Set Parquet isAdjustedToUTC to true for timestamps with non-empty tim…

…ezones
apache · Jun 18, 2019 · b5ebdbd · b5ebdbd
1 parent c5de420
commit b5ebdbd
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 12 deletions.
diff --git a/cpp/src/parquet/arrow/arrow-reader-writer-test.cc b/cpp/src/parquet/arrow/arrow-reader-writer-test.cc
@@ -115,7 +115,7 @@ std::shared_ptr<const LogicalAnnotation> get_logical_annotation(const ::DataType
  return LogicalAnnotation::Date();
  case ArrowId::TIMESTAMP: {
  const auto& ts_type = static_cast<const ::arrow::TimestampType&>(type);
- const bool adjusted_to_utc = (ts_type.timezone() == "UTC");
+ const bool adjusted_to_utc = !(ts_type.timezone().empty());
  switch (ts_type.unit()) {
  case TimeUnit::MILLI:
  return LogicalAnnotation::Timestamp(adjusted_to_utc,

diff --git a/cpp/src/parquet/arrow/arrow-schema-test.cc b/cpp/src/parquet/arrow/arrow-schema-test.cc
@@ -885,13 +885,13 @@ TEST_F(TestConvertArrowSchema, ArrowFields) {
  LogicalAnnotation::Timestamp(true, LogicalAnnotation::TimeUnit::NANOS),
  ParquetType::INT64, -1},
  {"timestamp(millisecond, CET)", ::arrow::timestamp(::arrow::TimeUnit::MILLI, "CET"),
- LogicalAnnotation::Timestamp(false, LogicalAnnotation::TimeUnit::MILLIS),
+ LogicalAnnotation::Timestamp(true, LogicalAnnotation::TimeUnit::MILLIS),
  ParquetType::INT64, -1},
  {"timestamp(microsecond, CET)", ::arrow::timestamp(::arrow::TimeUnit::MICRO, "CET"),
- LogicalAnnotation::Timestamp(false, LogicalAnnotation::TimeUnit::MICROS),
+ LogicalAnnotation::Timestamp(true, LogicalAnnotation::TimeUnit::MICROS),
  ParquetType::INT64, -1},
  {"timestamp(nanosecond, CET)", ::arrow::timestamp(::arrow::TimeUnit::NANO, "CET"),
- LogicalAnnotation::Timestamp(false, LogicalAnnotation::TimeUnit::NANOS),
+ LogicalAnnotation::Timestamp(true, LogicalAnnotation::TimeUnit::NANOS),
  ParquetType::INT64, -1},
  {"null", ::arrow::null(), LogicalAnnotation::Null(), ParquetType::INT32, -1}};
 

diff --git a/cpp/src/parquet/arrow/schema.cc b/cpp/src/parquet/arrow/schema.cc
@@ -135,7 +135,7 @@ static Status MakeArrowTime64(const std::shared_ptr<const LogicalAnnotation>& an
 static Status MakeArrowTimestamp(
  const std::shared_ptr<const LogicalAnnotation>& annotation,
  std::shared_ptr<ArrowType>* out) {
- static constexpr auto utc = "UTC";
+ static const char* utc = "UTC";
  const auto& timestamp = checked_cast<const TimestampAnnotation&>(*annotation);
  switch (timestamp.time_unit()) {
  case LogicalAnnotation::TimeUnit::MILLIS:
@@ -520,15 +520,9 @@ Status StructToNode(const std::shared_ptr<::arrow::StructType>& type,
  return Status::OK();
 }
 
-static bool HasUTCTimezone(const std::string& timezone) {
- static const std::vector<std::string> utczones{"UTC", "utc"};
- return std::any_of(utczones.begin(), utczones.end(),
- [timezone](const std::string& utc) { return timezone == utc; });
-}
-
 static std::shared_ptr<const LogicalAnnotation> TimestampAnnotationFromArrowTimestamp(
  const ::arrow::TimestampType& timestamp_type, ::arrow::TimeUnit::type time_unit) {
- const bool utc = HasUTCTimezone(timestamp_type.timezone());
+ const bool utc = !(timestamp_type.timezone().empty());
  switch (time_unit) {
  case ::arrow::TimeUnit::MILLI:
  return LogicalAnnotation::Timestamp(utc, LogicalAnnotation::TimeUnit::MILLIS);