotExistException) { |
|
|
|
|
194
|
|
|
$ex = new NotFoundException ( $this->l10n->t ( 'Cannot delete. Wrong ID.' ) ); |
195
|
|
|
$this->handleException ( $ex ); |
|
|
|
|
196
|
|
|
} else { |
197
|
|
|
$this->handleException ( $e ); |
198
|
|
|
} |
199
|
|
|
} |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* Gets all job objects for a specific user. |
204
|
|
|
* |
205
|
|
|
* @param string $userId |
206
|
|
|
* @return array |
207
|
|
|
*/ |
208
|
|
|
public function getAllJobsForUser($userId) { |
209
|
|
|
try { |
210
|
|
|
$jobs = $this->jobMapper->findAll ( $userId ); |
211
|
|
|
$jobsNew = array (); |
212
|
|
|
foreach ( $jobs as $job ) { |
213
|
|
|
$job->setTarget ( null ); |
214
|
|
|
$job->setSource ( null ); |
215
|
|
|
$job->setTempFile ( null ); |
216
|
|
|
$job->setType ( null ); |
217
|
|
|
$job->setUserId ( null ); |
218
|
|
|
$job->setErrorDisplayed ( null ); |
219
|
|
|
|
220
|
|
|
array_push ( $jobsNew, $job ); |
221
|
|
|
} |
222
|
|
|
return $jobsNew; |
223
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
224
|
|
|
$this->handleException ( $e ); |
225
|
|
|
} |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
/** |
229
|
|
|
* The function checks if there are finished jobs to process finally. |
230
|
|
|
* @throws NotFoundException |
231
|
|
|
*/ |
232
|
|
|
public function checkForFinishedJobs() { |
233
|
|
|
try { |
234
|
|
|
$finishedJobs = $this->redisService->readingFinishedJobs (); |
235
|
|
|
foreach ( $finishedJobs as $finishedJob ) { |
236
|
|
|
$fJob = $this->transformJob( $finishedJob ); |
237
|
|
|
$this->logger->debug('The following job finished: {job}', ['job' => $fJob]); |
238
|
|
|
$this->jobFinished ( $fJob->id, $fJob->error, $fJob->log ); |
239
|
|
|
} |
240
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
241
|
|
|
throw new NotFoundException ( $this->l10n->t ( 'Reading the finished jobs from redis went wrong.' ) ); |
242
|
|
|
} |
243
|
|
|
} |
244
|
|
|
|
245
|
|
|
/** |
246
|
|
|
* The function the worker will call in order to set the jobs status. |
247
|
|
|
* The worker should call it automatically after each processing step. |
248
|
|
|
* |
249
|
|
|
* @param integer $jobId |
250
|
|
|
* @param boolean $error |
251
|
|
|
* @param string $log |
252
|
|
|
*/ |
253
|
|
|
public function jobFinished($jobId, $error, $log) { |
254
|
|
|
try { |
255
|
|
|
$job = $this->jobMapper->find ( $jobId ); |
256
|
|
|
if (! $error) { |
257
|
|
|
$job->setStatus ( OcrConstants::STATUS_PROCESSED ); |
258
|
|
|
$this->jobMapper->update ( $job ); |
259
|
|
|
} else { |
260
|
|
|
$job->setStatus ( OcrConstants::STATUS_FAILED ); |
261
|
|
|
$job->setErrorLog( $log ); |
262
|
|
|
$this->jobMapper->update ( $job ); |
263
|
|
|
$this->logger->error ( $log); |
264
|
|
|
} |
265
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
266
|
|
|
$this->handleException ( $e ); |
267
|
|
|
} |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
/** |
271
|
|
|
* Finishes all Processed files by copying them to the right path and deleteing the temp files. |
272
|
|
|
* Returns the number of processed files. |
273
|
|
|
* |
274
|
|
|
* @return array |
275
|
|
|
*/ |
276
|
|
|
public function handleProcessed() { |
277
|
|
|
try { |
278
|
|
|
$this->logger->debug ( 'Check if files were processed by ocr and if so, put them to the right dirs.'); |
279
|
|
|
$processed = $this->jobMapper->findAllProcessed ( $this->userId ); |
280
|
|
|
foreach ( $processed as $job ) { |
281
|
|
|
if ($this->fileService->fileExists ( $job->getTempFile () )) { |
282
|
|
|
// Save the tmp file with newname |
283
|
|
|
$this->view->file_put_contents ( $job->getTarget (), $this->fileService->getFileContents ( $job->getTempFile () ) ); |
284
|
|
|
$this->jobMapper->delete ( $job ); |
285
|
|
|
$this->fileService->execRemove ( $job->getTempFile () ); |
286
|
|
|
} else { |
287
|
|
|
$job->setStatus ( OcrConstants::STATUS_FAILED ); |
288
|
|
|
$job->setErrorLog('Temp file does not exist.'); |
289
|
|
|
$this->jobMapper->update ( $job ); |
290
|
|
|
throw new NotFoundException ( $this->l10n->t ( 'Temp file does not exist.' ) ); |
291
|
|
|
} |
292
|
|
|
} |
293
|
|
|
return $processed; |
294
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
295
|
|
|
$this->handleException ( $e ); |
296
|
|
|
} |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
/** |
300
|
|
|
* Handles all failed orders of ocr processing queue and returns the job objects. |
301
|
|
|
* |
302
|
|
|
* @return array |
303
|
|
|
*/ |
304
|
|
|
public function handleFailed() { |
305
|
|
|
try { |
306
|
|
|
$failed = $this->jobMapper->findAllFailed ( $this->userId ); |
307
|
|
|
foreach ( $failed as $job ) { |
308
|
|
|
// clean the tempfile |
309
|
|
|
$this->fileService->execRemove ( $job->getTempFile () ); |
310
|
|
|
// set error displayed |
311
|
|
|
$job->setErrorDisplayed ( true ); |
312
|
|
|
$this->jobMapper->update ( $job ); |
313
|
|
|
} |
314
|
|
|
$this->logger->debug ( 'Following jobs failed: {failed}', [ |
315
|
|
|
'failed' => json_encode ( $failed ) |
316
|
|
|
] ); |
317
|
|
|
return $failed; |
318
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
319
|
|
|
$this->handleException ( $e ); |
320
|
|
|
} |
321
|
|
|
} |
322
|
|
|
|
323
|
|
|
/** |
324
|
|
|
* Gives a temp file name back depending on the type of the OCR. |
325
|
|
|
* Later in the worker this file is used as an output. |
326
|
|
|
* |
327
|
|
|
* @param integer $type |
328
|
|
|
* @return string |
329
|
|
|
*/ |
330
|
|
|
private function getTempFile($type) { |
331
|
|
|
if ($type == OcrConstants::TESSERACT) { |
332
|
|
|
$fileName = $this->phpUtil->tempnamWrapper($this->tempM->getTempBaseDir (), OcrConstants::TEMPFILE_PREFIX); |
333
|
|
|
$this->phpUtil->unlinkWrapper($fileName); |
334
|
|
|
$fileNameWithPostfix = $fileName . '.txt'; |
335
|
|
|
$this->phpUtil->touchWrapper( $fileNameWithPostfix ); |
336
|
|
|
$this->phpUtil->chmodWrapper( $fileNameWithPostfix, 0600 ); |
337
|
|
|
return $fileNameWithPostfix; |
338
|
|
|
} else { |
339
|
|
|
return $this->phpUtil->tempnamWrapper( $this->tempM->getTempBaseDir (), 'ocr_' ); |
340
|
|
|
} |
341
|
|
|
} |
342
|
|
|
|
343
|
|
|
/** |
344
|
|
|
* Takes care of transforming an incoming finished job into a php readable object. |
345
|
|
|
* |
346
|
|
|
* @param string $job |
347
|
|
|
* @throws NotFoundException |
348
|
|
|
* @return mixed |
349
|
|
|
*/ |
350
|
|
|
private function transformJob($job) { |
351
|
|
|
$decoded = json_decode ( $job ); |
352
|
|
|
if($decoded != null && isset($decoded->id)) { |
353
|
|
|
return $decoded; |
354
|
|
|
} else { |
355
|
|
|
throw new NotFoundException($this->l10n->t ( 'The finished job retrieved by Redis was corrupt.' )); |
356
|
|
|
} |
357
|
|
|
} |
358
|
|
|
|
359
|
|
|
/** |
360
|
|
|
* Checks if the given languages are supported or not. |
361
|
|
|
* |
362
|
|
|
* @param string[] $languages |
363
|
|
|
* @return boolean |
364
|
|
|
*/ |
365
|
|
|
private function checkForAcceptedLanguages($languages) { |
366
|
|
|
$installedLanguages = explode ( ';', $this->appConfigService->getAppValue ( 'languages' ) ); |
367
|
|
|
if (count ( array_diff ( $languages, $installedLanguages ) ) === 0) { |
368
|
|
|
return true; |
369
|
|
|
} else { |
370
|
|
|
return false; |
371
|
|
|
} |
372
|
|
|
} |
373
|
|
|
|
374
|
|
|
/** |
375
|
|
|
* Checks if the process should be initiated without any language specified. |
376
|
|
|
* |
377
|
|
|
* @param string[] $languages |
378
|
|
|
* @return boolean |
379
|
|
|
*/ |
380
|
|
|
private function noLanguage($languages) { |
381
|
|
|
if (in_array ( 'any', $languages )) { |
382
|
|
|
return true; |
383
|
|
|
} else { |
384
|
|
|
return false; |
385
|
|
|
} |
386
|
|
|
} |
387
|
|
|
|
388
|
|
|
/** |
389
|
|
|
* Handle the possible thrown Exceptions from all methods of this class. |
390
|
|
|
* |
391
|
|
|
* @param Exception $e |
392
|
|
|
* @throws Exception |
393
|
|
|
* @throws NotFoundException |
394
|
|
|
*/ |
395
|
|
|
private function handleException($e) { |
396
|
|
|
$this->logger->logException ( $e, [ |
397
|
|
|
'message' => 'Exception during job service function processing' |
398
|
|
|
] ); |
399
|
|
|
throw $e; |
400
|
|
|
} |
401
|
|
|
} |
If a method or function can return multiple different values and unless you are sure that you only can receive a single value in this context, we recommend to add an additional type check:
If this a common case that PHP Analyzer should handle natively, please let us know by opening an issue.