use of org.apache.parquet.filter2.predicate.Operators.FloatColumn in project Gaffer by gchq.
the class JavaPredicateToParquetPredicate method getIsMoreThanOrEqualToFilter.
private FilterPredicate getIsMoreThanOrEqualToFilter(final String colName, final Object[] parquetObjects, final String group, final SchemaUtils schemaUtils) {
String[] paths = schemaUtils.getPaths(group, colName);
if (null == paths) {
paths = new String[1];
paths[0] = colName;
}
FilterPredicate filter = null;
for (int i = 0; i < paths.length; i++) {
final String path = paths[i];
FilterPredicate tempFilter;
if (parquetObjects[i] instanceof String) {
tempFilter = gtEq(binaryColumn(path), Binary.fromString((String) parquetObjects[i]));
} else if (parquetObjects[i] instanceof Double) {
tempFilter = gtEq(doubleColumn(path), (Double) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Float) {
tempFilter = gtEq(floatColumn(path), (Float) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Integer) {
tempFilter = gtEq(intColumn(path), (Integer) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Long) {
tempFilter = gtEq(longColumn(path), (Long) parquetObjects[i]);
} else if (parquetObjects[i] instanceof java.util.Date) {
tempFilter = gtEq(longColumn(path), ((java.util.Date) parquetObjects[i]).getTime());
} else if (parquetObjects[i] instanceof java.sql.Date) {
tempFilter = gtEq(longColumn(path), ((java.sql.Date) parquetObjects[i]).getTime());
} else if (parquetObjects[i] instanceof Short) {
tempFilter = gtEq(intColumn(path), ((Short) parquetObjects[i]).intValue());
} else if (parquetObjects[i] instanceof byte[]) {
tempFilter = gtEq(binaryColumn(path), Binary.fromReusedByteArray((byte[]) parquetObjects[i]));
} else {
fullyApplied = false;
LOGGER.warn(parquetObjects[i].getClass().getCanonicalName() + " is not a natively supported type for the IsMoreThanOrEqualTo filter, therefore execution will take longer to perform this filter.");
return null;
}
if (null == filter) {
filter = tempFilter;
} else {
filter = and(filter, tempFilter);
}
}
return filter;
}
use of org.apache.parquet.filter2.predicate.Operators.FloatColumn in project presto by prestodb.
the class TestColumnIndexBuilder method testBuildFloat.
@Test
public void testBuildFloat() {
PrimitiveType type = Types.required(FLOAT).named("test_float");
ColumnIndexBuilder builder = ColumnIndexBuilder.getBuilder(type, Integer.MAX_VALUE);
// assertThat(builder, instanceOf(FloatColumnIndexBuilder.class));
assertNull(builder.build());
Operators.FloatColumn col = floatColumn("test_col");
StatsBuilder sb = new StatsBuilder();
builder.add(sb.stats(type, -4.2f, -4.1f));
builder.add(sb.stats(type, -11.7f, 7.0f, null));
builder.add(sb.stats(type, 2.2f, 2.2f, null, null));
builder.add(sb.stats(type, null, null, null));
builder.add(sb.stats(type, 1.9f, 2.32f));
builder.add(sb.stats(type, -21.0f, 8.1f));
assertEquals(6, builder.getPageCount());
assertEquals(sb.getMinMaxSize(), builder.getMinMaxSize());
ColumnIndex columnIndex = builder.build();
assertEquals(BoundaryOrder.UNORDERED, columnIndex.getBoundaryOrder());
assertCorrectNullCounts(columnIndex, 0, 1, 2, 3, 0, 0);
assertCorrectNullPages(columnIndex, false, false, false, true, false, false);
assertCorrectValues(columnIndex.getMaxValues(), -4.1f, 7.0f, 2.2f, null, 2.32f, 8.1f);
assertCorrectValues(columnIndex.getMinValues(), -4.2f, -11.7f, 2.2f, null, 1.9f, -21.0f);
assertCorrectFiltering(columnIndex, eq(col, 0.0f), 1, 5);
assertCorrectFiltering(columnIndex, eq(col, null), 1, 2, 3);
assertCorrectFiltering(columnIndex, notEq(col, 2.2f), 0, 1, 2, 3, 4, 5);
assertCorrectFiltering(columnIndex, notEq(col, null), 0, 1, 2, 4, 5);
assertCorrectFiltering(columnIndex, gt(col, 2.2f), 1, 4, 5);
assertCorrectFiltering(columnIndex, gtEq(col, 2.2f), 1, 2, 4, 5);
assertCorrectFiltering(columnIndex, lt(col, 0.0f), 0, 1, 5);
assertCorrectFiltering(columnIndex, ltEq(col, 1.9f), 0, 1, 4, 5);
assertCorrectFiltering(columnIndex, userDefined(col, FloatIsInteger.class), 1, 4, 5);
assertCorrectFiltering(columnIndex, invert(userDefined(col, FloatIsInteger.class)), 0, 1, 2, 3, 4, 5);
builder = ColumnIndexBuilder.getBuilder(type, Integer.MAX_VALUE);
sb = new StatsBuilder();
builder.add(sb.stats(type, null, null));
builder.add(sb.stats(type, -532.3f, -345.2f, null, null));
builder.add(sb.stats(type, -300.6f, -234.7f, null));
builder.add(sb.stats(type, null, null));
builder.add(sb.stats(type, null, null, null));
builder.add(sb.stats(type, -234.6f, 2.99999f));
builder.add(sb.stats(type, null, null));
builder.add(sb.stats(type, 3.0f, 42.83f));
builder.add(sb.stats(type, null, null));
assertEquals(9, builder.getPageCount());
assertEquals(sb.getMinMaxSize(), builder.getMinMaxSize());
columnIndex = builder.build();
assertEquals(BoundaryOrder.ASCENDING, columnIndex.getBoundaryOrder());
assertCorrectNullCounts(columnIndex, 2, 2, 1, 2, 3, 0, 2, 0, 2);
assertCorrectNullPages(columnIndex, true, false, false, true, true, false, true, false, true);
assertCorrectValues(columnIndex.getMaxValues(), null, -345.2f, -234.7f, null, null, 2.99999f, null, 42.83f, null);
assertCorrectValues(columnIndex.getMinValues(), null, -532.3f, -300.6f, null, null, -234.6f, null, 3.0f, null);
assertCorrectFiltering(columnIndex, eq(col, 0.0f), 5);
assertCorrectFiltering(columnIndex, eq(col, null), 0, 1, 2, 3, 4, 6, 8);
assertCorrectFiltering(columnIndex, notEq(col, 2.2f), 0, 1, 2, 3, 4, 5, 6, 7, 8);
assertCorrectFiltering(columnIndex, notEq(col, null), 1, 2, 5, 7);
assertCorrectFiltering(columnIndex, gt(col, 2.2f), 5, 7);
assertCorrectFiltering(columnIndex, gtEq(col, -234.7f), 2, 5, 7);
assertCorrectFiltering(columnIndex, lt(col, -234.6f), 1, 2);
assertCorrectFiltering(columnIndex, ltEq(col, -234.6f), 1, 2, 5);
assertCorrectFiltering(columnIndex, userDefined(col, FloatIsInteger.class), 1, 2, 5, 7);
assertCorrectFiltering(columnIndex, invert(userDefined(col, FloatIsInteger.class)), 0, 1, 2, 3, 4, 5, 6, 7, 8);
builder = ColumnIndexBuilder.getBuilder(type, Integer.MAX_VALUE);
sb = new StatsBuilder();
builder.add(sb.stats(type, null, null, null, null, null));
builder.add(sb.stats(type, 532.3f, 345.2f));
builder.add(sb.stats(type, null, null, null));
builder.add(sb.stats(type, 234.7f, 234.6f, null));
builder.add(sb.stats(type, null, null));
builder.add(sb.stats(type, 234.6f, -2.99999f));
builder.add(sb.stats(type, null, null));
builder.add(sb.stats(type, null, null));
builder.add(sb.stats(type, -3.0f, -42.83f));
assertEquals(9, builder.getPageCount());
assertEquals(sb.getMinMaxSize(), builder.getMinMaxSize());
columnIndex = builder.build();
assertEquals(BoundaryOrder.DESCENDING, columnIndex.getBoundaryOrder());
assertCorrectNullCounts(columnIndex, 5, 0, 3, 1, 2, 0, 2, 2, 0);
assertCorrectNullPages(columnIndex, true, false, true, false, true, false, true, true, false);
assertCorrectValues(columnIndex.getMaxValues(), null, 532.3f, null, 234.7f, null, 234.6f, null, null, -3.0f);
assertCorrectValues(columnIndex.getMinValues(), null, 345.2f, null, 234.6f, null, -2.99999f, null, null, -42.83f);
assertCorrectFiltering(columnIndex, eq(col, 234.65f), 3);
assertCorrectFiltering(columnIndex, eq(col, null), 0, 2, 3, 4, 6, 7);
assertCorrectFiltering(columnIndex, notEq(col, 2.2f), 0, 1, 2, 3, 4, 5, 6, 7, 8);
assertCorrectFiltering(columnIndex, notEq(col, null), 1, 3, 5, 8);
assertCorrectFiltering(columnIndex, gt(col, 2.2f), 1, 3, 5);
assertCorrectFiltering(columnIndex, gtEq(col, 2.2f), 1, 3, 5);
assertCorrectFiltering(columnIndex, lt(col, 0.0f), 5, 8);
assertCorrectFiltering(columnIndex, ltEq(col, 0.0f), 5, 8);
assertCorrectFiltering(columnIndex, userDefined(col, FloatIsInteger.class), 1, 5, 8);
assertCorrectFiltering(columnIndex, invert(userDefined(col, FloatIsInteger.class)), 0, 1, 2, 3, 4, 5, 6, 7, 8);
}
use of org.apache.parquet.filter2.predicate.Operators.FloatColumn in project Gaffer by gchq.
the class QueryGenerator method getIsEqualFilter.
private FilterPredicate getIsEqualFilter(final String colName, final Object[] parquetObjects, final String group) {
String[] paths = schemaUtils.getPaths(group, colName);
if (null == paths) {
paths = new String[1];
paths[0] = colName;
}
FilterPredicate filter = null;
for (int i = 0; i < paths.length; i++) {
final String path = paths[i];
FilterPredicate tempFilter;
if (parquetObjects[i] instanceof String) {
tempFilter = eq(binaryColumn(path), Binary.fromString((String) parquetObjects[i]));
} else if (parquetObjects[i] instanceof Boolean) {
tempFilter = eq(booleanColumn(path), (Boolean) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Double) {
tempFilter = eq(doubleColumn(path), (Double) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Float) {
tempFilter = eq(floatColumn(path), (Float) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Integer) {
tempFilter = eq(intColumn(path), (Integer) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Long) {
tempFilter = eq(longColumn(path), (Long) parquetObjects[i]);
} else if (parquetObjects[i] instanceof java.util.Date) {
tempFilter = eq(longColumn(path), ((java.util.Date) parquetObjects[i]).getTime());
} else if (parquetObjects[i] instanceof java.sql.Date) {
tempFilter = eq(longColumn(path), ((java.sql.Date) parquetObjects[i]).getTime());
} else if (parquetObjects[i] instanceof Short) {
tempFilter = eq(intColumn(path), ((Short) parquetObjects[i]).intValue());
} else if (parquetObjects[i] instanceof byte[]) {
tempFilter = eq(binaryColumn(path), Binary.fromReusedByteArray((byte[]) parquetObjects[i]));
} else {
LOGGER.warn(parquetObjects[i].getClass().getCanonicalName() + " is not a natively supported type for the IsEqual filter, therefore execution will take longer to perform this filter.");
return null;
}
if (null == filter) {
filter = tempFilter;
} else {
filter = and(filter, tempFilter);
}
}
return filter;
}
use of org.apache.parquet.filter2.predicate.Operators.FloatColumn in project Gaffer by gchq.
the class JavaPredicateToParquetPredicate method getIsLessThanFilter.
private FilterPredicate getIsLessThanFilter(final String colName, final Object[] parquetObjects, final String group, final SchemaUtils schemaUtils) {
String[] paths = schemaUtils.getPaths(group, colName);
if (null == paths) {
paths = new String[1];
paths[0] = colName;
}
FilterPredicate filter = null;
for (int i = 0; i < paths.length; i++) {
final String path = paths[i];
FilterPredicate tempFilter;
if (parquetObjects[i] instanceof String) {
tempFilter = lt(binaryColumn(path), Binary.fromString((String) parquetObjects[i]));
} else if (parquetObjects[i] instanceof Double) {
tempFilter = lt(doubleColumn(path), (Double) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Float) {
tempFilter = lt(floatColumn(path), (Float) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Integer) {
tempFilter = lt(intColumn(path), (Integer) parquetObjects[i]);
} else if (parquetObjects[i] instanceof Long) {
tempFilter = lt(longColumn(path), (Long) parquetObjects[i]);
} else if (parquetObjects[i] instanceof java.util.Date) {
tempFilter = lt(longColumn(path), ((java.util.Date) parquetObjects[i]).getTime());
} else if (parquetObjects[i] instanceof java.sql.Date) {
tempFilter = lt(longColumn(path), ((java.sql.Date) parquetObjects[i]).getTime());
} else if (parquetObjects[i] instanceof Short) {
tempFilter = lt(intColumn(path), ((Short) parquetObjects[i]).intValue());
} else if (parquetObjects[i] instanceof byte[]) {
tempFilter = lt(binaryColumn(path), Binary.fromReusedByteArray((byte[]) parquetObjects[i]));
} else {
fullyApplied = false;
LOGGER.warn(parquetObjects[i].getClass().getCanonicalName() + " is not a natively supported type for the IsLessThan filter, therefore execution will take longer to perform this filter.");
return null;
}
if (null == filter) {
filter = tempFilter;
} else {
filter = and(filter, tempFilter);
}
}
return filter;
}
Aggregations