mod lex_sort; mod row; use re_arrow2::array::*; use re_arrow2::compute::sort::*; use re_arrow2::datatypes::*; use re_arrow2::types::NativeType; fn to_indices_boolean_arrays(data: &[Option], options: SortOptions, expected_data: &[i32]) { let output = BooleanArray::from(data); let expected = Int32Array::from_slice(expected_data); let output = sort_to_indices(&output, &options, None).unwrap(); assert_eq!(output, expected) } fn primitive_arrays( data: &[Option], data_type: DataType, options: SortOptions, expected_data: &[Option], ) where T: NativeType, { let input = PrimitiveArray::::from(data).to(data_type.clone()); let expected = PrimitiveArray::::from(expected_data).to(data_type); let output = sort(&input, &options, None).unwrap(); assert_eq!(expected, output.as_ref()) } fn to_indices_string_arrays(data: &[Option<&str>], options: SortOptions, expected_data: &[i32]) { let input = Utf8Array::::from(data); let expected = Int32Array::from_slice(expected_data); let output = sort_to_indices(&input, &options, None).unwrap(); assert_eq!(output, expected) } fn string_arrays(data: &[Option<&str>], options: SortOptions, expected_data: &[Option<&str>]) { let input = Utf8Array::::from(data); let expected = Utf8Array::::from(expected_data); let output = sort(&input, &options, None).unwrap(); assert_eq!(expected, output.as_ref()) } fn string_dict_arrays(data: &[Option<&str>], options: SortOptions, expected_data: &[Option<&str>]) { let mut input = MutableDictionaryArray::>::new(); input.try_extend(data.iter().copied()).unwrap(); let input = input.into_arc(); let mut expected = MutableDictionaryArray::>::new(); expected.try_extend(expected_data.iter().copied()).unwrap(); let expected = expected.into_arc(); let output = sort(input.as_ref(), &options, None).unwrap(); assert_eq!(expected.as_ref(), output.as_ref()) } /* fn list_arrays( data: Vec>>>, options: Option, expected_data: Vec>>>, fixed_length: Option, ) where T: ArrowPrimitiveType, PrimitiveArray: From>>, { // for FixedSizedList if let Some(length) = fixed_length { let input = Arc::new(build_fixed_size_list_nullable(data.clone(), length)); let sorted = sort(&(input as ArrayRef), options).unwrap(); let expected = Arc::new(build_fixed_size_list_nullable( expected_data.clone(), length, )) as ArrayRef; assert_eq!(&sorted, &expected); } // for List let input = Arc::new(build_generic_list_nullable::(data.clone())); let sorted = sort(&(input as ArrayRef), options).unwrap(); let expected = Arc::new(build_generic_list_nullable::(expected_data.clone())) as ArrayRef; assert_eq!(&sorted, &expected); // for LargeList let input = Arc::new(build_generic_list_nullable::(data)); let sorted = sort(&(input as ArrayRef), options).unwrap(); let expected = Arc::new(build_generic_list_nullable::(expected_data)) as ArrayRef; assert_eq!(&sorted, &expected); } fn test_lex_sort_arrays(input: Vec, expected_output: Vec) { let sorted = lexsort(&input).unwrap(); for (result, expected) in sorted.iter().zip(expected_output.iter()) { assert_eq!(result, expected); } } */ #[test] fn boolean() { // boolean to_indices_boolean_arrays( &[None, Some(false), Some(true), Some(true), Some(false), None], SortOptions { descending: false, nulls_first: true, }, &[0, 5, 1, 4, 2, 3], ); // boolean, descending to_indices_boolean_arrays( &[None, Some(false), Some(true), Some(true), Some(false), None], SortOptions { descending: true, nulls_first: false, }, &[2, 3, 1, 4, 5, 0], ); // boolean, descending, nulls first to_indices_boolean_arrays( &[None, Some(false), Some(true), Some(true), Some(false), None], SortOptions { descending: true, nulls_first: true, }, &[5, 0, 2, 3, 1, 4], ); } #[test] #[ignore] // improve equality for NaN values. These are right but the equality fails fn test_nans() { primitive_arrays::( &[None, Some(0.0), Some(2.0), Some(-1.0), Some(f64::NAN), None], DataType::Float64, SortOptions { descending: true, nulls_first: true, }, &[None, None, Some(f64::NAN), Some(2.0), Some(0.0), Some(-1.0)], ); primitive_arrays::( &[Some(f64::NAN), Some(f64::NAN), Some(f64::NAN), Some(1.0)], DataType::Float64, SortOptions { descending: true, nulls_first: true, }, &[Some(f64::NAN), Some(f64::NAN), Some(f64::NAN), Some(1.0)], ); primitive_arrays::( &[None, Some(0.0), Some(2.0), Some(-1.0), Some(f64::NAN), None], DataType::Float64, SortOptions { descending: false, nulls_first: true, }, &[None, None, Some(-1.0), Some(0.0), Some(2.0), Some(f64::NAN)], ); // nans primitive_arrays::( &[Some(f64::NAN), Some(f64::NAN), Some(f64::NAN), Some(1.0)], DataType::Float64, SortOptions { descending: false, nulls_first: true, }, &[Some(1.0), Some(f64::NAN), Some(f64::NAN), Some(f64::NAN)], ); } #[test] fn to_indices_strings() { to_indices_string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: false, nulls_first: true, }, // &[3, 0, 5, 1, 4, 2] is also valid &[0, 3, 5, 1, 4, 2], ); to_indices_string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: true, nulls_first: false, }, // &[2, 4, 1, 5, 3, 0] is also valid &[2, 4, 1, 5, 0, 3], ); to_indices_string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: false, nulls_first: true, }, // &[3, 0, 5, 1, 4, 2] is also valid &[0, 3, 5, 1, 4, 2], ); to_indices_string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: true, nulls_first: true, }, // &[3, 0, 2, 4, 1, 5] is also valid &[0, 3, 2, 4, 1, 5], ); } #[test] fn strings() { string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: false, nulls_first: true, }, &[ None, None, Some("-ad"), Some("bad"), Some("glad"), Some("sad"), ], ); string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: true, nulls_first: false, }, &[ Some("sad"), Some("glad"), Some("bad"), Some("-ad"), None, None, ], ); string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: false, nulls_first: true, }, &[ None, None, Some("-ad"), Some("bad"), Some("glad"), Some("sad"), ], ); string_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: true, nulls_first: true, }, &[ None, None, Some("sad"), Some("glad"), Some("bad"), Some("-ad"), ], ); } #[test] fn string_dicts() { string_dict_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: false, nulls_first: true, }, &[ None, None, Some("-ad"), Some("bad"), Some("glad"), Some("sad"), ], ); string_dict_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: true, nulls_first: false, }, &[ Some("sad"), Some("glad"), Some("bad"), Some("-ad"), None, None, ], ); string_dict_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: false, nulls_first: true, }, &[ None, None, Some("-ad"), Some("bad"), Some("glad"), Some("sad"), ], ); string_dict_arrays( &[ None, Some("bad"), Some("sad"), None, Some("glad"), Some("-ad"), ], SortOptions { descending: true, nulls_first: true, }, &[ None, None, Some("sad"), Some("glad"), Some("bad"), Some("-ad"), ], ); } /* #[test] fn list() { list_arrays::( vec![ Some(vec![Some(1)]), Some(vec![Some(4)]), Some(vec![Some(2)]), Some(vec![Some(3)]), ], Some(SortOptions { descending: false, nulls_first: false, }), vec![ Some(vec![Some(1)]), Some(vec![Some(2)]), Some(vec![Some(3)]), Some(vec![Some(4)]), ], Some(1), ); list_arrays::( vec![ Some(vec![Some(1), Some(0)]), Some(vec![Some(4), Some(3), Some(2), Some(1)]), Some(vec![Some(2), Some(3), Some(4)]), Some(vec![Some(3), Some(3), Some(3), Some(3)]), Some(vec![Some(1), Some(1)]), ], Some(SortOptions { descending: false, nulls_first: false, }), vec![ Some(vec![Some(1), Some(0)]), Some(vec![Some(1), Some(1)]), Some(vec![Some(2), Some(3), Some(4)]), Some(vec![Some(3), Some(3), Some(3), Some(3)]), Some(vec![Some(4), Some(3), Some(2), Some(1)]), ], None, ); list_arrays::( vec![ None, Some(vec![Some(4), None, Some(2)]), Some(vec![Some(2), Some(3), Some(4)]), None, Some(vec![Some(3), Some(3), None]), ], Some(SortOptions { descending: false, nulls_first: false, }), vec![ Some(vec![Some(2), Some(3), Some(4)]), Some(vec![Some(3), Some(3), None]), Some(vec![Some(4), None, Some(2)]), None, None, ], Some(3), ); } #[test] fn test_lex_sort_single_column() { let input = vec![SortColumn { values: Arc::new(PrimitiveArray::::from(vec![ Some(17), Some(2), Some(-1), Some(0), ])) as ArrayRef, options: None, }]; let expected = vec![Arc::new(PrimitiveArray::::from(vec![ Some(-1), Some(0), Some(2), Some(17), ])) as ArrayRef]; test_lex_sort_arrays(input, expected); } #[test] fn test_lex_sort_unaligned_rows() { let input = vec![ SortColumn { values: Arc::new(PrimitiveArray::::from(vec![None, Some(-1)])) as ArrayRef, options: None, }, SortColumn { values: Arc::new(StringArray::from(vec![Some("foo")])) as ArrayRef, options: None, }, ]; assert!( lexsort(&input).is_err(), "lexsort should reject columns with different row counts" ); } */ #[test] fn consistency() { use re_arrow2::array::new_null_array; use re_arrow2::datatypes::DataType::*; use re_arrow2::datatypes::TimeUnit; let datatypes = vec![ Null, Boolean, UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64, Timestamp(TimeUnit::Second, None), Timestamp(TimeUnit::Millisecond, None), Timestamp(TimeUnit::Microsecond, None), Timestamp(TimeUnit::Nanosecond, None), Time64(TimeUnit::Microsecond), Time64(TimeUnit::Nanosecond), Date32, Time32(TimeUnit::Second), Time32(TimeUnit::Millisecond), Date64, Utf8, LargeUtf8, Binary, LargeBinary, Duration(TimeUnit::Second), Duration(TimeUnit::Millisecond), Duration(TimeUnit::Microsecond), Duration(TimeUnit::Nanosecond), ]; datatypes.into_iter().for_each(|d1| { let array = new_null_array(d1.clone(), 10); let options = SortOptions { descending: true, nulls_first: true, }; if can_sort(&d1) { assert!(sort(array.as_ref(), &options, None).is_ok()); } else { assert!(sort(array.as_ref(), &options, None).is_err()); } }); }