Current File : /www/varak.net/wiki.varak.net/extensions/CirrusSearch/includes/Query/RegexInSourceFeature.php
<?php
namespace CirrusSearch\Query;
use CirrusSearch\Extra\Query\SourceRegex;
use CirrusSearch\SearchConfig;
use CirrusSearch\Search\SearchContext;
use Elastica\Query\AbstractQuery;
/**
* Implements an insource: keyword supporting regular expression matching
* against wikitext source. Works best when combined with the wikimedia-extra
* plugin for elasticsearch, but can also fallback to a groovy based
* implementation. Can be really expensive, but mostly ok if you have the extra
* plugin enabled.
*
* Examples:
* insource:/abc?/
*/
class RegexInSourceFeature implements KeywordFeature {
/**
* @var bool Is this feature enabled?
*/
private $enabled;
/**
* @var string Locale used for case conversions. It's important that this
* matches the locale used for lowercasing in the ngram index.
*/
private $languageCode;
/**
* @var string[] Configuration flags for the regex plugin
*/
private $regexPlugin;
/**
* @var int The maximum number of automaton states that Lucene's regex
* compilation can expand to (even temporarily). Provides protection
* against overloading the search cluster. Only works when using the
* extra plugin, groovy based execution is unbounded.
*/
private $maxDeterminizedStates;
/**
* @param SearchConfig $config
*/
public function __construct( SearchConfig $config ) {
$this->enabled = $config->get( 'CirrusSearchEnableRegex' );
$this->languageCode = $config->get( 'LanguageCode' );
$this->regexPlugin = $config->getElement( 'CirrusSearchWikimediaExtraPlugin', 'regex' );
$this->maxDeterminizedStates = $config->get( 'CirrusSearchRegexMaxDeterminizedStates' );
}
/**
* @param SearchContext $context
* @param string $term
* @return string
*/
public function apply( SearchContext $context, $term ) {
return QueryHelper::extractSpecialSyntaxFromTerm(
$context,
$term,
'/(?<not>-)?insource:\/(?<pattern>(?:[^\\\\\/]|\\\\.)+)\/(?<insensitive>i)? ?/',
function ( $matches ) use ( $context ) {
if ( !$this->enabled ) {
return '';
}
$context->addSyntaxUsed( 'regex' );
$context->setSearchType( 'regex' );
$insensitive = !empty( $matches['insensitive'] );
$filter = $this->regexPlugin && in_array( 'use', $this->regexPlugin )
? $this->buildRegexWithPlugin( $matches['pattern'], $insensitive )
: $this->buildRegexWithGroovy( $matches['pattern'], $insensitive );
if ( empty( $matches['not'] ) ) {
$context->addFilter( $filter );
$context->addHighlightSource( [
'pattern' => $matches['pattern'],
'locale' => $this->languageCode,
'insensitive' => $insensitive,
] );
} else {
$context->addNotFilter( $filter );
}
}
);
}
/**
* Builds a regular expression query using the wikimedia-extra plugin.
*
* @param string $pattern The regular expression to match
* @param bool $insensitive Should the match be case insensitive?
* @return AbstractQuery Regular expression query
*/
private function buildRegexWithPlugin( $pattern, $insensitive ) {
$filter = new SourceRegex( $pattern, 'source_text', 'source_text.trigram' );
// set some defaults
$this->regexPlugin += [
'max_inspect' => 10000,
];
$filter->setMaxInspect( isset( $this->regexPlugin['max_inspect'] )
? $this->regexPlugin['max_inspect']
: 10000
);
$filter->setMaxDeterminizedStates( $this->maxDeterminizedStates );
if ( isset( $this->regexPlugin['max_ngrams_extracted'] ) ) {
$filter->setMaxNgramsExtracted( $this->regexPlugin['max_ngrams_extracted'] );
}
if ( isset( $this->regexPlugin['max_ngram_clauses'] ) && is_numeric( $this->regexPlugin['max_ngram_clauses'] ) ) {
$filter->setMaxNgramClauses( (int) $this->regexPlugin['max_ngram_clauses'] );
}
$filter->setCaseSensitive( !$insensitive );
$filter->setLocale( $this->languageCode );
return $filter;
}
/**
* Builds a regular expression query using groovy. It's significantly less
* good than the wikimedia-extra plugin, but it's something.
*
* @param string $pattern The regular expression to match
* @param bool $insensitive Should the match be case insensitive?
* @return AbstractQuery Regular expression query
*/
private function buildRegexWithGroovy( $pattern, $insensitive ) {
$script = <<<GROOVY
import org.apache.lucene.util.automaton.*;
sourceText = _source.get("source_text");
if (sourceText == null) {
false;
} else {
if (automaton == null) {
if (insensitive) {
locale = new Locale(language);
pattern = pattern.toLowerCase(locale);
}
regexp = new RegExp(pattern, RegExp.ALL ^ RegExp.AUTOMATON);
automaton = new CharacterRunAutomaton(regexp.toAutomaton());
}
if (insensitive) {
sourceText = sourceText.toLowerCase(locale);
}
automaton.run(sourceText);
}
GROOVY;
return new \Elastica\Query\Script( new \Elastica\Script\Script(
$script,
[
'pattern' => '.*(' . $pattern . ').*',
'insensitive' => $insensitive,
'language' => $this->languageCode,
// The null here creates a slot in which the script will shove
// an automaton while executing.
'automaton' => null,
'locale' => null,
],
'groovy'
) );
}
}
Mini Shell