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\Director; |
17
|
|
|
use SilverStripe\Control\HTTPRequest; |
18
|
|
|
use SilverStripe\Core\Injector\Injector; |
19
|
|
|
use SilverStripe\Dev\BuildTask; |
20
|
|
|
use SilverStripe\ORM\ArrayList; |
21
|
|
|
use SilverStripe\ORM\DataList; |
22
|
|
|
use SilverStripe\ORM\DataObject; |
23
|
|
|
use SilverStripe\ORM\ValidationException; |
24
|
|
|
use SilverStripe\Versioned\Versioned; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* Class SolrIndexTask |
28
|
|
|
* |
29
|
|
|
* @description Index items to Solr through a tasks |
30
|
|
|
* @package Firesphere\SolrSearch\Tasks |
31
|
|
|
*/ |
32
|
|
|
class SolrIndexTask extends BuildTask |
33
|
|
|
{ |
34
|
|
|
use LoggerTrait; |
35
|
|
|
/** |
36
|
|
|
* URLSegment of this task |
37
|
|
|
* |
38
|
|
|
* @var string |
39
|
|
|
*/ |
40
|
|
|
private static $segment = 'SolrIndexTask'; |
41
|
|
|
/** |
42
|
|
|
* Store the current states for all instances of SiteState |
43
|
|
|
* |
44
|
|
|
* @var array |
45
|
|
|
*/ |
46
|
|
|
public $currentStates; |
47
|
|
|
/** |
48
|
|
|
* My name |
49
|
|
|
* |
50
|
|
|
* @var string |
51
|
|
|
*/ |
52
|
|
|
protected $title = 'Solr Index update'; |
53
|
|
|
/** |
54
|
|
|
* What do I do? |
55
|
|
|
* |
56
|
|
|
* @var string |
57
|
|
|
*/ |
58
|
|
|
protected $description = 'Add or update documents to an existing Solr core.'; |
59
|
|
|
/** |
60
|
|
|
* Debug mode enabled, default false |
61
|
|
|
* |
62
|
|
|
* @var bool |
63
|
|
|
*/ |
64
|
|
|
protected $debug = false; |
65
|
|
|
/** |
66
|
|
|
* Singleton of {@link SolrCoreService} |
67
|
|
|
* |
68
|
|
|
* @var SolrCoreService |
69
|
|
|
*/ |
70
|
|
|
protected $service; |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* SolrIndexTask constructor. Sets up the document factory |
74
|
|
|
* |
75
|
|
|
* @throws ReflectionException |
76
|
|
|
*/ |
77
|
13 |
|
public function __construct() |
78
|
|
|
{ |
79
|
13 |
|
parent::__construct(); |
80
|
|
|
// Only index live items. |
81
|
|
|
// The old FTS module also indexed Draft items. This is unnecessary |
82
|
13 |
|
Versioned::set_reading_mode(Versioned::DEFAULT_MODE); |
83
|
13 |
|
$this->setService(Injector::inst()->get(SolrCoreService::class)); |
84
|
13 |
|
$this->setLogger(Injector::inst()->get(LoggerInterface::class)); |
85
|
13 |
|
$this->setDebug(Director::isDev() || Director::is_cli()); |
86
|
13 |
|
$this->currentStates = SiteState::currentStates(); |
87
|
13 |
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Set the {@link SolrCoreService} |
91
|
|
|
* |
92
|
|
|
* @param SolrCoreService $service |
93
|
|
|
* @return SolrIndexTask |
94
|
|
|
*/ |
95
|
13 |
|
public function setService(SolrCoreService $service): SolrIndexTask |
96
|
|
|
{ |
97
|
13 |
|
$this->service = $service; |
98
|
|
|
|
99
|
13 |
|
return $this; |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* Set the debug mode |
104
|
|
|
* |
105
|
|
|
* @param bool $debug |
106
|
|
|
* @return SolrIndexTask |
107
|
|
|
*/ |
108
|
13 |
|
public function setDebug(bool $debug): SolrIndexTask |
109
|
|
|
{ |
110
|
13 |
|
$this->debug = $debug; |
111
|
|
|
|
112
|
13 |
|
return $this; |
113
|
|
|
} |
114
|
|
|
|
115
|
|
|
/** |
116
|
|
|
* Implement this method in the task subclass to |
117
|
|
|
* execute via the TaskRunner |
118
|
|
|
* |
119
|
|
|
* @param HTTPRequest $request |
120
|
|
|
* @return int|bool |
121
|
|
|
* @throws Exception |
122
|
|
|
* @throws GuzzleException |
123
|
|
|
* @todo defer to background because it may run out of memory |
124
|
|
|
*/ |
125
|
12 |
|
public function run($request) |
126
|
|
|
{ |
127
|
12 |
|
$startTime = time(); |
128
|
12 |
|
list($vars, $group, $isGroup) = $this->taskSetup($request); |
129
|
12 |
|
$groups = 0; |
130
|
12 |
|
$indexes = $this->service->getValidIndexes($request->getVar('index')); |
131
|
|
|
|
132
|
12 |
|
foreach ($indexes as $indexName) { |
133
|
|
|
/** @var BaseIndex $index */ |
134
|
12 |
|
$index = Injector::inst()->get($indexName, false); |
135
|
|
|
|
136
|
12 |
|
$indexClasses = $index->getClasses(); |
137
|
12 |
|
$classes = $this->getClasses($vars, $indexClasses); |
138
|
12 |
|
if (!count($classes)) { |
139
|
9 |
|
continue; |
140
|
|
|
} |
141
|
|
|
|
142
|
12 |
|
$this->clearIndex($vars, $index); |
143
|
|
|
|
144
|
12 |
|
$groups = $this->indexClassForIndex($classes, $isGroup, $index, $group); |
145
|
|
|
} |
146
|
12 |
|
$this->getLogger()->info( |
147
|
12 |
|
sprintf('It took me %d seconds to do all the indexing%s', (time() - $startTime), PHP_EOL) |
148
|
|
|
); |
149
|
|
|
// Grab the latest logs from indexing if needed |
150
|
12 |
|
$solrLogger = new SolrLogger(); |
151
|
12 |
|
$solrLogger->saveSolrLog('Config'); |
152
|
|
|
|
153
|
12 |
|
return $groups; |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
/** |
157
|
|
|
* Set up the requirements for this task |
158
|
|
|
* |
159
|
|
|
* @param HTTPRequest $request |
160
|
|
|
* @return array |
161
|
|
|
*/ |
162
|
12 |
|
protected function taskSetup($request): array |
163
|
|
|
{ |
164
|
12 |
|
$vars = $request->getVars(); |
165
|
12 |
|
$this->debug = $this->debug || isset($vars['debug']); |
166
|
12 |
|
$group = $vars['group'] ?? 0; |
167
|
12 |
|
$start = $vars['start'] ?? 0; |
168
|
12 |
|
$group = ($start > $group) ? $start : $group; |
169
|
12 |
|
$isGroup = isset($vars['group']); |
170
|
|
|
|
171
|
12 |
|
return [$vars, $group, $isGroup]; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
/** |
175
|
|
|
* get the classes to run for this task execution |
176
|
|
|
* |
177
|
|
|
* @param $vars |
178
|
|
|
* @param array $classes |
179
|
|
|
* @return bool|array |
180
|
|
|
*/ |
181
|
12 |
|
protected function getClasses($vars, array $classes): array |
182
|
|
|
{ |
183
|
12 |
|
if (isset($vars['class'])) { |
184
|
1 |
|
return array_intersect($classes, [$vars['class']]); |
185
|
|
|
} |
186
|
|
|
|
187
|
11 |
|
return $classes; |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
/** |
191
|
|
|
* Clear the given index if a full re-index is needed |
192
|
|
|
* |
193
|
|
|
* @param $vars |
194
|
|
|
* @param BaseIndex $index |
195
|
|
|
* @throws Exception |
196
|
|
|
*/ |
197
|
12 |
|
public function clearIndex($vars, BaseIndex $index) |
198
|
|
|
{ |
199
|
12 |
|
if (!empty($vars['clear'])) { |
200
|
1 |
|
$this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName())); |
201
|
1 |
|
$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index); |
202
|
|
|
} |
203
|
12 |
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
|
|
* Index the classes for a specific index |
207
|
|
|
* |
208
|
|
|
* @param $classes |
209
|
|
|
* @param $isGroup |
210
|
|
|
* @param BaseIndex $index |
211
|
|
|
* @param $group |
212
|
|
|
* @return int |
213
|
|
|
* @throws Exception |
214
|
|
|
* @throws GuzzleException |
215
|
|
|
*/ |
216
|
12 |
|
protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int |
217
|
|
|
{ |
218
|
12 |
|
$groups = 0; |
219
|
12 |
|
foreach ($classes as $class) { |
220
|
12 |
|
$groups = $this->indexClass($isGroup, $class, $index, $group); |
221
|
|
|
} |
222
|
|
|
|
223
|
12 |
|
return $groups; |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
/** |
227
|
|
|
* Index a single class for a given index. {@link static::indexClassForIndex()} |
228
|
|
|
* |
229
|
|
|
* @param bool $isGroup |
230
|
|
|
* @param string $class |
231
|
|
|
* @param BaseIndex $index |
232
|
|
|
* @param int $group |
233
|
|
|
* @return int |
234
|
|
|
* @throws GuzzleException |
235
|
|
|
* @throws ValidationException |
236
|
|
|
*/ |
237
|
12 |
|
private function indexClass($isGroup, $class, BaseIndex $index, int $group): int |
238
|
|
|
{ |
239
|
12 |
|
$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()), []); |
240
|
|
|
|
241
|
12 |
|
$batchLength = DocumentFactory::config()->get('batchLength'); |
242
|
12 |
|
$groups = (int)ceil($class::get()->count() / $batchLength); |
243
|
12 |
|
$groups = $isGroup ? $group : $groups; |
244
|
12 |
|
while ($group <= $groups) { // Run from oldest to newest |
245
|
|
|
try { |
246
|
12 |
|
$this->doReindex($group, $class, $batchLength, $index); |
247
|
|
|
} catch (Exception $error) { |
248
|
|
|
$this->logException($index->getIndexName(), $group, $error); |
249
|
|
|
$group++; |
250
|
|
|
continue; |
251
|
|
|
} |
252
|
12 |
|
$group++; |
253
|
12 |
|
$this->getLogger()->info(sprintf('Indexed group %s', $group)); |
254
|
|
|
} |
255
|
|
|
|
256
|
12 |
|
return $groups; |
257
|
|
|
} |
258
|
|
|
|
259
|
|
|
/** |
260
|
|
|
* Reindex the given group, for each state |
261
|
|
|
* |
262
|
|
|
* @param int $group |
263
|
|
|
* @param string $class |
264
|
|
|
* @param int $batchLength |
265
|
|
|
* @param BaseIndex $index |
266
|
|
|
* @throws Exception |
267
|
|
|
*/ |
268
|
12 |
|
private function doReindex($group, $class, $batchLength, BaseIndex $index): void |
269
|
|
|
{ |
270
|
12 |
|
foreach (SiteState::getStates() as $state) { |
271
|
12 |
|
if ($state !== 'default') { |
272
|
|
|
SiteState::withState($state); |
273
|
|
|
} |
274
|
12 |
|
$this->stateReindex($group, $class, $batchLength, $index); |
275
|
|
|
} |
276
|
|
|
|
277
|
|
|
// Reset the variants back to it's original state for the next round |
278
|
12 |
|
foreach ($this->currentStates as $variant => $value) { |
279
|
11 |
|
singleton($variant)->activateState($value); |
280
|
|
|
} |
281
|
12 |
|
} |
282
|
|
|
|
283
|
|
|
/** |
284
|
|
|
* Index a group of a class for a specific state and index |
285
|
|
|
* |
286
|
|
|
* @param $group |
287
|
|
|
* @param $class |
288
|
|
|
* @param $batchLength |
289
|
|
|
* @param BaseIndex $index |
290
|
|
|
* @throws Exception |
291
|
|
|
*/ |
292
|
12 |
|
private function stateReindex($group, $class, $batchLength, BaseIndex $index): void |
293
|
|
|
{ |
294
|
|
|
// Generate filtered list of local records |
295
|
12 |
|
$baseClass = DataObject::getSchema()->baseDataClass($class); |
296
|
|
|
/** @var DataList|DataObject[] $items */ |
297
|
12 |
|
$items = DataObject::get($baseClass) |
298
|
12 |
|
->sort('ID ASC') |
299
|
12 |
|
->limit($batchLength, ($group * $batchLength)); |
300
|
12 |
|
if ($items->count()) { |
301
|
1 |
|
$this->updateIndex($index, $items); |
302
|
|
|
} |
303
|
12 |
|
} |
304
|
|
|
|
305
|
|
|
/** |
306
|
|
|
* Execute the update on the client |
307
|
|
|
* |
308
|
|
|
* @param BaseIndex $index |
309
|
|
|
* @param $items |
310
|
|
|
* @throws Exception |
311
|
|
|
*/ |
312
|
1 |
|
private function updateIndex(BaseIndex $index, $items): void |
313
|
|
|
{ |
314
|
1 |
|
$client = $index->getClient(); |
315
|
1 |
|
$update = $client->createUpdate(); |
316
|
1 |
|
$this->service->setInDebugMode($this->debug); |
317
|
1 |
|
$this->service->updateIndex($index, $items, $update); |
318
|
1 |
|
$update->addCommit(); |
319
|
1 |
|
$client->update($update); |
320
|
1 |
|
} |
321
|
|
|
|
322
|
|
|
/** |
323
|
|
|
* Log an exception if it happens. Most are catched, these logs are for the developers |
324
|
|
|
* to identify problems and fix them. |
325
|
|
|
* |
326
|
|
|
* @param string $index |
327
|
|
|
* @param int $group |
328
|
|
|
* @param Exception $exception |
329
|
|
|
* @throws GuzzleException |
330
|
|
|
* @throws ValidationException |
331
|
|
|
*/ |
332
|
|
|
private function logException($index, int $group, Exception $exception): void |
333
|
|
|
{ |
334
|
|
|
$this->getLogger()->error($exception->getMessage()); |
335
|
|
|
$msg = sprintf( |
336
|
|
|
'Error indexing core %s on group %s,' . PHP_EOL . |
337
|
|
|
'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL, |
338
|
|
|
$index, |
339
|
|
|
$group |
340
|
|
|
); |
341
|
|
|
SolrLogger::logMessage('ERROR', $msg, $index); |
342
|
|
|
} |
343
|
|
|
} |
344
|
|
|
|