Skip to content
Open
Show file tree
Hide file tree
Changes from 1 commit
Commits
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
18 changes: 11 additions & 7 deletions datafusion/substrait/src/logical_plan/consumer/types.rs
Original file line number Diff line number Diff line change
Expand Up @@ -24,13 +24,13 @@ use crate::variation_const::{
DEFAULT_CONTAINER_TYPE_VARIATION_REF, DEFAULT_INTERVAL_DAY_TYPE_VARIATION_REF,
DEFAULT_MAP_TYPE_VARIATION_REF, DEFAULT_TYPE_VARIATION_REF,
DICTIONARY_MAP_TYPE_VARIATION_REF, DURATION_INTERVAL_DAY_TYPE_VARIATION_REF,
INTERVAL_DAY_TIME_TYPE_REF, INTERVAL_MONTH_DAY_NANO_TYPE_NAME,
INTERVAL_MONTH_DAY_NANO_TYPE_REF, INTERVAL_YEAR_MONTH_TYPE_REF,
LARGE_CONTAINER_TYPE_VARIATION_REF, TIME_32_TYPE_VARIATION_REF,
TIME_64_TYPE_VARIATION_REF, TIMESTAMP_MICRO_TYPE_VARIATION_REF,
TIMESTAMP_MILLI_TYPE_VARIATION_REF, TIMESTAMP_NANO_TYPE_VARIATION_REF,
TIMESTAMP_SECOND_TYPE_VARIATION_REF, UNSIGNED_INTEGER_TYPE_VARIATION_REF,
VIEW_CONTAINER_TYPE_VARIATION_REF,
FIXED_SIZE_LIST_TYPE_VARIATION_REF, INTERVAL_DAY_TIME_TYPE_REF,
INTERVAL_MONTH_DAY_NANO_TYPE_NAME, INTERVAL_MONTH_DAY_NANO_TYPE_REF,
INTERVAL_YEAR_MONTH_TYPE_REF, LARGE_CONTAINER_TYPE_VARIATION_REF,
TIME_32_TYPE_VARIATION_REF, TIME_64_TYPE_VARIATION_REF,
TIMESTAMP_MICRO_TYPE_VARIATION_REF, TIMESTAMP_MILLI_TYPE_VARIATION_REF,
TIMESTAMP_NANO_TYPE_VARIATION_REF, TIMESTAMP_SECOND_TYPE_VARIATION_REF,
UNSIGNED_INTEGER_TYPE_VARIATION_REF, VIEW_CONTAINER_TYPE_VARIATION_REF,
};
use crate::variation_const::{FLOAT_16_TYPE_NAME, NULL_TYPE_NAME};
use datafusion::arrow::datatypes::{
Expand Down Expand Up @@ -167,6 +167,10 @@ pub fn from_substrait_type(
match list.type_variation_reference {
DEFAULT_CONTAINER_TYPE_VARIATION_REF => Ok(DataType::List(field)),
LARGE_CONTAINER_TYPE_VARIATION_REF => Ok(DataType::LargeList(field)),
v if v >= FIXED_SIZE_LIST_TYPE_VARIATION_REF => {
let size = (v - FIXED_SIZE_LIST_TYPE_VARIATION_REF) as i32;
Ok(DataType::FixedSizeList(field, size))
}
v => not_impl_err!(
"Unsupported Substrait type variation {v} of type {s_kind:?}"
)?,
Expand Down
23 changes: 20 additions & 3 deletions datafusion/substrait/src/logical_plan/producer/types.rs
Original file line number Diff line number Diff line change
Expand Up @@ -23,9 +23,10 @@ use crate::variation_const::{
DEFAULT_CONTAINER_TYPE_VARIATION_REF, DEFAULT_INTERVAL_DAY_TYPE_VARIATION_REF,
DEFAULT_MAP_TYPE_VARIATION_REF, DEFAULT_TYPE_VARIATION_REF,
DICTIONARY_MAP_TYPE_VARIATION_REF, DURATION_INTERVAL_DAY_TYPE_VARIATION_REF,
FLOAT_16_TYPE_NAME, LARGE_CONTAINER_TYPE_VARIATION_REF, NULL_TYPE_NAME,
TIME_32_TYPE_VARIATION_REF, TIME_64_TYPE_VARIATION_REF,
UNSIGNED_INTEGER_TYPE_VARIATION_REF, VIEW_CONTAINER_TYPE_VARIATION_REF,
FIXED_SIZE_LIST_TYPE_VARIATION_REF, FLOAT_16_TYPE_NAME,
LARGE_CONTAINER_TYPE_VARIATION_REF, NULL_TYPE_NAME, TIME_32_TYPE_VARIATION_REF,
TIME_64_TYPE_VARIATION_REF, UNSIGNED_INTEGER_TYPE_VARIATION_REF,
VIEW_CONTAINER_TYPE_VARIATION_REF,
};
use datafusion::arrow::datatypes::{DataType, IntervalUnit};
use datafusion::common::{DFSchemaRef, not_impl_err, plan_err};
Expand Down Expand Up @@ -286,6 +287,18 @@ pub(crate) fn to_substrait_type(
}))),
})
}
DataType::FixedSizeList(inner, size) => {
let inner_type =
to_substrait_type(producer, inner.data_type(), inner.is_nullable())?;
Ok(substrait::proto::Type {
kind: Some(r#type::Kind::List(Box::new(r#type::List {
r#type: Some(Box::new(inner_type)),
type_variation_reference: FIXED_SIZE_LIST_TYPE_VARIATION_REF
+ (*size as u32),
nullability,
}))),
})
}
DataType::Map(inner, _) => match inner.data_type() {
DataType::Struct(key_and_value) if key_and_value.len() == 2 => {
let key_type = to_substrait_type(
Expand Down Expand Up @@ -439,6 +452,10 @@ mod tests {
round_trip_type(DataType::LargeList(
Field::new_list_field(DataType::Int32, true).into(),
))?;
round_trip_type(DataType::FixedSizeList(
Field::new_list_field(DataType::Int64, true).into(),
10,
))?;

round_trip_type(DataType::Map(
Field::new_struct(
Expand Down
7 changes: 7 additions & 0 deletions datafusion/substrait/src/variation_const.rs
Original file line number Diff line number Diff line change
Expand Up @@ -55,6 +55,13 @@ pub const TIME_64_TYPE_VARIATION_REF: u32 = 1;
pub const DEFAULT_CONTAINER_TYPE_VARIATION_REF: u32 = 0;
pub const LARGE_CONTAINER_TYPE_VARIATION_REF: u32 = 1;
pub const VIEW_CONTAINER_TYPE_VARIATION_REF: u32 = 2;
/// Used for the arrow type [`DataType::FixedSizeList`].
///
/// The size of the fixed list is encoded by adding it to this base value.
/// For example, `FixedSizeList(10)` would use variation reference `3 + 10 = 13`.
///
/// [`DataType::FixedSizeList`]: datafusion::arrow::datatypes::DataType::FixedSizeList
pub const FIXED_SIZE_LIST_TYPE_VARIATION_REF: u32 = 3;
Copy link
Contributor

Choose a reason for hiding this comment

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

This is a smart way of representing the size of a list!

However, I imagine this can easily backfire in the future if different type variation refs are needed for lists, as this approach eliminates the option to add new ones.

I see two possible paths forward:

  1. Convince the Substrait community that adding a new FixedSizeList type to the Substrait specification is worth it. There's an open issue about this Support FixedSizeList type substrait-io/substrait#873
  2. Implement this as a user defined type, same as what we did for Float16: https://github.com/apache/datafusion/blob/main/datafusion/substrait/src/logical_plan/producer/types.rs#L111-L120

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Hmm, I didn't think of that. Thanks for pointing that out.

pub const DEFAULT_MAP_TYPE_VARIATION_REF: u32 = 0;
pub const DICTIONARY_MAP_TYPE_VARIATION_REF: u32 = 1;
pub const DECIMAL_128_TYPE_VARIATION_REF: u32 = 0;
Expand Down