|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* This file is part of the eZ Publish Kernel package. |
|
5
|
|
|
* |
|
6
|
|
|
* @copyright Copyright (C) eZ Systems AS. All rights reserved. |
|
7
|
|
|
* @license For full copyright and license information view LICENSE file distributed with this source code. |
|
8
|
|
|
*/ |
|
9
|
|
|
namespace eZ\Bundle\EzPublishCoreBundle\Command; |
|
10
|
|
|
|
|
11
|
|
|
use eZ\Publish\SPI\Persistence\Content\ContentInfo; |
|
12
|
|
|
use eZ\Publish\Core\Search\Common\Indexer; |
|
13
|
|
|
use eZ\Publish\Core\Search\Common\IncrementalIndexer; |
|
14
|
|
|
use Doctrine\DBAL\Driver\Statement; |
|
15
|
|
|
use Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand; |
|
16
|
|
|
use Symfony\Component\Console\Helper\ProgressBar; |
|
17
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
|
18
|
|
|
use Symfony\Component\Console\Input\InputOption; |
|
19
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
|
20
|
|
|
use Symfony\Component\Process\PhpExecutableFinder; |
|
21
|
|
|
use Symfony\Component\Process\ProcessBuilder; |
|
22
|
|
|
use RuntimeException; |
|
23
|
|
|
use DateTime; |
|
24
|
|
|
use PDO; |
|
25
|
|
|
|
|
26
|
|
|
class ReindexCommand extends ContainerAwareCommand |
|
27
|
|
|
{ |
|
28
|
|
|
/** |
|
29
|
|
|
* @var \eZ\Publish\Core\Search\Common\Indexer|\eZ\Publish\Core\Search\Common\IncrementalIndexer |
|
30
|
|
|
*/ |
|
31
|
|
|
private $searchIndexer; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @var \Doctrine\DBAL\Connection |
|
35
|
|
|
*/ |
|
36
|
|
|
private $connection; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @var string |
|
40
|
|
|
*/ |
|
41
|
|
|
private $phpPath; |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* Initialize objects required by {@see execute()}. |
|
45
|
|
|
* |
|
46
|
|
|
* @param InputInterface $input |
|
47
|
|
|
* @param OutputInterface $output |
|
48
|
|
|
*/ |
|
49
|
|
|
public function initialize(InputInterface $input, OutputInterface $output) |
|
50
|
|
|
{ |
|
51
|
|
|
parent::initialize($input, $output); |
|
52
|
|
|
$this->searchIndexer = $this->getContainer()->get('ezpublish.spi.search.indexer'); |
|
53
|
|
|
$this->connection = $this->getContainer()->get('ezpublish.api.storage_engine.legacy.connection'); |
|
54
|
|
|
if (!$this->searchIndexer instanceof Indexer) { |
|
55
|
|
|
throw new RuntimeException( |
|
56
|
|
|
sprintf( |
|
57
|
|
|
'Expected to find Search Engine Indexer but found "%s" instead', |
|
58
|
|
|
get_parent_class($this->searchIndexer) |
|
59
|
|
|
) |
|
60
|
|
|
); |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* {@inheritdoc} |
|
66
|
|
|
*/ |
|
67
|
|
|
protected function configure() |
|
68
|
|
|
{ |
|
69
|
|
|
$this |
|
70
|
|
|
->setName('ezplatform:reindex') |
|
71
|
|
|
->setDescription('Recreate or Refresh search engine index') |
|
72
|
|
|
->addOption( |
|
73
|
|
|
'iteration-count', |
|
74
|
|
|
'c', |
|
75
|
|
|
InputOption::VALUE_OPTIONAL, |
|
76
|
|
|
'Number of objects to be indexed in a single iteration, for avoiding using too much memory', |
|
77
|
|
|
50 |
|
78
|
|
|
)->addOption( |
|
79
|
|
|
'no-commit', |
|
80
|
|
|
null, |
|
81
|
|
|
InputOption::VALUE_NONE, |
|
82
|
|
|
'Do not commit after each iteration' |
|
83
|
|
|
)->addOption( |
|
84
|
|
|
'no-purge', |
|
85
|
|
|
null, |
|
86
|
|
|
InputOption::VALUE_NONE, |
|
87
|
|
|
'Do not purge before indexing' |
|
88
|
|
|
)->addOption( |
|
89
|
|
|
'since', |
|
90
|
|
|
null, |
|
91
|
|
|
InputOption::VALUE_OPTIONAL, |
|
92
|
|
|
'Refresh changes since a given time, any format understood by DateTime. Implies "no-purge", can not be combined with "content-ids" or "subtree"' |
|
93
|
|
|
)->addOption( |
|
94
|
|
|
'content-ids', |
|
95
|
|
|
null, |
|
96
|
|
|
InputOption::VALUE_OPTIONAL, |
|
97
|
|
|
'Comma separated list of content id\'s to refresh (deleted/updated/added). Implies "no-purge", can not be combined with "since" or "subtree"' |
|
98
|
|
|
)->addOption( |
|
99
|
|
|
'subtree', |
|
100
|
|
|
null, |
|
101
|
|
|
InputOption::VALUE_OPTIONAL, |
|
102
|
|
|
'Location Id to index subtree of (incl self). Implies "no-purge", can not be combined with "since" or "content-ids"' |
|
103
|
|
|
)->addOption( |
|
104
|
|
|
'processes', |
|
105
|
|
|
null, |
|
106
|
|
|
InputOption::VALUE_OPTIONAL, |
|
107
|
|
|
'Number of sub processes to spawn in parallel handling iterations, default number is number of CPU cores -1, set to 1 or 0 to disable', |
|
108
|
|
|
$this->getNumberOfCPUCores() |
|
109
|
|
|
)->setHelp( |
|
110
|
|
|
<<<EOT |
|
111
|
|
|
The command <info>%command.name%</info> indexes current configured database in configured search engine index. |
|
112
|
|
|
|
|
113
|
|
|
|
|
114
|
|
|
Example usage: |
|
115
|
|
|
- Refresh (add/update) index changes since yesterday: |
|
116
|
|
|
<comment>ezplatform:reindex --since=yesterday</comment> |
|
117
|
|
|
See: http://php.net/manual/en/datetime.formats.php |
|
118
|
|
|
|
|
119
|
|
|
- Refresh (add/update/remove) index on a set of content id's: |
|
120
|
|
|
<comment>ezplatform:reindex --content-ids=2,34,68</comment> |
|
121
|
|
|
|
|
122
|
|
|
- Refresh (add/update) index of a subtree: |
|
123
|
|
|
<comment>ezplatform:reindex --subtree=45</comment> |
|
124
|
|
|
|
|
125
|
|
|
- Refresh (add/update) the whole index using 3 processes, & let search engine handle commits itself using auto commit: |
|
126
|
|
|
<comment>ezplatform:reindex --no-purge --no-commit --processes=3</comment> |
|
127
|
|
|
|
|
128
|
|
|
EOT |
|
129
|
|
|
); |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
/** |
|
133
|
|
|
* {@inheritdoc} |
|
134
|
|
|
*/ |
|
135
|
|
|
protected function execute(InputInterface $input, OutputInterface $output) |
|
136
|
|
|
{ |
|
137
|
|
|
$commit = !$input->getOption('no-commit'); |
|
138
|
|
|
$iterationCount = $input->getOption('iteration-count'); |
|
139
|
|
|
if (!is_numeric($iterationCount) || (int) $iterationCount < 1) { |
|
140
|
|
|
throw new RuntimeException("'--iteration-count' option should be > 0, got '{$iterationCount}'"); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
if (!$this->searchIndexer instanceof IncrementalIndexer) { |
|
144
|
|
|
$output->writeln(<<<EOT |
|
145
|
|
|
DEPRECATED: |
|
146
|
|
|
Running indexing against an Indexer that has not been updated to use IncrementalIndexer abstract. |
|
147
|
|
|
|
|
148
|
|
|
Options that won't be taken into account: |
|
149
|
|
|
- since |
|
150
|
|
|
- content-ids |
|
151
|
|
|
- processes |
|
152
|
|
|
- no-purge |
|
153
|
|
|
EOT |
|
154
|
|
|
); |
|
155
|
|
|
$this->searchIndexer->createSearchIndex($output, (int) $iterationCount, !$commit); |
|
156
|
|
|
} else { |
|
157
|
|
|
$output->writeln('Re-indexing started for search engine: ' . $this->searchIndexer->getName()); |
|
158
|
|
|
$output->writeln(''); |
|
159
|
|
|
|
|
160
|
|
|
$return = $this->indexIncrementally($input, $output, $iterationCount, $commit); |
|
161
|
|
|
|
|
162
|
|
|
$output->writeln(''); |
|
163
|
|
|
$output->writeln('Finished re-indexing'); |
|
164
|
|
|
|
|
165
|
|
|
return $return; |
|
166
|
|
|
} |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
protected function indexIncrementally(InputInterface $input, OutputInterface $output, $iterationCount, $commit) |
|
170
|
|
|
{ |
|
171
|
|
|
if ($contentIds = $input->getOption('content-ids')) { |
|
172
|
|
|
$output->writeln('Indexing list of content id\'s'); |
|
173
|
|
|
|
|
174
|
|
|
return $this->searchIndexer->updateSearchIndex(explode(',', $contentIds), $commit); |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
if ($since = $input->getOption('since')) { |
|
178
|
|
|
$stmt = $this->getStatementContentSince(new DateTime($since)); |
|
179
|
|
|
$count = (int)$this->getStatementContentSince(new DateTime($since), true)->fetchColumn(); |
|
180
|
|
|
$purge = false; |
|
181
|
|
|
} elseif ($locationId = (int) $input->getOption('subtree')) { |
|
182
|
|
|
$stmt = $this->getStatementSubtree($locationId); |
|
183
|
|
|
$count = (int) $this->getStatementSubtree($locationId, true)->fetchColumn(); |
|
184
|
|
|
$purge = false; |
|
185
|
|
|
} else { |
|
186
|
|
|
$stmt = $this->getStatementContentAll(); |
|
187
|
|
|
$count = (int) $this->getStatementContentAll(true)->fetchColumn(); |
|
188
|
|
|
$purge = !$input->getOption('no-purge'); |
|
189
|
|
|
} |
|
190
|
|
|
|
|
191
|
|
|
if (!$count) { |
|
192
|
|
|
$output->writeln('<error>Could not find any items to index, aborting.</error>'); |
|
193
|
|
|
|
|
194
|
|
|
return 1; |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
|
|
$iterations = ceil($count / $iterationCount); |
|
198
|
|
|
$processCount = (int) $input->getOption('processes'); |
|
199
|
|
|
$processCount = $processCount > $iterations ? $iterations : $processCount; |
|
200
|
|
|
$processMessage = $processCount > 1 ? "using $processCount parallel processes" : 'using single process'; |
|
201
|
|
|
|
|
202
|
|
|
if ($purge) { |
|
203
|
|
|
$output->writeln('Purging index...'); |
|
204
|
|
|
$this->searchIndexer->purge(); |
|
205
|
|
|
|
|
206
|
|
|
$output->writeln( |
|
207
|
|
|
"<info>Re-Creating index for {$count} items across $iterations iteration(s), $processMessage:" |
|
208
|
|
|
); |
|
209
|
|
|
} else { |
|
210
|
|
|
$output->writeln( |
|
211
|
|
|
"<info>Refreshing index for {$count} items across $iterations iteration(s), $processMessage:" |
|
212
|
|
|
); |
|
213
|
|
|
} |
|
214
|
|
|
|
|
215
|
|
|
$progress = new ProgressBar($output); |
|
216
|
|
|
$progress->start($iterations); |
|
217
|
|
|
|
|
218
|
|
|
if ($processCount > 1) { |
|
219
|
|
|
$this->runParallelProcess($progress, $stmt, (int) $processCount, (int) $iterationCount, $commit); |
|
220
|
|
|
} else { |
|
221
|
|
|
// if we only have one process, or less iterations to warrant running several, we index it all inline |
|
222
|
|
|
foreach ($this->fetchIteration($stmt, $iterationCount) as $contentIds) { |
|
223
|
|
|
$this->searchIndexer->updateSearchIndex($contentIds, $commit); |
|
224
|
|
|
$progress->advance(1); |
|
225
|
|
|
} |
|
226
|
|
|
} |
|
227
|
|
|
|
|
228
|
|
|
$progress->finish(); |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
|
|
private function runParallelProcess(ProgressBar $progress, Statement $stmt, $processCount, $iterationCount, $commit) |
|
232
|
|
|
{ |
|
233
|
|
|
/** |
|
234
|
|
|
* @var $processes \Symfony\Component\Process\Process[]|null[] |
|
235
|
|
|
*/ |
|
236
|
|
|
$processes = array_fill(0, $processCount, null); |
|
237
|
|
|
$generator = $this->fetchIteration($stmt, $iterationCount); |
|
238
|
|
|
do { |
|
239
|
|
|
foreach ($processes as $key => $process) { |
|
240
|
|
|
if ($process !== null && $process->isRunning()) { |
|
241
|
|
|
continue; |
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
if ($process !== null) { |
|
245
|
|
|
// One of the processes just finished, so we increment progress bar |
|
246
|
|
|
$progress->advance(1); |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
if (!$generator->valid()) { |
|
250
|
|
|
unset($processes[$key]); |
|
251
|
|
|
continue; |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
$processes[$key] = $this->getPhpProcess($generator->current(), $commit); |
|
255
|
|
|
$processes[$key]->start(); |
|
256
|
|
|
$generator->next(); |
|
257
|
|
|
} |
|
258
|
|
|
|
|
259
|
|
|
if (!empty($processes)) { |
|
260
|
|
|
sleep(1); |
|
261
|
|
|
} |
|
262
|
|
|
} while (!empty($processes)); |
|
263
|
|
|
} |
|
264
|
|
|
|
|
265
|
|
|
/** |
|
266
|
|
|
* @param DateTime $since |
|
267
|
|
|
* @param bool $count |
|
268
|
|
|
* |
|
269
|
|
|
* @return \Doctrine\DBAL\Driver\Statement |
|
270
|
|
|
*/ |
|
271
|
|
|
private function getStatementContentSince(DateTime $since, $count = false) |
|
272
|
|
|
{ |
|
273
|
|
|
$q = $this->connection->createQueryBuilder() |
|
274
|
|
|
->select($count ? 'count(c.id)' : 'c.id') |
|
275
|
|
|
->from('ezcontentobject', 'c') |
|
276
|
|
|
->where('c.status = :status')->andWhere('c.modified >= :since') |
|
277
|
|
|
->orderBy('c.modified') |
|
278
|
|
|
->setParameter('status', ContentInfo::STATUS_PUBLISHED, PDO::PARAM_INT) |
|
279
|
|
|
->setParameter('since', $since->getTimestamp(), PDO::PARAM_INT); |
|
280
|
|
|
|
|
281
|
|
|
return $q->execute(); |
|
282
|
|
|
} |
|
283
|
|
|
|
|
284
|
|
|
/** |
|
285
|
|
|
* @param mixed $locationId |
|
286
|
|
|
* @param bool $count |
|
287
|
|
|
* |
|
288
|
|
|
* @return \Doctrine\DBAL\Driver\Statement |
|
289
|
|
|
*/ |
|
290
|
|
|
private function getStatementSubtree($locationId, $count = false) |
|
291
|
|
|
{ |
|
292
|
|
|
/** |
|
293
|
|
|
* @var $locationHandler \eZ\Publish\SPI\Persistence\Content\Location\Handler |
|
294
|
|
|
*/ |
|
295
|
|
|
$locationHandler = $this->getContainer()->get('ezpublish.spi.persistence.location_handler'); |
|
296
|
|
|
$location = $locationHandler->load($locationId); |
|
297
|
|
|
$q = $this->connection->createQueryBuilder() |
|
298
|
|
|
->select($count ? 'count(DISTINCT c.id)' : 'DISTINCT c.id') |
|
299
|
|
|
->from('ezcontentobject', 'c') |
|
300
|
|
|
->innerJoin('c', 'ezcontentobject_tree', 't', 't.contentobject_id = c.id') |
|
301
|
|
|
->where('c.status = :status') |
|
302
|
|
|
->andWhere('t.path_string LIKE :path') |
|
303
|
|
|
->orderBy('t.path_string') |
|
304
|
|
|
->setParameter('status', ContentInfo::STATUS_PUBLISHED, PDO::PARAM_INT) |
|
305
|
|
|
->setParameter('path', $location->pathString . '%', PDO::PARAM_STR); |
|
306
|
|
|
|
|
307
|
|
|
return $q->execute(); |
|
|
|
|
|
|
308
|
|
|
} |
|
309
|
|
|
|
|
310
|
|
|
/** |
|
311
|
|
|
* @param bool $count |
|
312
|
|
|
* |
|
313
|
|
|
* @return \Doctrine\DBAL\Driver\Statement |
|
314
|
|
|
*/ |
|
315
|
|
|
private function getStatementContentAll($count = false) |
|
316
|
|
|
{ |
|
317
|
|
|
$q = $this->connection->createQueryBuilder() |
|
318
|
|
|
->select($count ? 'count(c.id)' : 'c.id') |
|
319
|
|
|
->from('ezcontentobject', 'c') |
|
320
|
|
|
->where('c.status = :status') |
|
321
|
|
|
->setParameter('status', ContentInfo::STATUS_PUBLISHED, PDO::PARAM_INT); |
|
322
|
|
|
|
|
323
|
|
|
return $q->execute(); |
|
324
|
|
|
} |
|
325
|
|
|
|
|
326
|
|
|
/** |
|
327
|
|
|
* @param \Doctrine\DBAL\Driver\Statement $stmt |
|
328
|
|
|
* @param int $iterationCount |
|
329
|
|
|
* |
|
330
|
|
|
* @return \Generator Return an array of arrays, each array contains content id's of $iterationCount. |
|
331
|
|
|
*/ |
|
332
|
|
|
private function fetchIteration(Statement $stmt, $iterationCount) |
|
333
|
|
|
{ |
|
334
|
|
|
do { |
|
335
|
|
|
$contentIds = []; |
|
336
|
|
View Code Duplication |
for ($i = 0; $i < $iterationCount; ++$i) { |
|
|
|
|
|
|
337
|
|
|
if ($contentId = $stmt->fetch(PDO::FETCH_COLUMN)) { |
|
338
|
|
|
$contentIds[] = $contentId; |
|
339
|
|
|
} else { |
|
340
|
|
|
break; |
|
341
|
|
|
} |
|
342
|
|
|
} |
|
343
|
|
|
|
|
344
|
|
|
yield $contentIds; |
|
345
|
|
|
} while (!empty($contentId)); |
|
346
|
|
|
} |
|
347
|
|
|
|
|
348
|
|
|
/** |
|
349
|
|
|
* @param array $contentIds |
|
350
|
|
|
* @param bool $commit |
|
351
|
|
|
* |
|
352
|
|
|
* @return \Symfony\Component\Process\Process |
|
353
|
|
|
*/ |
|
354
|
|
|
private function getPhpProcess(array $contentIds, $commit) |
|
355
|
|
|
{ |
|
356
|
|
|
$process = new ProcessBuilder([ |
|
357
|
|
|
file_exists('bin/console') ? 'bin/console' : 'app/console', |
|
358
|
|
|
'ezplatform:reindex', |
|
359
|
|
|
'--content-ids=' . implode(',', $contentIds), |
|
360
|
|
|
]); |
|
361
|
|
|
$process->setTimeout(null); |
|
362
|
|
|
$process->setPrefix($this->getPhpPath()); |
|
363
|
|
|
|
|
364
|
|
|
if (!$commit) { |
|
365
|
|
|
$process->add('--no-commit'); |
|
366
|
|
|
} |
|
367
|
|
|
|
|
368
|
|
|
return $process->getProcess(); |
|
369
|
|
|
} |
|
370
|
|
|
|
|
371
|
|
|
/** |
|
372
|
|
|
* @return string |
|
373
|
|
|
*/ |
|
374
|
|
|
private function getPhpPath() |
|
375
|
|
|
{ |
|
376
|
|
|
if ($this->phpPath) { |
|
377
|
|
|
return $this->phpPath; |
|
378
|
|
|
} |
|
379
|
|
|
|
|
380
|
|
|
$phpFinder = new PhpExecutableFinder(); |
|
381
|
|
|
$this->phpPath = $phpFinder->find(); |
|
|
|
|
|
|
382
|
|
|
if (!$this->phpPath) { |
|
|
|
|
|
|
383
|
|
|
throw new \RuntimeException( |
|
384
|
|
|
'The php executable could not be found, it\'s needed for executing parable sub processes, so add it to your PATH environment variable and try again' |
|
385
|
|
|
); |
|
386
|
|
|
} |
|
387
|
|
|
} |
|
388
|
|
|
|
|
389
|
|
|
/** |
|
390
|
|
|
* @return int |
|
391
|
|
|
*/ |
|
392
|
|
|
private function getNumberOfCPUCores() |
|
393
|
|
|
{ |
|
394
|
|
|
$cores = 1; |
|
395
|
|
|
if (is_file('/proc/cpuinfo')) { |
|
396
|
|
|
// Linux (and potentially Windows with linux sub systems) |
|
397
|
|
|
$cpuinfo = file_get_contents('/proc/cpuinfo'); |
|
398
|
|
|
preg_match_all('/^processor/m', $cpuinfo, $matches); |
|
399
|
|
|
$cores = count($matches[0]); |
|
400
|
|
|
} elseif (DIRECTORY_SEPARATOR === '\\') { |
|
401
|
|
|
// Windows |
|
402
|
|
|
if (($process = @popen('wmic cpu get NumberOfCores', 'rb')) !== false) { |
|
403
|
|
|
fgets($process); |
|
404
|
|
|
$cores = (int) fgets($process); |
|
405
|
|
|
pclose($process); |
|
406
|
|
|
} |
|
407
|
|
|
} elseif (($process = @popen('sysctl -a', 'rb')) !== false) { |
|
408
|
|
|
// *nix (Linux, BSD and Mac) |
|
409
|
|
|
$output = stream_get_contents($process); |
|
410
|
|
|
if (preg_match('/hw.ncpu: (\d+)/', $output, $matches)) { |
|
411
|
|
|
$cores = (int) $matches[1][0]; |
|
412
|
|
|
} |
|
413
|
|
|
pclose($process); |
|
414
|
|
|
} |
|
415
|
|
|
|
|
416
|
|
|
return $cores; |
|
417
|
|
|
} |
|
418
|
|
|
} |
|
419
|
|
|
|