Examples with RangeStream datawave.query.index.lookup.RangeStream used on opensource projects

Example 1 with RangeStream

use of datawave.query.index.lookup.RangeStream in project datawave by NationalSecurityAgency.

the class DefaultQueryPlanner method initializeRangeStream.

/**
 * Initializes the range stream, whether it is configured to be a different class than the Default Range stream or not.
 *
 * @param config
 * @param scannerFactory
 * @param metadataHelper
 * @return
 */
private RangeStream initializeRangeStream(ShardQueryConfiguration config, ScannerFactory scannerFactory, MetadataHelper metadataHelper) {
    Class<? extends RangeStream> rstream;
    try {
        rstream = Class.forName(rangeStreamClass).asSubclass(RangeStream.class);
        RangeStream stream = rstream.getConstructor(ShardQueryConfiguration.class, ScannerFactory.class, MetadataHelper.class).newInstance(config, scannerFactory, metadataHelper);
        return stream.setUidIntersector(uidIntersector).setLimitScanners(limitScanners).setCreateUidsIteratorClass(createUidsIteratorClass);
    } catch (ClassNotFoundException | InstantiationException | IllegalAccessException | IllegalArgumentException | InvocationTargetException | NoSuchMethodException | SecurityException e) {
        throw new RuntimeException(e);
    }
}

Example 2 with RangeStream

use of datawave.query.index.lookup.RangeStream in project datawave by NationalSecurityAgency.

the class DefaultQueryPlanner method getQueryRanges.

/**
 * Returns a Tuple2<Iterable<Range>,Boolean> whose elements represent the Ranges to use for querying the shard table and whether or not this is
 * a "full-table-scan" query.
 *
 * @param scannerFactory
 * @param metadataHelper
 * @param config
 * @param queryTree
 * @return
 * @throws DatawaveQueryException
 */
public Tuple2<CloseableIterable<QueryPlan>, Boolean> getQueryRanges(ScannerFactory scannerFactory, MetadataHelper metadataHelper, ShardQueryConfiguration config, JexlNode queryTree) throws DatawaveQueryException {
    Preconditions.checkNotNull(queryTree);
    boolean needsFullTable = false;
    CloseableIterable<QueryPlan> ranges = null;
    // if the query has already been reduced to false there is no reason to do more
    if (QueryPruningVisitor.getState(queryTree) == QueryPruningVisitor.TruthState.FALSE) {
        return new Tuple2<>(emptyCloseableIterator(), false);
    }
    // if we still have an unexecutable tree, then a full table scan is
    // required
    List<String> debugOutput = null;
    if (log.isDebugEnabled()) {
        debugOutput = new ArrayList<>(32);
    }
    STATE state = ExecutableDeterminationVisitor.getState(queryTree, config, metadataHelper, debugOutput);
    if (log.isDebugEnabled()) {
        logDebug(debugOutput, "ExecutableDeterminationVisitor at getQueryRanges:");
    }
    if (state != STATE.EXECUTABLE) {
        if (state == STATE.ERROR) {
            log.warn("After expanding the query, it is determined that the query cannot be executed due to index-only fields mixed with expressions that cannot be run against the index.");
            BadRequestQueryException qe = new BadRequestQueryException(DatawaveErrorCode.INDEX_ONLY_FIELDS_MIXED_INVALID_EXPRESSIONS);
            throw new InvalidQueryException(qe);
        }
        log.warn("After expanding the query, it is determined that the query cannot be executed against the field index and a full table scan is required");
        needsFullTable = true;
    }
    // scan, then lets try to compute ranges
    if (!needsFullTable) {
        // count the terms
        int termCount = TermCountingVisitor.countTerms(queryTree);
        if (termCount >= pushdownThreshold) {
            if (log.isTraceEnabled()) {
                log.trace("pushing down query because it has " + termCount + " when our max is " + pushdownThreshold);
            }
            config.setCollapseUids(true);
        }
        TraceStopwatch stopwatch = config.getTimers().newStartedStopwatch("DefaultQueryPlanner - Begin stream of ranges from inverted index");
        RangeStream stream = initializeRangeStream(config, scannerFactory, metadataHelper);
        ranges = stream.streamPlans(queryTree);
        if (log.isTraceEnabled()) {
            log.trace("query stream is " + stream.context());
        }
        // if a term threshold is exceeded and we cannot handle that, then
        // throw unsupported
        boolean thresholdExceeded = StreamContext.EXCEEDED_TERM_THRESHOLD.equals(stream.context());
        if (thresholdExceeded && !config.canHandleExceededTermThreshold()) {
            throw new UnsupportedOperationException(EXCEED_TERM_EXPANSION_ERROR);
        }
        if (StreamContext.UNINDEXED.equals(stream.context())) {
            log.debug("Needs full table scan because of unindexed fields");
            needsFullTable = true;
        } else if (StreamContext.DELAYED_FIELD.equals(stream.context())) {
            log.debug("Needs full table scan because query consists of only delayed expressions");
            needsFullTable = true;
        } else // force a full table scan
        if (IvaratorRequiredVisitor.isIvaratorRequired(queryTree) && !config.canHandleExceededValueThreshold()) {
            log.debug("Needs full table scan because we exceeded the value threshold and config.canHandleExceededValueThreshold() is false");
            needsFullTable = true;
        }
        stopwatch.stop();
    }
    if (needsFullTable) {
        if (config.getFullTableScanEnabled()) {
            ranges = this.getFullScanRange(config, queryTree);
        } else {
            if (log.isTraceEnabled())
                log.trace("Full table scans are not enabled, query will not be run");
            QueryException qe = new QueryException(DatawaveErrorCode.FULL_TABLE_SCAN_REQUIRED_BUT_DISABLED);
            throw new FullTableScansDisallowedException(qe);
        }
        if (log.isTraceEnabled())
            log.trace("Ranges are " + ranges);
    }
    return new Tuple2<>(ranges, needsFullTable);
}