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