|
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 SilverStripe\Control\Director; |
|
16
|
|
|
use SilverStripe\Control\HTTPRequest; |
|
17
|
|
|
use SilverStripe\Core\Injector\Injector; |
|
18
|
|
|
use SilverStripe\Dev\BuildTask; |
|
19
|
|
|
use SilverStripe\ORM\ArrayList; |
|
20
|
|
|
use SilverStripe\ORM\DataList; |
|
21
|
|
|
use SilverStripe\ORM\DataObject; |
|
22
|
|
|
use SilverStripe\ORM\ValidationException; |
|
23
|
|
|
use SilverStripe\Versioned\Versioned; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Class SolrIndexTask |
|
27
|
|
|
* |
|
28
|
|
|
* @description Index items to Solr through a tasks |
|
29
|
|
|
* @package Firesphere\SolrSearch\Tasks |
|
30
|
|
|
*/ |
|
31
|
|
|
class SolrIndexTask extends BuildTask |
|
32
|
|
|
{ |
|
33
|
|
|
use LoggerTrait; |
|
34
|
|
|
/** |
|
35
|
|
|
* @var string |
|
36
|
|
|
*/ |
|
37
|
|
|
private static $segment = 'SolrIndexTask'; |
|
38
|
|
|
/** |
|
39
|
|
|
* @var string |
|
40
|
|
|
*/ |
|
41
|
|
|
protected $title = 'Solr Index update'; |
|
42
|
|
|
/** |
|
43
|
|
|
* @var string |
|
44
|
|
|
*/ |
|
45
|
|
|
protected $description = 'Add or update documents to an existing Solr core.'; |
|
46
|
|
|
/** |
|
47
|
|
|
* @var array Store the current states for all instances of SiteState |
|
48
|
|
|
*/ |
|
49
|
|
|
public $currentStates; |
|
50
|
|
|
/** |
|
51
|
|
|
* @var bool |
|
52
|
|
|
*/ |
|
53
|
|
|
protected $debug = false; |
|
54
|
|
|
/** |
|
55
|
|
|
* @var SolrCoreService |
|
56
|
|
|
*/ |
|
57
|
|
|
protected $service; |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* SolrIndexTask constructor. Sets up the document factory |
|
61
|
|
|
* |
|
62
|
|
|
* @throws \ReflectionException |
|
63
|
|
|
*/ |
|
64
|
12 |
|
public function __construct() |
|
65
|
|
|
{ |
|
66
|
12 |
|
parent::__construct(); |
|
67
|
|
|
// Only index live items. |
|
68
|
|
|
// The old FTS module also indexed Draft items. This is unnecessary |
|
69
|
12 |
|
Versioned::set_reading_mode(Versioned::DEFAULT_MODE); |
|
70
|
12 |
|
$this->setService(Injector::inst()->get(SolrCoreService::class)); |
|
71
|
12 |
|
$this->setLogger(Injector::inst()->get(LoggerInterface::class)); |
|
72
|
12 |
|
$this->setDebug(Director::isDev() || Director::is_cli()); |
|
73
|
12 |
|
$this->currentStates = SiteState::currentStates(); |
|
74
|
12 |
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* @param SolrCoreService $service |
|
78
|
|
|
* @return SolrIndexTask |
|
79
|
|
|
*/ |
|
80
|
12 |
|
public function setService(SolrCoreService $service): SolrIndexTask |
|
81
|
|
|
{ |
|
82
|
12 |
|
$this->service = $service; |
|
83
|
|
|
|
|
84
|
12 |
|
return $this; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
/** |
|
88
|
|
|
* @param bool $debug |
|
89
|
|
|
* @return SolrIndexTask |
|
90
|
|
|
*/ |
|
91
|
12 |
|
public function setDebug(bool $debug): SolrIndexTask |
|
92
|
|
|
{ |
|
93
|
12 |
|
$this->debug = $debug; |
|
94
|
|
|
|
|
95
|
12 |
|
return $this; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Implement this method in the task subclass to |
|
100
|
|
|
* execute via the TaskRunner |
|
101
|
|
|
* |
|
102
|
|
|
* @param HTTPRequest $request |
|
103
|
|
|
* @return int|bool |
|
104
|
|
|
* @throws Exception |
|
105
|
|
|
* @throws GuzzleException |
|
106
|
|
|
* @todo defer to background because it may run out of memory |
|
107
|
|
|
*/ |
|
108
|
11 |
|
public function run($request) |
|
109
|
|
|
{ |
|
110
|
11 |
|
$startTime = time(); |
|
111
|
11 |
|
[$vars, $group, $isGroup] = $this->taskSetup($request); |
|
112
|
11 |
|
$groups = 0; |
|
113
|
11 |
|
$indexes = $this->service->getValidIndexes($request->getVar('index')); |
|
114
|
|
|
|
|
115
|
11 |
|
foreach ($indexes as $indexName) { |
|
116
|
|
|
/** @var BaseIndex $index */ |
|
117
|
11 |
|
$index = Injector::inst()->get($indexName, false); |
|
118
|
|
|
|
|
119
|
11 |
|
$indexClasses = $index->getClasses(); |
|
120
|
11 |
|
$classes = $this->getClasses($vars, $indexClasses); |
|
121
|
11 |
|
if (!count($classes)) { |
|
122
|
8 |
|
continue; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
11 |
|
$this->clearIndex($vars, $index); |
|
126
|
|
|
|
|
127
|
11 |
|
$groups = $this->indexClassForIndex($classes, $isGroup, $index, $group); |
|
128
|
|
|
} |
|
129
|
11 |
|
$this->getLogger()->info( |
|
130
|
11 |
|
sprintf('It took me %d seconds to do all the indexing%s', (time() - $startTime), PHP_EOL) |
|
131
|
|
|
); |
|
132
|
|
|
// Grab the latest logs from indexing if needed |
|
133
|
11 |
|
$solrLogger = new SolrLogger(); |
|
134
|
11 |
|
$solrLogger->saveSolrLog('Config'); |
|
135
|
|
|
|
|
136
|
11 |
|
return $groups; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* @param HTTPRequest $request |
|
141
|
|
|
* @return array |
|
142
|
|
|
*/ |
|
143
|
11 |
|
protected function taskSetup($request): array |
|
144
|
|
|
{ |
|
145
|
11 |
|
$vars = $request->getVars(); |
|
146
|
11 |
|
$this->debug = $this->debug || isset($vars['debug']); |
|
147
|
11 |
|
$group = $vars['group'] ?? 0; |
|
148
|
11 |
|
$start = $vars['start'] ?? 0; |
|
149
|
11 |
|
$group = ($start > $group) ? $start : $group; |
|
150
|
11 |
|
$isGroup = isset($vars['group']); |
|
151
|
|
|
|
|
152
|
11 |
|
return [$vars, $group, $isGroup]; |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
/** |
|
156
|
|
|
* @param $vars |
|
157
|
|
|
* @param array $classes |
|
158
|
|
|
* @return bool|array |
|
159
|
|
|
*/ |
|
160
|
11 |
|
protected function getClasses($vars, array $classes): array |
|
161
|
|
|
{ |
|
162
|
11 |
|
if (isset($vars['class'])) { |
|
163
|
1 |
|
return array_intersect($classes, [$vars['class']]); |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
10 |
|
return $classes; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* Clear the given index if a full re-index is needed |
|
171
|
|
|
* |
|
172
|
|
|
* @param $vars |
|
173
|
|
|
* @param BaseIndex $index |
|
174
|
|
|
* @throws Exception |
|
175
|
|
|
*/ |
|
176
|
11 |
|
public function clearIndex($vars, BaseIndex $index) |
|
177
|
|
|
{ |
|
178
|
11 |
|
if (!empty($vars['clear'])) { |
|
179
|
1 |
|
$this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName())); |
|
180
|
1 |
|
$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index); |
|
181
|
|
|
} |
|
182
|
11 |
|
} |
|
183
|
|
|
|
|
184
|
|
|
/** |
|
185
|
|
|
* @param $classes |
|
186
|
|
|
* @param $isGroup |
|
187
|
|
|
* @param BaseIndex $index |
|
188
|
|
|
* @param $group |
|
189
|
|
|
* @return int |
|
190
|
|
|
* @throws Exception |
|
191
|
|
|
* @throws GuzzleException |
|
192
|
|
|
*/ |
|
193
|
11 |
|
protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int |
|
194
|
|
|
{ |
|
195
|
11 |
|
$groups = 0; |
|
196
|
11 |
|
foreach ($classes as $class) { |
|
197
|
11 |
|
$groups = $this->indexClass($isGroup, $class, $index, $group); |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
11 |
|
return $groups; |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
/** |
|
204
|
|
|
* @param bool $isGroup |
|
205
|
|
|
* @param string $class |
|
206
|
|
|
* @param BaseIndex $index |
|
207
|
|
|
* @param int $group |
|
208
|
|
|
* @return int |
|
209
|
|
|
* @throws GuzzleException |
|
210
|
|
|
* @throws ValidationException |
|
211
|
|
|
*/ |
|
212
|
11 |
|
private function indexClass($isGroup, $class, BaseIndex $index, int $group): int |
|
213
|
|
|
{ |
|
214
|
11 |
|
$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()), []); |
|
215
|
|
|
|
|
216
|
11 |
|
$batchLength = DocumentFactory::config()->get('batchLength'); |
|
217
|
11 |
|
$groups = (int)ceil($class::get()->count() / $batchLength); |
|
218
|
11 |
|
$groups = $isGroup ? $group : $groups; |
|
219
|
11 |
|
while ($group <= $groups) { // Run from oldest to newest |
|
220
|
|
|
try { |
|
221
|
11 |
|
$this->doReindex($group, $class, $batchLength, $index); |
|
222
|
|
|
} catch (Exception $error) { |
|
223
|
|
|
$this->logException($index->getIndexName(), $group, $error); |
|
224
|
|
|
$group++; |
|
225
|
|
|
continue; |
|
226
|
|
|
} |
|
227
|
11 |
|
$group++; |
|
228
|
11 |
|
$this->getLogger()->info(sprintf('Indexed group %s', $group)); |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
11 |
|
return $groups; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* Reindex the given group, for each state |
|
236
|
|
|
* |
|
237
|
|
|
* @param int $group |
|
238
|
|
|
* @param string $class |
|
239
|
|
|
* @param int $batchLength |
|
240
|
|
|
* @param BaseIndex $index |
|
241
|
|
|
* @throws Exception |
|
242
|
|
|
*/ |
|
243
|
11 |
|
private function doReindex($group, $class, $batchLength, BaseIndex $index): void |
|
244
|
|
|
{ |
|
245
|
11 |
|
foreach (SiteState::getStates() as $state) { |
|
246
|
11 |
|
if ($state !== 'default') { |
|
247
|
|
|
SiteState::withState($state); |
|
248
|
|
|
} |
|
249
|
|
|
|
|
250
|
|
|
// Generate filtered list of local records |
|
251
|
11 |
|
$baseClass = DataObject::getSchema()->baseDataClass($class); |
|
252
|
|
|
/** @var DataList|DataObject[] $items */ |
|
253
|
11 |
|
$items = DataObject::get($baseClass) |
|
254
|
11 |
|
->sort('ID ASC') |
|
255
|
11 |
|
->limit($batchLength, ($group * $batchLength)); |
|
256
|
11 |
|
if ($items->count()) { |
|
257
|
11 |
|
$this->updateIndex($index, $items); |
|
258
|
|
|
} |
|
259
|
|
|
} |
|
260
|
|
|
|
|
261
|
|
|
// Reset the variants back to it's original state for the next round |
|
262
|
11 |
|
foreach ($this->currentStates as $variant => $value) { |
|
263
|
10 |
|
singleton($variant)->activateState($value); |
|
264
|
|
|
} |
|
265
|
11 |
|
} |
|
266
|
|
|
|
|
267
|
|
|
/** |
|
268
|
|
|
* Execute the update on the client |
|
269
|
|
|
* |
|
270
|
|
|
* @param BaseIndex $index |
|
271
|
|
|
* @param $items |
|
272
|
|
|
* @throws Exception |
|
273
|
|
|
*/ |
|
274
|
1 |
|
private function updateIndex(BaseIndex $index, $items): void |
|
275
|
|
|
{ |
|
276
|
1 |
|
$client = $index->getClient(); |
|
277
|
1 |
|
$update = $client->createUpdate(); |
|
278
|
1 |
|
$this->service->setInDebugMode($this->debug); |
|
279
|
1 |
|
$this->service->updateIndex($index, $items, $update); |
|
280
|
1 |
|
$update->addCommit(); |
|
281
|
1 |
|
$client->update($update); |
|
282
|
1 |
|
} |
|
283
|
|
|
|
|
284
|
|
|
/** |
|
285
|
|
|
* @param string $index |
|
286
|
|
|
* @param int $group |
|
287
|
|
|
* @param Exception $exception |
|
288
|
|
|
* @throws GuzzleException |
|
289
|
|
|
* @throws ValidationException |
|
290
|
|
|
*/ |
|
291
|
|
|
private function logException($index, int $group, Exception $exception) |
|
292
|
|
|
{ |
|
293
|
|
|
$this->getLogger()->error($exception->getMessage()); |
|
294
|
|
|
$msg = sprintf( |
|
295
|
|
|
'Error indexing core %s on group %s,' . PHP_EOL . |
|
296
|
|
|
'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL, |
|
297
|
|
|
$index, |
|
298
|
|
|
$group |
|
299
|
|
|
); |
|
300
|
|
|
SolrLogger::logMessage('ERROR', $msg, $index); |
|
301
|
|
|
} |
|
302
|
|
|
} |
|
303
|
|
|
|