use of io.trino.spi.type.SqlVarbinary in project trino by trinodb.
the class AbstractTestHive method assertGetRecords.
protected void assertGetRecords(HiveStorageFormat hiveStorageFormat, ConnectorTableMetadata tableMetadata, HiveSplit hiveSplit, ConnectorPageSource pageSource, List<? extends ColumnHandle> columnHandles) throws IOException {
try {
MaterializedResult result = materializeSourceDataStream(newSession(), pageSource, getTypes(columnHandles));
assertPageSourceType(pageSource, hiveStorageFormat);
ImmutableMap<String, Integer> columnIndex = indexColumns(tableMetadata);
long rowNumber = 0;
long completedBytes = 0;
for (MaterializedRow row : result) {
try {
assertValueTypes(row, tableMetadata.getColumns());
} catch (RuntimeException e) {
throw new RuntimeException("row " + rowNumber, e);
}
rowNumber++;
Integer index;
Object value;
// STRING
index = columnIndex.get("t_string");
value = row.getField(index);
if (rowNumber % 19 == 0) {
assertNull(value);
} else if (rowNumber % 19 == 1) {
assertEquals(value, "");
} else {
assertEquals(value, "test");
}
// NUMBERS
assertEquals(row.getField(columnIndex.get("t_tinyint")), (byte) (1 + rowNumber));
assertEquals(row.getField(columnIndex.get("t_smallint")), (short) (2 + rowNumber));
assertEquals(row.getField(columnIndex.get("t_int")), (int) (3 + rowNumber));
index = columnIndex.get("t_bigint");
if ((rowNumber % 13) == 0) {
assertNull(row.getField(index));
} else {
assertEquals(row.getField(index), 4 + rowNumber);
}
assertEquals((Float) row.getField(columnIndex.get("t_float")), 5.1f + rowNumber, 0.001);
assertEquals(row.getField(columnIndex.get("t_double")), 6.2 + rowNumber);
// BOOLEAN
index = columnIndex.get("t_boolean");
if ((rowNumber % 3) == 2) {
assertNull(row.getField(index));
} else {
assertEquals(row.getField(index), (rowNumber % 3) != 0);
}
// TIMESTAMP
index = columnIndex.get("t_timestamp");
if (index != null) {
if ((rowNumber % 17) == 0) {
assertNull(row.getField(index));
} else {
SqlTimestamp expected = sqlTimestampOf(3, 2011, 5, 6, 7, 8, 9, 123);
assertEquals(row.getField(index), expected);
}
}
// BINARY
index = columnIndex.get("t_binary");
if (index != null) {
if ((rowNumber % 23) == 0) {
assertNull(row.getField(index));
} else {
assertEquals(row.getField(index), new SqlVarbinary("test binary".getBytes(UTF_8)));
}
}
// DATE
index = columnIndex.get("t_date");
if (index != null) {
if ((rowNumber % 37) == 0) {
assertNull(row.getField(index));
} else {
SqlDate expected = new SqlDate(toIntExact(MILLISECONDS.toDays(new DateTime(2013, 8, 9, 0, 0, 0, UTC).getMillis())));
assertEquals(row.getField(index), expected);
}
}
// VARCHAR(50)
index = columnIndex.get("t_varchar");
if (index != null) {
value = row.getField(index);
if (rowNumber % 39 == 0) {
assertNull(value);
} else if (rowNumber % 39 == 1) {
// RCBINARY reads empty VARCHAR as null
if (hiveStorageFormat == RCBINARY) {
assertNull(value);
} else {
assertEquals(value, "");
}
} else {
assertEquals(value, "test varchar");
}
}
// CHAR(25)
index = columnIndex.get("t_char");
if (index != null) {
value = row.getField(index);
if ((rowNumber % 41) == 0) {
assertNull(value);
} else {
assertEquals(value, (rowNumber % 41) == 1 ? " " : "test char ");
}
}
// MAP<STRING, STRING>
index = columnIndex.get("t_map");
if (index != null) {
if ((rowNumber % 27) == 0) {
assertNull(row.getField(index));
} else {
assertEquals(row.getField(index), ImmutableMap.of("test key", "test value"));
}
}
// ARRAY<STRING>
index = columnIndex.get("t_array_string");
if (index != null) {
if ((rowNumber % 29) == 0) {
assertNull(row.getField(index));
} else {
assertEquals(row.getField(index), ImmutableList.of("abc", "xyz", "data"));
}
}
// ARRAY<TIMESTAMP>
index = columnIndex.get("t_array_timestamp");
if (index != null) {
if ((rowNumber % 43) == 0) {
assertNull(row.getField(index));
} else {
SqlTimestamp expected = sqlTimestampOf(3, LocalDateTime.of(2011, 5, 6, 7, 8, 9, 123_000_000));
assertEquals(row.getField(index), ImmutableList.of(expected));
}
}
// ARRAY<STRUCT<s_string: STRING, s_double:DOUBLE>>
index = columnIndex.get("t_array_struct");
if (index != null) {
if ((rowNumber % 31) == 0) {
assertNull(row.getField(index));
} else {
List<Object> expected1 = ImmutableList.of("test abc", 0.1);
List<Object> expected2 = ImmutableList.of("test xyz", 0.2);
assertEquals(row.getField(index), ImmutableList.of(expected1, expected2));
}
}
// STRUCT<s_string: STRING, s_double:DOUBLE>
index = columnIndex.get("t_struct");
if (index != null) {
if ((rowNumber % 31) == 0) {
assertNull(row.getField(index));
} else {
assertTrue(row.getField(index) instanceof List);
List<?> values = (List<?>) row.getField(index);
assertEquals(values.size(), 2);
assertEquals(values.get(0), "test abc");
assertEquals(values.get(1), 0.1);
}
}
// MAP<INT, ARRAY<STRUCT<s_string: STRING, s_double:DOUBLE>>>
index = columnIndex.get("t_complex");
if (index != null) {
if ((rowNumber % 33) == 0) {
assertNull(row.getField(index));
} else {
List<Object> expected1 = ImmutableList.of("test abc", 0.1);
List<Object> expected2 = ImmutableList.of("test xyz", 0.2);
assertEquals(row.getField(index), ImmutableMap.of(1, ImmutableList.of(expected1, expected2)));
}
}
// NEW COLUMN
assertNull(row.getField(columnIndex.get("new_column")));
long newCompletedBytes = pageSource.getCompletedBytes();
assertTrue(newCompletedBytes >= completedBytes);
// some formats (e.g., parquet) over read the data by a bit
assertLessThanOrEqual(newCompletedBytes, hiveSplit.getLength() + (100 * 1024));
completedBytes = newCompletedBytes;
}
assertLessThanOrEqual(completedBytes, hiveSplit.getLength() + (100 * 1024));
assertEquals(rowNumber, 100);
} finally {
pageSource.close();
}
}
use of io.trino.spi.type.SqlVarbinary in project trino by trinodb.
the class ParquetTester method writeValue.
private static void writeValue(Type type, BlockBuilder blockBuilder, Object value) {
if (value == null) {
blockBuilder.appendNull();
} else {
if (BOOLEAN.equals(type)) {
type.writeBoolean(blockBuilder, (Boolean) value);
} else if (TINYINT.equals(type) || SMALLINT.equals(type) || INTEGER.equals(type) || BIGINT.equals(type)) {
type.writeLong(blockBuilder, ((Number) value).longValue());
} else if (Decimals.isShortDecimal(type)) {
type.writeLong(blockBuilder, ((SqlDecimal) value).getUnscaledValue().longValue());
} else if (Decimals.isLongDecimal(type)) {
if (Decimals.overflows(((SqlDecimal) value).getUnscaledValue(), MAX_PRECISION_INT64)) {
type.writeObject(blockBuilder, Int128.valueOf(((SqlDecimal) value).toBigDecimal().unscaledValue()));
} else {
type.writeObject(blockBuilder, Int128.valueOf(((SqlDecimal) value).getUnscaledValue().longValue()));
}
} else if (DOUBLE.equals(type)) {
type.writeDouble(blockBuilder, ((Number) value).doubleValue());
} else if (REAL.equals(type)) {
float floatValue = ((Number) value).floatValue();
type.writeLong(blockBuilder, Float.floatToIntBits(floatValue));
} else if (type instanceof VarcharType) {
Slice slice = truncateToLength(utf8Slice((String) value), type);
type.writeSlice(blockBuilder, slice);
} else if (type instanceof CharType) {
Slice slice = truncateToLengthAndTrimSpaces(utf8Slice((String) value), type);
type.writeSlice(blockBuilder, slice);
} else if (VARBINARY.equals(type)) {
type.writeSlice(blockBuilder, Slices.wrappedBuffer(((SqlVarbinary) value).getBytes()));
} else if (DATE.equals(type)) {
long days = ((SqlDate) value).getDays();
type.writeLong(blockBuilder, days);
} else if (TIMESTAMP_MILLIS.equals(type)) {
type.writeLong(blockBuilder, ((SqlTimestamp) value).getEpochMicros());
} else {
if (type instanceof ArrayType) {
List<?> array = (List<?>) value;
Type elementType = type.getTypeParameters().get(0);
BlockBuilder arrayBlockBuilder = blockBuilder.beginBlockEntry();
for (Object elementValue : array) {
writeValue(elementType, arrayBlockBuilder, elementValue);
}
blockBuilder.closeEntry();
} else if (type instanceof MapType) {
Map<?, ?> map = (Map<?, ?>) value;
Type keyType = type.getTypeParameters().get(0);
Type valueType = type.getTypeParameters().get(1);
BlockBuilder mapBlockBuilder = blockBuilder.beginBlockEntry();
for (Map.Entry<?, ?> entry : map.entrySet()) {
writeValue(keyType, mapBlockBuilder, entry.getKey());
writeValue(valueType, mapBlockBuilder, entry.getValue());
}
blockBuilder.closeEntry();
} else if (type instanceof RowType) {
List<?> array = (List<?>) value;
List<Type> fieldTypes = type.getTypeParameters();
BlockBuilder rowBlockBuilder = blockBuilder.beginBlockEntry();
for (int fieldId = 0; fieldId < fieldTypes.size(); fieldId++) {
Type fieldType = fieldTypes.get(fieldId);
writeValue(fieldType, rowBlockBuilder, array.get(fieldId));
}
blockBuilder.closeEntry();
} else {
throw new IllegalArgumentException("Unsupported type " + type);
}
}
}
}
use of io.trino.spi.type.SqlVarbinary in project trino by trinodb.
the class TestQuantileDigestAggregationFunction method assertPercentilesWithinError.
private void assertPercentilesWithinError(String type, SqlVarbinary binary, double error, List<? extends Number> rows, double[] percentiles) {
List<Double> boxedPercentiles = Arrays.stream(percentiles).sorted().boxed().collect(toImmutableList());
List<Number> lowerBounds = boxedPercentiles.stream().map(percentile -> getLowerBound(error, rows, percentile)).collect(toImmutableList());
List<Number> upperBounds = boxedPercentiles.stream().map(percentile -> getUpperBound(error, rows, percentile)).collect(toImmutableList());
// Ensure that the lower bound of each item in the distribution is not greater than the chosen quantiles
functionAssertions.assertFunction(format("zip_with(values_at_quantiles(CAST(X'%s' AS qdigest(%s)), ARRAY[%s]), ARRAY[%s], (value, lowerbound) -> value >= lowerbound)", binary.toString().replaceAll("\\s+", " "), type, ARRAY_JOINER.join(boxedPercentiles), ARRAY_JOINER.join(lowerBounds)), new ArrayType(BOOLEAN), Collections.nCopies(percentiles.length, true));
// Ensure that the upper bound of each item in the distribution is not less than the chosen quantiles
functionAssertions.assertFunction(format("zip_with(values_at_quantiles(CAST(X'%s' AS qdigest(%s)), ARRAY[%s]), ARRAY[%s], (value, upperbound) -> value <= upperbound)", binary.toString().replaceAll("\\s+", " "), type, ARRAY_JOINER.join(boxedPercentiles), ARRAY_JOINER.join(upperBounds)), new ArrayType(BOOLEAN), Collections.nCopies(percentiles.length, true));
}
use of io.trino.spi.type.SqlVarbinary in project trino by trinodb.
the class TestQuantileDigestAggregationFunction method testAggregationDoubles.
private void testAggregationDoubles(List<TypeSignatureProvider> parameterTypes, Page page, double maxError, double... inputs) {
assertAggregation(FUNCTION_RESOLUTION, NAME, parameterTypes, QDIGEST_EQUALITY, "test multiple positions", page, getExpectedValueDoubles(maxError, inputs));
// test scalars
List<Double> rows = Arrays.stream(inputs).sorted().boxed().collect(Collectors.toList());
SqlVarbinary returned = (SqlVarbinary) AggregationTestUtils.aggregation(FUNCTION_RESOLUTION.getAggregateFunction(NAME, parameterTypes), page);
assertPercentileWithinError(StandardTypes.DOUBLE, returned, maxError, rows, 0.1, 0.5, 0.9, 0.99);
}
use of io.trino.spi.type.SqlVarbinary in project trino by trinodb.
the class TestQuantileDigestAggregationFunction method testAggregationReal.
private void testAggregationReal(List<TypeSignatureProvider> parameterTypes, Page page, double maxError, float... inputs) {
assertAggregation(FUNCTION_RESOLUTION, NAME, parameterTypes, QDIGEST_EQUALITY, "test multiple positions", page, getExpectedValuesFloats(maxError, inputs));
// test scalars
List<Double> rows = Floats.asList(inputs).stream().sorted().map(Float::doubleValue).collect(Collectors.toList());
SqlVarbinary returned = (SqlVarbinary) AggregationTestUtils.aggregation(FUNCTION_RESOLUTION.getAggregateFunction(NAME, parameterTypes), page);
assertPercentileWithinError(StandardTypes.REAL, returned, maxError, rows, 0.1, 0.5, 0.9, 0.99);
}
Aggregations