Skip to content

feat(parquet): Allow reading a wider integer as a narrower one #15343

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+82 −43
Open

feat(parquet): Allow reading a wider integer as a narrower one #15343

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
54 changes: 11 additions & 43 deletions velox/dwio/parquet/reader/ParquetReader.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -759,6 +759,11 @@ TypePtr ReaderBase::convertType(
"Converted type {} is not allowed for requested type {}";
const bool isRepeated = schemaElement.__isset.repetition_type &&
schemaElement.repetition_type == thrift::FieldRepetitionType::REPEATED;
const auto isInteger = [](const TypePtr& type) {
return type->kind() == TypeKind::TINYINT ||
type->kind() == TypeKind::SMALLINT ||
type->kind() == TypeKind::INTEGER || type->kind() == TypeKind::BIGINT;
};
if (schemaElement.__isset.converted_type) {
switch (schemaElement.converted_type) {
case thrift::ConvertedType::INT_8:
Expand All @@ -770,15 +775,7 @@ TypePtr ReaderBase::convertType(
schemaElement.converted_type);
VELOX_CHECK(
!requestedType ||
isCompatible(
requestedType,
isRepeated,
[](const TypePtr& type) {
return type->kind() == TypeKind::TINYINT ||
type->kind() == TypeKind::SMALLINT ||
type->kind() == TypeKind::INTEGER ||
type->kind() == TypeKind::BIGINT;
}),
isCompatible(requestedType, isRepeated, isInteger),
kTypeMappingErrorFmtStr,
"TINYINT",
requestedType->toString());
Expand All @@ -793,14 +790,7 @@ TypePtr ReaderBase::convertType(
schemaElement.converted_type);
VELOX_CHECK(
!requestedType ||
isCompatible(
requestedType,
isRepeated,
[](const TypePtr& type) {
return type->kind() == TypeKind::SMALLINT ||
type->kind() == TypeKind::INTEGER ||
type->kind() == TypeKind::BIGINT;
}),
isCompatible(requestedType, isRepeated, isInteger),
kTypeMappingErrorFmtStr,
"SMALLINT",
requestedType->toString());
Expand All @@ -815,13 +805,7 @@ TypePtr ReaderBase::convertType(
schemaElement.converted_type);
VELOX_CHECK(
!requestedType ||
isCompatible(
requestedType,
isRepeated,
[](const TypePtr& type) {
return type->kind() == TypeKind::INTEGER ||
type->kind() == TypeKind::BIGINT;
}),
isCompatible(requestedType, isRepeated, isInteger),
kTypeMappingErrorFmtStr,
"INTEGER",
requestedType->toString());
Expand All @@ -836,12 +820,7 @@ TypePtr ReaderBase::convertType(
schemaElement.converted_type);
VELOX_CHECK(
!requestedType ||
isCompatible(
requestedType,
isRepeated,
[](const TypePtr& type) {
return type->kind() == TypeKind::BIGINT;
}),
isCompatible(requestedType, isRepeated, isInteger),
kTypeMappingErrorFmtStr,
"BIGINT",
requestedType->toString());
Expand Down Expand Up @@ -1007,13 +986,7 @@ TypePtr ReaderBase::convertType(
case thrift::Type::type::INT32:
VELOX_CHECK(
!requestedType ||
isCompatible(
requestedType,
isRepeated,
[](const TypePtr& type) {
return type->kind() == TypeKind::INTEGER ||
type->kind() == TypeKind::BIGINT;
}),
isCompatible(requestedType, isRepeated, isInteger),
kTypeMappingErrorFmtStr,
"INTEGER",
requestedType->toString());
Expand All @@ -1037,12 +1010,7 @@ TypePtr ReaderBase::convertType(
}
VELOX_CHECK(
!requestedType ||
isCompatible(
requestedType,
isRepeated,
[](const TypePtr& type) {
return type->kind() == TypeKind::BIGINT;
}),
isCompatible(requestedType, isRepeated, isInteger),
kTypeMappingErrorFmtStr,
"BIGINT",
requestedType->toString());
Expand Down
71 changes: 71 additions & 0 deletions velox/dwio/parquet/tests/reader/ParquetTableScanTest.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1400,6 +1400,77 @@ TEST_F(ParquetTableScanTest, intToBigintRead) {
assertEqualVectors(bigintDataFileVectors->childAt(0), rows->childAt(0));
}

TEST_F(ParquetTableScanTest, intReadWithNarrowerType) {
// Reading a wider integer as a narrower one causes unchecked truncation and
// two’s complement reinterpretation, resulting in values INT_MAX becoming -1.
RowVectorPtr intVectors = makeRowVector(
{"c1", "c2", "c3"},
{
makeFlatVector<int16_t>(
{123,
std::numeric_limits<int8_t>::max(),
std::numeric_limits<int8_t>::min(),
std::numeric_limits<int16_t>::max(),
std::numeric_limits<int16_t>::min()}),
makeFlatVector<int32_t>(
{123,
std::numeric_limits<int16_t>::max(),
std::numeric_limits<int16_t>::min(),
std::numeric_limits<int32_t>::max(),
std::numeric_limits<int32_t>::min()}),
makeFlatVector<int64_t>(
{123,
std::numeric_limits<int32_t>::max(),
std::numeric_limits<int32_t>::min(),
std::numeric_limits<int64_t>::max(),
std::numeric_limits<int64_t>::min()}),
});

RowVectorPtr smallerIntVectors = makeRowVector(
{"c1", "c2", "c3"},
{
makeFlatVector<int8_t>({
123,
std::numeric_limits<int8_t>::max(),
std::numeric_limits<int8_t>::min(),
-1,
0,
}),
makeFlatVector<int16_t>({{
123,
std::numeric_limits<int16_t>::max(),
std::numeric_limits<int16_t>::min(),
-1,
0,
}}),
makeFlatVector<int32_t>({
123,
std::numeric_limits<int32_t>::max(),
std::numeric_limits<int32_t>::min(),
-1,
0,
}),
});

auto dataFile = TempFilePath::create();
WriterOptions options;
writeToParquetFile(dataFile->getPath(), {intVectors}, options);

auto rowType = ROW({"c1", "c2", "c3"}, {TINYINT(), SMALLINT(), INTEGER()});
auto op = PlanBuilder()
.startTableScan()
.outputType(rowType)
.dataColumns(rowType)
.endTableScan()
.planNode();

auto split = makeSplit(dataFile->getPath());
auto result = AssertQueryBuilder(op).split(split).copyResults(pool());
auto rows = result->as<RowVector>();

assertEqualVectors(smallerIntVectors->childAt(0), rows->childAt(0));
}

TEST_F(ParquetTableScanTest, shortAndLongDecimalReadWithLargerPrecision) {
// decimal.parquet holds two columns (a: DECIMAL(5, 2), b: DECIMAL(20, 5)) and
// 20 rows (10 rows per group). Data is in plain uncompressed format:
Expand Down
Loading