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 = ConcreteDataType::try_from(&array.value_type())?;
172 let mut builder = ListVectorBuilder::with_type_capacity(item_type.clone(), 1);
173 let values = ScalarValue::convert_array_to_scalar_vec(array.as_ref())
174 .context(ConvertArrowArrayToScalarsSnafu)?
175 .into_iter()
176 .flatten()
177 .map(ScalarValue::try_into)
178 .collect::<Result<Vec<Value>>>()?;
179 builder.push(Some(ListValueRef::Ref {
180 val: &ListValue::new(values, item_type),
181 }));
182 let list_vector = builder.to_vector();
183 ConstantVector::new(list_vector, length)
184 }
185 ScalarValue::Date32(v) => {
186 ConstantVector::new(Arc::new(DateVector::from(vec![v])), length)
187 }
188 ScalarValue::TimestampSecond(v, _) => {
189 ConstantVector::new(Arc::new(TimestampSecondVector::from(vec![v])), length)
191 }
192 ScalarValue::TimestampMillisecond(v, _) => {
193 ConstantVector::new(Arc::new(TimestampMillisecondVector::from(vec![v])), length)
195 }
196 ScalarValue::TimestampMicrosecond(v, _) => {
197 ConstantVector::new(Arc::new(TimestampMicrosecondVector::from(vec![v])), length)
199 }
200 ScalarValue::TimestampNanosecond(v, _) => {
201 ConstantVector::new(Arc::new(TimestampNanosecondVector::from(vec![v])), length)
203 }
204 ScalarValue::Time32Second(v) => {
205 ConstantVector::new(Arc::new(TimeSecondVector::from(vec![v])), length)
206 }
207 ScalarValue::Time32Millisecond(v) => {
208 ConstantVector::new(Arc::new(TimeMillisecondVector::from(vec![v])), length)
209 }
210 ScalarValue::Time64Microsecond(v) => {
211 ConstantVector::new(Arc::new(TimeMicrosecondVector::from(vec![v])), length)
212 }
213 ScalarValue::Time64Nanosecond(v) => {
214 ConstantVector::new(Arc::new(TimeNanosecondVector::from(vec![v])), length)
215 }
216 ScalarValue::IntervalYearMonth(v) => {
217 ConstantVector::new(Arc::new(IntervalYearMonthVector::from(vec![v])), length)
218 }
219 ScalarValue::IntervalDayTime(v) => {
220 ConstantVector::new(Arc::new(IntervalDayTimeVector::from(vec![v])), length)
221 }
222 ScalarValue::IntervalMonthDayNano(v) => {
223 ConstantVector::new(Arc::new(IntervalMonthDayNanoVector::from(vec![v])), length)
224 }
225 ScalarValue::DurationSecond(v) => {
226 ConstantVector::new(Arc::new(DurationSecondVector::from(vec![v])), length)
227 }
228 ScalarValue::DurationMillisecond(v) => {
229 ConstantVector::new(Arc::new(DurationMillisecondVector::from(vec![v])), length)
230 }
231 ScalarValue::DurationMicrosecond(v) => {
232 ConstantVector::new(Arc::new(DurationMicrosecondVector::from(vec![v])), length)
233 }
234 ScalarValue::DurationNanosecond(v) => {
235 ConstantVector::new(Arc::new(DurationNanosecondVector::from(vec![v])), length)
236 }
237 ScalarValue::Decimal128(v, p, s) => {
238 let vector = Decimal128Vector::from(vec![v]).with_precision_and_scale(p, s)?;
239 ConstantVector::new(Arc::new(vector), length)
240 }
241 ScalarValue::Struct(v) => {
242 let struct_type = StructType::try_from(v.fields())?;
243 ConstantVector::new(
244 Arc::new(StructVector::try_new(struct_type, (*v).clone())?),
245 length,
246 )
247 }
248 ScalarValue::Decimal256(_, _, _)
249 | ScalarValue::FixedSizeList(_)
250 | ScalarValue::LargeList(_)
251 | ScalarValue::Dictionary(_, _)
252 | ScalarValue::Union(_, _, _)
253 | ScalarValue::Utf8View(_)
254 | ScalarValue::BinaryView(_)
255 | ScalarValue::Map(_)
256 | ScalarValue::Date64(_) => {
257 return error::ConversionSnafu {
258 from: format!("Unsupported scalar value: {value}"),
259 }
260 .fail();
261 }
262 };
263
264 Ok(Arc::new(vector))
265 }
266
267 pub fn try_into_vector(array: impl AsRef<dyn Array>) -> Result<VectorRef> {
272 Ok(match array.as_ref().data_type() {
273 ArrowDataType::Null => Arc::new(NullVector::try_from_arrow_array(array)?),
274 ArrowDataType::Boolean => Arc::new(BooleanVector::try_from_arrow_array(array)?),
275 ArrowDataType::Binary => Arc::new(BinaryVector::try_from_arrow_array(array)?),
276 ArrowDataType::LargeBinary
277 | ArrowDataType::FixedSizeBinary(_)
278 | ArrowDataType::BinaryView => {
279 let array = arrow::compute::cast(array.as_ref(), &ArrowDataType::Binary)
280 .context(crate::error::ArrowComputeSnafu)?;
281 Arc::new(BinaryVector::try_from_arrow_array(array)?)
282 }
283 ArrowDataType::Int8 => Arc::new(Int8Vector::try_from_arrow_array(array)?),
284 ArrowDataType::Int16 => Arc::new(Int16Vector::try_from_arrow_array(array)?),
285 ArrowDataType::Int32 => Arc::new(Int32Vector::try_from_arrow_array(array)?),
286 ArrowDataType::Int64 => Arc::new(Int64Vector::try_from_arrow_array(array)?),
287 ArrowDataType::UInt8 => Arc::new(UInt8Vector::try_from_arrow_array(array)?),
288 ArrowDataType::UInt16 => Arc::new(UInt16Vector::try_from_arrow_array(array)?),
289 ArrowDataType::UInt32 => Arc::new(UInt32Vector::try_from_arrow_array(array)?),
290 ArrowDataType::UInt64 => Arc::new(UInt64Vector::try_from_arrow_array(array)?),
291 ArrowDataType::Float32 => Arc::new(Float32Vector::try_from_arrow_array(array)?),
292 ArrowDataType::Float64 => Arc::new(Float64Vector::try_from_arrow_array(array)?),
293 ArrowDataType::Utf8 => Arc::new(StringVector::try_from_arrow_array(array)?),
294 ArrowDataType::LargeUtf8 | ArrowDataType::Utf8View => {
295 let array = arrow::compute::cast(array.as_ref(), &ArrowDataType::Utf8)
296 .context(crate::error::ArrowComputeSnafu)?;
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)]
465mod tests {
466 use arrow::array::{
467 ArrayRef, BooleanArray, Date32Array, Float32Array, Float64Array, Int8Array, Int16Array,
468 Int32Array, Int64Array, LargeBinaryArray, ListArray, NullArray, Time32MillisecondArray,
469 Time32SecondArray, Time64MicrosecondArray, Time64NanosecondArray,
470 TimestampMicrosecondArray, TimestampMillisecondArray, TimestampNanosecondArray,
471 TimestampSecondArray, UInt8Array, UInt16Array, UInt32Array, UInt64Array,
472 };
473 use arrow::buffer::Buffer;
474 use arrow::datatypes::{Int32Type, IntervalMonthDayNano};
475 use arrow_array::{BinaryArray, DictionaryArray, FixedSizeBinaryArray, LargeStringArray};
476 use arrow_schema::DataType;
477 use common_decimal::Decimal128;
478 use common_time::time::Time;
479 use common_time::timestamp::TimeUnit;
480 use common_time::{Date, Duration};
481
482 use super::*;
483 use crate::value::Value;
484 use crate::vectors::ConcreteDataType;
485
486 #[test]
487 fn test_try_into_vectors() {
488 let arrays: Vec<ArrayRef> = vec![
489 Arc::new(Int32Array::from(vec![1])),
490 Arc::new(Int32Array::from(vec![2])),
491 Arc::new(Int32Array::from(vec![3])),
492 ];
493 let vectors = Helper::try_into_vectors(&arrays).unwrap();
494 vectors.iter().for_each(|v| assert_eq!(1, v.len()));
495 assert_eq!(Value::Int32(1), vectors[0].get(0));
496 assert_eq!(Value::Int32(2), vectors[1].get(0));
497 assert_eq!(Value::Int32(3), vectors[2].get(0));
498 }
499
500 #[test]
501 fn test_try_into_date_vector() {
502 let vector = DateVector::from(vec![Some(1), Some(2), None]);
503 let arrow_array = vector.to_arrow_array();
504 assert_eq!(&ArrowDataType::Date32, arrow_array.data_type());
505 let vector_converted = Helper::try_into_vector(arrow_array).unwrap();
506 assert_eq!(vector.len(), vector_converted.len());
507 for i in 0..vector_converted.len() {
508 assert_eq!(vector.get(i), vector_converted.get(i));
509 }
510 }
511
512 #[test]
513 fn test_try_from_scalar_date_value() {
514 let vector = Helper::try_from_scalar_value(ScalarValue::Date32(Some(42)), 3).unwrap();
515 assert_eq!(ConcreteDataType::date_datatype(), vector.data_type());
516 assert_eq!(3, vector.len());
517 for i in 0..vector.len() {
518 assert_eq!(Value::Date(Date::new(42)), vector.get(i));
519 }
520 }
521
522 #[test]
523 fn test_try_from_scalar_duration_value() {
524 let vector =
525 Helper::try_from_scalar_value(ScalarValue::DurationSecond(Some(42)), 3).unwrap();
526 assert_eq!(
527 ConcreteDataType::duration_second_datatype(),
528 vector.data_type()
529 );
530 assert_eq!(3, vector.len());
531 for i in 0..vector.len() {
532 assert_eq!(
533 Value::Duration(Duration::new(42, TimeUnit::Second)),
534 vector.get(i)
535 );
536 }
537 }
538
539 #[test]
540 fn test_try_from_scalar_decimal128_value() {
541 let vector =
542 Helper::try_from_scalar_value(ScalarValue::Decimal128(Some(42), 3, 1), 3).unwrap();
543 assert_eq!(
544 ConcreteDataType::decimal128_datatype(3, 1),
545 vector.data_type()
546 );
547 assert_eq!(3, vector.len());
548 for i in 0..vector.len() {
549 assert_eq!(Value::Decimal128(Decimal128::new(42, 3, 1)), vector.get(i));
550 }
551 }
552
553 #[test]
554 fn test_try_from_list_value() {
555 let value = ScalarValue::List(ScalarValue::new_list(
556 &[ScalarValue::Int32(Some(1)), ScalarValue::Int32(Some(2))],
557 &ArrowDataType::Int32,
558 true,
559 ));
560 let vector = Helper::try_from_scalar_value(value, 3).unwrap();
561 assert_eq!(
562 ConcreteDataType::list_datatype(ConcreteDataType::int32_datatype()),
563 vector.data_type()
564 );
565 assert_eq!(3, vector.len());
566 for i in 0..vector.len() {
567 let v = vector.get(i);
568 let items = v.as_list().unwrap().unwrap().items();
569 assert_eq!(vec![Value::Int32(1), Value::Int32(2)], items);
570 }
571 }
572
573 #[test]
574 fn test_like_utf8() {
575 fn assert_vector(expected: Vec<&str>, actual: &VectorRef) {
576 let actual = actual.as_any().downcast_ref::<StringVector>().unwrap();
577 assert_eq!(*actual, StringVector::from(expected));
578 }
579
580 let names: Vec<String> = vec!["greptime", "hello", "public", "world"]
581 .into_iter()
582 .map(|x| x.to_string())
583 .collect();
584
585 let ret = Helper::like_utf8(names.clone(), "%ll%").unwrap();
586 assert_vector(vec!["hello"], &ret);
587
588 let ret = Helper::like_utf8(names.clone(), "%time").unwrap();
589 assert_vector(vec!["greptime"], &ret);
590
591 let ret = Helper::like_utf8(names.clone(), "%ld").unwrap();
592 assert_vector(vec!["world"], &ret);
593
594 let ret = Helper::like_utf8(names, "%").unwrap();
595 assert_vector(vec!["greptime", "hello", "public", "world"], &ret);
596 }
597
598 #[test]
599 fn test_like_utf8_filter() {
600 fn assert_vector(expected: Vec<&str>, actual: &VectorRef) {
601 let actual = actual.as_any().downcast_ref::<StringVector>().unwrap();
602 assert_eq!(*actual, StringVector::from(expected));
603 }
604
605 fn assert_filter(array: Vec<String>, s: &str, expected_filter: &BooleanVector) {
606 let array = StringArray::from(array);
607 let s = StringArray::new_scalar(s);
608 let actual_filter = comparison::like(&array, &s).unwrap();
609 assert_eq!(BooleanVector::from(actual_filter), *expected_filter);
610 }
611
612 let names: Vec<String> = vec!["greptime", "timeseries", "cloud", "database"]
613 .into_iter()
614 .map(|x| x.to_string())
615 .collect();
616
617 let (table, filter) = Helper::like_utf8_filter(names.clone(), "%ti%").unwrap();
618 assert_vector(vec!["greptime", "timeseries"], &table);
619 assert_filter(names.clone(), "%ti%", &filter);
620
621 let (tables, filter) = Helper::like_utf8_filter(names.clone(), "%lou").unwrap();
622 assert_vector(vec![], &tables);
623 assert_filter(names.clone(), "%lou", &filter);
624
625 let (tables, filter) = Helper::like_utf8_filter(names.clone(), "%d%").unwrap();
626 assert_vector(vec!["cloud", "database"], &tables);
627 assert_filter(names.clone(), "%d%", &filter);
628 }
629
630 fn check_try_into_vector(array: impl Array + 'static) {
631 let array: ArrayRef = Arc::new(array);
632 let vector = Helper::try_into_vector(array.clone()).unwrap();
633 assert_eq!(&array, &vector.to_arrow_array());
634 }
635
636 #[test]
637 fn test_try_into_vector() {
638 check_try_into_vector(NullArray::new(2));
639 check_try_into_vector(BooleanArray::from(vec![true, false]));
640 check_try_into_vector(Int8Array::from(vec![1, 2, 3]));
641 check_try_into_vector(Int16Array::from(vec![1, 2, 3]));
642 check_try_into_vector(Int32Array::from(vec![1, 2, 3]));
643 check_try_into_vector(Int64Array::from(vec![1, 2, 3]));
644 check_try_into_vector(UInt8Array::from(vec![1, 2, 3]));
645 check_try_into_vector(UInt16Array::from(vec![1, 2, 3]));
646 check_try_into_vector(UInt32Array::from(vec![1, 2, 3]));
647 check_try_into_vector(UInt64Array::from(vec![1, 2, 3]));
648 check_try_into_vector(Float32Array::from(vec![1.0, 2.0, 3.0]));
649 check_try_into_vector(Float64Array::from(vec![1.0, 2.0, 3.0]));
650 check_try_into_vector(StringArray::from(vec!["hello", "world"]));
651 check_try_into_vector(Date32Array::from(vec![1, 2, 3]));
652 let data = vec![None, Some(vec![Some(6), Some(7)])];
653 let list_array = ListArray::from_iter_primitive::<Int32Type, _, _>(data);
654 check_try_into_vector(list_array);
655 check_try_into_vector(TimestampSecondArray::from(vec![1, 2, 3]));
656 check_try_into_vector(TimestampMillisecondArray::from(vec![1, 2, 3]));
657 check_try_into_vector(TimestampMicrosecondArray::from(vec![1, 2, 3]));
658 check_try_into_vector(TimestampNanosecondArray::from(vec![1, 2, 3]));
659 check_try_into_vector(Time32SecondArray::from(vec![1, 2, 3]));
660 check_try_into_vector(Time32MillisecondArray::from(vec![1, 2, 3]));
661 check_try_into_vector(Time64MicrosecondArray::from(vec![1, 2, 3]));
662 check_try_into_vector(Time64NanosecondArray::from(vec![1, 2, 3]));
663
664 let values = StringArray::from_iter_values(["a", "b", "c"]);
666
667 let keys = Int8Array::from_iter_values([0, 0, 1, 2]);
669 let array: ArrayRef =
670 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
671 Helper::try_into_vector(array).unwrap();
672
673 let keys = Int16Array::from_iter_values([0, 0, 1, 2]);
675 let array: ArrayRef =
676 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
677 Helper::try_into_vector(array).unwrap();
678
679 let keys = Int32Array::from_iter_values([0, 0, 1, 2]);
681 let array: ArrayRef =
682 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
683 Helper::try_into_vector(array).unwrap();
684
685 let keys = Int64Array::from_iter_values([0, 0, 1, 2]);
687 let array: ArrayRef =
688 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
689 Helper::try_into_vector(array).unwrap();
690
691 let keys = UInt8Array::from_iter_values([0, 0, 1, 2]);
693 let array: ArrayRef =
694 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
695 Helper::try_into_vector(array).unwrap();
696
697 let keys = UInt16Array::from_iter_values([0, 0, 1, 2]);
699 let array: ArrayRef =
700 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
701 Helper::try_into_vector(array).unwrap();
702
703 let keys = UInt32Array::from_iter_values([0, 0, 1, 2]);
705 let array: ArrayRef =
706 Arc::new(DictionaryArray::try_new(keys, Arc::new(values.clone())).unwrap());
707 Helper::try_into_vector(array).unwrap();
708
709 let keys = UInt64Array::from_iter_values([0, 0, 1, 2]);
711 let array: ArrayRef = Arc::new(DictionaryArray::try_new(keys, Arc::new(values)).unwrap());
712 Helper::try_into_vector(array).unwrap();
713 }
714
715 #[test]
716 fn test_try_binary_array_into_vector() {
717 let input_vec: Vec<&[u8]> = vec!["hello".as_bytes(), "world".as_bytes()];
718 let assertion_vector = BinaryVector::from(input_vec.clone());
719
720 let input_arrays: Vec<ArrayRef> = vec![
721 Arc::new(LargeBinaryArray::from(input_vec.clone())) as ArrayRef,
722 Arc::new(BinaryArray::from(input_vec.clone())) as ArrayRef,
723 Arc::new(FixedSizeBinaryArray::new(
724 5,
725 Buffer::from_vec("helloworld".as_bytes().to_vec()),
726 None,
727 )) as ArrayRef,
728 ];
729
730 for input_array in input_arrays {
731 let vector = Helper::try_into_vector(input_array).unwrap();
732
733 assert_eq!(2, vector.len());
734 assert_eq!(0, vector.null_count());
735
736 let output_arrow_array: ArrayRef = vector.to_arrow_array();
737 assert_eq!(&DataType::Binary, output_arrow_array.data_type());
738 assert_eq!(&assertion_vector.to_arrow_array(), &output_arrow_array);
739 }
740 }
741
742 #[test]
743 fn test_large_string_array_into_vector() {
744 let input_vec = vec!["a", "b"];
745 let assertion_array = StringArray::from(input_vec.clone());
746
747 let large_string_array: ArrayRef = Arc::new(LargeStringArray::from(input_vec));
748 let vector = Helper::try_into_vector(large_string_array).unwrap();
749 assert_eq!(2, vector.len());
750 assert_eq!(0, vector.null_count());
751
752 let output_arrow_array: StringArray = vector
753 .to_arrow_array()
754 .as_any()
755 .downcast_ref::<StringArray>()
756 .unwrap()
757 .clone();
758 assert_eq!(&assertion_array, &output_arrow_array);
759 }
760
761 #[test]
762 fn test_try_from_scalar_time_value() {
763 let vector = Helper::try_from_scalar_value(ScalarValue::Time32Second(Some(42)), 3).unwrap();
764 assert_eq!(ConcreteDataType::time_second_datatype(), vector.data_type());
765 assert_eq!(3, vector.len());
766 for i in 0..vector.len() {
767 assert_eq!(Value::Time(Time::new_second(42)), vector.get(i));
768 }
769 }
770
771 #[test]
772 fn test_try_from_scalar_interval_value() {
773 let vector = Helper::try_from_scalar_value(
774 ScalarValue::IntervalMonthDayNano(Some(IntervalMonthDayNano::new(1, 1, 2000))),
775 3,
776 )
777 .unwrap();
778
779 assert_eq!(
780 ConcreteDataType::interval_month_day_nano_datatype(),
781 vector.data_type()
782 );
783 assert_eq!(3, vector.len());
784 for i in 0..vector.len() {
785 assert_eq!(
786 Value::IntervalMonthDayNano(IntervalMonthDayNano::new(1, 1, 2000).into()),
787 vector.get(i)
788 );
789 }
790 }
791
792 fn check_try_from_row_to_vector(row: Vec<Value>, dt: &ConcreteDataType) {
793 let vector = Helper::try_from_row_into_vector(&row, dt).unwrap();
794 for (i, item) in row.iter().enumerate().take(vector.len()) {
795 assert_eq!(*item, vector.get(i));
796 }
797 }
798
799 fn check_into_and_from(array: impl Array + 'static) {
800 let array: ArrayRef = Arc::new(array);
801 let vector = Helper::try_into_vector(array.clone()).unwrap();
802 assert_eq!(&array, &vector.to_arrow_array());
803 let row: Vec<Value> = (0..array.len()).map(|i| vector.get(i)).collect();
804 let dt = vector.data_type();
805 check_try_from_row_to_vector(row, &dt);
806 }
807
808 #[test]
809 fn test_try_from_row_to_vector() {
810 check_into_and_from(NullArray::new(2));
811 check_into_and_from(BooleanArray::from(vec![true, false]));
812 check_into_and_from(Int8Array::from(vec![1, 2, 3]));
813 check_into_and_from(Int16Array::from(vec![1, 2, 3]));
814 check_into_and_from(Int32Array::from(vec![1, 2, 3]));
815 check_into_and_from(Int64Array::from(vec![1, 2, 3]));
816 check_into_and_from(UInt8Array::from(vec![1, 2, 3]));
817 check_into_and_from(UInt16Array::from(vec![1, 2, 3]));
818 check_into_and_from(UInt32Array::from(vec![1, 2, 3]));
819 check_into_and_from(UInt64Array::from(vec![1, 2, 3]));
820 check_into_and_from(Float32Array::from(vec![1.0, 2.0, 3.0]));
821 check_into_and_from(Float64Array::from(vec![1.0, 2.0, 3.0]));
822 check_into_and_from(StringArray::from(vec!["hello", "world"]));
823 check_into_and_from(Date32Array::from(vec![1, 2, 3]));
824
825 check_into_and_from(TimestampSecondArray::from(vec![1, 2, 3]));
826 check_into_and_from(TimestampMillisecondArray::from(vec![1, 2, 3]));
827 check_into_and_from(TimestampMicrosecondArray::from(vec![1, 2, 3]));
828 check_into_and_from(TimestampNanosecondArray::from(vec![1, 2, 3]));
829 check_into_and_from(Time32SecondArray::from(vec![1, 2, 3]));
830 check_into_and_from(Time32MillisecondArray::from(vec![1, 2, 3]));
831 check_into_and_from(Time64MicrosecondArray::from(vec![1, 2, 3]));
832 check_into_and_from(Time64NanosecondArray::from(vec![1, 2, 3]));
833 }
834}