You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@arrow.apache.org by GitBox <gi...@apache.org> on 2021/01/25 07:36:10 UTC
[GitHub] [arrow] jorgecarleitao commented on a change in pull request #9305: ARROW-11362:[Rust][DataFusion] Use iterator APIs in to_array_of_size to improve performance
jorgecarleitao commented on a change in pull request #9305:
URL: https://github.com/apache/arrow/pull/9305#discussion_r563510962
##########
File path: rust/datafusion/src/scalar.rs
##########
@@ -205,28 +205,104 @@ impl ScalarValue {
ScalarValue::Boolean(e) => {
Arc::new(BooleanArray::from(vec![*e; size])) as ArrayRef
}
- ScalarValue::Float64(e) => {
- Arc::new(Float64Array::from(vec![*e; size])) as ArrayRef
- }
- ScalarValue::Float32(e) => Arc::new(Float32Array::from(vec![*e; size])),
- ScalarValue::Int8(e) => Arc::new(Int8Array::from(vec![*e; size])),
- ScalarValue::Int16(e) => Arc::new(Int16Array::from(vec![*e; size])),
- ScalarValue::Int32(e) => Arc::new(Int32Array::from(vec![*e; size])),
- ScalarValue::Int64(e) => Arc::new(Int64Array::from(vec![*e; size])),
- ScalarValue::UInt8(e) => Arc::new(UInt8Array::from(vec![*e; size])),
- ScalarValue::UInt16(e) => Arc::new(UInt16Array::from(vec![*e; size])),
- ScalarValue::UInt32(e) => Arc::new(UInt32Array::from(vec![*e; size])),
- ScalarValue::UInt64(e) => Arc::new(UInt64Array::from(vec![*e; size])),
- ScalarValue::TimeMicrosecond(e) => {
- Arc::new(TimestampMicrosecondArray::from(vec![*e]))
- }
- ScalarValue::TimeNanosecond(e) => {
- Arc::new(TimestampNanosecondArray::from_opt_vec(vec![*e], None))
- }
- ScalarValue::Utf8(e) => Arc::new(StringArray::from(vec![e.as_deref(); size])),
- ScalarValue::LargeUtf8(e) => {
- Arc::new(LargeStringArray::from(vec![e.as_deref(); size]))
- }
+ ScalarValue::Float64(e) => match e {
+ Some(value) => {
+ Arc::new(Float64Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<Float64Array>()),
+ },
+ ScalarValue::Float32(e) => match e {
+ Some(value) => {
+ Arc::new(Float32Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<Float32Array>()),
+ },
+ ScalarValue::Int8(e) => match e {
+ Some(value) => {
+ Arc::new(Int8Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<Int8Array>()),
+ },
+ ScalarValue::Int16(e) => match e {
+ Some(value) => {
+ Arc::new(Int16Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<Int16Array>()),
+ },
+ ScalarValue::Int32(e) => match e {
+ Some(value) => {
+ Arc::new(Int32Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<Int32Array>()),
+ },
+ ScalarValue::Int64(e) => match e {
+ Some(value) => {
+ Arc::new(Int64Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<Int64Array>()),
+ },
+ ScalarValue::UInt8(e) => match e {
+ Some(value) => {
+ Arc::new(UInt8Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<UInt8Array>()),
+ },
+ ScalarValue::UInt16(e) => match e {
+ Some(value) => {
+ Arc::new(UInt16Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<UInt16Array>()),
+ },
+ ScalarValue::UInt32(e) => match e {
+ Some(value) => {
+ Arc::new(UInt32Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<UInt32Array>()),
+ },
+ ScalarValue::UInt64(e) => match e {
+ Some(value) => {
+ Arc::new(UInt64Array::from_iter_values(repeat(*value).take(size)))
+ }
+ None => Arc::new(repeat(None).take(size).collect::<UInt64Array>()),
+ },
+ ScalarValue::TimeMicrosecond(e) => match e {
+ Some(value) => Arc::new(TimestampMicrosecondArray::from_iter_values(
+ repeat(*value).take(size),
+ )),
+ None => Arc::new(
+ repeat(None)
+ .take(size)
+ .collect::<TimestampMicrosecondArray>(),
+ ),
+ },
+ ScalarValue::TimeNanosecond(e) => match e {
+ Some(value) => Arc::new(TimestampNanosecondArray::from_iter_values(
+ repeat(*value).take(size),
+ )),
+ None => Arc::new(
+ repeat(None)
+ .take(size)
+ .collect::<TimestampNanosecondArray>(),
+ ),
+ },
+ ScalarValue::Utf8(e) => match e {
+ Some(value) => {
+ Arc::new(StringArray::from_iter_values(repeat(value).take(size)))
+ }
+ None => {
+ Arc::new(repeat(e.as_deref()).take(size).collect::<StringArray>())
Review comment:
any reason to use `e.as_deref` instead of `None`? Just curious.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org