|
| 1 | +use arrow::array::{Int64Array, RecordBatch, StructArray}; |
| 2 | +use arrow::datatypes::{DataType, Field, Fields, Schema}; |
| 3 | +use datafusion::execution::context::SessionContext; |
| 4 | +use datafusion::logical_expr::{cast, col}; |
| 5 | +use std::sync::Arc; |
| 6 | + |
| 7 | +#[tokio::main] |
| 8 | +async fn main() -> Result<(), Box<dyn std::error::Error>> { |
| 9 | + let ctx = SessionContext::new(); |
| 10 | + |
| 11 | + // Source: struct with fields [b=3, a=4] |
| 12 | + let source_fields = Fields::from(vec![ |
| 13 | + Field::new("b", DataType::Int64, false), |
| 14 | + Field::new("a", DataType::Int64, false), |
| 15 | + ]); |
| 16 | + |
| 17 | + let source_struct = StructArray::new( |
| 18 | + source_fields.clone(), |
| 19 | + vec![ |
| 20 | + Arc::new(Int64Array::from(vec![3i64])), // b = 3 |
| 21 | + Arc::new(Int64Array::from(vec![4i64])), // a = 4 |
| 22 | + ], |
| 23 | + None, |
| 24 | + ); |
| 25 | + |
| 26 | + let batch = RecordBatch::try_new( |
| 27 | + Arc::new(Schema::new(vec![Field::new( |
| 28 | + "s", |
| 29 | + DataType::Struct(source_fields), |
| 30 | + false, |
| 31 | + )])), |
| 32 | + vec![Arc::new(source_struct)], |
| 33 | + )?; |
| 34 | + |
| 35 | + let table = datafusion::datasource::memory::MemTable::try_new( |
| 36 | + batch.schema(), |
| 37 | + vec![vec![batch]], |
| 38 | + )?; |
| 39 | + |
| 40 | + ctx.register_table("t", Arc::new(table))?; |
| 41 | + |
| 42 | + // Validate source data: should be b=3, a=4 |
| 43 | + let source_data = ctx.table("t").await?.collect().await?; |
| 44 | + use arrow::array::AsArray; |
| 45 | + let src_struct = source_data[0].column(0).as_struct(); |
| 46 | + let src_a = src_struct |
| 47 | + .column_by_name("a") |
| 48 | + .unwrap() |
| 49 | + .as_primitive::<arrow::array::types::Int64Type>() |
| 50 | + .value(0); |
| 51 | + let src_b = src_struct |
| 52 | + .column_by_name("b") |
| 53 | + .unwrap() |
| 54 | + .as_primitive::<arrow::array::types::Int64Type>() |
| 55 | + .value(0); |
| 56 | + assert_eq!(src_a, 4, "Source field 'a' should be 4"); |
| 57 | + assert_eq!(src_b, 3, "Source field 'b' should be 3"); |
| 58 | + println!("✓ Source validation passed: b={}, a={}", src_b, src_a); |
| 59 | + |
| 60 | + // Target: reorder fields to [a, b] |
| 61 | + let target_type = DataType::Struct(Fields::from(vec![ |
| 62 | + Field::new("a", DataType::Int64, false), |
| 63 | + Field::new("b", DataType::Int64, false), |
| 64 | + ])); |
| 65 | + |
| 66 | + // Execute cast |
| 67 | + let result = ctx |
| 68 | + .table("t") |
| 69 | + .await? |
| 70 | + .select(vec![cast(col("s"), target_type)])? |
| 71 | + .collect() |
| 72 | + .await?; |
| 73 | + |
| 74 | + // Validate result |
| 75 | + let res_struct = result[0].column(0).as_struct(); |
| 76 | + let res_a = res_struct |
| 77 | + .column_by_name("a") |
| 78 | + .unwrap() |
| 79 | + .as_primitive::<arrow::array::types::Int64Type>() |
| 80 | + .value(0); |
| 81 | + let res_b = res_struct |
| 82 | + .column_by_name("b") |
| 83 | + .unwrap() |
| 84 | + .as_primitive::<arrow::array::types::Int64Type>() |
| 85 | + .value(0); |
| 86 | + |
| 87 | + if res_a == 4 && res_b == 3 { |
| 88 | + println!("✓ Cast result passed: a={}, b={}", res_a, res_b); |
| 89 | + } else { |
| 90 | + println!( |
| 91 | + "✗ Bug: Cast maps by position, not name. Expected a=4,b=3 but got a={}, b={}", |
| 92 | + res_a, res_b |
| 93 | + ); |
| 94 | + } |
| 95 | + |
| 96 | + Ok(()) |
| 97 | +} |
0 commit comments