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 | ScalarValue::RunEndEncoded(_, _, _) => {
261 return error::ConversionSnafu {
262 from: format!("Unsupported scalar value: {value}"),
263 }
264 .fail();
265 }
266 };
267
268 Ok(Arc::new(vector))
269 }
270
271 pub fn try_into_vector(array: impl AsRef<dyn Array>) -> Result<VectorRef> {
276 Ok(match array.as_ref().data_type() {
277 ArrowDataType::Null => Arc::new(NullVector::try_from_arrow_array(array)?),
278 ArrowDataType::Boolean => Arc::new(BooleanVector::try_from_arrow_array(array)?),
279 ArrowDataType::Binary | ArrowDataType::BinaryView => {
280 Arc::new(BinaryVector::try_from_arrow_array(array)?)
281 }
282 ArrowDataType::LargeBinary | ArrowDataType::FixedSizeBinary(_) => {
283 let array = arrow::compute::cast(array.as_ref(), &ArrowDataType::Binary)
284 .context(crate::error::ArrowComputeSnafu)?;
285 Arc::new(BinaryVector::try_from_arrow_array(array)?)
286 }
287 ArrowDataType::Int8 => Arc::new(Int8Vector::try_from_arrow_array(array)?),
288 ArrowDataType::Int16 => Arc::new(Int16Vector::try_from_arrow_array(array)?),
289 ArrowDataType::Int32 => Arc::new(Int32Vector::try_from_arrow_array(array)?),
290 ArrowDataType::Int64 => Arc::new(Int64Vector::try_from_arrow_array(array)?),
291 ArrowDataType::UInt8 => Arc::new(UInt8Vector::try_from_arrow_array(array)?),
292 ArrowDataType::UInt16 => Arc::new(UInt16Vector::try_from_arrow_array(array)?),
293 ArrowDataType::UInt32 => Arc::new(UInt32Vector::try_from_arrow_array(array)?),
294 ArrowDataType::UInt64 => Arc::new(UInt64Vector::try_from_arrow_array(array)?),
295 ArrowDataType::Float32 => Arc::new(Float32Vector::try_from_arrow_array(array)?),
296 ArrowDataType::Float64 => Arc::new(Float64Vector::try_from_arrow_array(array)?),
297 ArrowDataType::Utf8 | ArrowDataType::LargeUtf8 | ArrowDataType::Utf8View => {
298 Arc::new(StringVector::try_from_arrow_array(array)?)
299 }
300 ArrowDataType::Date32 => Arc::new(DateVector::try_from_arrow_array(array)?),
301 ArrowDataType::List(_) => Arc::new(ListVector::try_from_arrow_array(array)?),
302 ArrowDataType::Timestamp(unit, _) => match unit {
303 TimeUnit::Second => Arc::new(TimestampSecondVector::try_from_arrow_array(array)?),
304 TimeUnit::Millisecond => {
305 Arc::new(TimestampMillisecondVector::try_from_arrow_array(array)?)
306 }
307 TimeUnit::Microsecond => {
308 Arc::new(TimestampMicrosecondVector::try_from_arrow_array(array)?)
309 }
310 TimeUnit::Nanosecond => {
311 Arc::new(TimestampNanosecondVector::try_from_arrow_array(array)?)
312 }
313 },
314 ArrowDataType::Time32(unit) => match unit {
315 TimeUnit::Second => Arc::new(TimeSecondVector::try_from_arrow_array(array)?),
316 TimeUnit::Millisecond => {
317 Arc::new(TimeMillisecondVector::try_from_arrow_array(array)?)
318 }
319 _ => unreachable!(
321 "unexpected arrow array datatype: {:?}",
322 array.as_ref().data_type()
323 ),
324 },
325 ArrowDataType::Time64(unit) => match unit {
326 TimeUnit::Microsecond => {
327 Arc::new(TimeMicrosecondVector::try_from_arrow_array(array)?)
328 }
329 TimeUnit::Nanosecond => {
330 Arc::new(TimeNanosecondVector::try_from_arrow_array(array)?)
331 }
332 _ => unreachable!(
334 "unexpected arrow array datatype: {:?}",
335 array.as_ref().data_type()
336 ),
337 },
338 ArrowDataType::Interval(unit) => match unit {
339 IntervalUnit::YearMonth => {
340 Arc::new(IntervalYearMonthVector::try_from_arrow_array(array)?)
341 }
342 IntervalUnit::DayTime => {
343 Arc::new(IntervalDayTimeVector::try_from_arrow_array(array)?)
344 }
345 IntervalUnit::MonthDayNano => {
346 Arc::new(IntervalMonthDayNanoVector::try_from_arrow_array(array)?)
347 }
348 },
349 ArrowDataType::Duration(unit) => match unit {
350 TimeUnit::Second => Arc::new(DurationSecondVector::try_from_arrow_array(array)?),
351 TimeUnit::Millisecond => {
352 Arc::new(DurationMillisecondVector::try_from_arrow_array(array)?)
353 }
354 TimeUnit::Microsecond => {
355 Arc::new(DurationMicrosecondVector::try_from_arrow_array(array)?)
356 }
357 TimeUnit::Nanosecond => {
358 Arc::new(DurationNanosecondVector::try_from_arrow_array(array)?)
359 }
360 },
361 ArrowDataType::Decimal128(_, _) => {
362 Arc::new(Decimal128Vector::try_from_arrow_array(array)?)
363 }
364 ArrowDataType::Dictionary(key, value) => {
365 macro_rules! handle_dictionary_key_type {
366 ($key_type:ident) => {{
367 let array = array
368 .as_ref()
369 .as_any()
370 .downcast_ref::<DictionaryArray<$key_type>>()
371 .unwrap(); Arc::new(DictionaryVector::new(
373 array.clone(),
374 ConcreteDataType::try_from(value.as_ref())?,
375 )?)
376 }};
377 }
378
379 match key.as_ref() {
380 ArrowDataType::Int8 => handle_dictionary_key_type!(Int8Type),
381 ArrowDataType::Int16 => handle_dictionary_key_type!(Int16Type),
382 ArrowDataType::Int32 => handle_dictionary_key_type!(Int32Type),
383 ArrowDataType::Int64 => handle_dictionary_key_type!(Int64Type),
384 ArrowDataType::UInt8 => handle_dictionary_key_type!(UInt8Type),
385 ArrowDataType::UInt16 => handle_dictionary_key_type!(UInt16Type),
386 ArrowDataType::UInt32 => handle_dictionary_key_type!(UInt32Type),
387 ArrowDataType::UInt64 => handle_dictionary_key_type!(UInt64Type),
388 _ => {
389 return error::UnsupportedArrowTypeSnafu {
390 arrow_type: array.as_ref().data_type().clone(),
391 }
392 .fail();
393 }
394 }
395 }
396
397 ArrowDataType::Struct(fields) => {
398 let array = array
399 .as_ref()
400 .as_any()
401 .downcast_ref::<StructArray>()
402 .unwrap();
403 Arc::new(StructVector::try_new(
404 StructType::try_from(fields)?,
405 array.clone(),
406 )?)
407 }
408 ArrowDataType::Float16
409 | ArrowDataType::LargeList(_)
410 | ArrowDataType::FixedSizeList(_, _)
411 | ArrowDataType::Union(_, _)
412 | ArrowDataType::Decimal256(_, _)
413 | ArrowDataType::Map(_, _)
414 | ArrowDataType::RunEndEncoded(_, _)
415 | ArrowDataType::ListView(_)
416 | ArrowDataType::LargeListView(_)
417 | ArrowDataType::Date64
418 | ArrowDataType::Decimal32(_, _)
419 | ArrowDataType::Decimal64(_, _) => {
420 return error::UnsupportedArrowTypeSnafu {
421 arrow_type: array.as_ref().data_type().clone(),
422 }
423 .fail();
424 }
425 })
426 }
427
428 pub fn try_from_row_into_vector(row: &[Value], dt: &ConcreteDataType) -> Result<VectorRef> {
430 let mut builder = dt.create_mutable_vector(row.len());
431 for val in row {
432 builder.try_push_value_ref(&val.as_value_ref())?;
433 }
434 let vector = builder.to_vector();
435 Ok(vector)
436 }
437
438 pub fn try_into_vectors(arrays: &[ArrayRef]) -> Result<Vec<VectorRef>> {
440 arrays.iter().map(Self::try_into_vector).collect()
441 }
442
443 pub fn like_utf8(names: Vec<String>, s: &str) -> Result<VectorRef> {
445 let array = StringArray::from(names);
446
447 let s = StringArray::new_scalar(s);
448 let filter = comparison::like(&array, &s).context(error::ArrowComputeSnafu)?;
449
450 let result = compute::filter(&array, &filter).context(error::ArrowComputeSnafu)?;
451 Helper::try_into_vector(result)
452 }
453
454 pub fn like_utf8_filter(names: Vec<String>, s: &str) -> Result<(VectorRef, BooleanVector)> {
455 let array = StringArray::from(names);
456 let s = StringArray::new_scalar(s);
457 let filter = comparison::like(&array, &s).context(error::ArrowComputeSnafu)?;
458 let result = compute::filter(&array, &filter).context(error::ArrowComputeSnafu)?;
459 let vector = Helper::try_into_vector(result)?;
460
461 Ok((vector, BooleanVector::from(filter)))
462 }
463}
464
465#[cfg(test)]
466mod tests {
467 use arrow::array::{
468 ArrayRef, BooleanArray, Date32Array, Float32Array, Float64Array, Int8Array, Int16Array,
469 Int32Array, Int64Array, LargeBinaryArray, ListArray, NullArray, Time32MillisecondArray,
470 Time32SecondArray, Time64MicrosecondArray, Time64NanosecondArray,
471 TimestampMicrosecondArray, TimestampMillisecondArray, TimestampNanosecondArray,
472 TimestampSecondArray, UInt8Array, UInt16Array, UInt32Array, UInt64Array,
473 };
474 use arrow::buffer::Buffer;
475 use arrow::datatypes::{Int32Type, IntervalMonthDayNano};
476 use arrow_array::{BinaryArray, DictionaryArray, FixedSizeBinaryArray, LargeStringArray};
477 use arrow_schema::DataType;
478 use common_decimal::Decimal128;
479 use common_time::time::Time;
480 use common_time::timestamp::TimeUnit;
481 use common_time::{Date, Duration};
482
483 use super::*;
484 use crate::value::Value;
485 use crate::vectors::ConcreteDataType;
486
487 #[test]
488 fn test_try_into_vectors() {
489 let arrays: Vec<ArrayRef> = vec![
490 Arc::new(Int32Array::from(vec![1])),
491 Arc::new(Int32Array::from(vec![2])),
492 Arc::new(Int32Array::from(vec![3])),
493 ];
494 let vectors = Helper::try_into_vectors(&arrays).unwrap();
495 vectors.iter().for_each(|v| assert_eq!(1, v.len()));
496 assert_eq!(Value::Int32(1), vectors[0].get(0));
497 assert_eq!(Value::Int32(2), vectors[1].get(0));
498 assert_eq!(Value::Int32(3), vectors[2].get(0));
499 }
500
501 #[test]
502 fn test_try_into_date_vector() {
503 let vector = DateVector::from(vec![Some(1), Some(2), None]);
504 let arrow_array = vector.to_arrow_array();
505 assert_eq!(&ArrowDataType::Date32, arrow_array.data_type());
506 let vector_converted = Helper::try_into_vector(arrow_array).unwrap();
507 assert_eq!(vector.len(), vector_converted.len());
508 for i in 0..vector_converted.len() {
509 assert_eq!(vector.get(i), vector_converted.get(i));
510 }
511 }
512
513 #[test]
514 fn test_try_from_scalar_date_value() {
515 let vector = Helper::try_from_scalar_value(ScalarValue::Date32(Some(42)), 3).unwrap();
516 assert_eq!(ConcreteDataType::date_datatype(), vector.data_type());
517 assert_eq!(3, vector.len());
518 for i in 0..vector.len() {
519 assert_eq!(Value::Date(Date::new(42)), vector.get(i));
520 }
521 }
522
523 #[test]
524 fn test_try_from_scalar_duration_value() {
525 let vector =
526 Helper::try_from_scalar_value(ScalarValue::DurationSecond(Some(42)), 3).unwrap();
527 assert_eq!(
528 ConcreteDataType::duration_second_datatype(),
529 vector.data_type()
530 );
531 assert_eq!(3, vector.len());
532 for i in 0..vector.len() {
533 assert_eq!(
534 Value::Duration(Duration::new(42, TimeUnit::Second)),
535 vector.get(i)
536 );
537 }
538 }
539
540 #[test]
541 fn test_try_from_scalar_decimal128_value() {
542 let vector =
543 Helper::try_from_scalar_value(ScalarValue::Decimal128(Some(42), 3, 1), 3).unwrap();
544 assert_eq!(
545 ConcreteDataType::decimal128_datatype(3, 1),
546 vector.data_type()
547 );
548 assert_eq!(3, vector.len());
549 for i in 0..vector.len() {
550 assert_eq!(Value::Decimal128(Decimal128::new(42, 3, 1)), vector.get(i));
551 }
552 }
553
554 #[test]
555 fn test_try_from_list_value() {
556 let value = ScalarValue::List(ScalarValue::new_list(
557 &[ScalarValue::Int32(Some(1)), ScalarValue::Int32(Some(2))],
558 &ArrowDataType::Int32,
559 true,
560 ));
561 let vector = Helper::try_from_scalar_value(value, 3).unwrap();
562 assert_eq!(
563 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype())),
564 vector.data_type()
565 );
566 assert_eq!(3, vector.len());
567 for i in 0..vector.len() {
568 let v = vector.get(i);
569 let items = v.as_list().unwrap().unwrap().items();
570 assert_eq!(vec![Value::Int32(1), Value::Int32(2)], items);
571 }
572 }
573
574 #[test]
575 fn test_like_utf8() {
576 fn assert_vector(expected: Vec<&str>, actual: &VectorRef) {
577 let actual = actual.as_any().downcast_ref::<StringVector>().unwrap();
578 assert_eq!(*actual, StringVector::from(expected));
579 }
580
581 let names: Vec<String> = vec!["greptime", "hello", "public", "world"]
582 .into_iter()
583 .map(|x| x.to_string())
584 .collect();
585
586 let ret = Helper::like_utf8(names.clone(), "%ll%").unwrap();
587 assert_vector(vec!["hello"], &ret);
588
589 let ret = Helper::like_utf8(names.clone(), "%time").unwrap();
590 assert_vector(vec!["greptime"], &ret);
591
592 let ret = Helper::like_utf8(names.clone(), "%ld").unwrap();
593 assert_vector(vec!["world"], &ret);
594
595 let ret = Helper::like_utf8(names, "%").unwrap();
596 assert_vector(vec!["greptime", "hello", "public", "world"], &ret);
597 }
598
599 #[test]
600 fn test_like_utf8_filter() {
601 fn assert_vector(expected: Vec<&str>, actual: &VectorRef) {
602 let actual = actual.as_any().downcast_ref::<StringVector>().unwrap();
603 assert_eq!(*actual, StringVector::from(expected));
604 }
605
606 fn assert_filter(array: Vec<String>, s: &str, expected_filter: &BooleanVector) {
607 let array = StringArray::from(array);
608 let s = StringArray::new_scalar(s);
609 let actual_filter = comparison::like(&array, &s).unwrap();
610 assert_eq!(BooleanVector::from(actual_filter), *expected_filter);
611 }
612
613 let names: Vec<String> = vec!["greptime", "timeseries", "cloud", "database"]
614 .into_iter()
615 .map(|x| x.to_string())
616 .collect();
617
618 let (table, filter) = Helper::like_utf8_filter(names.clone(), "%ti%").unwrap();
619 assert_vector(vec!["greptime", "timeseries"], &table);
620 assert_filter(names.clone(), "%ti%", &filter);
621
622 let (tables, filter) = Helper::like_utf8_filter(names.clone(), "%lou").unwrap();
623 assert_vector(vec![], &tables);
624 assert_filter(names.clone(), "%lou", &filter);
625
626 let (tables, filter) = Helper::like_utf8_filter(names.clone(), "%d%").unwrap();
627 assert_vector(vec!["cloud", "database"], &tables);
628 assert_filter(names.clone(), "%d%", &filter);
629 }
630
631 fn check_try_into_vector(array: impl Array + 'static) {
632 let array: ArrayRef = Arc::new(array);
633 let vector = Helper::try_into_vector(array.clone()).unwrap();
634 assert_eq!(&array, &vector.to_arrow_array());
635 }
636
637 #[test]
638 fn test_try_into_vector() {
639 check_try_into_vector(NullArray::new(2));
640 check_try_into_vector(BooleanArray::from(vec![true, false]));
641 check_try_into_vector(Int8Array::from(vec![1, 2, 3]));
642 check_try_into_vector(Int16Array::from(vec![1, 2, 3]));
643 check_try_into_vector(Int32Array::from(vec![1, 2, 3]));
644 check_try_into_vector(Int64Array::from(vec![1, 2, 3]));
645 check_try_into_vector(UInt8Array::from(vec![1, 2, 3]));
646 check_try_into_vector(UInt16Array::from(vec![1, 2, 3]));
647 check_try_into_vector(UInt32Array::from(vec![1, 2, 3]));
648 check_try_into_vector(UInt64Array::from(vec![1, 2, 3]));
649 check_try_into_vector(Float32Array::from(vec![1.0, 2.0, 3.0]));
650 check_try_into_vector(Float64Array::from(vec![1.0, 2.0, 3.0]));
651 check_try_into_vector(StringArray::from(vec!["hello", "world"]));
652 check_try_into_vector(Date32Array::from(vec![1, 2, 3]));
653 let data = vec![None, Some(vec![Some(6), Some(7)])];
654 let list_array = ListArray::from_iter_primitive::<Int32Type, _, _>(data);
655 check_try_into_vector(list_array);
656 check_try_into_vector(TimestampSecondArray::from(vec![1, 2, 3]));
657 check_try_into_vector(TimestampMillisecondArray::from(vec![1, 2, 3]));
658 check_try_into_vector(TimestampMicrosecondArray::from(vec![1, 2, 3]));
659 check_try_into_vector(TimestampNanosecondArray::from(vec![1, 2, 3]));
660 check_try_into_vector(Time32SecondArray::from(vec![1, 2, 3]));
661 check_try_into_vector(Time32MillisecondArray::from(vec![1, 2, 3]));
662 check_try_into_vector(Time64MicrosecondArray::from(vec![1, 2, 3]));
663 check_try_into_vector(Time64NanosecondArray::from(vec![1, 2, 3]));
664
665 let values = StringArray::from_iter_values(["a", "b", "c"]);
667
668 let keys = Int8Array::from_iter_values([0, 0, 1, 2]);
670 let array: ArrayRef =
671 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
672 Helper::try_into_vector(array).unwrap();
673
674 let keys = Int16Array::from_iter_values([0, 0, 1, 2]);
676 let array: ArrayRef =
677 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
678 Helper::try_into_vector(array).unwrap();
679
680 let keys = Int32Array::from_iter_values([0, 0, 1, 2]);
682 let array: ArrayRef =
683 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
684 Helper::try_into_vector(array).unwrap();
685
686 let keys = Int64Array::from_iter_values([0, 0, 1, 2]);
688 let array: ArrayRef =
689 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
690 Helper::try_into_vector(array).unwrap();
691
692 let keys = UInt8Array::from_iter_values([0, 0, 1, 2]);
694 let array: ArrayRef =
695 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
696 Helper::try_into_vector(array).unwrap();
697
698 let keys = UInt16Array::from_iter_values([0, 0, 1, 2]);
700 let array: ArrayRef =
701 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
702 Helper::try_into_vector(array).unwrap();
703
704 let keys = UInt32Array::from_iter_values([0, 0, 1, 2]);
706 let array: ArrayRef =
707 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
708 Helper::try_into_vector(array).unwrap();
709
710 let keys = UInt64Array::from_iter_values([0, 0, 1, 2]);
712 let array: ArrayRef = Arc::new(DictionaryArray::try_new(keys, Arc::new(values)).unwrap());
713 Helper::try_into_vector(array).unwrap();
714 }
715
716 #[test]
717 fn test_try_binary_array_into_vector() {
718 let input_vec: Vec<&[u8]> = vec!["hello".as_bytes(), "world".as_bytes()];
719 let assertion_vector = BinaryVector::from(input_vec.clone());
720
721 let input_arrays: Vec<ArrayRef> = vec![
722 Arc::new(LargeBinaryArray::from(input_vec.clone())) as ArrayRef,
723 Arc::new(BinaryArray::from(input_vec.clone())) as ArrayRef,
724 Arc::new(FixedSizeBinaryArray::new(
725 5,
726 Buffer::from_vec("helloworld".as_bytes().to_vec()),
727 None,
728 )) as ArrayRef,
729 ];
730
731 for input_array in input_arrays {
732 let vector = Helper::try_into_vector(input_array).unwrap();
733
734 assert_eq!(2, vector.len());
735 assert_eq!(0, vector.null_count());
736
737 let output_arrow_array: ArrayRef = vector.to_arrow_array();
738 assert_eq!(&DataType::Binary, output_arrow_array.data_type());
739 assert_eq!(&assertion_vector.to_arrow_array(), &output_arrow_array);
740 }
741 }
742
743 #[test]
744 fn test_large_string_array_into_vector() {
745 let input_vec = vec!["a", "b"];
746 let assertion_array = LargeStringArray::from(input_vec.clone());
747
748 let large_string_array: ArrayRef = Arc::new(LargeStringArray::from(input_vec));
749 let vector = Helper::try_into_vector(large_string_array).unwrap();
750 assert_eq!(2, vector.len());
751 assert_eq!(0, vector.null_count());
752
753 let output_arrow_array: LargeStringArray = vector
754 .to_arrow_array()
755 .as_any()
756 .downcast_ref::<LargeStringArray>()
757 .unwrap()
758 .clone();
759 assert_eq!(&assertion_array, &output_arrow_array);
760 }
761
762 #[test]
763 fn test_try_from_scalar_time_value() {
764 let vector = Helper::try_from_scalar_value(ScalarValue::Time32Second(Some(42)), 3).unwrap();
765 assert_eq!(ConcreteDataType::time_second_datatype(), vector.data_type());
766 assert_eq!(3, vector.len());
767 for i in 0..vector.len() {
768 assert_eq!(Value::Time(Time::new_second(42)), vector.get(i));
769 }
770 }
771
772 #[test]
773 fn test_try_from_scalar_interval_value() {
774 let vector = Helper::try_from_scalar_value(
775 ScalarValue::IntervalMonthDayNano(Some(IntervalMonthDayNano::new(1, 1, 2000))),
776 3,
777 )
778 .unwrap();
779
780 assert_eq!(
781 ConcreteDataType::interval_month_day_nano_datatype(),
782 vector.data_type()
783 );
784 assert_eq!(3, vector.len());
785 for i in 0..vector.len() {
786 assert_eq!(
787 Value::IntervalMonthDayNano(IntervalMonthDayNano::new(1, 1, 2000).into()),
788 vector.get(i)
789 );
790 }
791 }
792
793 fn check_try_from_row_to_vector(row: Vec<Value>, dt: &ConcreteDataType) {
794 let vector = Helper::try_from_row_into_vector(&row, dt).unwrap();
795 for (i, item) in row.iter().enumerate().take(vector.len()) {
796 assert_eq!(*item, vector.get(i));
797 }
798 }
799
800 fn check_into_and_from(array: impl Array + 'static) {
801 let array: ArrayRef = Arc::new(array);
802 let vector = Helper::try_into_vector(array.clone()).unwrap();
803 assert_eq!(&array, &vector.to_arrow_array());
804 let row: Vec<Value> = (0..array.len()).map(|i| vector.get(i)).collect();
805 let dt = vector.data_type();
806 check_try_from_row_to_vector(row, &dt);
807 }
808
809 #[test]
810 fn test_try_from_row_to_vector() {
811 check_into_and_from(NullArray::new(2));
812 check_into_and_from(BooleanArray::from(vec![true, false]));
813 check_into_and_from(Int8Array::from(vec![1, 2, 3]));
814 check_into_and_from(Int16Array::from(vec![1, 2, 3]));
815 check_into_and_from(Int32Array::from(vec![1, 2, 3]));
816 check_into_and_from(Int64Array::from(vec![1, 2, 3]));
817 check_into_and_from(UInt8Array::from(vec![1, 2, 3]));
818 check_into_and_from(UInt16Array::from(vec![1, 2, 3]));
819 check_into_and_from(UInt32Array::from(vec![1, 2, 3]));
820 check_into_and_from(UInt64Array::from(vec![1, 2, 3]));
821 check_into_and_from(Float32Array::from(vec![1.0, 2.0, 3.0]));
822 check_into_and_from(Float64Array::from(vec![1.0, 2.0, 3.0]));
823 check_into_and_from(StringArray::from(vec!["hello", "world"]));
824 check_into_and_from(Date32Array::from(vec![1, 2, 3]));
825
826 check_into_and_from(TimestampSecondArray::from(vec![1, 2, 3]));
827 check_into_and_from(TimestampMillisecondArray::from(vec![1, 2, 3]));
828 check_into_and_from(TimestampMicrosecondArray::from(vec![1, 2, 3]));
829 check_into_and_from(TimestampNanosecondArray::from(vec![1, 2, 3]));
830 check_into_and_from(Time32SecondArray::from(vec![1, 2, 3]));
831 check_into_and_from(Time32MillisecondArray::from(vec![1, 2, 3]));
832 check_into_and_from(Time64MicrosecondArray::from(vec![1, 2, 3]));
833 check_into_and_from(Time64NanosecondArray::from(vec![1, 2, 3]));
834 }
835}