Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 103 |
|
0.00% |
0 / 5 |
CRAP | |
0.00% |
0 / 1 |
| ApiFeatureUsageQueryEngineElastica | |
0.00% |
0 / 103 |
|
0.00% |
0 / 5 |
380 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
2 | |||
| getClient | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getIndexNames | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
12 | |||
| execute | |
0.00% |
0 / 66 |
|
0.00% |
0 / 1 |
110 | |||
| suggestDateRange | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\ApiFeatureUsage; |
| 4 | |
| 5 | use DateInterval; |
| 6 | use Elastica\Aggregation\DateHistogram; |
| 7 | use Elastica\Aggregation\Terms as AggregationTerms; |
| 8 | use Elastica\Client; |
| 9 | use Elastica\Query; |
| 10 | use Elastica\Query\BoolQuery; |
| 11 | use Elastica\Query\Prefix; |
| 12 | use Elastica\Query\Range; |
| 13 | use Elastica\Query\Terms as QueryTerms; |
| 14 | use Elastica\Search; |
| 15 | use MediaWiki\Status\Status; |
| 16 | use MediaWiki\Utils\MWTimestamp; |
| 17 | use RuntimeException; |
| 18 | |
| 19 | /** |
| 20 | * Query feature usage data from Elasticsearch. |
| 21 | * |
| 22 | * Config fields are: |
| 23 | * serverList: Array of servers to connect to |
| 24 | * maxConnectionAttempts: Maximum connection attempts |
| 25 | * indexPrefix: Index prefix |
| 26 | * indexFormat: Date format string for index |
| 27 | * featureField: Name of the field holding $feature |
| 28 | * timestampField: Name of the field holding the timestamp |
| 29 | * agentField: Name of the field holding the user agent |
| 30 | */ |
| 31 | class ApiFeatureUsageQueryEngineElastica extends ApiFeatureUsageQueryEngine { |
| 32 | /** @var Client|null */ |
| 33 | private $client = null; |
| 34 | /** @var string[]|null */ |
| 35 | private $indexNames = null; |
| 36 | |
| 37 | /** |
| 38 | * @param array $options |
| 39 | */ |
| 40 | public function __construct( array $options ) { |
| 41 | $options += [ |
| 42 | 'indexPrefix' => 'apifeatureusage-', |
| 43 | 'indexFormat' => 'Y.m.d', |
| 44 | 'featureField' => 'feature', |
| 45 | 'featureFieldAggSize' => 10000, |
| 46 | 'timestampField' => '@timestamp', |
| 47 | 'agentField' => 'agent', |
| 48 | ]; |
| 49 | |
| 50 | parent::__construct( $options ); |
| 51 | } |
| 52 | |
| 53 | /** |
| 54 | * @return Client |
| 55 | */ |
| 56 | protected function getClient() { |
| 57 | if ( !$this->client ) { |
| 58 | $connection = new ApiFeatureUsageQueryEngineElasticaConnection( $this->options ); |
| 59 | $this->client = $connection->getClient(); |
| 60 | } |
| 61 | return $this->client; |
| 62 | } |
| 63 | |
| 64 | /** |
| 65 | * @return string[] |
| 66 | */ |
| 67 | protected function getIndexNames() { |
| 68 | if ( !$this->indexNames ) { |
| 69 | $response = $this->getClient()->request( |
| 70 | urlencode( $this->options['indexPrefix'] ) . '*/_alias' |
| 71 | ); |
| 72 | if ( $response->isOK() ) { |
| 73 | $this->indexNames = array_keys( $response->getData() ); |
| 74 | } else { |
| 75 | throw new RuntimeException( __METHOD__ . |
| 76 | ': Cannot fetch index names from elasticsearch: ' . |
| 77 | $response->getError() |
| 78 | ); |
| 79 | } |
| 80 | } |
| 81 | return $this->indexNames; |
| 82 | } |
| 83 | |
| 84 | /** @inheritDoc */ |
| 85 | public function execute( $agent, MWTimestamp $start, MWTimestamp $end, array $features = null ) { |
| 86 | $status = Status::newGood( [] ); |
| 87 | |
| 88 | # Force $start and $end to day boundaries |
| 89 | $oneDay = new DateInterval( 'P1D' ); |
| 90 | $start = clone $start; |
| 91 | $start->timestamp = clone $start->timestamp; |
| 92 | $start->timestamp->setTime( 0, 0, 0 ); |
| 93 | $end = clone $end; |
| 94 | $end->timestamp = clone $end->timestamp; |
| 95 | $end->timestamp->setTime( 0, 0, 0 ); |
| 96 | $end->timestamp->add( $oneDay )->sub( new DateInterval( 'PT1S' ) ); |
| 97 | |
| 98 | $query = new Query(); |
| 99 | |
| 100 | $bools = new BoolQuery(); |
| 101 | |
| 102 | $prefix = new Prefix(); |
| 103 | $prefix->setPrefix( $this->options['agentField'], $agent ); |
| 104 | $bools->addMust( $prefix ); |
| 105 | |
| 106 | $bools->addMust( new Range( $this->options['timestampField'], [ |
| 107 | 'gte' => $start->getTimestamp( TS_ISO_8601 ), |
| 108 | 'lte' => $end->getTimestamp( TS_ISO_8601 ), |
| 109 | ] ) ); |
| 110 | |
| 111 | if ( $features !== null ) { |
| 112 | $bools->addMust( new QueryTerms( $this->options['featureField'], $features ) ); |
| 113 | } |
| 114 | |
| 115 | $query->setQuery( $bools ); |
| 116 | |
| 117 | $termsAgg = new AggregationTerms( 'feature' ); |
| 118 | $termsAgg->setField( $this->options['featureField'] ); |
| 119 | $termsAgg->setSize( $this->options['featureFieldAggSize'] ); |
| 120 | |
| 121 | $datesAgg = new DateHistogram( |
| 122 | 'date', $this->options['timestampField'], 'day' |
| 123 | ); |
| 124 | $datesAgg->setFormat( '8uuuu-MM-dd' ); |
| 125 | |
| 126 | $termsAgg->addAggregation( $datesAgg ); |
| 127 | $query->addAggregation( $termsAgg ); |
| 128 | |
| 129 | $search = new Search( $this->getClient() ); |
| 130 | $search->setOption( Search::OPTION_SIZE, 0 ); |
| 131 | |
| 132 | $allIndexes = $this->getIndexNames(); |
| 133 | $indexAvailable = false; |
| 134 | $skippedAny = false; |
| 135 | $s = clone $start->timestamp; |
| 136 | while ( $s <= $end->timestamp ) { |
| 137 | $index = $this->options['indexPrefix'] . $s->format( $this->options['indexFormat'] ); |
| 138 | if ( in_array( $index, $allIndexes ) ) { |
| 139 | $indexAvailable = true; |
| 140 | } else { |
| 141 | $skippedAny = true; |
| 142 | } |
| 143 | $s->add( $oneDay ); |
| 144 | } |
| 145 | if ( !$indexAvailable ) { |
| 146 | // No dates in range |
| 147 | $status->warning( 'apifeatureusage-no-indexes' ); |
| 148 | return $status; |
| 149 | } |
| 150 | if ( $skippedAny ) { |
| 151 | $status->warning( 'apifeatureusage-missing-indexes' ); |
| 152 | } |
| 153 | |
| 154 | $search->setQuery( $query ); |
| 155 | // Prefer the wildcard approach over using an explicit list of indices to avoid building a |
| 156 | // list that might be too long to encode in the search URL. |
| 157 | // This feature is rarely used so that it's probably fine to hit all these indices and let |
| 158 | // the date filtering quickly skip unrelated ones. |
| 159 | $search->addIndexByName( $this->options['indexPrefix'] . '*' ); |
| 160 | |
| 161 | $res = $search->search(); |
| 162 | |
| 163 | if ( $res->getResponse()->hasError() ) { |
| 164 | return Status::newFatal( |
| 165 | 'apifeatureusage-elasticsearch-error', $res->getResponse()->getError() |
| 166 | ); |
| 167 | } |
| 168 | |
| 169 | $ret = []; |
| 170 | $aggs = $res->getAggregations(); |
| 171 | if ( isset( $aggs['feature'] ) ) { |
| 172 | foreach ( $aggs['feature']['buckets'] as $feature ) { |
| 173 | foreach ( $feature['date']['buckets'] as $date ) { |
| 174 | $ret[] = [ |
| 175 | 'feature' => $feature['key'], |
| 176 | 'date' => $date['key_as_string'], |
| 177 | 'count' => $date['doc_count'], |
| 178 | ]; |
| 179 | } |
| 180 | } |
| 181 | } |
| 182 | $status->value = $ret; |
| 183 | |
| 184 | return $status; |
| 185 | } |
| 186 | |
| 187 | public function suggestDateRange() { |
| 188 | $start = new MWTimestamp(); |
| 189 | $start->setTimezone( 'UTC' ); |
| 190 | $start->timestamp->setTime( 0, 0, 0 ); |
| 191 | $end = new MWTimestamp(); |
| 192 | $end->setTimezone( 'UTC' ); |
| 193 | |
| 194 | $oneDay = new DateInterval( 'P1D' ); |
| 195 | $allIndexes = $this->getIndexNames(); |
| 196 | while ( true ) { |
| 197 | $start->timestamp->sub( $oneDay ); |
| 198 | $index = $this->options['indexPrefix'] . $start->format( $this->options['indexFormat'] ); |
| 199 | if ( !in_array( $index, $allIndexes ) ) { |
| 200 | $start->timestamp->add( $oneDay ); |
| 201 | return [ $start, $end ]; |
| 202 | } |
| 203 | } |
| 204 | } |
| 205 | } |