1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Queryr\Replicator\Cli\Command; |
4
|
|
|
|
5
|
|
|
use Queryr\Replicator\Cli\Import\PagesImporterCli; |
6
|
|
|
use Symfony\Component\Console\Input\InputArgument; |
7
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
8
|
|
|
use Symfony\Component\Console\Input\InputOption; |
9
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
10
|
|
|
use Wikibase\JsonDumpReader\JsonDumpFactory; |
11
|
|
|
use Wikibase\JsonDumpReader\SeekableDumpReader; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* @licence GNU GPL v2+ |
15
|
|
|
* @author Jeroen De Dauw < [email protected] > |
16
|
|
|
*/ |
17
|
|
|
class GzJsonImportCommand extends ImportCommandBase { |
18
|
|
|
|
19
|
|
|
protected function configure() { |
20
|
|
|
$this->setName( 'import:gz' ); |
21
|
|
|
$this->setDescription( 'Imports entities from a gzip compressed JSON dump' ); |
22
|
|
|
|
23
|
|
|
$this->addArgument( |
24
|
|
|
'file', |
25
|
|
|
InputArgument::REQUIRED, |
26
|
|
|
'Full path of the gz JSON dump file' |
27
|
|
|
); |
28
|
|
|
|
29
|
|
|
$this->addOption( |
30
|
|
|
'continue', |
31
|
|
|
'c', |
32
|
|
|
InputOption::VALUE_OPTIONAL, |
33
|
|
|
'The position to resume import from' |
34
|
|
|
); |
35
|
|
|
|
36
|
|
|
$this->addOption( |
37
|
|
|
'max', |
38
|
|
|
'm', |
39
|
|
|
InputOption::VALUE_OPTIONAL, |
40
|
|
|
'The maximum number of entities to import' |
41
|
|
|
); |
42
|
|
|
} |
43
|
|
|
|
44
|
|
|
protected function executeCommand( InputInterface $input, OutputInterface $output ) { |
45
|
|
|
$dumpReader = ( new JsonDumpFactory() )->newGzDumpReader( |
46
|
|
|
$input->getArgument( 'file' ), |
47
|
|
|
is_numeric( $input->getOption( 'continue' ) ) ? (int)$input->getOption( 'continue' ) : 0 |
48
|
|
|
); |
49
|
|
|
|
50
|
|
|
$importer = new PagesImporterCli( |
51
|
|
|
$input, |
52
|
|
|
$output, |
53
|
|
|
$this->factory, |
|
|
|
|
54
|
|
|
function() use ( $output, $dumpReader ) { |
55
|
|
|
$output->writeln( "\n" ); |
56
|
|
|
$output->writeln( "<info>Import process aborted</info>" ); |
57
|
|
|
$output->writeln( "<comment>To resume, run with</comment> --continue=" . $dumpReader->getPosition() ); |
58
|
|
|
} |
59
|
|
|
); |
60
|
|
|
|
61
|
|
|
$iterator = $this->factory->newJsonEntityPageIterator( $dumpReader ); |
62
|
|
|
|
63
|
|
|
if ( is_numeric( $input->getOption( 'max' ) ) ) { |
64
|
|
|
$iterator = new \LimitIterator( $iterator, 0, (int)$input->getOption( 'max' ) ); |
65
|
|
|
} |
66
|
|
|
|
67
|
|
|
$importer->runImport( $iterator ); |
68
|
|
|
|
69
|
|
|
$this->outputMaxContinuation( $input, $output, $dumpReader ); |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
private function outputMaxContinuation( InputInterface $input, OutputInterface $output, SeekableDumpReader $reader ) { |
73
|
|
|
if ( is_numeric( $input->getOption( 'max' ) ) ) { |
74
|
|
|
$output->writeln( |
75
|
|
|
"\n<comment>To continue from current position, run with</comment> --continue=" . $reader->getPosition() |
76
|
|
|
); |
77
|
|
|
} |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
} |
81
|
|
|
|
Unless you are absolutely sure that the expression can never be null because of other conditions, we strongly recommend to add an additional type check to your code: