Added API for Box<dyn Array>

src/io/orc/read/mod.rs

@@ -1,7 +1,7 @@
 //! APIs to read from [ORC format](https://orc.apache.org).
 use std::io::{Read, Seek, SeekFrom};
 
-use crate::array::{BooleanArray, Float32Array, Int32Array, Int64Array};
+use crate::array::{Array, BooleanArray, Float32Array, Int32Array, Int64Array};
 use crate::bitmap::{Bitmap, MutableBitmap};
 use crate::datatypes::{DataType, Field, Schema};
 use crate::error::Error;
@@ -102,7 +102,7 @@ fn deserialize_validity(
 }
 
 /// Deserializes column `column` from `stripe`, assumed to represent a f32
-pub fn deserialize_f32(
+fn deserialize_f32(
     data_type: DataType,
     stripe: &Stripe,
     column: usize,
@@ -138,7 +138,7 @@ pub fn deserialize_f32(
 }
 
 /// Deserializes column `column` from `stripe`, assumed to represent a boolean array
-pub fn deserialize_bool(
+fn deserialize_bool(
     data_type: DataType,
     stripe: &Stripe,
     column: usize,
@@ -219,7 +219,7 @@ impl<'a> Iterator for IntIter<'a> {
 }
 
 /// Deserializes column `column` from `stripe`, assumed to represent a boolean array
-pub fn deserialize_i64(
+fn deserialize_i64(
     data_type: DataType,
     stripe: &Stripe,
     column: usize,
@@ -266,7 +266,7 @@ pub fn deserialize_i64(
 }
 
 /// Deserializes column `column` from `stripe`, assumed to represent a boolean array
-pub fn deserialize_i32(
+fn deserialize_i32(
     data_type: DataType,
     stripe: &Stripe,
     column: usize,
@@ -317,3 +317,19 @@ pub fn deserialize_i32(
 
     Int32Array::try_new(data_type, values.into(), validity)
 }
+
+/// Deserializes column `column` from `stripe`, assumed
+/// to represent an array of `data_type`.
+pub fn deserialize(
+    data_type: DataType,
+    stripe: &Stripe,
+    column: usize,
+) -> Result<Box<dyn Array>, Error> {
+    match data_type {
+        DataType::Boolean => deserialize_bool(data_type, stripe, column).map(|x| x.boxed()),
+        DataType::Int32 => deserialize_i32(data_type, stripe, column).map(|x| x.boxed()),
+        DataType::Int64 => deserialize_i64(data_type, stripe, column).map(|x| x.boxed()),
+        DataType::Float32 => deserialize_f32(data_type, stripe, column).map(|x| x.boxed()),
+        dt => return Err(Error::nyi(format!("Reading {dt:?} from ORC"))),
+    }
+}

tests/it/io/orc/read.rs

@@ -19,16 +19,14 @@ fn float32() -> Result<(), Error> {
     let (ps, footer, _) = format::read::read_metadata(&mut reader)?;
     let stripe = read::read_stripe(&mut reader, footer.stripes[0].clone(), ps.compression())?;
 
-    let array = read::deserialize_f32(DataType::Float32, &stripe, 1)?;
     assert_eq!(
-        array,
-        Float32Array::from([Some(1.0), Some(2.0), None, Some(4.0), Some(5.0)])
+        read::deserialize(DataType::Float32, &stripe, 1)?,
+        Float32Array::from([Some(1.0), Some(2.0), None, Some(4.0), Some(5.0)]).boxed()
     );
 
-    let array = read::deserialize_f32(DataType::Float32, &stripe, 2)?;
     assert_eq!(
-        array,
-        Float32Array::from([Some(1.0), Some(2.0), Some(3.0), Some(4.0), Some(5.0)])
+        read::deserialize(DataType::Float32, &stripe, 2)?,
+        Float32Array::from([Some(1.0), Some(2.0), Some(3.0), Some(4.0), Some(5.0)]).boxed()
     );
     Ok(())
 }
@@ -39,16 +37,14 @@ fn boolean() -> Result<(), Error> {
     let (ps, footer, _) = format::read::read_metadata(&mut reader)?;
     let stripe = read::read_stripe(&mut reader, footer.stripes[0].clone(), ps.compression())?;
 
-    let array = read::deserialize_bool(DataType::Boolean, &stripe, 3)?;
     assert_eq!(
-        array,
-        BooleanArray::from([Some(true), Some(false), None, Some(true), Some(false)])
+        read::deserialize(DataType::Boolean, &stripe, 3)?,
+        BooleanArray::from([Some(true), Some(false), None, Some(true), Some(false)]).boxed()
     );
 
-    let array = read::deserialize_bool(DataType::Boolean, &stripe, 4)?;
     assert_eq!(
-        array,
-        BooleanArray::from([Some(true), Some(false), Some(true), Some(true), Some(false)])
+        read::deserialize(DataType::Boolean, &stripe, 4)?,
+        BooleanArray::from([Some(true), Some(false), Some(true), Some(true), Some(false)]).boxed()
     );
     Ok(())
 }
@@ -59,16 +55,14 @@ fn int() -> Result<(), Error> {
     let (ps, footer, _) = format::read::read_metadata(&mut reader)?;
     let stripe = read::read_stripe(&mut reader, footer.stripes[0].clone(), ps.compression())?;
 
-    let array = read::deserialize_i32(DataType::Int32, &stripe, 5)?;
     assert_eq!(
-        array,
-        Int32Array::from([Some(5), Some(-5), None, Some(5), Some(5)])
+        read::deserialize(DataType::Int32, &stripe, 5)?,
+        Int32Array::from([Some(5), Some(-5), None, Some(5), Some(5)]).boxed()
     );
 
-    let array = read::deserialize_i32(DataType::Int32, &stripe, 6)?;
     assert_eq!(
-        array,
-        Int32Array::from([Some(5), Some(-5), Some(1), Some(5), Some(5)])
+        read::deserialize(DataType::Int32, &stripe, 6)?,
+        Int32Array::from([Some(5), Some(-5), Some(1), Some(5), Some(5)]).boxed()
     );
     Ok(())
 }