1use std::any::Any;
18use std::sync::Arc;
19
20use arrow::array::{Array, ArrayRef, StringArray};
21use arrow::compute;
22use arrow::compute::kernels::comparison;
23use arrow::datatypes::{
24 DataType as ArrowDataType, Int8Type, Int16Type, Int32Type, Int64Type, TimeUnit, UInt8Type,
25 UInt16Type, UInt32Type, UInt64Type,
26};
27use arrow_array::{DictionaryArray, StructArray};
28use arrow_schema::IntervalUnit;
29use datafusion_common::ScalarValue;
30use snafu::{OptionExt, ResultExt};
31
32use crate::data_type::ConcreteDataType;
33use crate::error::{self, ConvertArrowArrayToScalarsSnafu, Result};
34use crate::prelude::DataType;
35use crate::scalars::{Scalar, ScalarVectorBuilder};
36use crate::types::StructType;
37use crate::value::{ListValue, ListValueRef, Value};
38use crate::vectors::struct_vector::StructVector;
39use crate::vectors::{
40 BinaryVector, BooleanVector, ConstantVector, DateVector, Decimal128Vector, DictionaryVector,
41 DurationMicrosecondVector, DurationMillisecondVector, DurationNanosecondVector,
42 DurationSecondVector, Float32Vector, Float64Vector, Int8Vector, Int16Vector, Int32Vector,
43 Int64Vector, IntervalDayTimeVector, IntervalMonthDayNanoVector, IntervalYearMonthVector,
44 ListVector, ListVectorBuilder, MutableVector, NullVector, StringVector, TimeMicrosecondVector,
45 TimeMillisecondVector, TimeNanosecondVector, TimeSecondVector, TimestampMicrosecondVector,
46 TimestampMillisecondVector, TimestampNanosecondVector, TimestampSecondVector, UInt8Vector,
47 UInt16Vector, UInt32Vector, UInt64Vector, Vector, VectorRef,
48};
49
50pub struct Helper;
52
53impl Helper {
54 pub unsafe fn static_cast<T: Any>(vector: &VectorRef) -> &T {
59 let object = vector.as_ref();
60 debug_assert!(object.as_any().is::<T>());
61 unsafe { &*(object as *const dyn Vector as *const T) }
62 }
63
64 pub fn check_get_scalar<T: Scalar>(vector: &VectorRef) -> Result<&<T as Scalar>::VectorType> {
65 vector
66 .as_any()
67 .downcast_ref::<<T as Scalar>::VectorType>()
68 .with_context(|| error::UnknownVectorSnafu {
69 msg: format!(
70 "downcast vector error, vector type: {:?}, expected vector: {:?}",
71 vector.vector_type_name(),
72 std::any::type_name::<T>(),
73 ),
74 })
75 }
76
77 pub fn check_get<T: 'static + Vector>(vector: &VectorRef) -> Result<&T> {
78 vector
79 .as_any()
80 .downcast_ref::<T>()
81 .with_context(|| error::UnknownVectorSnafu {
82 msg: format!(
83 "downcast vector error, vector type: {:?}, expected vector: {:?}",
84 vector.vector_type_name(),
85 std::any::type_name::<T>(),
86 ),
87 })
88 }
89
90 pub fn check_get_mutable_vector<T: 'static + MutableVector>(
91 vector: &mut dyn MutableVector,
92 ) -> Result<&mut T> {
93 let ty = vector.data_type();
94 vector
95 .as_mut_any()
96 .downcast_mut()
97 .with_context(|| error::UnknownVectorSnafu {
98 msg: format!(
99 "downcast vector error, vector type: {:?}, expected vector: {:?}",
100 ty,
101 std::any::type_name::<T>(),
102 ),
103 })
104 }
105
106 pub fn check_get_scalar_vector<T: Scalar>(
107 vector: &VectorRef,
108 ) -> Result<&<T as Scalar>::VectorType> {
109 vector
110 .as_any()
111 .downcast_ref::<<T as Scalar>::VectorType>()
112 .with_context(|| error::UnknownVectorSnafu {
113 msg: format!(
114 "downcast vector error, vector type: {:?}, expected vector: {:?}",
115 vector.vector_type_name(),
116 std::any::type_name::<T>(),
117 ),
118 })
119 }
120
121 pub fn try_from_scalar_value(value: ScalarValue, length: usize) -> Result<VectorRef> {
123 let vector = match value {
124 ScalarValue::Null => ConstantVector::new(Arc::new(NullVector::new(1)), length),
125 ScalarValue::Boolean(v) => {
126 ConstantVector::new(Arc::new(BooleanVector::from(vec![v])), length)
127 }
128 ScalarValue::Float16(v) => ConstantVector::new(
129 Arc::new(Float32Vector::from(vec![v.map(f32::from)])),
130 length,
131 ),
132 ScalarValue::Float32(v) => {
133 ConstantVector::new(Arc::new(Float32Vector::from(vec![v])), length)
134 }
135 ScalarValue::Float64(v) => {
136 ConstantVector::new(Arc::new(Float64Vector::from(vec![v])), length)
137 }
138 ScalarValue::Int8(v) => {
139 ConstantVector::new(Arc::new(Int8Vector::from(vec![v])), length)
140 }
141 ScalarValue::Int16(v) => {
142 ConstantVector::new(Arc::new(Int16Vector::from(vec![v])), length)
143 }
144 ScalarValue::Int32(v) => {
145 ConstantVector::new(Arc::new(Int32Vector::from(vec![v])), length)
146 }
147 ScalarValue::Int64(v) => {
148 ConstantVector::new(Arc::new(Int64Vector::from(vec![v])), length)
149 }
150 ScalarValue::UInt8(v) => {
151 ConstantVector::new(Arc::new(UInt8Vector::from(vec![v])), length)
152 }
153 ScalarValue::UInt16(v) => {
154 ConstantVector::new(Arc::new(UInt16Vector::from(vec![v])), length)
155 }
156 ScalarValue::UInt32(v) => {
157 ConstantVector::new(Arc::new(UInt32Vector::from(vec![v])), length)
158 }
159 ScalarValue::UInt64(v) => {
160 ConstantVector::new(Arc::new(UInt64Vector::from(vec![v])), length)
161 }
162 ScalarValue::Utf8(v) | ScalarValue::LargeUtf8(v) => {
163 ConstantVector::new(Arc::new(StringVector::from(vec![v])), length)
164 }
165 ScalarValue::Binary(v)
166 | ScalarValue::LargeBinary(v)
167 | ScalarValue::FixedSizeBinary(_, v) => {
168 ConstantVector::new(Arc::new(BinaryVector::from(vec![v])), length)
169 }
170 ScalarValue::List(array) => {
171 let item_type = Arc::new(ConcreteDataType::try_from(&array.value_type())?);
172 let mut builder = ListVectorBuilder::with_type_capacity(item_type.clone(), 1);
173 let scalar_values = ScalarValue::convert_array_to_scalar_vec(array.as_ref())
174 .context(ConvertArrowArrayToScalarsSnafu)?;
175 let values = scalar_values
176 .into_iter()
177 .flat_map(|v| v.unwrap_or_else(|| vec![ScalarValue::Null]))
178 .map(ScalarValue::try_into)
179 .collect::<Result<Vec<Value>>>()?;
180 builder.push(Some(ListValueRef::Ref {
181 val: &ListValue::new(values, item_type),
182 }));
183 let list_vector = builder.to_vector();
184 ConstantVector::new(list_vector, length)
185 }
186 ScalarValue::Date32(v) => {
187 ConstantVector::new(Arc::new(DateVector::from(vec![v])), length)
188 }
189 ScalarValue::TimestampSecond(v, _) => {
190 ConstantVector::new(Arc::new(TimestampSecondVector::from(vec![v])), length)
192 }
193 ScalarValue::TimestampMillisecond(v, _) => {
194 ConstantVector::new(Arc::new(TimestampMillisecondVector::from(vec![v])), length)
196 }
197 ScalarValue::TimestampMicrosecond(v, _) => {
198 ConstantVector::new(Arc::new(TimestampMicrosecondVector::from(vec![v])), length)
200 }
201 ScalarValue::TimestampNanosecond(v, _) => {
202 ConstantVector::new(Arc::new(TimestampNanosecondVector::from(vec![v])), length)
204 }
205 ScalarValue::Time32Second(v) => {
206 ConstantVector::new(Arc::new(TimeSecondVector::from(vec![v])), length)
207 }
208 ScalarValue::Time32Millisecond(v) => {
209 ConstantVector::new(Arc::new(TimeMillisecondVector::from(vec![v])), length)
210 }
211 ScalarValue::Time64Microsecond(v) => {
212 ConstantVector::new(Arc::new(TimeMicrosecondVector::from(vec![v])), length)
213 }
214 ScalarValue::Time64Nanosecond(v) => {
215 ConstantVector::new(Arc::new(TimeNanosecondVector::from(vec![v])), length)
216 }
217 ScalarValue::IntervalYearMonth(v) => {
218 ConstantVector::new(Arc::new(IntervalYearMonthVector::from(vec![v])), length)
219 }
220 ScalarValue::IntervalDayTime(v) => {
221 ConstantVector::new(Arc::new(IntervalDayTimeVector::from(vec![v])), length)
222 }
223 ScalarValue::IntervalMonthDayNano(v) => {
224 ConstantVector::new(Arc::new(IntervalMonthDayNanoVector::from(vec![v])), length)
225 }
226 ScalarValue::DurationSecond(v) => {
227 ConstantVector::new(Arc::new(DurationSecondVector::from(vec![v])), length)
228 }
229 ScalarValue::DurationMillisecond(v) => {
230 ConstantVector::new(Arc::new(DurationMillisecondVector::from(vec![v])), length)
231 }
232 ScalarValue::DurationMicrosecond(v) => {
233 ConstantVector::new(Arc::new(DurationMicrosecondVector::from(vec![v])), length)
234 }
235 ScalarValue::DurationNanosecond(v) => {
236 ConstantVector::new(Arc::new(DurationNanosecondVector::from(vec![v])), length)
237 }
238 ScalarValue::Decimal128(v, p, s) => {
239 let vector = Decimal128Vector::from(vec![v]).with_precision_and_scale(p, s)?;
240 ConstantVector::new(Arc::new(vector), length)
241 }
242 ScalarValue::Struct(v) => {
243 let struct_type = StructType::try_from(v.fields())?;
244 ConstantVector::new(
245 Arc::new(StructVector::try_new(struct_type, (*v).clone())?),
246 length,
247 )
248 }
249 ScalarValue::Decimal32(_, _, _)
250 | ScalarValue::Decimal64(_, _, _)
251 | ScalarValue::Decimal256(_, _, _)
252 | ScalarValue::FixedSizeList(_)
253 | ScalarValue::LargeList(_)
254 | ScalarValue::Dictionary(_, _)
255 | ScalarValue::Union(_, _, _)
256 | ScalarValue::Utf8View(_)
257 | ScalarValue::BinaryView(_)
258 | ScalarValue::Map(_)
259 | ScalarValue::Date64(_) => {
260 return error::ConversionSnafu {
261 from: format!("Unsupported scalar value: {value}"),
262 }
263 .fail();
264 }
265 };
266
267 Ok(Arc::new(vector))
268 }
269
270 pub fn try_into_vector(array: impl AsRef<dyn Array>) -> Result<VectorRef> {
275 Ok(match array.as_ref().data_type() {
276 ArrowDataType::Null => Arc::new(NullVector::try_from_arrow_array(array)?),
277 ArrowDataType::Boolean => Arc::new(BooleanVector::try_from_arrow_array(array)?),
278 ArrowDataType::Binary | ArrowDataType::BinaryView => {
279 Arc::new(BinaryVector::try_from_arrow_array(array)?)
280 }
281 ArrowDataType::LargeBinary | ArrowDataType::FixedSizeBinary(_) => {
282 let array = arrow::compute::cast(array.as_ref(), &ArrowDataType::Binary)
283 .context(crate::error::ArrowComputeSnafu)?;
284 Arc::new(BinaryVector::try_from_arrow_array(array)?)
285 }
286 ArrowDataType::Int8 => Arc::new(Int8Vector::try_from_arrow_array(array)?),
287 ArrowDataType::Int16 => Arc::new(Int16Vector::try_from_arrow_array(array)?),
288 ArrowDataType::Int32 => Arc::new(Int32Vector::try_from_arrow_array(array)?),
289 ArrowDataType::Int64 => Arc::new(Int64Vector::try_from_arrow_array(array)?),
290 ArrowDataType::UInt8 => Arc::new(UInt8Vector::try_from_arrow_array(array)?),
291 ArrowDataType::UInt16 => Arc::new(UInt16Vector::try_from_arrow_array(array)?),
292 ArrowDataType::UInt32 => Arc::new(UInt32Vector::try_from_arrow_array(array)?),
293 ArrowDataType::UInt64 => Arc::new(UInt64Vector::try_from_arrow_array(array)?),
294 ArrowDataType::Float32 => Arc::new(Float32Vector::try_from_arrow_array(array)?),
295 ArrowDataType::Float64 => Arc::new(Float64Vector::try_from_arrow_array(array)?),
296 ArrowDataType::Utf8 | ArrowDataType::LargeUtf8 | ArrowDataType::Utf8View => {
297 Arc::new(StringVector::try_from_arrow_array(array)?)
298 }
299 ArrowDataType::Date32 => Arc::new(DateVector::try_from_arrow_array(array)?),
300 ArrowDataType::List(_) => Arc::new(ListVector::try_from_arrow_array(array)?),
301 ArrowDataType::Timestamp(unit, _) => match unit {
302 TimeUnit::Second => Arc::new(TimestampSecondVector::try_from_arrow_array(array)?),
303 TimeUnit::Millisecond => {
304 Arc::new(TimestampMillisecondVector::try_from_arrow_array(array)?)
305 }
306 TimeUnit::Microsecond => {
307 Arc::new(TimestampMicrosecondVector::try_from_arrow_array(array)?)
308 }
309 TimeUnit::Nanosecond => {
310 Arc::new(TimestampNanosecondVector::try_from_arrow_array(array)?)
311 }
312 },
313 ArrowDataType::Time32(unit) => match unit {
314 TimeUnit::Second => Arc::new(TimeSecondVector::try_from_arrow_array(array)?),
315 TimeUnit::Millisecond => {
316 Arc::new(TimeMillisecondVector::try_from_arrow_array(array)?)
317 }
318 _ => unreachable!(
320 "unexpected arrow array datatype: {:?}",
321 array.as_ref().data_type()
322 ),
323 },
324 ArrowDataType::Time64(unit) => match unit {
325 TimeUnit::Microsecond => {
326 Arc::new(TimeMicrosecondVector::try_from_arrow_array(array)?)
327 }
328 TimeUnit::Nanosecond => {
329 Arc::new(TimeNanosecondVector::try_from_arrow_array(array)?)
330 }
331 _ => unreachable!(
333 "unexpected arrow array datatype: {:?}",
334 array.as_ref().data_type()
335 ),
336 },
337 ArrowDataType::Interval(unit) => match unit {
338 IntervalUnit::YearMonth => {
339 Arc::new(IntervalYearMonthVector::try_from_arrow_array(array)?)
340 }
341 IntervalUnit::DayTime => {
342 Arc::new(IntervalDayTimeVector::try_from_arrow_array(array)?)
343 }
344 IntervalUnit::MonthDayNano => {
345 Arc::new(IntervalMonthDayNanoVector::try_from_arrow_array(array)?)
346 }
347 },
348 ArrowDataType::Duration(unit) => match unit {
349 TimeUnit::Second => Arc::new(DurationSecondVector::try_from_arrow_array(array)?),
350 TimeUnit::Millisecond => {
351 Arc::new(DurationMillisecondVector::try_from_arrow_array(array)?)
352 }
353 TimeUnit::Microsecond => {
354 Arc::new(DurationMicrosecondVector::try_from_arrow_array(array)?)
355 }
356 TimeUnit::Nanosecond => {
357 Arc::new(DurationNanosecondVector::try_from_arrow_array(array)?)
358 }
359 },
360 ArrowDataType::Decimal128(_, _) => {
361 Arc::new(Decimal128Vector::try_from_arrow_array(array)?)
362 }
363 ArrowDataType::Dictionary(key, value) => {
364 macro_rules! handle_dictionary_key_type {
365 ($key_type:ident) => {{
366 let array = array
367 .as_ref()
368 .as_any()
369 .downcast_ref::<DictionaryArray<$key_type>>()
370 .unwrap(); Arc::new(DictionaryVector::new(
372 array.clone(),
373 ConcreteDataType::try_from(value.as_ref())?,
374 )?)
375 }};
376 }
377
378 match key.as_ref() {
379 ArrowDataType::Int8 => handle_dictionary_key_type!(Int8Type),
380 ArrowDataType::Int16 => handle_dictionary_key_type!(Int16Type),
381 ArrowDataType::Int32 => handle_dictionary_key_type!(Int32Type),
382 ArrowDataType::Int64 => handle_dictionary_key_type!(Int64Type),
383 ArrowDataType::UInt8 => handle_dictionary_key_type!(UInt8Type),
384 ArrowDataType::UInt16 => handle_dictionary_key_type!(UInt16Type),
385 ArrowDataType::UInt32 => handle_dictionary_key_type!(UInt32Type),
386 ArrowDataType::UInt64 => handle_dictionary_key_type!(UInt64Type),
387 _ => {
388 return error::UnsupportedArrowTypeSnafu {
389 arrow_type: array.as_ref().data_type().clone(),
390 }
391 .fail();
392 }
393 }
394 }
395
396 ArrowDataType::Struct(fields) => {
397 let array = array
398 .as_ref()
399 .as_any()
400 .downcast_ref::<StructArray>()
401 .unwrap();
402 Arc::new(StructVector::try_new(
403 StructType::try_from(fields)?,
404 array.clone(),
405 )?)
406 }
407 ArrowDataType::Float16
408 | ArrowDataType::LargeList(_)
409 | ArrowDataType::FixedSizeList(_, _)
410 | ArrowDataType::Union(_, _)
411 | ArrowDataType::Decimal256(_, _)
412 | ArrowDataType::Map(_, _)
413 | ArrowDataType::RunEndEncoded(_, _)
414 | ArrowDataType::ListView(_)
415 | ArrowDataType::LargeListView(_)
416 | ArrowDataType::Date64
417 | ArrowDataType::Decimal32(_, _)
418 | ArrowDataType::Decimal64(_, _) => {
419 return error::UnsupportedArrowTypeSnafu {
420 arrow_type: array.as_ref().data_type().clone(),
421 }
422 .fail();
423 }
424 })
425 }
426
427 pub fn try_from_row_into_vector(row: &[Value], dt: &ConcreteDataType) -> Result<VectorRef> {
429 let mut builder = dt.create_mutable_vector(row.len());
430 for val in row {
431 builder.try_push_value_ref(&val.as_value_ref())?;
432 }
433 let vector = builder.to_vector();
434 Ok(vector)
435 }
436
437 pub fn try_into_vectors(arrays: &[ArrayRef]) -> Result<Vec<VectorRef>> {
439 arrays.iter().map(Self::try_into_vector).collect()
440 }
441
442 pub fn like_utf8(names: Vec<String>, s: &str) -> Result<VectorRef> {
444 let array = StringArray::from(names);
445
446 let s = StringArray::new_scalar(s);
447 let filter = comparison::like(&array, &s).context(error::ArrowComputeSnafu)?;
448
449 let result = compute::filter(&array, &filter).context(error::ArrowComputeSnafu)?;
450 Helper::try_into_vector(result)
451 }
452
453 pub fn like_utf8_filter(names: Vec<String>, s: &str) -> Result<(VectorRef, BooleanVector)> {
454 let array = StringArray::from(names);
455 let s = StringArray::new_scalar(s);
456 let filter = comparison::like(&array, &s).context(error::ArrowComputeSnafu)?;
457 let result = compute::filter(&array, &filter).context(error::ArrowComputeSnafu)?;
458 let vector = Helper::try_into_vector(result)?;
459
460 Ok((vector, BooleanVector::from(filter)))
461 }
462}
463
464#[cfg(test)]
465pub(crate) fn pretty_print(vector: VectorRef) -> String {
466 let array = vector.to_arrow_array();
467 arrow::util::pretty::pretty_format_columns(&vector.vector_type_name(), &[array])
468 .map(|x| x.to_string())
469 .unwrap_or_else(|e| e.to_string())
470}
471
472#[cfg(test)]
473mod tests {
474 use arrow::array::{
475 ArrayRef, BooleanArray, Date32Array, Float32Array, Float64Array, Int8Array, Int16Array,
476 Int32Array, Int64Array, LargeBinaryArray, ListArray, NullArray, Time32MillisecondArray,
477 Time32SecondArray, Time64MicrosecondArray, Time64NanosecondArray,
478 TimestampMicrosecondArray, TimestampMillisecondArray, TimestampNanosecondArray,
479 TimestampSecondArray, UInt8Array, UInt16Array, UInt32Array, UInt64Array,
480 };
481 use arrow::buffer::Buffer;
482 use arrow::datatypes::{Int32Type, IntervalMonthDayNano};
483 use arrow_array::{BinaryArray, DictionaryArray, FixedSizeBinaryArray, LargeStringArray};
484 use arrow_schema::DataType;
485 use common_decimal::Decimal128;
486 use common_time::time::Time;
487 use common_time::timestamp::TimeUnit;
488 use common_time::{Date, Duration};
489
490 use super::*;
491 use crate::value::Value;
492 use crate::vectors::ConcreteDataType;
493
494 #[test]
495 fn test_try_into_vectors() {
496 let arrays: Vec<ArrayRef> = vec![
497 Arc::new(Int32Array::from(vec![1])),
498 Arc::new(Int32Array::from(vec![2])),
499 Arc::new(Int32Array::from(vec![3])),
500 ];
501 let vectors = Helper::try_into_vectors(&arrays).unwrap();
502 vectors.iter().for_each(|v| assert_eq!(1, v.len()));
503 assert_eq!(Value::Int32(1), vectors[0].get(0));
504 assert_eq!(Value::Int32(2), vectors[1].get(0));
505 assert_eq!(Value::Int32(3), vectors[2].get(0));
506 }
507
508 #[test]
509 fn test_try_into_date_vector() {
510 let vector = DateVector::from(vec![Some(1), Some(2), None]);
511 let arrow_array = vector.to_arrow_array();
512 assert_eq!(&ArrowDataType::Date32, arrow_array.data_type());
513 let vector_converted = Helper::try_into_vector(arrow_array).unwrap();
514 assert_eq!(vector.len(), vector_converted.len());
515 for i in 0..vector_converted.len() {
516 assert_eq!(vector.get(i), vector_converted.get(i));
517 }
518 }
519
520 #[test]
521 fn test_try_from_scalar_date_value() {
522 let vector = Helper::try_from_scalar_value(ScalarValue::Date32(Some(42)), 3).unwrap();
523 assert_eq!(ConcreteDataType::date_datatype(), vector.data_type());
524 assert_eq!(3, vector.len());
525 for i in 0..vector.len() {
526 assert_eq!(Value::Date(Date::new(42)), vector.get(i));
527 }
528 }
529
530 #[test]
531 fn test_try_from_scalar_duration_value() {
532 let vector =
533 Helper::try_from_scalar_value(ScalarValue::DurationSecond(Some(42)), 3).unwrap();
534 assert_eq!(
535 ConcreteDataType::duration_second_datatype(),
536 vector.data_type()
537 );
538 assert_eq!(3, vector.len());
539 for i in 0..vector.len() {
540 assert_eq!(
541 Value::Duration(Duration::new(42, TimeUnit::Second)),
542 vector.get(i)
543 );
544 }
545 }
546
547 #[test]
548 fn test_try_from_scalar_decimal128_value() {
549 let vector =
550 Helper::try_from_scalar_value(ScalarValue::Decimal128(Some(42), 3, 1), 3).unwrap();
551 assert_eq!(
552 ConcreteDataType::decimal128_datatype(3, 1),
553 vector.data_type()
554 );
555 assert_eq!(3, vector.len());
556 for i in 0..vector.len() {
557 assert_eq!(Value::Decimal128(Decimal128::new(42, 3, 1)), vector.get(i));
558 }
559 }
560
561 #[test]
562 fn test_try_from_list_value() {
563 let value = ScalarValue::List(ScalarValue::new_list(
564 &[ScalarValue::Int32(Some(1)), ScalarValue::Int32(Some(2))],
565 &ArrowDataType::Int32,
566 true,
567 ));
568 let vector = Helper::try_from_scalar_value(value, 3).unwrap();
569 assert_eq!(
570 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype())),
571 vector.data_type()
572 );
573 assert_eq!(3, vector.len());
574 for i in 0..vector.len() {
575 let v = vector.get(i);
576 let items = v.as_list().unwrap().unwrap().items();
577 assert_eq!(vec![Value::Int32(1), Value::Int32(2)], items);
578 }
579 }
580
581 #[test]
582 fn test_like_utf8() {
583 fn assert_vector(expected: Vec<&str>, actual: &VectorRef) {
584 let actual = actual.as_any().downcast_ref::<StringVector>().unwrap();
585 assert_eq!(*actual, StringVector::from(expected));
586 }
587
588 let names: Vec<String> = vec!["greptime", "hello", "public", "world"]
589 .into_iter()
590 .map(|x| x.to_string())
591 .collect();
592
593 let ret = Helper::like_utf8(names.clone(), "%ll%").unwrap();
594 assert_vector(vec!["hello"], &ret);
595
596 let ret = Helper::like_utf8(names.clone(), "%time").unwrap();
597 assert_vector(vec!["greptime"], &ret);
598
599 let ret = Helper::like_utf8(names.clone(), "%ld").unwrap();
600 assert_vector(vec!["world"], &ret);
601
602 let ret = Helper::like_utf8(names, "%").unwrap();
603 assert_vector(vec!["greptime", "hello", "public", "world"], &ret);
604 }
605
606 #[test]
607 fn test_like_utf8_filter() {
608 fn assert_vector(expected: Vec<&str>, actual: &VectorRef) {
609 let actual = actual.as_any().downcast_ref::<StringVector>().unwrap();
610 assert_eq!(*actual, StringVector::from(expected));
611 }
612
613 fn assert_filter(array: Vec<String>, s: &str, expected_filter: &BooleanVector) {
614 let array = StringArray::from(array);
615 let s = StringArray::new_scalar(s);
616 let actual_filter = comparison::like(&array, &s).unwrap();
617 assert_eq!(BooleanVector::from(actual_filter), *expected_filter);
618 }
619
620 let names: Vec<String> = vec!["greptime", "timeseries", "cloud", "database"]
621 .into_iter()
622 .map(|x| x.to_string())
623 .collect();
624
625 let (table, filter) = Helper::like_utf8_filter(names.clone(), "%ti%").unwrap();
626 assert_vector(vec!["greptime", "timeseries"], &table);
627 assert_filter(names.clone(), "%ti%", &filter);
628
629 let (tables, filter) = Helper::like_utf8_filter(names.clone(), "%lou").unwrap();
630 assert_vector(vec![], &tables);
631 assert_filter(names.clone(), "%lou", &filter);
632
633 let (tables, filter) = Helper::like_utf8_filter(names.clone(), "%d%").unwrap();
634 assert_vector(vec!["cloud", "database"], &tables);
635 assert_filter(names.clone(), "%d%", &filter);
636 }
637
638 fn check_try_into_vector(array: impl Array + 'static) {
639 let array: ArrayRef = Arc::new(array);
640 let vector = Helper::try_into_vector(array.clone()).unwrap();
641 assert_eq!(&array, &vector.to_arrow_array());
642 }
643
644 #[test]
645 fn test_try_into_vector() {
646 check_try_into_vector(NullArray::new(2));
647 check_try_into_vector(BooleanArray::from(vec![true, false]));
648 check_try_into_vector(Int8Array::from(vec![1, 2, 3]));
649 check_try_into_vector(Int16Array::from(vec![1, 2, 3]));
650 check_try_into_vector(Int32Array::from(vec![1, 2, 3]));
651 check_try_into_vector(Int64Array::from(vec![1, 2, 3]));
652 check_try_into_vector(UInt8Array::from(vec![1, 2, 3]));
653 check_try_into_vector(UInt16Array::from(vec![1, 2, 3]));
654 check_try_into_vector(UInt32Array::from(vec![1, 2, 3]));
655 check_try_into_vector(UInt64Array::from(vec![1, 2, 3]));
656 check_try_into_vector(Float32Array::from(vec![1.0, 2.0, 3.0]));
657 check_try_into_vector(Float64Array::from(vec![1.0, 2.0, 3.0]));
658 check_try_into_vector(StringArray::from(vec!["hello", "world"]));
659 check_try_into_vector(Date32Array::from(vec![1, 2, 3]));
660 let data = vec![None, Some(vec![Some(6), Some(7)])];
661 let list_array = ListArray::from_iter_primitive::<Int32Type, _, _>(data);
662 check_try_into_vector(list_array);
663 check_try_into_vector(TimestampSecondArray::from(vec![1, 2, 3]));
664 check_try_into_vector(TimestampMillisecondArray::from(vec![1, 2, 3]));
665 check_try_into_vector(TimestampMicrosecondArray::from(vec![1, 2, 3]));
666 check_try_into_vector(TimestampNanosecondArray::from(vec![1, 2, 3]));
667 check_try_into_vector(Time32SecondArray::from(vec![1, 2, 3]));
668 check_try_into_vector(Time32MillisecondArray::from(vec![1, 2, 3]));
669 check_try_into_vector(Time64MicrosecondArray::from(vec![1, 2, 3]));
670 check_try_into_vector(Time64NanosecondArray::from(vec![1, 2, 3]));
671
672 let values = StringArray::from_iter_values(["a", "b", "c"]);
674
675 let keys = Int8Array::from_iter_values([0, 0, 1, 2]);
677 let array: ArrayRef =
678 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
679 Helper::try_into_vector(array).unwrap();
680
681 let keys = Int16Array::from_iter_values([0, 0, 1, 2]);
683 let array: ArrayRef =
684 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
685 Helper::try_into_vector(array).unwrap();
686
687 let keys = Int32Array::from_iter_values([0, 0, 1, 2]);
689 let array: ArrayRef =
690 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
691 Helper::try_into_vector(array).unwrap();
692
693 let keys = Int64Array::from_iter_values([0, 0, 1, 2]);
695 let array: ArrayRef =
696 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
697 Helper::try_into_vector(array).unwrap();
698
699 let keys = UInt8Array::from_iter_values([0, 0, 1, 2]);
701 let array: ArrayRef =
702 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
703 Helper::try_into_vector(array).unwrap();
704
705 let keys = UInt16Array::from_iter_values([0, 0, 1, 2]);
707 let array: ArrayRef =
708 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
709 Helper::try_into_vector(array).unwrap();
710
711 let keys = UInt32Array::from_iter_values([0, 0, 1, 2]);
713 let array: ArrayRef =
714 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
715 Helper::try_into_vector(array).unwrap();
716
717 let keys = UInt64Array::from_iter_values([0, 0, 1, 2]);
719 let array: ArrayRef = Arc::new(DictionaryArray::try_new(keys, Arc::new(values)).unwrap());
720 Helper::try_into_vector(array).unwrap();
721 }
722
723 #[test]
724 fn test_try_binary_array_into_vector() {
725 let input_vec: Vec<&[u8]> = vec!["hello".as_bytes(), "world".as_bytes()];
726 let assertion_vector = BinaryVector::from(input_vec.clone());
727
728 let input_arrays: Vec<ArrayRef> = vec![
729 Arc::new(LargeBinaryArray::from(input_vec.clone())) as ArrayRef,
730 Arc::new(BinaryArray::from(input_vec.clone())) as ArrayRef,
731 Arc::new(FixedSizeBinaryArray::new(
732 5,
733 Buffer::from_vec("helloworld".as_bytes().to_vec()),
734 None,
735 )) as ArrayRef,
736 ];
737
738 for input_array in input_arrays {
739 let vector = Helper::try_into_vector(input_array).unwrap();
740
741 assert_eq!(2, vector.len());
742 assert_eq!(0, vector.null_count());
743
744 let output_arrow_array: ArrayRef = vector.to_arrow_array();
745 assert_eq!(&DataType::Binary, output_arrow_array.data_type());
746 assert_eq!(&assertion_vector.to_arrow_array(), &output_arrow_array);
747 }
748 }
749
750 #[test]
751 fn test_large_string_array_into_vector() {
752 let input_vec = vec!["a", "b"];
753 let assertion_array = LargeStringArray::from(input_vec.clone());
754
755 let large_string_array: ArrayRef = Arc::new(LargeStringArray::from(input_vec));
756 let vector = Helper::try_into_vector(large_string_array).unwrap();
757 assert_eq!(2, vector.len());
758 assert_eq!(0, vector.null_count());
759
760 let output_arrow_array: LargeStringArray = vector
761 .to_arrow_array()
762 .as_any()
763 .downcast_ref::<LargeStringArray>()
764 .unwrap()
765 .clone();
766 assert_eq!(&assertion_array, &output_arrow_array);
767 }
768
769 #[test]
770 fn test_try_from_scalar_time_value() {
771 let vector = Helper::try_from_scalar_value(ScalarValue::Time32Second(Some(42)), 3).unwrap();
772 assert_eq!(ConcreteDataType::time_second_datatype(), vector.data_type());
773 assert_eq!(3, vector.len());
774 for i in 0..vector.len() {
775 assert_eq!(Value::Time(Time::new_second(42)), vector.get(i));
776 }
777 }
778
779 #[test]
780 fn test_try_from_scalar_interval_value() {
781 let vector = Helper::try_from_scalar_value(
782 ScalarValue::IntervalMonthDayNano(Some(IntervalMonthDayNano::new(1, 1, 2000))),
783 3,
784 )
785 .unwrap();
786
787 assert_eq!(
788 ConcreteDataType::interval_month_day_nano_datatype(),
789 vector.data_type()
790 );
791 assert_eq!(3, vector.len());
792 for i in 0..vector.len() {
793 assert_eq!(
794 Value::IntervalMonthDayNano(IntervalMonthDayNano::new(1, 1, 2000).into()),
795 vector.get(i)
796 );
797 }
798 }
799
800 fn check_try_from_row_to_vector(row: Vec<Value>, dt: &ConcreteDataType) {
801 let vector = Helper::try_from_row_into_vector(&row, dt).unwrap();
802 for (i, item) in row.iter().enumerate().take(vector.len()) {
803 assert_eq!(*item, vector.get(i));
804 }
805 }
806
807 fn check_into_and_from(array: impl Array + 'static) {
808 let array: ArrayRef = Arc::new(array);
809 let vector = Helper::try_into_vector(array.clone()).unwrap();
810 assert_eq!(&array, &vector.to_arrow_array());
811 let row: Vec<Value> = (0..array.len()).map(|i| vector.get(i)).collect();
812 let dt = vector.data_type();
813 check_try_from_row_to_vector(row, &dt);
814 }
815
816 #[test]
817 fn test_try_from_row_to_vector() {
818 check_into_and_from(NullArray::new(2));
819 check_into_and_from(BooleanArray::from(vec![true, false]));
820 check_into_and_from(Int8Array::from(vec![1, 2, 3]));
821 check_into_and_from(Int16Array::from(vec![1, 2, 3]));
822 check_into_and_from(Int32Array::from(vec![1, 2, 3]));
823 check_into_and_from(Int64Array::from(vec![1, 2, 3]));
824 check_into_and_from(UInt8Array::from(vec![1, 2, 3]));
825 check_into_and_from(UInt16Array::from(vec![1, 2, 3]));
826 check_into_and_from(UInt32Array::from(vec![1, 2, 3]));
827 check_into_and_from(UInt64Array::from(vec![1, 2, 3]));
828 check_into_and_from(Float32Array::from(vec![1.0, 2.0, 3.0]));
829 check_into_and_from(Float64Array::from(vec![1.0, 2.0, 3.0]));
830 check_into_and_from(StringArray::from(vec!["hello", "world"]));
831 check_into_and_from(Date32Array::from(vec![1, 2, 3]));
832
833 check_into_and_from(TimestampSecondArray::from(vec![1, 2, 3]));
834 check_into_and_from(TimestampMillisecondArray::from(vec![1, 2, 3]));
835 check_into_and_from(TimestampMicrosecondArray::from(vec![1, 2, 3]));
836 check_into_and_from(TimestampNanosecondArray::from(vec![1, 2, 3]));
837 check_into_and_from(Time32SecondArray::from(vec![1, 2, 3]));
838 check_into_and_from(Time32MillisecondArray::from(vec![1, 2, 3]));
839 check_into_and_from(Time64MicrosecondArray::from(vec![1, 2, 3]));
840 check_into_and_from(Time64NanosecondArray::from(vec![1, 2, 3]));
841 }
842}