fix

Author: alamb

parquet/src/arrow/array_reader/primitive_array.rs

@@ -23,7 +23,12 @@ use crate::column::page::PageIterator;
 use crate::data_type::{DataType, Int96};
 use crate::errors::Result;
 use crate::schema::types::ColumnDescPtr;
-use arrow_array::{Array, ArrayRef, Date64Array, Decimal64Array, Decimal128Array, Decimal256Array, Int8Array, Int16Array, Int32Array, Int64Array, PrimitiveArray, UInt8Array, UInt16Array, builder::PrimitiveDictionaryBuilder, cast::AsArray, downcast_integer,  types::*, BooleanArray, Decimal32Array, Float64Array, Float32Array};
+use arrow_array::{
+    Array, ArrayRef, BooleanArray, Date64Array, Decimal64Array, Decimal128Array,
+    Decimal256Array, Float32Array, Float64Array, Int8Array, Int16Array, Int32Array, Int64Array,
+    PrimitiveArray, UInt8Array, UInt16Array, builder::PrimitiveDictionaryBuilder, cast::AsArray,
+    downcast_integer, types::*,
+};
 use arrow_array::{
     TimestampMicrosecondArray, TimestampMillisecondArray, TimestampNanosecondArray,
     TimestampSecondArray, UInt32Array, UInt64Array,
@@ -146,19 +151,9 @@ where
 
     fn consume_batch(&mut self) -> Result<ArrayRef> {
         let target_type = &self.data_type;
-        let arrow_data_type = match T::get_physical_type() {
-            PhysicalType::BOOLEAN => ArrowType::Boolean,
-            PhysicalType::INT32 => ArrowType::Int32,
-            PhysicalType::INT64 => ArrowType::Int64,
-            PhysicalType::FLOAT => ArrowType::Float32,
-            PhysicalType::DOUBLE => ArrowType::Float64,
-            PhysicalType::INT96 => ArrowType::Int64,
-            PhysicalType::BYTE_ARRAY | PhysicalType::FIXED_LEN_BYTE_ARRAY => {
-                unreachable!("PrimitiveArrayReaders don't support complex physical types");
-            }
-        };
 
-        // Convert to equivalent arrow type to parquet physical type
+        // Convert physical data to equivalent arrow type, and then perform
+        // coercion as needed
         let record_data = self
             .record_reader
             .consume_record_data()
@@ -175,36 +170,14 @@ where
                 BooleanBuffer::new(record_data, 0, len),
                 nulls,
             )),
-            PhysicalType::INT32 => match arrow_data_type {
-                ArrowType::UInt32 => Arc::new(UInt32Array::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                ArrowType::Int32 => Arc::new(Int32Array::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                ArrowType::Decimal32(_, _) => Arc::new(Decimal32Array::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                _ => unreachable!(),
-            },
-            PhysicalType::INT64 => match arrow_data_type {
-                ArrowType::UInt64 => Arc::new(UInt64Array::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                ArrowType::Int64 => Arc::new(Int64Array::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                ArrowType::Decimal64(_, _) => Arc::new(Decimal64Array::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                _ => unreachable!(),
-            },
+            PhysicalType::INT32 => Arc::new(Int32Array::new(
+                ScalarBuffer::new(record_data, 0, len),
+                nulls,
+            )),
+            PhysicalType::INT64 => Arc::new(Int64Array::new(
+                ScalarBuffer::new(record_data, 0, len),
+                nulls,
+            )),
             PhysicalType::FLOAT => Arc::new(Float32Array::new(
                 ScalarBuffer::new(record_data, 0, len),
                 nulls,
@@ -213,22 +186,10 @@ where
                 ScalarBuffer::new(record_data, 0, len),
                 nulls,
             )),
-            PhysicalType::INT96 => match target_type {
-                ArrowType::Timestamp(TimeUnit::Second, _) => Arc::new(TimestampSecondArray::new(
-                    ScalarBuffer::new(record_data, 0, len),
-                    nulls,
-                )),
-                ArrowType::Timestamp(TimeUnit::Millisecond, _) => Arc::new(
-                    TimestampMillisecondArray::new(ScalarBuffer::new(record_data, 0, len), nulls),
-                ),
-                ArrowType::Timestamp(TimeUnit::Microsecond, _) => Arc::new(
-                    TimestampMicrosecondArray::new(ScalarBuffer::new(record_data, 0, len), nulls),
-                ),
-                ArrowType::Timestamp(TimeUnit::Nanosecond, _) => Arc::new(
-                    TimestampNanosecondArray::new(ScalarBuffer::new(record_data, 0, len), nulls),
-                ),
-                _ => unreachable!("INT96 must be a timestamp."),
-            },
+            PhysicalType::INT96 => Arc::new(Int64Array::new(
+                ScalarBuffer::new(record_data, 0, len),
+                nulls,
+            )),
             PhysicalType::BYTE_ARRAY | PhysicalType::FIXED_LEN_BYTE_ARRAY => {
                 unreachable!("PrimitiveArrayReaders don't support complex physical types");
             }
@@ -275,7 +236,7 @@ fn coerce_array(array: ArrayRef, target_type: &ArrowType) -> Result<ArrayRef> {
         ArrowType::Int32 => coerce_i32(array.as_primitive(), target_type),
         ArrowType::Int64 => coerce_i64(array.as_primitive(), target_type),
         ArrowType::Boolean | ArrowType::Float32 | ArrowType::Float64 => Ok(array),
-        _ => unreachable!(),
+        _ => unreachable!("Cannot coerce array of type {}", array.data_type()),
     }
 }