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
|
|
|
* @var string |
37
|
|
|
*/ |
38
|
|
|
private static $segment = 'SolrIndexTask'; |
39
|
|
|
/** |
40
|
|
|
* @var array Store the current states for all instances of SiteState |
41
|
|
|
*/ |
42
|
|
|
public $currentStates; |
43
|
|
|
/** |
44
|
|
|
* @var string |
45
|
|
|
*/ |
46
|
|
|
protected $title = 'Solr Index update'; |
47
|
|
|
/** |
48
|
|
|
* @var string |
49
|
|
|
*/ |
50
|
|
|
protected $description = 'Add or update documents to an existing Solr core.'; |
51
|
|
|
/** |
52
|
|
|
* @var bool |
53
|
|
|
*/ |
54
|
|
|
protected $debug = false; |
55
|
|
|
/** |
56
|
|
|
* @var SolrCoreService |
57
|
|
|
*/ |
58
|
|
|
protected $service; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* SolrIndexTask constructor. Sets up the document factory |
62
|
|
|
* |
63
|
|
|
* @throws ReflectionException |
64
|
|
|
*/ |
65
|
12 |
|
public function __construct() |
66
|
|
|
{ |
67
|
12 |
|
parent::__construct(); |
68
|
|
|
// Only index live items. |
69
|
|
|
// The old FTS module also indexed Draft items. This is unnecessary |
70
|
12 |
|
Versioned::set_reading_mode(Versioned::DEFAULT_MODE); |
71
|
12 |
|
$this->setService(Injector::inst()->get(SolrCoreService::class)); |
72
|
12 |
|
$this->setLogger(Injector::inst()->get(LoggerInterface::class)); |
73
|
12 |
|
$this->setDebug(Director::isDev() || Director::is_cli()); |
74
|
12 |
|
$this->currentStates = SiteState::currentStates(); |
75
|
12 |
|
} |
76
|
|
|
|
77
|
|
|
/** |
78
|
|
|
* @param SolrCoreService $service |
79
|
|
|
* @return SolrIndexTask |
80
|
|
|
*/ |
81
|
12 |
|
public function setService(SolrCoreService $service): SolrIndexTask |
82
|
|
|
{ |
83
|
12 |
|
$this->service = $service; |
84
|
|
|
|
85
|
12 |
|
return $this; |
86
|
|
|
} |
87
|
|
|
|
88
|
|
|
/** |
89
|
|
|
* @param bool $debug |
90
|
|
|
* @return SolrIndexTask |
91
|
|
|
*/ |
92
|
12 |
|
public function setDebug(bool $debug): SolrIndexTask |
93
|
|
|
{ |
94
|
12 |
|
$this->debug = $debug; |
95
|
|
|
|
96
|
12 |
|
return $this; |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* Implement this method in the task subclass to |
101
|
|
|
* execute via the TaskRunner |
102
|
|
|
* |
103
|
|
|
* @param HTTPRequest $request |
104
|
|
|
* @return int|bool |
105
|
|
|
* @throws Exception |
106
|
|
|
* @throws GuzzleException |
107
|
|
|
* @todo defer to background because it may run out of memory |
108
|
|
|
*/ |
109
|
11 |
|
public function run($request) |
110
|
|
|
{ |
111
|
11 |
|
$startTime = time(); |
112
|
11 |
|
[$vars, $group, $isGroup] = $this->taskSetup($request); |
113
|
11 |
|
$groups = 0; |
114
|
11 |
|
$indexes = $this->service->getValidIndexes($request->getVar('index')); |
115
|
|
|
|
116
|
11 |
|
foreach ($indexes as $indexName) { |
117
|
|
|
/** @var BaseIndex $index */ |
118
|
11 |
|
$index = Injector::inst()->get($indexName, false); |
119
|
|
|
|
120
|
11 |
|
$indexClasses = $index->getClasses(); |
121
|
11 |
|
$classes = $this->getClasses($vars, $indexClasses); |
122
|
11 |
|
if (!count($classes)) { |
123
|
8 |
|
continue; |
124
|
|
|
} |
125
|
|
|
|
126
|
11 |
|
$this->clearIndex($vars, $index); |
127
|
|
|
|
128
|
11 |
|
$groups = $this->indexClassForIndex($classes, $isGroup, $index, $group); |
129
|
|
|
} |
130
|
11 |
|
$this->getLogger()->info( |
131
|
11 |
|
sprintf('It took me %d seconds to do all the indexing%s', (time() - $startTime), PHP_EOL) |
132
|
|
|
); |
133
|
|
|
// Grab the latest logs from indexing if needed |
134
|
11 |
|
$solrLogger = new SolrLogger(); |
135
|
11 |
|
$solrLogger->saveSolrLog('Config'); |
136
|
|
|
|
137
|
11 |
|
return $groups; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* @param HTTPRequest $request |
142
|
|
|
* @return array |
143
|
|
|
*/ |
144
|
11 |
|
protected function taskSetup($request): array |
145
|
|
|
{ |
146
|
11 |
|
$vars = $request->getVars(); |
147
|
11 |
|
$this->debug = $this->debug || isset($vars['debug']); |
148
|
11 |
|
$group = $vars['group'] ?? 0; |
149
|
11 |
|
$start = $vars['start'] ?? 0; |
150
|
11 |
|
$group = ($start > $group) ? $start : $group; |
151
|
11 |
|
$isGroup = isset($vars['group']); |
152
|
|
|
|
153
|
11 |
|
return [$vars, $group, $isGroup]; |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
/** |
157
|
|
|
* @param $vars |
158
|
|
|
* @param array $classes |
159
|
|
|
* @return bool|array |
160
|
|
|
*/ |
161
|
11 |
|
protected function getClasses($vars, array $classes): array |
162
|
|
|
{ |
163
|
11 |
|
if (isset($vars['class'])) { |
164
|
1 |
|
return array_intersect($classes, [$vars['class']]); |
165
|
|
|
} |
166
|
|
|
|
167
|
10 |
|
return $classes; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* Clear the given index if a full re-index is needed |
172
|
|
|
* |
173
|
|
|
* @param $vars |
174
|
|
|
* @param BaseIndex $index |
175
|
|
|
* @throws Exception |
176
|
|
|
*/ |
177
|
11 |
|
public function clearIndex($vars, BaseIndex $index) |
178
|
|
|
{ |
179
|
11 |
|
if (!empty($vars['clear'])) { |
180
|
1 |
|
$this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName())); |
181
|
1 |
|
$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index); |
182
|
|
|
} |
183
|
11 |
|
} |
184
|
|
|
|
185
|
|
|
/** |
186
|
|
|
* @param $classes |
187
|
|
|
* @param $isGroup |
188
|
|
|
* @param BaseIndex $index |
189
|
|
|
* @param $group |
190
|
|
|
* @return int |
191
|
|
|
* @throws Exception |
192
|
|
|
* @throws GuzzleException |
193
|
|
|
*/ |
194
|
11 |
|
protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int |
195
|
|
|
{ |
196
|
11 |
|
$groups = 0; |
197
|
11 |
|
foreach ($classes as $class) { |
198
|
11 |
|
$groups = $this->indexClass($isGroup, $class, $index, $group); |
199
|
|
|
} |
200
|
|
|
|
201
|
11 |
|
return $groups; |
202
|
|
|
} |
203
|
|
|
|
204
|
|
|
/** |
205
|
|
|
* @param bool $isGroup |
206
|
|
|
* @param string $class |
207
|
|
|
* @param BaseIndex $index |
208
|
|
|
* @param int $group |
209
|
|
|
* @return int |
210
|
|
|
* @throws GuzzleException |
211
|
|
|
* @throws ValidationException |
212
|
|
|
*/ |
213
|
11 |
|
private function indexClass($isGroup, $class, BaseIndex $index, int $group): int |
214
|
|
|
{ |
215
|
11 |
|
$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()), []); |
216
|
|
|
|
217
|
11 |
|
$batchLength = DocumentFactory::config()->get('batchLength'); |
218
|
11 |
|
$groups = (int)ceil($class::get()->count() / $batchLength); |
219
|
11 |
|
$groups = $isGroup ? $group : $groups; |
220
|
11 |
|
while ($group <= $groups) { // Run from oldest to newest |
221
|
|
|
try { |
222
|
11 |
|
$this->doReindex($group, $class, $batchLength, $index); |
223
|
|
|
} catch (Exception $error) { |
224
|
|
|
$this->logException($index->getIndexName(), $group, $error); |
225
|
|
|
$group++; |
226
|
|
|
continue; |
227
|
|
|
} |
228
|
11 |
|
$group++; |
229
|
11 |
|
$this->getLogger()->info(sprintf('Indexed group %s', $group)); |
230
|
|
|
} |
231
|
|
|
|
232
|
11 |
|
return $groups; |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
/** |
236
|
|
|
* Reindex the given group, for each state |
237
|
|
|
* |
238
|
|
|
* @param int $group |
239
|
|
|
* @param string $class |
240
|
|
|
* @param int $batchLength |
241
|
|
|
* @param BaseIndex $index |
242
|
|
|
* @throws Exception |
243
|
|
|
*/ |
244
|
11 |
|
private function doReindex($group, $class, $batchLength, BaseIndex $index): void |
245
|
|
|
{ |
246
|
11 |
|
foreach (SiteState::getStates() as $state) { |
247
|
11 |
|
if ($state !== 'default') { |
248
|
|
|
SiteState::withState($state); |
249
|
|
|
} |
250
|
11 |
|
$this->stateReindex($group, $class, $batchLength, $index); |
251
|
|
|
} |
252
|
|
|
|
253
|
|
|
// Reset the variants back to it's original state for the next round |
254
|
11 |
|
foreach ($this->currentStates as $variant => $value) { |
255
|
10 |
|
singleton($variant)->activateState($value); |
256
|
|
|
} |
257
|
11 |
|
} |
258
|
|
|
|
259
|
|
|
/** |
260
|
|
|
* @param $group |
261
|
|
|
* @param $class |
262
|
|
|
* @param $batchLength |
263
|
|
|
* @param BaseIndex $index |
264
|
|
|
* @throws Exception |
265
|
|
|
*/ |
266
|
11 |
|
private function stateReindex($group, $class, $batchLength, BaseIndex $index): void |
267
|
|
|
{ |
268
|
|
|
// Generate filtered list of local records |
269
|
11 |
|
$baseClass = DataObject::getSchema()->baseDataClass($class); |
270
|
|
|
/** @var DataList|DataObject[] $items */ |
271
|
11 |
|
$items = DataObject::get($baseClass) |
272
|
11 |
|
->sort('ID ASC') |
273
|
11 |
|
->limit($batchLength, ($group * $batchLength)); |
274
|
11 |
|
if ($items->count()) { |
275
|
1 |
|
$this->updateIndex($index, $items); |
276
|
|
|
} |
277
|
11 |
|
} |
278
|
|
|
|
279
|
|
|
/** |
280
|
|
|
* Execute the update on the client |
281
|
|
|
* |
282
|
|
|
* @param BaseIndex $index |
283
|
|
|
* @param $items |
284
|
|
|
* @throws Exception |
285
|
|
|
*/ |
286
|
1 |
|
private function updateIndex(BaseIndex $index, $items): void |
287
|
|
|
{ |
288
|
1 |
|
$client = $index->getClient(); |
289
|
1 |
|
$update = $client->createUpdate(); |
290
|
1 |
|
$this->service->setInDebugMode($this->debug); |
291
|
1 |
|
$this->service->updateIndex($index, $items, $update); |
292
|
1 |
|
$update->addCommit(); |
293
|
1 |
|
$client->update($update); |
294
|
1 |
|
} |
295
|
|
|
|
296
|
|
|
/** |
297
|
|
|
* @param string $index |
298
|
|
|
* @param int $group |
299
|
|
|
* @param Exception $exception |
300
|
|
|
* @throws GuzzleException |
301
|
|
|
* @throws ValidationException |
302
|
|
|
*/ |
303
|
|
|
private function logException($index, int $group, Exception $exception): void |
304
|
|
|
{ |
305
|
|
|
$this->getLogger()->error($exception->getMessage()); |
306
|
|
|
$msg = sprintf( |
307
|
|
|
'Error indexing core %s on group %s,' . PHP_EOL . |
308
|
|
|
'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL, |
309
|
|
|
$index, |
310
|
|
|
$group |
311
|
|
|
); |
312
|
|
|
SolrLogger::logMessage('ERROR', $msg, $index); |
313
|
|
|
} |
314
|
|
|
} |
315
|
|
|
|