Skip to content

Use timestamp statistics #1999

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
ptkool wants to merge 2 commits into from
Closed

Use timestamp statistics #1999

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
370cd61
Use timestamp statistics
ptkool Oct 21, 2019
e912245
Create timestamp statistics based on original type
ptkool Nov 17, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 20 additions & 0 deletions presto-parquet/src/main/java/io/prestosql/parquet/predicate/TupleDomainParquetPredicate.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,11 +52,14 @@
import static io.prestosql.spi.type.IntegerType.INTEGER;
import static io.prestosql.spi.type.RealType.REAL;
import static io.prestosql.spi.type.SmallintType.SMALLINT;
import static io.prestosql.spi.type.TimestampType.TIMESTAMP;
import static io.prestosql.spi.type.TinyintType.TINYINT;
import static io.prestosql.spi.type.Varchars.isVarcharType;
import static java.lang.Float.floatToRawIntBits;
import static java.lang.String.format;
import static java.util.Objects.requireNonNull;
import static java.util.concurrent.TimeUnit.MICROSECONDS;
import static org.apache.parquet.schema.OriginalType.TIMESTAMP_MICROS;

public class TupleDomainParquetPredicate
implements Predicate
Expand Down Expand Up @@ -229,6 +232,23 @@ public static Domain getDomain(Type type, long rowCount, Statistics<?> statistic
return createDomain(type, hasNullValue, parquetIntegerStatistics);
}

if (type.equals(TIMESTAMP) && statistics instanceof LongStatistics) {
LongStatistics longStatistics = (LongStatistics) statistics;
if (longStatistics.genericGetMin() > longStatistics.genericGetMax()) {
failWithCorruptionException(failOnCorruptedParquetStatistics, column, id, longStatistics);
return Domain.create(ValueSet.all(type), hasNullValue);
}

long min = longStatistics.getMin();
long max = longStatistics.getMax();

ParquetIntegerStatistics parquetIntegerStatistics =
statistics.type().getOriginalType() == TIMESTAMP_MICROS
? new ParquetIntegerStatistics(MICROSECONDS.toMillis(min), MICROSECONDS.toMillis(max))
: new ParquetIntegerStatistics(min, max);
Copy link
Member

@ryanrupp ryanrupp Dec 2, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For TIMESTAMP_MILLIS I think do an explicit check here on original type because it looks like timestamp nanosecond support was added in apache/parquet-java#519 and it's using the new LogicalTypeAnnotation which will replace OriginalType in the next release of parquet-mr (see apache/parquet-java#463). So to be forward compatible if the check is explicit here it won't accidentally go into this path if a newer file is using nanoseconds.

return createDomain(type, hasNullValue, parquetIntegerStatistics);
}

return Domain.create(ValueSet.all(type), hasNullValue);
}

Expand Down
38 changes: 38 additions & 0 deletions presto-parquet/src/test/java/io/prestosql/parquet/TestTupleDomainParquetPredicate.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,6 +32,7 @@
import org.apache.parquet.column.statistics.LongStatistics;
import org.apache.parquet.column.statistics.Statistics;
import org.apache.parquet.io.api.Binary;
import org.apache.parquet.schema.OriginalType;
import org.apache.parquet.schema.PrimitiveType;
import org.testng.annotations.DataProvider;
import org.testng.annotations.Test;
Expand All @@ -56,6 +57,7 @@
import static io.prestosql.spi.type.IntegerType.INTEGER;
import static io.prestosql.spi.type.RealType.REAL;
import static io.prestosql.spi.type.SmallintType.SMALLINT;
import static io.prestosql.spi.type.TimestampType.TIMESTAMP;
import static io.prestosql.spi.type.TinyintType.TINYINT;
import static io.prestosql.spi.type.VarcharType.createUnboundedVarcharType;
import static io.prestosql.spi.type.VarcharType.createVarcharType;
Expand All @@ -68,6 +70,7 @@
import static org.apache.parquet.schema.PrimitiveType.PrimitiveTypeName.INT32;
import static org.apache.parquet.schema.PrimitiveType.PrimitiveTypeName.INT64;
import static org.apache.parquet.schema.Type.Repetition.OPTIONAL;
import static org.apache.parquet.schema.Type.Repetition.REQUIRED;
import static org.assertj.core.api.Assertions.assertThatExceptionOfType;
import static org.testng.Assert.assertEquals;
import static org.testng.Assert.assertFalse;
Expand Down Expand Up @@ -271,6 +274,34 @@ public void testDate()
.withMessage("Corrupted statistics for column \"DateColumn\" in Parquet file \"testFile\": [min: 200, max: 100, num_nulls: 0]");
}

@Test
public void testTimestamp()
throws ParquetCorruptionException
{
String column = "TimestampColumn";
assertEquals(getDomain(TIMESTAMP, 0, null, ID, column, true), all(TIMESTAMP));
assertEquals(getDomain(TIMESTAMP, 10, timestampColumnStats(1574026172386L, 1574026172386L, OriginalType.TIMESTAMP_MILLIS), ID, column, true), singleValue(TIMESTAMP, 1574026172386L));
assertEquals(getDomain(TIMESTAMP, 10, timestampColumnStats(1546318800000L, 1574026172386L, OriginalType.TIMESTAMP_MILLIS), ID, column, true), create(ValueSet.ofRanges(range(TIMESTAMP, 1546318800000L, true, 1574026172386L, true)), false));

// ignore corrupted statistics
assertEquals(getDomain(TIMESTAMP, 10, timestampColumnStats(1574026172386L, 1546318800000L, OriginalType.TIMESTAMP_MILLIS), ID, column, false), create(ValueSet.all(TIMESTAMP), false));
// fail on corrupted statistics
assertThatExceptionOfType(ParquetCorruptionException.class)
.isThrownBy(() -> getDomain(TIMESTAMP, 10, timestampColumnStats(1574026172386L, 1546318800000L, OriginalType.TIMESTAMP_MILLIS), ID, column, true))
.withMessage("Corrupted statistics for column \"TimestampColumn\" in Parquet file \"testFile\": [min: 2019-11-17T21:29:32.386, max: 2019-01-01T05:00:00.000, num_nulls: 0]");

assertEquals(getDomain(TIMESTAMP, 0, null, ID, column, true), all(TIMESTAMP));
assertEquals(getDomain(TIMESTAMP, 10, timestampColumnStats(1574026172386123L, 1574026172386123L, OriginalType.TIMESTAMP_MICROS), ID, column, true), singleValue(TIMESTAMP, 1574026172386L));
assertEquals(getDomain(TIMESTAMP, 10, timestampColumnStats(1546318800000123L, 1574026172386123L, OriginalType.TIMESTAMP_MICROS), ID, column, true), create(ValueSet.ofRanges(range(TIMESTAMP, 1546318800000L, true, 1574026172386L, true)), false));

// ignore corrupted statistics
assertEquals(getDomain(TIMESTAMP, 10, timestampColumnStats(1574026172386000L, 1546318800000000L, OriginalType.TIMESTAMP_MICROS), ID, column, false), create(ValueSet.all(TIMESTAMP), false));
// fail on corrupted statistics
assertThatExceptionOfType(ParquetCorruptionException.class)
.isThrownBy(() -> getDomain(TIMESTAMP, 10, timestampColumnStats(1574026172386123L, 1546318800000123L, OriginalType.TIMESTAMP_MICROS), ID, column, true))
.withMessage("Corrupted statistics for column \"TimestampColumn\" in Parquet file \"testFile\": [min: 2019-11-17T21:29:32.386123, max: 2019-01-01T05:00:00.000123, num_nulls: 0]");
}

@Test
public void testVarcharMatchesWithStatistics()
throws ParquetCorruptionException
Expand Down Expand Up @@ -369,4 +400,11 @@ private static LongStatistics longColumnStats(long minimum, long maximum)
statistics.setMinMax(minimum, maximum);
return statistics;
}

private static LongStatistics timestampColumnStats(long minimum, long maximum, OriginalType type)
{
LongStatistics statistics = (LongStatistics) Statistics.createStats(new PrimitiveType(REQUIRED, INT64, "timestampColumn", type));
statistics.setMinMax(minimum, maximum);
return statistics;
}
}