Search in sources :

Example 1 with IsNull

use of org.apache.spark.sql.sources.IsNull in project Gaffer by gchq.

the class FiltersToOperationConverter method getFunctionsFromFilter.

/**
 * Converts a Spark {@link Filter} to a map from group to a list of Gaffer {@link TupleAdaptedPredicate}s.
 * <p>
 * Note that Spark also applies all the filters provided to the {@code buildScan(String[], Filter[])} method
 * so not implementing some of the provided {@link Filter}s in Gaffer will not cause errors. However, as many as
 * possible should be implemented so that as much filtering as possible happens in iterators running in Accumulo's
 * tablet servers (this avoids unnecessary data transfer from Accumulo to Spark).
 *
 * @param filter The {@link Filter} to transform.
 * @return A map from {@link String} to {@link TupleAdaptedPredicate}s implementing the provided {@link Filter}.
 */
private Map<String, List<TupleAdaptedPredicate<String, ?>>> getFunctionsFromFilter(final Filter filter) {
    final Map<String, List<TupleAdaptedPredicate<String, ?>>> map = new HashMap<>();
    if (filter instanceof EqualTo) {
    // Not dealt with as requires a Predicate<?> that returns null if either the controlValue or the
    // test value is null - the API of Predicate<?> doesn't permit this.
    } else if (filter instanceof EqualNullSafe) {
        final EqualNullSafe equalNullSafe = (EqualNullSafe) filter;
        final Predicate<?> isEqual = new IsEqual(equalNullSafe.value());
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(isEqual, new String[] { equalNullSafe.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to IsEqual ({})", filter, equalNullSafe.attribute());
    } else if (filter instanceof GreaterThan) {
        final GreaterThan greaterThan = (GreaterThan) filter;
        final Predicate<?> isMoreThan = new IsMoreThan((Comparable<?>) greaterThan.value(), false);
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(isMoreThan, new String[] { greaterThan.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to isMoreThan ({})", filter, greaterThan.attribute());
    } else if (filter instanceof GreaterThanOrEqual) {
        final GreaterThanOrEqual greaterThan = (GreaterThanOrEqual) filter;
        final Predicate<?> isMoreThan = new IsMoreThan((Comparable<?>) greaterThan.value(), true);
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(isMoreThan, new String[] { greaterThan.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to IsMoreThan ({})", filter, greaterThan.attribute());
    } else if (filter instanceof LessThan) {
        final LessThan lessThan = (LessThan) filter;
        final Predicate<?> isLessThan = new IsLessThan((Comparable<?>) lessThan.value(), false);
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(isLessThan, new String[] { lessThan.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to IsLessThan ({})", filter, lessThan.attribute());
    } else if (filter instanceof LessThanOrEqual) {
        final LessThanOrEqual lessThan = (LessThanOrEqual) filter;
        final Predicate<?> isLessThan = new IsLessThan((Comparable<?>) lessThan.value(), true);
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(isLessThan, new String[] { lessThan.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to LessThanOrEqual ({})", filter, lessThan.attribute());
    } else if (filter instanceof In) {
        final In in = (In) filter;
        final Predicate<?> isIn = new IsIn(new HashSet<>(Arrays.asList(in.values())));
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(isIn, new String[] { in.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to IsIn ({})", filter, in.attribute());
    } else if (filter instanceof IsNull) {
        final IsNull isNull = (IsNull) filter;
        final Predicate<?> doesntExist = new Not<>(new Exists());
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(doesntExist, new String[] { isNull.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to Not(Exists) ({})", filter, isNull.attribute());
    } else if (filter instanceof IsNotNull) {
        final IsNotNull isNotNull = (IsNotNull) filter;
        final Predicate<?> exists = new Exists();
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).add(new TupleAdaptedPredicate<>(exists, new String[] { isNotNull.attribute() }));
            }
        }
        LOGGER.debug("Converted {} to Exists ({})", filter, isNotNull.attribute());
    } else if (filter instanceof And) {
        final And and = (And) filter;
        final Map<String, List<TupleAdaptedPredicate<String, ?>>> left = getFunctionsFromFilter(and.left());
        final Map<String, List<TupleAdaptedPredicate<String, ?>>> right = getFunctionsFromFilter(and.right());
        final Set<String> relevantGroups = getGroupsFromFilter(filter);
        if (null != relevantGroups) {
            for (final String group : relevantGroups) {
                final List<TupleAdaptedPredicate<String, ?>> concatFilters = new ArrayList<>();
                if (null != left.get(group)) {
                    concatFilters.addAll(left.get(group));
                }
                if (null != right.get(group)) {
                    concatFilters.addAll(right.get(group));
                }
                if (!map.containsKey(group)) {
                    map.put(group, new ArrayList<>());
                }
                map.get(group).addAll(concatFilters);
            }
        }
        LOGGER.debug("Converted {} to list of filters ({})", filter, StringUtils.join(map.entrySet(), ','));
    }
    return map;
}
Also used : HashSet(java.util.HashSet) Set(java.util.Set) EqualNullSafe(org.apache.spark.sql.sources.EqualNullSafe) LessThanOrEqual(org.apache.spark.sql.sources.LessThanOrEqual) HashMap(java.util.HashMap) In(org.apache.spark.sql.sources.In) IsIn(uk.gov.gchq.koryphe.impl.predicate.IsIn) ArrayList(java.util.ArrayList) GreaterThanOrEqual(org.apache.spark.sql.sources.GreaterThanOrEqual) TupleAdaptedPredicate(uk.gov.gchq.koryphe.tuple.predicate.TupleAdaptedPredicate) Predicate(java.util.function.Predicate) IsLessThan(uk.gov.gchq.koryphe.impl.predicate.IsLessThan) LessThan(org.apache.spark.sql.sources.LessThan) IsNotNull(org.apache.spark.sql.sources.IsNotNull) IsLessThan(uk.gov.gchq.koryphe.impl.predicate.IsLessThan) GreaterThan(org.apache.spark.sql.sources.GreaterThan) ArrayList(java.util.ArrayList) List(java.util.List) TupleAdaptedPredicate(uk.gov.gchq.koryphe.tuple.predicate.TupleAdaptedPredicate) IsIn(uk.gov.gchq.koryphe.impl.predicate.IsIn) EqualTo(org.apache.spark.sql.sources.EqualTo) IsEqual(uk.gov.gchq.koryphe.impl.predicate.IsEqual) Exists(uk.gov.gchq.koryphe.impl.predicate.Exists) And(org.apache.spark.sql.sources.And) IsNull(org.apache.spark.sql.sources.IsNull) IsMoreThan(uk.gov.gchq.koryphe.impl.predicate.IsMoreThan) HashMap(java.util.HashMap) Map(java.util.Map)

Aggregations

ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 HashSet (java.util.HashSet)1 List (java.util.List)1 Map (java.util.Map)1 Set (java.util.Set)1 Predicate (java.util.function.Predicate)1 And (org.apache.spark.sql.sources.And)1 EqualNullSafe (org.apache.spark.sql.sources.EqualNullSafe)1 EqualTo (org.apache.spark.sql.sources.EqualTo)1 GreaterThan (org.apache.spark.sql.sources.GreaterThan)1 GreaterThanOrEqual (org.apache.spark.sql.sources.GreaterThanOrEqual)1 In (org.apache.spark.sql.sources.In)1 IsNotNull (org.apache.spark.sql.sources.IsNotNull)1 IsNull (org.apache.spark.sql.sources.IsNull)1 LessThan (org.apache.spark.sql.sources.LessThan)1 LessThanOrEqual (org.apache.spark.sql.sources.LessThanOrEqual)1 Exists (uk.gov.gchq.koryphe.impl.predicate.Exists)1 IsEqual (uk.gov.gchq.koryphe.impl.predicate.IsEqual)1 IsIn (uk.gov.gchq.koryphe.impl.predicate.IsIn)1