1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Class SolrIndexTask|Firesphere\SolrSearch\Tasks\SolrIndexTask Index Solr cores |
4
|
|
|
* |
5
|
|
|
* @package Firesphere\SolrSearch\Tasks |
6
|
|
|
* @author Simon `Firesphere` Erkelens; Marco `Sheepy` Hermo |
7
|
|
|
* @copyright Copyright (c) 2018 - now() Firesphere & Sheepy |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
namespace Firesphere\SolrSearch\Tasks; |
11
|
|
|
|
12
|
|
|
use Exception; |
13
|
|
|
use Firesphere\SolrSearch\Factories\DocumentFactory; |
14
|
|
|
use Firesphere\SolrSearch\Helpers\SolrLogger; |
15
|
|
|
use Firesphere\SolrSearch\Indexes\BaseIndex; |
16
|
|
|
use Firesphere\SolrSearch\Services\SolrCoreService; |
17
|
|
|
use Firesphere\SolrSearch\States\SiteState; |
18
|
|
|
use Firesphere\SolrSearch\Traits\LoggerTrait; |
19
|
|
|
use Firesphere\SolrSearch\Traits\SolrIndexTrait; |
20
|
|
|
use GuzzleHttp\Exception\GuzzleException; |
21
|
|
|
use Psr\Log\LoggerInterface; |
22
|
|
|
use ReflectionException; |
23
|
|
|
use SilverStripe\Control\Controller; |
24
|
|
|
use SilverStripe\Control\Director; |
25
|
|
|
use SilverStripe\Control\HTTPRequest; |
26
|
|
|
use SilverStripe\Core\Injector\Injector; |
27
|
|
|
use SilverStripe\Dev\BuildTask; |
28
|
|
|
use SilverStripe\ORM\ArrayList; |
29
|
|
|
use SilverStripe\ORM\DataList; |
30
|
|
|
use SilverStripe\ORM\DataObject; |
31
|
|
|
use SilverStripe\ORM\DB; |
32
|
|
|
use SilverStripe\ORM\SS_List; |
33
|
|
|
use SilverStripe\ORM\ValidationException; |
34
|
|
|
use SilverStripe\Versioned\Versioned; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Class SolrIndexTask |
38
|
|
|
* |
39
|
|
|
* @description Index items to Solr through a tasks |
40
|
|
|
* @package Firesphere\SolrSearch\Tasks |
41
|
|
|
*/ |
42
|
|
|
class SolrIndexTask extends BuildTask |
43
|
|
|
{ |
44
|
|
|
use LoggerTrait; |
45
|
|
|
use SolrIndexTrait; |
46
|
|
|
/** |
47
|
|
|
* URLSegment of this task |
48
|
|
|
* |
49
|
|
|
* @var string |
50
|
|
|
*/ |
51
|
|
|
private static $segment = 'SolrIndexTask'; |
52
|
|
|
/** |
53
|
|
|
* Store the current states for all instances of SiteState |
54
|
|
|
* |
55
|
|
|
* @var array |
56
|
|
|
*/ |
57
|
|
|
public $currentStates; |
58
|
|
|
/** |
59
|
|
|
* My name |
60
|
|
|
* |
61
|
|
|
* @var string |
62
|
|
|
*/ |
63
|
|
|
protected $title = 'Solr Index update'; |
64
|
|
|
/** |
65
|
|
|
* What do I do? |
66
|
|
|
* |
67
|
|
|
* @var string |
68
|
|
|
*/ |
69
|
|
|
protected $description = 'Add or update documents to an existing Solr core.'; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* SolrIndexTask constructor. Sets up the document factory |
73
|
|
|
* |
74
|
|
|
* @throws ReflectionException |
75
|
|
|
*/ |
76
|
16 |
|
public function __construct() |
77
|
|
|
{ |
78
|
16 |
|
parent::__construct(); |
79
|
|
|
// Only index live items. |
80
|
|
|
// The old FTS module also indexed Draft items. This is unnecessary |
81
|
|
|
// If versioned is needed, a separate Versioned Search module is required |
82
|
16 |
|
Versioned::set_reading_mode(Versioned::DEFAULT_MODE); |
83
|
16 |
|
$this->setService(Injector::inst()->get(SolrCoreService::class)); |
84
|
16 |
|
$this->setLogger(Injector::inst()->get(LoggerInterface::class)); |
85
|
16 |
|
$this->setDebug(Director::isDev() || Director::is_cli()); |
86
|
16 |
|
$this->setBatchLength(DocumentFactory::config()->get('batchLength')); |
87
|
16 |
|
$cores = SolrCoreService::config()->get('cpucores') ?: 1; |
88
|
16 |
|
$this->setCores($cores); |
89
|
16 |
|
$currentStates = SiteState::currentStates(); |
90
|
16 |
|
SiteState::setDefaultStates($currentStates); |
91
|
16 |
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* Implement this method in the task subclass to |
95
|
|
|
* execute via the TaskRunner |
96
|
|
|
* |
97
|
|
|
* @param HTTPRequest $request Current request |
98
|
|
|
* @return int|bool |
99
|
|
|
* @throws Exception |
100
|
|
|
* @throws GuzzleException |
101
|
|
|
*/ |
102
|
13 |
|
public function run($request) |
103
|
|
|
{ |
104
|
13 |
|
$start = time(); |
105
|
13 |
|
$this->getLogger()->info(date('Y-m-d H:i:s')); |
106
|
13 |
|
list($vars, $group, $isGroup) = $this->taskSetup($request); |
107
|
13 |
|
$groups = 0; |
108
|
13 |
|
$indexes = $this->service->getValidIndexes($request->getVar('index')); |
109
|
|
|
|
110
|
13 |
|
foreach ($indexes as $indexName) { |
111
|
|
|
/** @var BaseIndex $index */ |
112
|
13 |
|
$index = Injector::inst()->get($indexName, false); |
113
|
13 |
|
$this->setIndex($index); |
114
|
|
|
|
115
|
13 |
|
$indexClasses = $this->index->getClasses(); |
116
|
13 |
|
$classes = $this->getClasses($vars, $indexClasses); |
117
|
13 |
|
if (!count($classes)) { |
118
|
10 |
|
continue; |
119
|
|
|
} |
120
|
|
|
|
121
|
13 |
|
$this->clearIndex($vars); |
122
|
|
|
|
123
|
13 |
|
$groups = $this->indexClassForIndex($classes, $isGroup, $group); |
124
|
|
|
} |
125
|
|
|
|
126
|
13 |
|
$this->getLogger()->info(date('Y-m-d H:i:s')); |
127
|
13 |
|
$this->getLogger()->info(sprintf('Time taken: %s minutes', (time() - $start) / 60)); |
128
|
|
|
|
129
|
13 |
|
return $groups; |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
/** |
133
|
|
|
* Set up the requirements for this task |
134
|
|
|
* |
135
|
|
|
* @param HTTPRequest $request Current request |
136
|
|
|
* @return array |
137
|
|
|
*/ |
138
|
13 |
|
protected function taskSetup($request): array |
139
|
|
|
{ |
140
|
13 |
|
$vars = $request->getVars(); |
141
|
13 |
|
$debug = $this->isDebug() || isset($vars['debug']); |
142
|
|
|
// Forcefully set the debugging to whatever the outcome of the above is |
143
|
13 |
|
$this->setDebug($debug, true); |
144
|
13 |
|
$group = $vars['group'] ?? 0; |
145
|
13 |
|
$start = $vars['start'] ?? 0; |
146
|
13 |
|
$group = ($start > $group) ? $start : $group; |
147
|
13 |
|
$isGroup = isset($vars['group']); |
148
|
|
|
|
149
|
13 |
|
return [$vars, $group, $isGroup]; |
150
|
|
|
} |
151
|
|
|
|
152
|
|
|
/** |
153
|
|
|
* get the classes to run for this task execution |
154
|
|
|
* |
155
|
|
|
* @param array $vars URL GET Parameters |
156
|
|
|
* @param array $classes Classes to index |
157
|
|
|
* @return bool|array |
158
|
|
|
*/ |
159
|
13 |
|
protected function getClasses($vars, array $classes): array |
160
|
|
|
{ |
161
|
13 |
|
if (isset($vars['class'])) { |
162
|
1 |
|
return array_intersect($classes, [$vars['class']]); |
163
|
|
|
} |
164
|
|
|
|
165
|
12 |
|
return $classes; |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
/** |
169
|
|
|
* Clear the given index if a full re-index is needed |
170
|
|
|
* |
171
|
|
|
* @param array $vars URL GET Parameters |
172
|
|
|
* @throws Exception |
173
|
|
|
*/ |
174
|
13 |
|
public function clearIndex($vars) |
175
|
|
|
{ |
176
|
13 |
|
if (!empty($vars['clear'])) { |
177
|
1 |
|
$this->getLogger()->info(sprintf('Clearing index %s', $this->index->getIndexName())); |
178
|
1 |
|
$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $this->index); |
179
|
|
|
} |
180
|
13 |
|
} |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* Index the classes for a specific index |
184
|
|
|
* |
185
|
|
|
* @param array $classes Classes that need indexing |
186
|
|
|
* @param bool $isGroup Indexing a specific group? |
187
|
|
|
* @param int $group Group to index |
188
|
|
|
* @return int |
189
|
|
|
* @throws Exception |
190
|
|
|
* @throws GuzzleException |
191
|
|
|
*/ |
192
|
13 |
|
protected function indexClassForIndex($classes, $isGroup, $group): int |
193
|
|
|
{ |
194
|
13 |
|
$groups = 0; |
195
|
13 |
|
foreach ($classes as $class) { |
196
|
13 |
|
$groups = $this->indexClass($isGroup, $class, $group); |
197
|
|
|
} |
198
|
|
|
|
199
|
13 |
|
return $groups; |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* Index a single class for a given index. {@link static::indexClassForIndex()} |
204
|
|
|
* |
205
|
|
|
* @param bool $isGroup Is a specific group indexed |
206
|
|
|
* @param string $class Class to index |
207
|
|
|
* @param int $group Group to index |
208
|
|
|
* @return int |
209
|
|
|
* @throws GuzzleException |
210
|
|
|
* @throws ValidationException |
211
|
|
|
*/ |
212
|
13 |
|
private function indexClass($isGroup, $class, int $group): int |
213
|
|
|
{ |
214
|
13 |
|
$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $this->getIndex()->getIndexName())); |
215
|
13 |
|
list($totalGroups, $groups) = $this->getGroupSettings($isGroup, $class, $group); |
216
|
13 |
|
$this->getLogger()->info(sprintf('Total groups %s', $totalGroups)); |
217
|
|
|
do { // Run from oldest to newest |
218
|
|
|
try { |
219
|
13 |
|
if ($this->hasPCNTL()) { |
220
|
|
|
// @codeCoverageIgnoreStart |
221
|
|
|
$group = $this->spawnChildren($class, $group, $groups); |
222
|
|
|
// @codeCoverageIgnoreEnd |
223
|
|
|
} else { |
224
|
13 |
|
$this->doReindex($group, $class); |
225
|
|
|
} |
226
|
13 |
|
$group++; |
227
|
|
|
} catch (Exception $error) { |
228
|
|
|
// @codeCoverageIgnoreStart |
229
|
|
|
$this->logException($this->index->getIndexName(), $group, $error); |
230
|
|
|
$group++; |
231
|
|
|
continue; |
232
|
|
|
// @codeCoverageIgnoreEnd |
233
|
|
|
} |
234
|
13 |
|
} while ($group <= $groups); |
235
|
|
|
|
236
|
13 |
|
return $totalGroups; |
237
|
|
|
} |
238
|
|
|
|
239
|
|
|
/** |
240
|
|
|
* Check the amount of groups and the total against the isGroup check. |
241
|
|
|
* |
242
|
|
|
* @param bool $isGroup Is it a specific group |
243
|
|
|
* @param string $class Class to check |
244
|
|
|
* @param int $group Current group to index |
245
|
|
|
* @return array |
246
|
|
|
*/ |
247
|
13 |
|
private function getGroupSettings($isGroup, $class, int $group): array |
248
|
|
|
{ |
249
|
13 |
|
$totalGroups = (int)ceil($class::get()->count() / $this->getBatchLength()); |
250
|
13 |
|
$groups = $isGroup ? ($group + $this->getCores() - 1) : $totalGroups; |
251
|
|
|
|
252
|
13 |
|
return [$totalGroups, $groups]; |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* Check if PCNTL is available and/or useable. |
257
|
|
|
* The unittest param is from phpunit.xml.dist, meant to bypass the exit(0) call |
258
|
|
|
* The pcntl parameter check is for unit tests, but PHPUnit does not support PCNTL (yet) |
259
|
|
|
* |
260
|
|
|
* @return bool |
261
|
|
|
*/ |
262
|
13 |
|
private function hasPCNTL() |
263
|
|
|
{ |
264
|
13 |
|
return Director::is_cli() && |
265
|
13 |
|
function_exists('pcntl_fork') && |
266
|
|
|
(Controller::curr()->getRequest()->getVar('unittest') === 'pcntl' || |
267
|
13 |
|
!Controller::curr()->getRequest()->getVar('unittest')); |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
/** |
271
|
|
|
* For each core, spawn a child process that will handle a separate group. |
272
|
|
|
* This speeds up indexing through CLI massively. |
273
|
|
|
* |
274
|
|
|
* @codeCoverageIgnore Can't be tested because PCNTL is not available |
275
|
|
|
* @param string $class Class to index |
276
|
|
|
* @param int $group Group to index |
277
|
|
|
* @param int $groups Total amount of groups |
278
|
|
|
* @return int Last group indexed |
279
|
|
|
* @throws Exception |
280
|
|
|
* @throws GuzzleException |
281
|
|
|
*/ |
282
|
|
|
private function spawnChildren($class, int $group, int $groups): int |
283
|
|
|
{ |
284
|
|
|
$start = $group; |
285
|
|
|
$pids = []; |
286
|
|
|
$cores = $this->getCores(); |
287
|
|
|
// for each core, start a grouped indexing |
288
|
|
|
for ($i = 0; $i < $cores; $i++) { |
289
|
|
|
$start = $group + $i; |
290
|
|
|
if ($start < $groups) { |
291
|
|
|
$this->runForkedChild($class, $pids, $start); |
292
|
|
|
} |
293
|
|
|
} |
294
|
|
|
// Wait for each child to finish |
295
|
|
|
// It needs to wait for them independently, |
296
|
|
|
// or it runs out of memory for some reason |
297
|
|
|
foreach ($pids as $pid) { |
298
|
|
|
pcntl_waitpid($pid, $status); |
299
|
|
|
} |
300
|
|
|
$commit = $this->index->getClient()->createUpdate(); |
301
|
|
|
$commit->addCommit(); |
302
|
|
|
|
303
|
|
|
$this->index->getClient()->update($commit); |
304
|
|
|
|
305
|
|
|
return $start; |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
/** |
309
|
|
|
* Create a fork and run the child |
310
|
|
|
* |
311
|
|
|
* @codeCoverageIgnore Can't be tested because PCNTL is not available |
312
|
|
|
* @param string $class Class to index |
313
|
|
|
* @param array $pids Array of all the child Process IDs |
314
|
|
|
* @param int $start Start point for the objects |
315
|
|
|
* @return void |
316
|
|
|
* @throws GuzzleException |
317
|
|
|
* @throws ValidationException |
318
|
|
|
*/ |
319
|
|
|
private function runForkedChild($class, array &$pids, int $start): void |
320
|
|
|
{ |
321
|
|
|
$pid = pcntl_fork(); |
322
|
|
|
// PID needs to be pushed before anything else, for some reason |
323
|
|
|
$pids[] = $pid; |
324
|
|
|
$config = DB::getConfig(); |
325
|
|
|
DB::connect($config); |
326
|
|
|
$this->runChild($class, $pid, $start); |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
/** |
330
|
|
|
* Ren a single child index operation |
331
|
|
|
* |
332
|
|
|
* @codeCoverageIgnore Can't be tested because PCNTL is not available |
333
|
|
|
* @param string $class Class to index |
334
|
|
|
* @param int $pid PID of the child |
335
|
|
|
* @param int $start Position to start |
336
|
|
|
* @throws GuzzleException |
337
|
|
|
* @throws ValidationException |
338
|
|
|
* @throws Exception |
339
|
|
|
*/ |
340
|
|
|
private function runChild($class, int $pid, int $start): void |
341
|
|
|
{ |
342
|
|
|
try { |
343
|
|
|
$this->doReindex($start, $class, $pid); |
344
|
|
|
} catch (Exception $error) { |
345
|
|
|
SolrLogger::logMessage('ERROR', $error, $this->index->getIndexName()); |
346
|
|
|
$msg = sprintf( |
347
|
|
|
'Something went wrong while indexing %s on %s, see the logs for details', |
348
|
|
|
$start, |
349
|
|
|
$this->index->getIndexName() |
350
|
|
|
); |
351
|
|
|
throw new Exception($msg); |
352
|
|
|
} |
353
|
|
|
} |
354
|
|
|
|
355
|
|
|
/** |
356
|
|
|
* Reindex the given group, for each state |
357
|
|
|
* |
358
|
|
|
* @param int $group Group to index |
359
|
|
|
* @param string $class Class to index |
360
|
|
|
* @param bool|int $pcntl Are we a child process or not |
361
|
|
|
* @throws Exception |
362
|
|
|
*/ |
363
|
13 |
|
private function doReindex($group, $class, $pcntl = false) |
364
|
|
|
{ |
365
|
13 |
|
foreach (SiteState::getStates() as $state) { |
366
|
13 |
|
if ($state !== 'default' && !empty($state)) { |
367
|
|
|
SiteState::withState($state); |
368
|
|
|
} |
369
|
13 |
|
$this->stateReindex($group, $class); |
370
|
|
|
} |
371
|
|
|
|
372
|
13 |
|
SiteState::withState(SiteState::DEFAULT_STATE); |
373
|
13 |
|
$this->getLogger()->info(sprintf('Indexed group %s', $group)); |
374
|
|
|
|
375
|
|
|
// @codeCoverageIgnoreStart |
376
|
|
|
if ($pcntl !== false) { |
377
|
|
|
exit(0); |
|
|
|
|
378
|
|
|
} |
379
|
|
|
// @codeCoverageIgnoreEnd |
380
|
13 |
|
} |
381
|
|
|
|
382
|
|
|
/** |
383
|
|
|
* Index a group of a class for a specific state and index |
384
|
|
|
* |
385
|
|
|
* @param string $group Group to index |
386
|
|
|
* @param string $class Class to index |
387
|
|
|
* @throws Exception |
388
|
|
|
*/ |
389
|
13 |
|
private function stateReindex($group, $class): void |
390
|
|
|
{ |
391
|
|
|
// Generate filtered list of local records |
392
|
13 |
|
$baseClass = DataObject::getSchema()->baseDataClass($class); |
393
|
|
|
/** @var DataList|DataObject[] $items */ |
394
|
13 |
|
$items = DataObject::get($baseClass) |
395
|
13 |
|
->sort('ID ASC') |
396
|
13 |
|
->limit($this->getBatchLength(), ($group * $this->getBatchLength())); |
397
|
13 |
|
if ($items->count()) { |
398
|
1 |
|
$this->updateIndex($items); |
399
|
|
|
} |
400
|
13 |
|
} |
401
|
|
|
|
402
|
|
|
/** |
403
|
|
|
* Execute the update on the client |
404
|
|
|
* |
405
|
|
|
* @param SS_List $items Items to index |
406
|
|
|
* @throws Exception |
407
|
|
|
*/ |
408
|
1 |
|
private function updateIndex($items): void |
409
|
|
|
{ |
410
|
1 |
|
$index = $this->getIndex(); |
411
|
1 |
|
$client = $index->getClient(); |
412
|
1 |
|
$update = $client->createUpdate(); |
413
|
1 |
|
$this->service->setDebug($this->debug); |
414
|
1 |
|
$this->service->updateIndex($index, $items, $update); |
415
|
1 |
|
$client->update($update); |
416
|
1 |
|
} |
417
|
|
|
|
418
|
|
|
/** |
419
|
|
|
* Log an exception if it happens. Most are catched, these logs are for the developers |
420
|
|
|
* to identify problems and fix them. |
421
|
|
|
* |
422
|
|
|
* @codeCoverageIgnore This is actually tested through reflection |
423
|
|
|
* @param string $index Index that is currently running |
424
|
|
|
* @param int $group Group currently attempted to index |
425
|
|
|
* @param Exception $exception Exception that's been thrown |
426
|
|
|
* @throws GuzzleException |
427
|
|
|
* @throws ValidationException |
428
|
|
|
*/ |
429
|
|
|
private function logException($index, int $group, Exception $exception): void |
430
|
|
|
{ |
431
|
|
|
$this->getLogger()->error($exception->getMessage()); |
432
|
|
|
$msg = sprintf( |
433
|
|
|
'Error indexing core %s on group %s,' . PHP_EOL . |
434
|
|
|
'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL, |
435
|
|
|
$index, |
436
|
|
|
$group |
437
|
|
|
); |
438
|
|
|
SolrLogger::logMessage('ERROR', $msg, $index); |
439
|
|
|
} |
440
|
|
|
} |
441
|
|
|
|
In general, usage of exit should be done with care and only when running in a scripting context like a CLI script.