1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
|
4
|
|
|
namespace Firesphere\SolrSearch\Tasks; |
5
|
|
|
|
6
|
|
|
use Exception; |
7
|
|
|
use Firesphere\SolrSearch\Factories\DocumentFactory; |
8
|
|
|
use Firesphere\SolrSearch\Helpers\SolrLogger; |
9
|
|
|
use Firesphere\SolrSearch\Indexes\BaseIndex; |
10
|
|
|
use Firesphere\SolrSearch\Services\SolrCoreService; |
11
|
|
|
use Firesphere\SolrSearch\States\SiteState; |
12
|
|
|
use Firesphere\SolrSearch\Traits\LoggerTrait; |
13
|
|
|
use GuzzleHttp\Exception\GuzzleException; |
14
|
|
|
use Psr\Log\LoggerInterface; |
15
|
|
|
use ReflectionException; |
16
|
|
|
use SilverStripe\Control\Controller; |
17
|
|
|
use SilverStripe\Control\Director; |
18
|
|
|
use SilverStripe\Control\HTTPRequest; |
19
|
|
|
use SilverStripe\Core\Injector\Injector; |
20
|
|
|
use SilverStripe\Dev\BuildTask; |
21
|
|
|
use SilverStripe\ORM\ArrayList; |
22
|
|
|
use SilverStripe\ORM\DataList; |
23
|
|
|
use SilverStripe\ORM\DataObject; |
24
|
|
|
use SilverStripe\ORM\DB; |
25
|
|
|
use SilverStripe\ORM\ValidationException; |
26
|
|
|
use SilverStripe\Versioned\Versioned; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* Class SolrIndexTask |
30
|
|
|
* |
31
|
|
|
* @description Index items to Solr through a tasks |
32
|
|
|
* @package Firesphere\SolrSearch\Tasks |
33
|
|
|
*/ |
34
|
|
|
class SolrIndexTask extends BuildTask |
35
|
|
|
{ |
36
|
|
|
use LoggerTrait; |
37
|
|
|
/** |
38
|
|
|
* URLSegment of this task |
39
|
|
|
* |
40
|
|
|
* @var string |
41
|
|
|
*/ |
42
|
|
|
private static $segment = 'SolrIndexTask'; |
43
|
|
|
/** |
44
|
|
|
* Store the current states for all instances of SiteState |
45
|
|
|
* |
46
|
|
|
* @var array |
47
|
|
|
*/ |
48
|
|
|
public $currentStates; |
49
|
|
|
/** |
50
|
|
|
* My name |
51
|
|
|
* |
52
|
|
|
* @var string |
53
|
|
|
*/ |
54
|
|
|
protected $title = 'Solr Index update'; |
55
|
|
|
/** |
56
|
|
|
* What do I do? |
57
|
|
|
* |
58
|
|
|
* @var string |
59
|
|
|
*/ |
60
|
|
|
protected $description = 'Add or update documents to an existing Solr core.'; |
61
|
|
|
/** |
62
|
|
|
* Debug mode enabled, default false |
63
|
|
|
* |
64
|
|
|
* @var bool |
65
|
|
|
*/ |
66
|
|
|
protected $debug = false; |
67
|
|
|
/** |
68
|
|
|
* Singleton of {@link SolrCoreService} |
69
|
|
|
* |
70
|
|
|
* @var SolrCoreService |
71
|
|
|
*/ |
72
|
|
|
protected $service; |
73
|
|
|
/** |
74
|
|
|
* @var int Number of CPU cores available |
75
|
|
|
*/ |
76
|
|
|
protected $cores = 1; |
77
|
|
|
/** |
78
|
|
|
* Default batch length |
79
|
|
|
* |
80
|
|
|
* @var int |
81
|
|
|
*/ |
82
|
|
|
protected $batchLength = 1; |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @var BaseIndex Current core being indexed |
86
|
|
|
*/ |
87
|
|
|
protected $index; |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* SolrIndexTask constructor. Sets up the document factory |
91
|
|
|
* |
92
|
|
|
* @throws ReflectionException |
93
|
|
|
*/ |
94
|
15 |
|
public function __construct() |
95
|
|
|
{ |
96
|
15 |
|
parent::__construct(); |
97
|
|
|
// Only index live items. |
98
|
|
|
// The old FTS module also indexed Draft items. This is unnecessary |
99
|
15 |
|
Versioned::set_reading_mode(Versioned::DEFAULT_MODE); |
100
|
|
|
// If versioned is needed, a separate Versioned Search module is required |
101
|
15 |
|
$this->setService(Injector::inst()->get(SolrCoreService::class)); |
102
|
15 |
|
$this->setLogger(Injector::inst()->get(LoggerInterface::class)); |
103
|
15 |
|
$this->setDebug(Director::isDev() || Director::is_cli()); |
104
|
15 |
|
$this->setBatchLength(DocumentFactory::config()->get('batchLength')); |
105
|
15 |
|
$cores = SolrCoreService::config()->get('cpucores') ?: 1; |
106
|
15 |
|
$this->setCores($cores); |
107
|
15 |
|
$currentStates = SiteState::currentStates(); |
108
|
15 |
|
SiteState::setDefaultStates($currentStates); |
109
|
15 |
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* Set the {@link SolrCoreService} |
113
|
|
|
* |
114
|
|
|
* @param SolrCoreService $service |
115
|
|
|
* @return SolrIndexTask |
116
|
|
|
*/ |
117
|
15 |
|
public function setService(SolrCoreService $service): SolrIndexTask |
118
|
|
|
{ |
119
|
15 |
|
$this->service = $service; |
120
|
|
|
|
121
|
15 |
|
return $this; |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* Set the debug mode |
126
|
|
|
* |
127
|
|
|
* @param bool $debug |
128
|
|
|
* @return SolrIndexTask |
129
|
|
|
*/ |
130
|
15 |
|
public function setDebug(bool $debug): SolrIndexTask |
131
|
|
|
{ |
132
|
15 |
|
$this->debug = $debug; |
133
|
|
|
|
134
|
15 |
|
return $this; |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
/** |
138
|
|
|
* Implement this method in the task subclass to |
139
|
|
|
* execute via the TaskRunner |
140
|
|
|
* |
141
|
|
|
* @param HTTPRequest $request |
142
|
|
|
* @return int|bool |
143
|
|
|
* @throws Exception |
144
|
|
|
* @throws GuzzleException |
145
|
|
|
*/ |
146
|
13 |
|
public function run($request) |
147
|
|
|
{ |
148
|
13 |
|
$start = time(); |
149
|
13 |
|
$this->getLogger()->info(date('Y-m-d H:i:s')); |
150
|
13 |
|
list($vars, $group, $isGroup) = $this->taskSetup($request); |
151
|
13 |
|
$groups = 0; |
152
|
13 |
|
$indexes = $this->service->getValidIndexes($request->getVar('index')); |
153
|
|
|
|
154
|
13 |
|
foreach ($indexes as $indexName) { |
155
|
|
|
/** @var BaseIndex $index */ |
156
|
13 |
|
$index = Injector::inst()->get($indexName, false); |
157
|
13 |
|
$this->setIndex($index); |
158
|
|
|
|
159
|
13 |
|
$indexClasses = $this->index->getClasses(); |
160
|
13 |
|
$classes = $this->getClasses($vars, $indexClasses); |
161
|
13 |
|
if (!count($classes)) { |
162
|
10 |
|
continue; |
163
|
|
|
} |
164
|
|
|
|
165
|
13 |
|
$this->clearIndex($vars); |
166
|
|
|
|
167
|
13 |
|
$groups = $this->indexClassForIndex($classes, $isGroup, $group); |
168
|
|
|
} |
169
|
|
|
|
170
|
13 |
|
$this->getLogger()->info(date('Y-m-d H:i:s')); |
171
|
13 |
|
$this->getLogger()->info(sprintf('Time taken: %s minutes', (time() - $start) / 60)); |
172
|
|
|
|
173
|
13 |
|
return $groups; |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
/** |
177
|
|
|
* Set up the requirements for this task |
178
|
|
|
* |
179
|
|
|
* @param HTTPRequest $request |
180
|
|
|
* @return array |
181
|
|
|
*/ |
182
|
13 |
|
protected function taskSetup($request): array |
183
|
|
|
{ |
184
|
13 |
|
$vars = $request->getVars(); |
185
|
13 |
|
$this->debug = $this->debug || isset($vars['debug']); |
186
|
13 |
|
$group = $vars['group'] ?? 0; |
187
|
13 |
|
$start = $vars['start'] ?? 0; |
188
|
13 |
|
$group = ($start > $group) ? $start : $group; |
189
|
13 |
|
$isGroup = isset($vars['group']); |
190
|
|
|
|
191
|
13 |
|
return [$vars, $group, $isGroup]; |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
/** |
195
|
|
|
* get the classes to run for this task execution |
196
|
|
|
* |
197
|
|
|
* @param $vars |
198
|
|
|
* @param array $classes |
199
|
|
|
* @return bool|array |
200
|
|
|
*/ |
201
|
13 |
|
protected function getClasses($vars, array $classes): array |
202
|
|
|
{ |
203
|
13 |
|
if (isset($vars['class'])) { |
204
|
1 |
|
return array_intersect($classes, [$vars['class']]); |
205
|
|
|
} |
206
|
|
|
|
207
|
12 |
|
return $classes; |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
/** |
211
|
|
|
* Clear the given index if a full re-index is needed |
212
|
|
|
* |
213
|
|
|
* @param $vars |
214
|
|
|
* @throws Exception |
215
|
|
|
*/ |
216
|
13 |
|
public function clearIndex($vars) |
217
|
|
|
{ |
218
|
13 |
|
if (!empty($vars['clear'])) { |
219
|
1 |
|
$this->getLogger()->info(sprintf('Clearing index %s', $this->index->getIndexName())); |
220
|
1 |
|
$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $this->index); |
221
|
|
|
} |
222
|
13 |
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Index the classes for a specific index |
226
|
|
|
* |
227
|
|
|
* @param $classes |
228
|
|
|
* @param $isGroup |
229
|
|
|
* @param $group |
230
|
|
|
* @return int |
231
|
|
|
* @throws Exception |
232
|
|
|
* @throws GuzzleException |
233
|
|
|
*/ |
234
|
13 |
|
protected function indexClassForIndex($classes, $isGroup, $group): int |
235
|
|
|
{ |
236
|
13 |
|
$groups = 0; |
237
|
13 |
|
foreach ($classes as $class) { |
238
|
13 |
|
$groups = $this->indexClass($isGroup, $class, $group); |
239
|
|
|
} |
240
|
|
|
|
241
|
13 |
|
return $groups; |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Index a single class for a given index. {@link static::indexClassForIndex()} |
246
|
|
|
* |
247
|
|
|
* @param bool $isGroup |
248
|
|
|
* @param string $class |
249
|
|
|
* @param BaseIndex $index |
250
|
|
|
* @param int $group |
251
|
|
|
* @return int |
252
|
|
|
* @throws GuzzleException |
253
|
|
|
* @throws ValidationException |
254
|
|
|
*/ |
255
|
13 |
|
private function indexClass($isGroup, $class, int $group): int |
256
|
|
|
{ |
257
|
13 |
|
$index = $this->getIndex(); |
258
|
13 |
|
$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName())); |
259
|
13 |
|
$totalGroups = (int)ceil($class::get()->count() / $this->batchLength); |
260
|
13 |
|
$groups = $isGroup ? ($group + $this->cores - 1) : $totalGroups; |
261
|
13 |
|
$this->getLogger()->info(sprintf('Total groups %s', $totalGroups)); |
262
|
|
|
do { // Run from oldest to newest |
263
|
|
|
try { |
264
|
|
|
// The unittest param is from phpunit.xml.dist, meant to bypass the exit(0) call |
265
|
13 |
|
if (function_exists('pcntl_fork') && |
266
|
13 |
|
!Controller::curr()->getRequest()->getVar('unittest') |
267
|
|
|
) { |
268
|
|
|
$group = $this->spawnChildren($class, $group, $groups); |
269
|
|
|
} else { |
270
|
13 |
|
$this->doReindex($group, $class); |
271
|
|
|
} |
272
|
|
|
} catch (Exception $error) { |
273
|
|
|
$this->logException($index->getIndexName(), $group, $error); |
274
|
|
|
$group++; |
275
|
|
|
continue; |
276
|
|
|
} |
277
|
13 |
|
$group++; |
278
|
13 |
|
} while ($group <= $groups); |
279
|
|
|
|
280
|
13 |
|
return $totalGroups; |
281
|
|
|
} |
282
|
|
|
|
283
|
|
|
/** |
284
|
|
|
* @return BaseIndex |
285
|
|
|
*/ |
286
|
14 |
|
public function getIndex(): BaseIndex |
287
|
|
|
{ |
288
|
14 |
|
return $this->index; |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
/** |
292
|
|
|
* @param BaseIndex $index |
293
|
|
|
*/ |
294
|
14 |
|
public function setIndex(BaseIndex $index): void |
295
|
|
|
{ |
296
|
14 |
|
$this->index = $index; |
297
|
14 |
|
} |
298
|
|
|
|
299
|
|
|
/** |
300
|
|
|
* For each core, spawn a child process that will handle a separate group. |
301
|
|
|
* This speeds up indexing through CLI massively. |
302
|
|
|
* |
303
|
|
|
* @param string $class Class to index |
304
|
|
|
* @param int $group Group to index |
305
|
|
|
* @param int $groups Total amount of groups |
306
|
|
|
* @return int Last group indexed |
307
|
|
|
* @throws Exception |
308
|
|
|
* @throws GuzzleException |
309
|
|
|
*/ |
310
|
|
|
private function spawnChildren($class, int $group, int $groups): int |
311
|
|
|
{ |
312
|
|
|
$start = $group; |
313
|
|
|
$pids = []; |
314
|
|
|
$cores = $this->getCores(); |
315
|
|
|
// for each core, start a grouped indexing |
316
|
|
|
for ($i = 0; $i < $cores; $i++) { |
317
|
|
|
$start = $group + $i; |
318
|
|
|
if ($start < $groups) { |
319
|
|
|
$this->runForkedChild($class, $pids, $i, $start); |
320
|
|
|
} |
321
|
|
|
} |
322
|
|
|
// Wait for each child to finish |
323
|
|
|
foreach ($pids as $key => $pid) { |
324
|
|
|
pcntl_waitpid($pid, $status); |
325
|
|
|
if ($status === 0) { |
326
|
|
|
unset($pids[$key]); |
327
|
|
|
} |
328
|
|
|
} |
329
|
|
|
$commit = $this->index->getClient()->createUpdate(); |
330
|
|
|
$commit->addCommit(); |
331
|
|
|
|
332
|
|
|
$this->index->getClient()->update($commit); |
333
|
|
|
|
334
|
|
|
return $start; |
335
|
|
|
} |
336
|
|
|
|
337
|
|
|
/** |
338
|
|
|
* Create a fork and run the child |
339
|
|
|
* |
340
|
|
|
* @param string $class Class to index |
341
|
|
|
* @param array $pids Array of all the child PID's |
342
|
|
|
* @param int $coreNumber Core number |
343
|
|
|
* @param int $start Start point for the objects |
344
|
|
|
* @return void |
345
|
|
|
* @throws GuzzleException |
346
|
|
|
* @throws ValidationException |
347
|
|
|
*/ |
348
|
|
|
private function runForkedChild($class, array &$pids, int $coreNumber, int $start): void |
349
|
|
|
{ |
350
|
|
|
$pid = pcntl_fork(); |
351
|
|
|
// PID needs to be pushed before anything else, for some reason |
352
|
|
|
$pids[$coreNumber] = $pid; |
353
|
|
|
$config = DB::getConfig(); |
354
|
|
|
DB::connect($config); |
355
|
|
|
$this->runChild($class, $pid, $start); |
356
|
|
|
} |
357
|
|
|
|
358
|
|
|
/** |
359
|
|
|
* Ren a single child index operation |
360
|
|
|
* |
361
|
|
|
* @param string $class Class to index |
362
|
|
|
* @param int $pid PID of the child |
363
|
|
|
* @param int $start Position to start |
364
|
|
|
* @throws GuzzleException |
365
|
|
|
* @throws ValidationException |
366
|
|
|
* @throws Exception |
367
|
|
|
*/ |
368
|
|
|
private function runChild($class, int $pid, int $start): void |
369
|
|
|
{ |
370
|
|
|
if (!$pid) { |
371
|
|
|
try { |
372
|
|
|
$this->doReindex($start, $class, $pid); |
373
|
|
|
} catch (Exception $e) { |
374
|
|
|
SolrLogger::logMessage('ERROR', $e, $this->index->getIndexName()); |
375
|
|
|
$msg = sprintf( |
376
|
|
|
'Something went wrong while indexing %s on %s, see the logs for details', |
377
|
|
|
$start, |
378
|
|
|
$this->index->getIndexName() |
379
|
|
|
); |
380
|
|
|
throw new Exception($msg); |
381
|
|
|
} |
382
|
|
|
} |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
/** |
386
|
|
|
* Reindex the given group, for each state |
387
|
|
|
* |
388
|
|
|
* @param int $group |
389
|
|
|
* @param string $class |
390
|
|
|
* @param bool|int $pcntl |
391
|
|
|
* @throws Exception |
392
|
|
|
*/ |
393
|
13 |
|
private function doReindex($group, $class, $pcntl = false) |
394
|
|
|
{ |
395
|
13 |
|
foreach (SiteState::getStates() as $state) { |
396
|
13 |
|
if ($state !== 'default' && !empty($state)) { |
397
|
|
|
SiteState::withState($state); |
398
|
|
|
} |
399
|
13 |
|
$this->stateReindex($group, $class); |
400
|
|
|
} |
401
|
|
|
|
402
|
13 |
|
SiteState::withState(SiteState::DEFAULT_STATE); |
403
|
13 |
|
$this->getLogger()->info(sprintf('Indexed group %s', $group)); |
404
|
|
|
|
405
|
13 |
|
if ($pcntl !== false) { |
406
|
|
|
exit(0); |
|
|
|
|
407
|
|
|
} |
408
|
13 |
|
} |
409
|
|
|
|
410
|
|
|
/** |
411
|
|
|
* Index a group of a class for a specific state and index |
412
|
|
|
* |
413
|
|
|
* @param $group |
414
|
|
|
* @param $class |
415
|
|
|
* @throws Exception |
416
|
|
|
*/ |
417
|
13 |
|
private function stateReindex($group, $class): void |
418
|
|
|
{ |
419
|
|
|
// Generate filtered list of local records |
420
|
13 |
|
$baseClass = DataObject::getSchema()->baseDataClass($class); |
421
|
|
|
/** @var DataList|DataObject[] $items */ |
422
|
13 |
|
$items = DataObject::get($baseClass) |
423
|
13 |
|
->sort('ID ASC') |
424
|
13 |
|
->limit($this->batchLength, ($group * $this->batchLength)); |
425
|
13 |
|
if ($items->count()) { |
426
|
1 |
|
$this->updateIndex($items); |
427
|
|
|
} |
428
|
13 |
|
} |
429
|
|
|
|
430
|
|
|
/** |
431
|
|
|
* Execute the update on the client |
432
|
|
|
* |
433
|
|
|
* @param $items |
434
|
|
|
* @throws Exception |
435
|
|
|
*/ |
436
|
1 |
|
private function updateIndex($items): void |
437
|
|
|
{ |
438
|
1 |
|
$index = $this->getIndex(); |
439
|
1 |
|
$client = $index->getClient(); |
440
|
1 |
|
$update = $client->createUpdate(); |
441
|
1 |
|
$this->service->setInDebugMode($this->debug); |
442
|
1 |
|
$this->service->updateIndex($index, $items, $update); |
443
|
1 |
|
$client->update($update); |
444
|
1 |
|
} |
445
|
|
|
|
446
|
|
|
/** |
447
|
|
|
* Log an exception if it happens. Most are catched, these logs are for the developers |
448
|
|
|
* to identify problems and fix them. |
449
|
|
|
* |
450
|
|
|
* @param string $index |
451
|
|
|
* @param int $group |
452
|
|
|
* @param Exception $exception |
453
|
|
|
* @throws GuzzleException |
454
|
|
|
* @throws ValidationException |
455
|
|
|
*/ |
456
|
|
|
private function logException($index, int $group, Exception $exception): void |
457
|
|
|
{ |
458
|
|
|
$this->getLogger()->error($exception->getMessage()); |
459
|
|
|
$msg = sprintf( |
460
|
|
|
'Error indexing core %s on group %s,' . PHP_EOL . |
461
|
|
|
'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL, |
462
|
|
|
$index, |
463
|
|
|
$group |
464
|
|
|
); |
465
|
|
|
SolrLogger::logMessage('ERROR', $msg, $index); |
466
|
|
|
} |
467
|
|
|
|
468
|
|
|
/** |
469
|
|
|
* @return int |
470
|
|
|
*/ |
471
|
1 |
|
public function getCores(): int |
472
|
|
|
{ |
473
|
1 |
|
return $this->cores; |
474
|
|
|
} |
475
|
|
|
|
476
|
|
|
/** |
477
|
|
|
* @param int $cores |
478
|
|
|
*/ |
479
|
15 |
|
public function setCores(int $cores): void |
480
|
|
|
{ |
481
|
15 |
|
$this->cores = $cores; |
482
|
15 |
|
} |
483
|
|
|
|
484
|
|
|
/** |
485
|
|
|
* @return int |
486
|
|
|
*/ |
487
|
|
|
public function getBatchLength(): int |
488
|
|
|
{ |
489
|
|
|
return $this->batchLength; |
490
|
|
|
} |
491
|
|
|
|
492
|
|
|
/** |
493
|
|
|
* @param int $batchLength |
494
|
|
|
*/ |
495
|
15 |
|
public function setBatchLength(int $batchLength): void |
496
|
|
|
{ |
497
|
15 |
|
$this->batchLength = $batchLength; |
498
|
15 |
|
} |
499
|
|
|
} |
500
|
|
|
|
In general, usage of exit should be done with care and only when running in a scripting context like a CLI script.