Examples with IndexRangeIteratorBuilder datawave.query.iterator.builder.IndexRangeIteratorBuilder used on opensource projects

Example 1 with IndexRangeIteratorBuilder

use of datawave.query.iterator.builder.IndexRangeIteratorBuilder in project datawave by NationalSecurityAgency.

the class IteratorBuildingVisitor method visit.

@Override
public Object visit(ASTAndNode and, Object data) {
    QueryPropertyMarker.Instance instance = QueryPropertyMarker.findInstance(and);
    if (instance.isType(ExceededOrThresholdMarkerJexlNode.class)) {
        JexlNode source = instance.getSource();
        // Ivarator to get the job done
        if (source instanceof ASTAndNode) {
            try {
                ivarateList(and, source, data);
            } catch (IOException ioe) {
                throw new DatawaveFatalQueryException(ioe);
            }
        } else {
            QueryException qe = new QueryException(DatawaveErrorCode.UNEXPECTED_SOURCE_NODE, MessageFormat.format("{0}", "Limited ExceededOrThresholdMarkerJexlNode"));
            throw new DatawaveFatalQueryException(qe);
        }
    } else if (data instanceof IndexRangeIteratorBuilder) {
        // index checking has already been done, otherwise we would not have
        // an "ExceededValueThresholdMarker"
        // hence the "IndexAgnostic" method can be used here
        LiteralRange range = JexlASTHelper.findRange().recursively().getRange(and);
        if (range == null) {
            QueryException qe = new QueryException(DatawaveErrorCode.MULTIPLE_RANGES_IN_EXPRESSION);
            throw new DatawaveFatalQueryException(qe);
        }
        ((IndexRangeIteratorBuilder) data).setRange(range);
    } else if (instance.isType(ExceededValueThresholdMarkerJexlNode.class)) {
        // if the parent is our ExceededValueThreshold marker, then use an
        // Ivarator to get the job done unless we don't have to
        JexlNode source = instance.getSource();
        String identifier = null;
        LiteralRange<?> range = null;
        boolean negatedLocal = false;
        if (source instanceof ASTAndNode) {
            range = buildLiteralRange(source, null);
            identifier = range.getFieldName();
        } else {
            if (source instanceof ASTNRNode || source instanceof ASTNotNode)
                negatedLocal = true;
            range = buildLiteralRange(source);
            identifier = JexlASTHelper.getIdentifier(source);
        }
        boolean negatedOverall = negatedLocal;
        if (data instanceof AbstractIteratorBuilder) {
            AbstractIteratorBuilder oib = (AbstractIteratorBuilder) data;
            if (oib.isInANot()) {
                negatedOverall = !negatedOverall;
            }
        }
        // or the field is index only but not in the term frequencies, then we must ivarate
        if (!limitLookup || !allowTermFrequencyLookup || (indexOnlyFields.contains(identifier) && !termFrequencyFields.contains(identifier))) {
            if (source instanceof ASTAndNode) {
                try {
                    List<ASTFunctionNode> functionNodes = JexlASTHelper.getFunctionNodes(source).stream().filter(node -> JexlFunctionArgumentDescriptorFactory.F.getArgumentDescriptor(node).allowIvaratorFiltering()).collect(Collectors.toList());
                    if (functionNodes.isEmpty()) {
                        ivarateRange(and, source, data);
                    } else {
                        ivarateFilter(and, source, data, functionNodes);
                    }
                } catch (IOException ioe) {
                    throw new DatawaveFatalQueryException("Unable to ivarate", ioe);
                }
            } else if (source instanceof ASTERNode || source instanceof ASTNRNode) {
                try {
                    ivarateRegex(and, source, data);
                } catch (IOException ioe) {
                    throw new DatawaveFatalQueryException("Unable to ivarate", ioe);
                }
            } else {
                QueryException qe = new QueryException(DatawaveErrorCode.UNEXPECTED_SOURCE_NODE, MessageFormat.format("{0}", "ExceededValueThresholdMarkerJexlNode"));
                throw new DatawaveFatalQueryException(qe);
            }
        } else {
            NestedIterator<Key> nested = null;
            if (termFrequencyFields.contains(identifier)) {
                nested = buildExceededFromTermFrequency(identifier, and, source, range, data);
            } else {
                /**
                 * This is okay since 1) We are doc specific 2) We are not index only or tf 3) Therefore, we must evaluate against the document for this
                 * expression 4) Return a stubbed range in case we have a disjunction that breaks the current doc.
                 */
                if (!limitOverride && !negatedOverall)
                    nested = createExceededCheck(identifier, range, and);
            }
            if (null != nested && null != data && data instanceof AbstractIteratorBuilder) {
                AbstractIteratorBuilder iterators = (AbstractIteratorBuilder) data;
                if (negatedLocal) {
                    iterators.addExclude(nested);
                } else {
                    iterators.addInclude(nested);
                }
            } else {
                if (isQueryFullySatisfied == true) {
                    log.warn("Determined that isQueryFullySatisfied should be false, but it was not preset to false in the SatisfactionVisitor");
                }
                return nested;
            }
        }
    } else if (null != data && data instanceof AndIteratorBuilder) {
        and.childrenAccept(this, data);
    } else {
        // Create an AndIterator and recursively add the children
        AbstractIteratorBuilder andItr = new AndIteratorBuilder();
        andItr.negateAsNeeded(data);
        and.childrenAccept(this, andItr);
        // If there is no parent
        if (data == null) {
            // Make this AndIterator the root node
            if (!andItr.includes().isEmpty()) {
                root = andItr.build();
            }
        } else {
            // Otherwise, add this AndIterator to its parent
            AbstractIteratorBuilder parent = (AbstractIteratorBuilder) data;
            if (!andItr.includes().isEmpty()) {
                parent.addInclude(andItr.build());
            }
        }
        if (log.isTraceEnabled()) {
            log.trace("ASTAndNode visit: pretty formatting of:\nparent.includes:" + formatIncludesOrExcludes(andItr.includes()) + "\nparent.excludes:" + formatIncludesOrExcludes(andItr.excludes()));
        }
    }
    return null;
}

Example 2 with IndexRangeIteratorBuilder

use of datawave.query.iterator.builder.IndexRangeIteratorBuilder in project datawave by NationalSecurityAgency.

the class IteratorBuildingVisitor method ivarateRange.

/**
 * Build the iterator stack using the regex ivarator (field index caching regex iterator)
 *
 * @param rootNode
 *            the node that was processed to generated this builder
 * @param sourceNode
 *            the source node derived from the root
 * @param data
 */
public void ivarateRange(JexlNode rootNode, JexlNode sourceNode, Object data) throws IOException {
    IndexRangeIteratorBuilder builder = new IndexRangeIteratorBuilder();
    builder.negateAsNeeded(data);
    // hence the "IndexAgnostic" method can be used here
    if (sourceNode instanceof ASTAndNode) {
        LiteralRange range = JexlASTHelper.findRange().recursively().getRange(sourceNode);
        if (range == null) {
            QueryException qe = new QueryException(DatawaveErrorCode.MULTIPLE_RANGES_IN_EXPRESSION);
            throw new DatawaveFatalQueryException(qe);
        }
        builder.setRange(range);
    } else {
        QueryException qe = new QueryException(DatawaveErrorCode.UNEXPECTED_SOURCE_NODE, MessageFormat.format("{0}", "ExceededValueThresholdMarkerJexlNode"));
        throw new DatawaveFatalQueryException(qe);
    }
    builder.forceDocumentBuild(!limitLookup && this.isQueryFullySatisfied);
    ivarate(builder, rootNode, sourceNode, data);
}

Example 3 with IndexRangeIteratorBuilder

use of datawave.query.iterator.builder.IndexRangeIteratorBuilder in project datawave by NationalSecurityAgency.

the class IteratorBuildingVisitor method ivarateList.

/**
 * Build the iterator stack using the regex ivarator (field index caching regex iterator)
 *
 * @param rootNode
 *            the node that was processed to generated this builder
 * @param sourceNode
 *            the source node derived from the root
 * @param data
 */
public void ivarateList(JexlNode rootNode, JexlNode sourceNode, Object data) throws IOException {
    IvaratorBuilder builder = null;
    try {
        String id = ExceededOrThresholdMarkerJexlNode.getId(sourceNode);
        String field = JexlASTHelper.deconstructIdentifier(ExceededOrThresholdMarkerJexlNode.getField(sourceNode));
        ExceededOrThresholdMarkerJexlNode.ExceededOrParams params = ExceededOrThresholdMarkerJexlNode.getParameters(sourceNode);
        if (params.getRanges() != null && !params.getRanges().isEmpty()) {
            IndexRangeIteratorBuilder rangeIterBuilder = new IndexRangeIteratorBuilder();
            builder = rangeIterBuilder;
            SortedSet<Range> ranges = params.getSortedAccumuloRanges();
            rangeIterBuilder.setSubRanges(params.getSortedAccumuloRanges());
            // cache these ranges for use during Jexl Evaluation
            if (exceededOrEvaluationCache != null)
                exceededOrEvaluationCache.put(id, ranges);
            LiteralRange<?> fullRange = new LiteralRange<>(String.valueOf(ranges.first().getStartKey().getRow()), ranges.first().isStartKeyInclusive(), String.valueOf(ranges.last().getEndKey().getRow()), ranges.last().isEndKeyInclusive(), field, NodeOperand.AND);
            rangeIterBuilder.setRange(fullRange);
        } else {
            IndexListIteratorBuilder listIterBuilder = new IndexListIteratorBuilder();
            builder = listIterBuilder;
            if (params.getValues() != null && !params.getValues().isEmpty()) {
                Set<String> values = new TreeSet<>(params.getValues());
                listIterBuilder.setValues(values);
                // cache these values for use during Jexl Evaluation
                if (exceededOrEvaluationCache != null)
                    exceededOrEvaluationCache.put(id, values);
            } else if (params.getFstURI() != null) {
                URI fstUri = new URI(params.getFstURI());
                FST fst;
                // only recompute this if not already set since this is potentially expensive
                if (exceededOrEvaluationCache.containsKey(id)) {
                    fst = (FST) exceededOrEvaluationCache.get(id);
                } else {
                    fst = DatawaveFieldIndexListIteratorJexl.FSTManager.get(new Path(fstUri), hdfsFileCompressionCodec, hdfsFileSystem.getFileSystem(fstUri));
                }
                listIterBuilder.setFst(fst);
                // cache this fst for use during JexlEvaluation.
                if (exceededOrEvaluationCache != null)
                    exceededOrEvaluationCache.put(id, fst);
            }
            // If this is actually negated, then this will be added to excludes. Do not negate in the ivarator
            listIterBuilder.setNegated(false);
        }
        builder.setField(field);
    } catch (IOException | URISyntaxException | NullPointerException e) {
        QueryException qe = new QueryException(DatawaveErrorCode.UNPARSEABLE_EXCEEDED_OR_PARAMS, e, MessageFormat.format("Class: {0}", ExceededOrThresholdMarkerJexlNode.class.getSimpleName()));
        throw new DatawaveFatalQueryException(qe);
    }
    builder.negateAsNeeded(data);
    builder.forceDocumentBuild(!limitLookup && this.isQueryFullySatisfied);
    ivarate(builder, rootNode, sourceNode, data);
}