1use std::fmt;
16use std::sync::Arc;
17
18use arrow::datatypes::{
19 DataType as ArrowDataType, IntervalUnit as ArrowIntervalUnit, TimeUnit as ArrowTimeUnit,
20};
21use arrow_schema::DECIMAL_DEFAULT_SCALE;
22use common_decimal::decimal128::DECIMAL128_MAX_PRECISION;
23use common_time::interval::IntervalUnit;
24use common_time::timestamp::TimeUnit;
25use enum_dispatch::enum_dispatch;
26use paste::paste;
27use serde::{Deserialize, Serialize};
28
29use crate::error::{self, Error, Result};
30use crate::type_id::LogicalTypeId;
31use crate::types::json_type::JsonNativeType;
32use crate::types::{
33 BinaryType, BooleanType, DateType, Decimal128Type, DictionaryType, DurationMicrosecondType,
34 DurationMillisecondType, DurationNanosecondType, DurationSecondType, DurationType, Float32Type,
35 Float64Type, Int8Type, Int16Type, Int32Type, Int64Type, IntervalDayTimeType,
36 IntervalMonthDayNanoType, IntervalType, IntervalYearMonthType, JsonType, ListType, NullType,
37 StringType, StructType, TimeMillisecondType, TimeType, TimestampMicrosecondType,
38 TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType, TimestampType,
39 UInt8Type, UInt16Type, UInt32Type, UInt64Type, VectorType,
40};
41use crate::value::Value;
42use crate::vectors::MutableVector;
43
44#[derive(Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
45#[enum_dispatch(DataType)]
46pub enum ConcreteDataType {
47 Null(NullType),
48 Boolean(BooleanType),
49
50 Int8(Int8Type),
52 Int16(Int16Type),
53 Int32(Int32Type),
54 Int64(Int64Type),
55 UInt8(UInt8Type),
56 UInt16(UInt16Type),
57 UInt32(UInt32Type),
58 UInt64(UInt64Type),
59 Float32(Float32Type),
60 Float64(Float64Type),
61
62 Decimal128(Decimal128Type),
64
65 Binary(BinaryType),
67 String(StringType),
68
69 Date(DateType),
71 Timestamp(TimestampType),
72 Time(TimeType),
73
74 Duration(DurationType),
76
77 Interval(IntervalType),
79
80 List(ListType),
82 Dictionary(DictionaryType),
83 Struct(StructType),
84
85 Json(JsonType),
87
88 Vector(VectorType),
90}
91
92impl fmt::Display for ConcreteDataType {
93 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
94 match self {
95 ConcreteDataType::Null(v) => write!(f, "{}", v.name()),
96 ConcreteDataType::Boolean(v) => write!(f, "{}", v.name()),
97 ConcreteDataType::Int8(v) => write!(f, "{}", v.name()),
98 ConcreteDataType::Int16(v) => write!(f, "{}", v.name()),
99 ConcreteDataType::Int32(v) => write!(f, "{}", v.name()),
100 ConcreteDataType::Int64(v) => write!(f, "{}", v.name()),
101 ConcreteDataType::UInt8(v) => write!(f, "{}", v.name()),
102 ConcreteDataType::UInt16(v) => write!(f, "{}", v.name()),
103 ConcreteDataType::UInt32(v) => write!(f, "{}", v.name()),
104 ConcreteDataType::UInt64(v) => write!(f, "{}", v.name()),
105 ConcreteDataType::Float32(v) => write!(f, "{}", v.name()),
106 ConcreteDataType::Float64(v) => write!(f, "{}", v.name()),
107 ConcreteDataType::Binary(v) => write!(f, "{}", v.name()),
108 ConcreteDataType::String(v) => write!(f, "{}", v.name()),
109 ConcreteDataType::Date(v) => write!(f, "{}", v.name()),
110 ConcreteDataType::Timestamp(t) => match t {
111 TimestampType::Second(v) => write!(f, "{}", v.name()),
112 TimestampType::Millisecond(v) => write!(f, "{}", v.name()),
113 TimestampType::Microsecond(v) => write!(f, "{}", v.name()),
114 TimestampType::Nanosecond(v) => write!(f, "{}", v.name()),
115 },
116 ConcreteDataType::Time(t) => match t {
117 TimeType::Second(v) => write!(f, "{}", v.name()),
118 TimeType::Millisecond(v) => write!(f, "{}", v.name()),
119 TimeType::Microsecond(v) => write!(f, "{}", v.name()),
120 TimeType::Nanosecond(v) => write!(f, "{}", v.name()),
121 },
122 ConcreteDataType::Interval(i) => match i {
123 IntervalType::YearMonth(v) => write!(f, "{}", v.name()),
124 IntervalType::DayTime(v) => write!(f, "{}", v.name()),
125 IntervalType::MonthDayNano(v) => write!(f, "{}", v.name()),
126 },
127 ConcreteDataType::Duration(d) => match d {
128 DurationType::Second(v) => write!(f, "{}", v.name()),
129 DurationType::Millisecond(v) => write!(f, "{}", v.name()),
130 DurationType::Microsecond(v) => write!(f, "{}", v.name()),
131 DurationType::Nanosecond(v) => write!(f, "{}", v.name()),
132 },
133 ConcreteDataType::Decimal128(v) => write!(f, "{}", v.name()),
134 ConcreteDataType::List(v) => write!(f, "{}", v.name()),
135 ConcreteDataType::Struct(v) => write!(f, "{}", v.name()),
136 ConcreteDataType::Dictionary(v) => write!(f, "{}", v.name()),
137 ConcreteDataType::Json(v) => write!(f, "{}", v.name()),
138 ConcreteDataType::Vector(v) => write!(f, "{}", v.name()),
139 }
140 }
141}
142
143impl ConcreteDataType {
146 pub fn is_float(&self) -> bool {
147 matches!(
148 self,
149 ConcreteDataType::Float64(_) | ConcreteDataType::Float32(_)
150 )
151 }
152
153 pub fn is_boolean(&self) -> bool {
154 matches!(self, ConcreteDataType::Boolean(_))
155 }
156
157 pub fn is_string(&self) -> bool {
158 matches!(self, ConcreteDataType::String(_))
159 }
160
161 pub fn is_stringifiable(&self) -> bool {
162 matches!(
163 self,
164 ConcreteDataType::String(_)
165 | ConcreteDataType::Date(_)
166 | ConcreteDataType::Timestamp(_)
167 | ConcreteDataType::Time(_)
168 | ConcreteDataType::Interval(_)
169 | ConcreteDataType::Duration(_)
170 | ConcreteDataType::Decimal128(_)
171 | ConcreteDataType::Binary(_)
172 | ConcreteDataType::Json(_)
173 | ConcreteDataType::Vector(_)
174 )
175 }
176
177 pub fn is_signed(&self) -> bool {
178 matches!(
179 self,
180 ConcreteDataType::Int8(_)
181 | ConcreteDataType::Int16(_)
182 | ConcreteDataType::Int32(_)
183 | ConcreteDataType::Int64(_)
184 | ConcreteDataType::Date(_)
185 | ConcreteDataType::Timestamp(_)
186 | ConcreteDataType::Time(_)
187 | ConcreteDataType::Interval(_)
188 | ConcreteDataType::Duration(_)
189 | ConcreteDataType::Decimal128(_)
190 )
191 }
192
193 pub fn is_unsigned(&self) -> bool {
194 matches!(
195 self,
196 ConcreteDataType::UInt8(_)
197 | ConcreteDataType::UInt16(_)
198 | ConcreteDataType::UInt32(_)
199 | ConcreteDataType::UInt64(_)
200 )
201 }
202
203 pub fn is_numeric(&self) -> bool {
204 matches!(
205 self,
206 ConcreteDataType::Int8(_)
207 | ConcreteDataType::Int16(_)
208 | ConcreteDataType::Int32(_)
209 | ConcreteDataType::Int64(_)
210 | ConcreteDataType::UInt8(_)
211 | ConcreteDataType::UInt16(_)
212 | ConcreteDataType::UInt32(_)
213 | ConcreteDataType::UInt64(_)
214 | ConcreteDataType::Float32(_)
215 | ConcreteDataType::Float64(_)
216 )
217 }
218
219 pub fn is_timestamp(&self) -> bool {
220 matches!(self, ConcreteDataType::Timestamp(_))
221 }
222
223 pub fn is_decimal(&self) -> bool {
224 matches!(self, ConcreteDataType::Decimal128(_))
225 }
226
227 pub fn is_json(&self) -> bool {
228 matches!(self, ConcreteDataType::Json(_))
229 }
230
231 pub fn is_json2(&self) -> bool {
233 self.as_json().is_some_and(|json_type| json_type.is_json2())
234 }
235
236 pub fn is_vector(&self) -> bool {
237 matches!(self, ConcreteDataType::Vector(_))
238 }
239
240 pub fn numerics() -> Vec<ConcreteDataType> {
241 vec![
242 ConcreteDataType::int8_datatype(),
243 ConcreteDataType::int16_datatype(),
244 ConcreteDataType::int32_datatype(),
245 ConcreteDataType::int64_datatype(),
246 ConcreteDataType::uint8_datatype(),
247 ConcreteDataType::uint16_datatype(),
248 ConcreteDataType::uint32_datatype(),
249 ConcreteDataType::uint64_datatype(),
250 ConcreteDataType::float32_datatype(),
251 ConcreteDataType::float64_datatype(),
252 ]
253 }
254
255 pub fn unsigned_integers() -> Vec<ConcreteDataType> {
256 vec![
257 ConcreteDataType::uint8_datatype(),
258 ConcreteDataType::uint16_datatype(),
259 ConcreteDataType::uint32_datatype(),
260 ConcreteDataType::uint64_datatype(),
261 ]
262 }
263
264 pub fn timestamps() -> Vec<ConcreteDataType> {
265 vec![
266 ConcreteDataType::timestamp_second_datatype(),
267 ConcreteDataType::timestamp_millisecond_datatype(),
268 ConcreteDataType::timestamp_microsecond_datatype(),
269 ConcreteDataType::timestamp_nanosecond_datatype(),
270 ]
271 }
272
273 pub fn from_arrow_type(dt: &ArrowDataType) -> Self {
278 ConcreteDataType::try_from(dt).expect("Unimplemented type")
279 }
280
281 pub fn is_null(&self) -> bool {
282 matches!(self, ConcreteDataType::Null(NullType))
283 }
284
285 pub fn as_list(&self) -> Option<&ListType> {
287 match self {
288 ConcreteDataType::List(t) => Some(t),
289 _ => None,
290 }
291 }
292
293 pub fn as_struct(&self) -> Option<&StructType> {
294 match self {
295 ConcreteDataType::Struct(s) => Some(s),
296 _ => None,
297 }
298 }
299
300 pub fn as_timestamp(&self) -> Option<TimestampType> {
302 match self {
303 ConcreteDataType::Timestamp(t) => Some(*t),
304 _ => None,
305 }
306 }
307
308 pub fn numeric_precision(&self) -> Option<u8> {
310 match self {
311 ConcreteDataType::Int8(_) | ConcreteDataType::UInt8(_) => Some(3),
312 ConcreteDataType::Int16(_) | ConcreteDataType::UInt16(_) => Some(5),
313 ConcreteDataType::Int32(_) | ConcreteDataType::UInt32(_) => Some(10),
314 ConcreteDataType::Int64(_) => Some(19),
315 ConcreteDataType::UInt64(_) => Some(20),
316 ConcreteDataType::Float32(_) => Some(12),
317 ConcreteDataType::Float64(_) => Some(22),
318 ConcreteDataType::Decimal128(decimal_type) => Some(decimal_type.precision()),
319 _ => None,
320 }
321 }
322
323 pub fn numeric_scale(&self) -> Option<i8> {
325 match self {
326 ConcreteDataType::Int8(_)
327 | ConcreteDataType::UInt8(_)
328 | ConcreteDataType::Int16(_)
329 | ConcreteDataType::UInt16(_)
330 | ConcreteDataType::Int32(_)
331 | ConcreteDataType::UInt32(_)
332 | ConcreteDataType::Int64(_)
333 | ConcreteDataType::UInt64(_) => Some(0),
334 ConcreteDataType::Float32(_) | ConcreteDataType::Float64(_) => None,
335 ConcreteDataType::Decimal128(decimal_type) => Some(decimal_type.scale()),
336 _ => None,
337 }
338 }
339
340 pub fn as_time(&self) -> Option<TimeType> {
342 match self {
343 ConcreteDataType::Int64(_) => Some(TimeType::Millisecond(TimeMillisecondType)),
344 ConcreteDataType::Time(t) => Some(*t),
345 _ => None,
346 }
347 }
348
349 pub fn as_decimal128(&self) -> Option<Decimal128Type> {
350 match self {
351 ConcreteDataType::Decimal128(d) => Some(*d),
352 _ => None,
353 }
354 }
355
356 pub fn as_json(&self) -> Option<&JsonType> {
357 match self {
358 ConcreteDataType::Json(j) => Some(j),
359 _ => None,
360 }
361 }
362
363 pub fn as_vector(&self) -> Option<VectorType> {
364 match self {
365 ConcreteDataType::Vector(v) => Some(*v),
366 _ => None,
367 }
368 }
369
370 pub fn can_arrow_type_cast_to(&self, to_type: &ConcreteDataType) -> bool {
372 match (self, to_type) {
373 (ConcreteDataType::Json(this), ConcreteDataType::Json(that)) => that.is_include(this),
374 _ => arrow::compute::can_cast_types(&self.as_arrow_type(), &to_type.as_arrow_type()),
375 }
376 }
377
378 pub fn is_timestamp_unit_widening_to(&self, to_type: &ConcreteDataType) -> bool {
383 match (self.as_timestamp(), to_type.as_timestamp()) {
384 (Some(from), Some(to)) => to.unit() > from.unit(),
385 _ => false,
386 }
387 }
388
389 pub fn as_duration(&self) -> Option<DurationType> {
391 match self {
392 ConcreteDataType::Duration(d) => Some(*d),
393 _ => None,
394 }
395 }
396
397 pub fn postgres_datatype_name(&self) -> &'static str {
399 match self {
400 &ConcreteDataType::Null(_) => "UNKNOWN",
401 &ConcreteDataType::Boolean(_) => "BOOL",
402 &ConcreteDataType::Int8(_) | &ConcreteDataType::UInt8(_) => "CHAR",
403 &ConcreteDataType::Int16(_) | &ConcreteDataType::UInt16(_) => "INT2",
404 &ConcreteDataType::Int32(_) | &ConcreteDataType::UInt32(_) => "INT4",
405 &ConcreteDataType::Int64(_) | &ConcreteDataType::UInt64(_) => "INT8",
406 &ConcreteDataType::Float32(_) => "FLOAT4",
407 &ConcreteDataType::Float64(_) => "FLOAT8",
408 &ConcreteDataType::Binary(_) | &ConcreteDataType::Vector(_) => "BYTEA",
409 &ConcreteDataType::String(_) => "VARCHAR",
410 &ConcreteDataType::Date(_) => "DATE",
411 &ConcreteDataType::Timestamp(_) => "TIMESTAMP",
412 &ConcreteDataType::Time(_) => "TIME",
413 &ConcreteDataType::Interval(_) => "INTERVAL",
414 &ConcreteDataType::Decimal128(_) => "NUMERIC",
415 &ConcreteDataType::Json(_) => "JSON",
416 ConcreteDataType::List(list) => match list.item_type() {
417 &ConcreteDataType::Null(_) => "UNKNOWN",
418 &ConcreteDataType::Boolean(_) => "_BOOL",
419 &ConcreteDataType::Int8(_) | &ConcreteDataType::UInt8(_) => "_CHAR",
420 &ConcreteDataType::Int16(_) | &ConcreteDataType::UInt16(_) => "_INT2",
421 &ConcreteDataType::Int32(_) | &ConcreteDataType::UInt32(_) => "_INT4",
422 &ConcreteDataType::Int64(_) | &ConcreteDataType::UInt64(_) => "_INT8",
423 &ConcreteDataType::Float32(_) => "_FLOAT4",
424 &ConcreteDataType::Float64(_) => "_FLOAT8",
425 &ConcreteDataType::Binary(_) => "_BYTEA",
426 &ConcreteDataType::String(_) => "_VARCHAR",
427 &ConcreteDataType::Date(_) => "_DATE",
428 &ConcreteDataType::Timestamp(_) => "_TIMESTAMP",
429 &ConcreteDataType::Time(_) => "_TIME",
430 &ConcreteDataType::Interval(_) => "_INTERVAL",
431 &ConcreteDataType::Decimal128(_) => "_NUMERIC",
432 &ConcreteDataType::Json(_) => "_JSON",
433 &ConcreteDataType::Duration(_)
434 | &ConcreteDataType::Dictionary(_)
435 | &ConcreteDataType::Vector(_)
436 | &ConcreteDataType::List(_)
437 | &ConcreteDataType::Struct(_) => "UNKNOWN",
438 },
439 &ConcreteDataType::Duration(_)
440 | &ConcreteDataType::Dictionary(_)
441 | &ConcreteDataType::Struct(_) => "UNKNOWN",
442 }
443 }
444}
445
446impl From<&ConcreteDataType> for ConcreteDataType {
447 fn from(t: &ConcreteDataType) -> Self {
448 t.clone()
449 }
450}
451
452impl TryFrom<&ArrowDataType> for ConcreteDataType {
453 type Error = Error;
454
455 fn try_from(dt: &ArrowDataType) -> Result<ConcreteDataType> {
456 let concrete_type = match dt {
457 ArrowDataType::Null => Self::null_datatype(),
458 ArrowDataType::Boolean => Self::boolean_datatype(),
459 ArrowDataType::UInt8 => Self::uint8_datatype(),
460 ArrowDataType::UInt16 => Self::uint16_datatype(),
461 ArrowDataType::UInt32 => Self::uint32_datatype(),
462 ArrowDataType::UInt64 => Self::uint64_datatype(),
463 ArrowDataType::Int8 => Self::int8_datatype(),
464 ArrowDataType::Int16 => Self::int16_datatype(),
465 ArrowDataType::Int32 => Self::int32_datatype(),
466 ArrowDataType::Int64 => Self::int64_datatype(),
467 ArrowDataType::Float32 => Self::float32_datatype(),
468 ArrowDataType::Float64 => Self::float64_datatype(),
469 ArrowDataType::Date32 => Self::date_datatype(),
470 ArrowDataType::Timestamp(u, _) => ConcreteDataType::from_arrow_time_unit(u),
471 ArrowDataType::Interval(u) => ConcreteDataType::from_arrow_interval_unit(u),
472 ArrowDataType::Binary | ArrowDataType::LargeBinary => {
473 Self::Binary(BinaryType::binary())
474 }
475 ArrowDataType::BinaryView => Self::Binary(BinaryType::binary_view()),
476 ArrowDataType::Utf8 => Self::String(StringType::utf8()),
477 ArrowDataType::Utf8View => Self::String(StringType::utf8_view()),
478 ArrowDataType::LargeUtf8 => Self::String(StringType::large_utf8()),
479 ArrowDataType::List(field) => Self::List(ListType::new(Arc::new(
480 ConcreteDataType::from_arrow_type(field.data_type()),
481 ))),
482 ArrowDataType::Dictionary(key_type, value_type) => {
483 let key_type = ConcreteDataType::from_arrow_type(key_type);
484 let value_type = ConcreteDataType::from_arrow_type(value_type);
485 Self::Dictionary(DictionaryType::new(key_type, value_type))
486 }
487 ArrowDataType::Time32(u) => ConcreteDataType::Time(TimeType::from_unit(u.into())),
488 ArrowDataType::Time64(u) => ConcreteDataType::Time(TimeType::from_unit(u.into())),
489 ArrowDataType::Duration(u) => {
490 ConcreteDataType::Duration(DurationType::from_unit(u.into()))
491 }
492 ArrowDataType::Decimal128(precision, scale) => {
493 ConcreteDataType::decimal128_datatype(*precision, *scale)
494 }
495 ArrowDataType::Struct(fields) => ConcreteDataType::Struct(StructType::from(fields)),
496 ArrowDataType::Float16
497 | ArrowDataType::Date64
498 | ArrowDataType::FixedSizeBinary(_)
499 | ArrowDataType::ListView(_)
500 | ArrowDataType::FixedSizeList(_, _)
501 | ArrowDataType::LargeList(_)
502 | ArrowDataType::LargeListView(_)
503 | ArrowDataType::Union(_, _)
504 | ArrowDataType::Decimal256(_, _)
505 | ArrowDataType::Map(_, _)
506 | ArrowDataType::RunEndEncoded(_, _)
507 | ArrowDataType::Decimal32(_, _)
508 | ArrowDataType::Decimal64(_, _) => {
509 return error::UnsupportedArrowTypeSnafu {
510 arrow_type: dt.clone(),
511 }
512 .fail();
513 }
514 };
515
516 Ok(concrete_type)
517 }
518}
519
520macro_rules! impl_new_concrete_type_functions {
521 ($($Type: ident), +) => {
522 paste! {
523 impl ConcreteDataType {
524 $(
525 pub fn [<$Type:lower _datatype>]() -> ConcreteDataType {
526 ConcreteDataType::$Type([<$Type Type>]::default())
527 }
528 )+
529 }
530 }
531 }
532}
533
534impl_new_concrete_type_functions!(
535 Null, Boolean, UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64,
536 Binary, Date, String, Json
537);
538
539impl ConcreteDataType {
540 pub fn large_string_datatype() -> Self {
541 ConcreteDataType::String(StringType::large_utf8())
542 }
543
544 pub fn utf8_view_datatype() -> Self {
545 ConcreteDataType::String(StringType::utf8_view())
546 }
547
548 pub fn binary_view_datatype() -> Self {
549 ConcreteDataType::Binary(BinaryType::binary_view())
550 }
551
552 pub fn timestamp_second_datatype() -> Self {
553 ConcreteDataType::Timestamp(TimestampType::Second(TimestampSecondType))
554 }
555
556 pub fn timestamp_millisecond_datatype() -> Self {
557 ConcreteDataType::Timestamp(TimestampType::Millisecond(TimestampMillisecondType))
558 }
559
560 pub fn timestamp_microsecond_datatype() -> Self {
561 ConcreteDataType::Timestamp(TimestampType::Microsecond(TimestampMicrosecondType))
562 }
563
564 pub fn timestamp_nanosecond_datatype() -> Self {
565 ConcreteDataType::Timestamp(TimestampType::Nanosecond(TimestampNanosecondType))
566 }
567
568 pub fn time_datatype(unit: TimeUnit) -> Self {
570 ConcreteDataType::Time(TimeType::from_unit(unit))
571 }
572
573 pub fn time_second_datatype() -> Self {
575 Self::time_datatype(TimeUnit::Second)
576 }
577
578 pub fn time_millisecond_datatype() -> Self {
580 Self::time_datatype(TimeUnit::Millisecond)
581 }
582
583 pub fn time_microsecond_datatype() -> Self {
585 Self::time_datatype(TimeUnit::Microsecond)
586 }
587
588 pub fn time_nanosecond_datatype() -> Self {
590 Self::time_datatype(TimeUnit::Nanosecond)
591 }
592
593 pub fn duration_second_datatype() -> Self {
595 ConcreteDataType::Duration(DurationType::Second(DurationSecondType))
596 }
597
598 pub fn duration_millisecond_datatype() -> Self {
600 ConcreteDataType::Duration(DurationType::Millisecond(DurationMillisecondType))
601 }
602
603 pub fn duration_microsecond_datatype() -> Self {
605 ConcreteDataType::Duration(DurationType::Microsecond(DurationMicrosecondType))
606 }
607
608 pub fn duration_nanosecond_datatype() -> Self {
610 ConcreteDataType::Duration(DurationType::Nanosecond(DurationNanosecondType))
611 }
612
613 pub fn interval_month_day_nano_datatype() -> Self {
615 ConcreteDataType::Interval(IntervalType::MonthDayNano(IntervalMonthDayNanoType))
616 }
617
618 pub fn interval_year_month_datatype() -> Self {
620 ConcreteDataType::Interval(IntervalType::YearMonth(IntervalYearMonthType))
621 }
622
623 pub fn interval_day_time_datatype() -> Self {
625 ConcreteDataType::Interval(IntervalType::DayTime(IntervalDayTimeType))
626 }
627
628 pub fn timestamp_datatype(unit: TimeUnit) -> Self {
629 match unit {
630 TimeUnit::Second => Self::timestamp_second_datatype(),
631 TimeUnit::Millisecond => Self::timestamp_millisecond_datatype(),
632 TimeUnit::Microsecond => Self::timestamp_microsecond_datatype(),
633 TimeUnit::Nanosecond => Self::timestamp_nanosecond_datatype(),
634 }
635 }
636
637 pub fn from_arrow_time_unit(t: &ArrowTimeUnit) -> Self {
639 match t {
640 ArrowTimeUnit::Second => Self::timestamp_second_datatype(),
641 ArrowTimeUnit::Millisecond => Self::timestamp_millisecond_datatype(),
642 ArrowTimeUnit::Microsecond => Self::timestamp_microsecond_datatype(),
643 ArrowTimeUnit::Nanosecond => Self::timestamp_nanosecond_datatype(),
644 }
645 }
646
647 pub fn duration_datatype(unit: TimeUnit) -> Self {
648 match unit {
649 TimeUnit::Second => Self::duration_second_datatype(),
650 TimeUnit::Millisecond => Self::duration_millisecond_datatype(),
651 TimeUnit::Microsecond => Self::duration_microsecond_datatype(),
652 TimeUnit::Nanosecond => Self::duration_nanosecond_datatype(),
653 }
654 }
655
656 pub fn interval_datatype(unit: IntervalUnit) -> Self {
657 match unit {
658 IntervalUnit::YearMonth => Self::interval_year_month_datatype(),
659 IntervalUnit::DayTime => Self::interval_day_time_datatype(),
660 IntervalUnit::MonthDayNano => Self::interval_month_day_nano_datatype(),
661 }
662 }
663
664 pub fn from_arrow_interval_unit(u: &ArrowIntervalUnit) -> Self {
665 match u {
666 ArrowIntervalUnit::YearMonth => Self::interval_year_month_datatype(),
667 ArrowIntervalUnit::DayTime => Self::interval_day_time_datatype(),
668 ArrowIntervalUnit::MonthDayNano => Self::interval_month_day_nano_datatype(),
669 }
670 }
671
672 pub fn list_datatype(item_type: Arc<ConcreteDataType>) -> ConcreteDataType {
673 ConcreteDataType::List(ListType::new(item_type))
674 }
675
676 pub fn struct_datatype(fields: StructType) -> ConcreteDataType {
677 ConcreteDataType::Struct(fields)
678 }
679
680 pub fn dictionary_datatype(
681 key_type: ConcreteDataType,
682 value_type: ConcreteDataType,
683 ) -> ConcreteDataType {
684 ConcreteDataType::Dictionary(DictionaryType::new(key_type, value_type))
685 }
686
687 pub fn decimal128_datatype(precision: u8, scale: i8) -> ConcreteDataType {
688 ConcreteDataType::Decimal128(Decimal128Type::new(precision, scale))
689 }
690
691 pub fn decimal128_default_datatype() -> ConcreteDataType {
692 Self::decimal128_datatype(DECIMAL128_MAX_PRECISION, DECIMAL_DEFAULT_SCALE)
693 }
694
695 pub fn vector_datatype(dim: u32) -> ConcreteDataType {
696 ConcreteDataType::Vector(VectorType::new(dim))
697 }
698
699 pub fn vector_default_datatype() -> ConcreteDataType {
700 Self::vector_datatype(0)
701 }
702
703 pub fn json2(native_type: JsonNativeType) -> ConcreteDataType {
704 ConcreteDataType::Json(JsonType::json2(Arc::new(native_type)))
705 }
706}
707
708#[enum_dispatch::enum_dispatch]
710pub trait DataType: std::fmt::Debug + Send + Sync {
711 fn name(&self) -> String;
713
714 fn logical_type_id(&self) -> LogicalTypeId;
716
717 fn default_value(&self) -> Value;
719
720 fn as_arrow_type(&self) -> ArrowDataType;
722
723 fn create_mutable_vector(&self, capacity: usize) -> Box<dyn MutableVector>;
725
726 fn try_cast(&self, from: Value) -> Option<Value>;
729}
730
731pub type DataTypeRef = Arc<dyn DataType>;
732
733#[cfg(test)]
734mod tests {
735 use arrow::datatypes::Field;
736
737 use super::*;
738
739 #[test]
740 fn test_is_timestamp_unit_widening_to() {
741 let second = ConcreteDataType::timestamp_second_datatype();
742 let milli = ConcreteDataType::timestamp_millisecond_datatype();
743 let nano = ConcreteDataType::timestamp_nanosecond_datatype();
744
745 assert!(second.is_timestamp_unit_widening_to(&milli));
746 assert!(second.is_timestamp_unit_widening_to(&nano));
747 assert!(!milli.is_timestamp_unit_widening_to(&milli));
749 assert!(!nano.is_timestamp_unit_widening_to(&milli));
750 assert!(!milli.is_timestamp_unit_widening_to(&ConcreteDataType::int64_datatype()));
752 assert!(!ConcreteDataType::int64_datatype().is_timestamp_unit_widening_to(&nano));
753 }
754
755 #[test]
756 fn test_concrete_type_as_datatype_trait() {
757 let concrete_type = ConcreteDataType::boolean_datatype();
758
759 assert_eq!("Boolean", concrete_type.to_string());
760 assert_eq!(Value::Boolean(false), concrete_type.default_value());
761 assert_eq!(LogicalTypeId::Boolean, concrete_type.logical_type_id());
762 assert_eq!(ArrowDataType::Boolean, concrete_type.as_arrow_type());
763 }
764
765 #[test]
766 fn test_from_arrow_type() {
767 assert!(matches!(
768 ConcreteDataType::from_arrow_type(&ArrowDataType::Null),
769 ConcreteDataType::Null(_)
770 ));
771 assert!(matches!(
772 ConcreteDataType::from_arrow_type(&ArrowDataType::Boolean),
773 ConcreteDataType::Boolean(_)
774 ));
775 assert!(matches!(
776 ConcreteDataType::from_arrow_type(&ArrowDataType::Binary),
777 ConcreteDataType::Binary(_)
778 ));
779 assert!(matches!(
780 ConcreteDataType::from_arrow_type(&ArrowDataType::LargeBinary),
781 ConcreteDataType::Binary(_)
782 ));
783 assert!(matches!(
784 ConcreteDataType::from_arrow_type(&ArrowDataType::Int8),
785 ConcreteDataType::Int8(_)
786 ));
787 assert!(matches!(
788 ConcreteDataType::from_arrow_type(&ArrowDataType::Int16),
789 ConcreteDataType::Int16(_)
790 ));
791 assert!(matches!(
792 ConcreteDataType::from_arrow_type(&ArrowDataType::Int32),
793 ConcreteDataType::Int32(_)
794 ));
795 assert!(matches!(
796 ConcreteDataType::from_arrow_type(&ArrowDataType::Int64),
797 ConcreteDataType::Int64(_)
798 ));
799 assert!(matches!(
800 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt8),
801 ConcreteDataType::UInt8(_)
802 ));
803 assert!(matches!(
804 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt16),
805 ConcreteDataType::UInt16(_)
806 ));
807 assert!(matches!(
808 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt32),
809 ConcreteDataType::UInt32(_)
810 ));
811 assert!(matches!(
812 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt64),
813 ConcreteDataType::UInt64(_)
814 ));
815 assert!(matches!(
816 ConcreteDataType::from_arrow_type(&ArrowDataType::Float32),
817 ConcreteDataType::Float32(_)
818 ));
819 assert!(matches!(
820 ConcreteDataType::from_arrow_type(&ArrowDataType::Float64),
821 ConcreteDataType::Float64(_)
822 ));
823 assert!(matches!(
824 ConcreteDataType::from_arrow_type(&ArrowDataType::Utf8),
825 ConcreteDataType::String(_)
826 ));
827 let utf8_view_string_type = ConcreteDataType::from_arrow_type(&ArrowDataType::Utf8View);
828 assert!(matches!(utf8_view_string_type, ConcreteDataType::String(_)));
829 assert_eq!(
830 ArrowDataType::Utf8View,
831 utf8_view_string_type.as_arrow_type()
832 );
833 let large_string_type = ConcreteDataType::from_arrow_type(&ArrowDataType::LargeUtf8);
835 assert!(matches!(large_string_type, ConcreteDataType::String(_)));
836 if let ConcreteDataType::String(string_type) = &large_string_type {
837 assert!(string_type.is_large());
838 } else {
839 panic!("Expected a String type");
840 }
841 assert_eq!(
842 ConcreteDataType::from_arrow_type(&ArrowDataType::List(Arc::new(Field::new(
843 "item",
844 ArrowDataType::Int32,
845 true,
846 )))),
847 ConcreteDataType::List(ListType::new(Arc::new(ConcreteDataType::int32_datatype())))
848 );
849 assert!(matches!(
850 ConcreteDataType::from_arrow_type(&ArrowDataType::Date32),
851 ConcreteDataType::Date(_)
852 ));
853 }
854
855 #[test]
856 fn test_view_round_trip() {
857 let utf8_view_arrow = ArrowDataType::Utf8View;
858 let concrete_type = ConcreteDataType::from_arrow_type(&utf8_view_arrow);
859 let back_to_arrow = concrete_type.as_arrow_type();
860 assert_eq!(utf8_view_arrow, back_to_arrow);
861
862 let binary_view_arrow = ArrowDataType::BinaryView;
863 let concrete_type = ConcreteDataType::from_arrow_type(&binary_view_arrow);
864 let back_to_arrow = concrete_type.as_arrow_type();
865 assert_eq!(binary_view_arrow, back_to_arrow);
866 }
867
868 #[test]
869 fn test_large_utf8_round_trip() {
870 let large_utf8_arrow = ArrowDataType::LargeUtf8;
872 let concrete_type = ConcreteDataType::from_arrow_type(&large_utf8_arrow);
873 let back_to_arrow = concrete_type.as_arrow_type();
874
875 assert!(matches!(concrete_type, ConcreteDataType::String(_)));
876 assert_eq!(large_utf8_arrow, back_to_arrow);
878
879 let utf8_concrete = ConcreteDataType::from_arrow_type(&ArrowDataType::Utf8);
881 let large_utf8_concrete = ConcreteDataType::from_arrow_type(&ArrowDataType::LargeUtf8);
882
883 assert!(matches!(utf8_concrete, ConcreteDataType::String(_)));
884 assert!(matches!(large_utf8_concrete, ConcreteDataType::String(_)));
885
886 if let (ConcreteDataType::String(utf8_type), ConcreteDataType::String(large_type)) =
888 (&utf8_concrete, &large_utf8_concrete)
889 {
890 assert!(!utf8_type.is_large());
891 assert!(large_type.is_large());
892 } else {
893 panic!("Expected both to be String types");
894 }
895
896 let view_concrete = ConcreteDataType::from_arrow_type(&ArrowDataType::Utf8View);
898 assert_ne!(utf8_concrete, view_concrete);
899 assert_ne!(large_utf8_concrete, view_concrete);
900
901 assert_ne!(utf8_concrete, large_utf8_concrete);
903 }
904
905 #[test]
906 fn test_from_arrow_timestamp() {
907 assert_eq!(
908 ConcreteDataType::timestamp_millisecond_datatype(),
909 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Millisecond)
910 );
911 assert_eq!(
912 ConcreteDataType::timestamp_microsecond_datatype(),
913 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Microsecond)
914 );
915 assert_eq!(
916 ConcreteDataType::timestamp_nanosecond_datatype(),
917 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Nanosecond)
918 );
919 assert_eq!(
920 ConcreteDataType::timestamp_second_datatype(),
921 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Second)
922 );
923 }
924
925 #[test]
926 fn test_is_null() {
927 assert!(ConcreteDataType::null_datatype().is_null());
928 assert!(!ConcreteDataType::int32_datatype().is_null());
929 }
930
931 #[test]
932 fn test_is_float() {
933 assert!(!ConcreteDataType::int32_datatype().is_float());
934 assert!(ConcreteDataType::float32_datatype().is_float());
935 assert!(ConcreteDataType::float64_datatype().is_float());
936 }
937
938 #[test]
939 fn test_is_boolean() {
940 assert!(!ConcreteDataType::int32_datatype().is_boolean());
941 assert!(!ConcreteDataType::float32_datatype().is_boolean());
942 assert!(ConcreteDataType::boolean_datatype().is_boolean());
943 }
944
945 #[test]
946 fn test_is_decimal() {
947 assert!(!ConcreteDataType::int32_datatype().is_decimal());
948 assert!(!ConcreteDataType::float32_datatype().is_decimal());
949 assert!(ConcreteDataType::decimal128_datatype(10, 2).is_decimal());
950 assert!(ConcreteDataType::decimal128_datatype(18, 6).is_decimal());
951 }
952
953 #[test]
954 fn test_is_stringifiable() {
955 assert!(!ConcreteDataType::int32_datatype().is_stringifiable());
956 assert!(!ConcreteDataType::float32_datatype().is_stringifiable());
957 assert!(ConcreteDataType::string_datatype().is_stringifiable());
958 assert!(ConcreteDataType::binary_datatype().is_stringifiable());
959 assert!(ConcreteDataType::date_datatype().is_stringifiable());
960 assert!(ConcreteDataType::timestamp_second_datatype().is_stringifiable());
961 assert!(ConcreteDataType::timestamp_millisecond_datatype().is_stringifiable());
962 assert!(ConcreteDataType::timestamp_microsecond_datatype().is_stringifiable());
963 assert!(ConcreteDataType::timestamp_nanosecond_datatype().is_stringifiable());
964 assert!(ConcreteDataType::time_second_datatype().is_stringifiable());
965 assert!(ConcreteDataType::time_millisecond_datatype().is_stringifiable());
966 assert!(ConcreteDataType::time_microsecond_datatype().is_stringifiable());
967 assert!(ConcreteDataType::time_nanosecond_datatype().is_stringifiable());
968
969 assert!(ConcreteDataType::interval_year_month_datatype().is_stringifiable());
970 assert!(ConcreteDataType::interval_day_time_datatype().is_stringifiable());
971 assert!(ConcreteDataType::interval_month_day_nano_datatype().is_stringifiable());
972
973 assert!(ConcreteDataType::duration_second_datatype().is_stringifiable());
974 assert!(ConcreteDataType::duration_millisecond_datatype().is_stringifiable());
975 assert!(ConcreteDataType::duration_microsecond_datatype().is_stringifiable());
976 assert!(ConcreteDataType::duration_nanosecond_datatype().is_stringifiable());
977 assert!(ConcreteDataType::decimal128_datatype(10, 2).is_stringifiable());
978 assert!(ConcreteDataType::vector_default_datatype().is_stringifiable());
979 }
980
981 #[test]
982 fn test_is_signed() {
983 assert!(ConcreteDataType::int8_datatype().is_signed());
984 assert!(ConcreteDataType::int16_datatype().is_signed());
985 assert!(ConcreteDataType::int32_datatype().is_signed());
986 assert!(ConcreteDataType::int64_datatype().is_signed());
987 assert!(ConcreteDataType::date_datatype().is_signed());
988 assert!(ConcreteDataType::timestamp_second_datatype().is_signed());
989 assert!(ConcreteDataType::timestamp_millisecond_datatype().is_signed());
990 assert!(ConcreteDataType::timestamp_microsecond_datatype().is_signed());
991 assert!(ConcreteDataType::timestamp_nanosecond_datatype().is_signed());
992 assert!(ConcreteDataType::time_second_datatype().is_signed());
993 assert!(ConcreteDataType::time_millisecond_datatype().is_signed());
994 assert!(ConcreteDataType::time_microsecond_datatype().is_signed());
995 assert!(ConcreteDataType::time_nanosecond_datatype().is_signed());
996 assert!(ConcreteDataType::interval_year_month_datatype().is_signed());
997 assert!(ConcreteDataType::interval_day_time_datatype().is_signed());
998 assert!(ConcreteDataType::interval_month_day_nano_datatype().is_signed());
999 assert!(ConcreteDataType::duration_second_datatype().is_signed());
1000 assert!(ConcreteDataType::duration_millisecond_datatype().is_signed());
1001 assert!(ConcreteDataType::duration_microsecond_datatype().is_signed());
1002 assert!(ConcreteDataType::duration_nanosecond_datatype().is_signed());
1003
1004 assert!(!ConcreteDataType::uint8_datatype().is_signed());
1005 assert!(!ConcreteDataType::uint16_datatype().is_signed());
1006 assert!(!ConcreteDataType::uint32_datatype().is_signed());
1007 assert!(!ConcreteDataType::uint64_datatype().is_signed());
1008
1009 assert!(!ConcreteDataType::float32_datatype().is_signed());
1010 assert!(!ConcreteDataType::float64_datatype().is_signed());
1011
1012 assert!(ConcreteDataType::decimal128_datatype(10, 2).is_signed());
1013 }
1014
1015 #[test]
1016 fn test_is_unsigned() {
1017 assert!(!ConcreteDataType::int8_datatype().is_unsigned());
1018 assert!(!ConcreteDataType::int16_datatype().is_unsigned());
1019 assert!(!ConcreteDataType::int32_datatype().is_unsigned());
1020 assert!(!ConcreteDataType::int64_datatype().is_unsigned());
1021 assert!(!ConcreteDataType::date_datatype().is_unsigned());
1022 assert!(!ConcreteDataType::timestamp_second_datatype().is_unsigned());
1023 assert!(!ConcreteDataType::timestamp_millisecond_datatype().is_unsigned());
1024 assert!(!ConcreteDataType::timestamp_microsecond_datatype().is_unsigned());
1025 assert!(!ConcreteDataType::timestamp_nanosecond_datatype().is_unsigned());
1026 assert!(!ConcreteDataType::time_second_datatype().is_unsigned());
1027 assert!(!ConcreteDataType::time_millisecond_datatype().is_unsigned());
1028 assert!(!ConcreteDataType::time_microsecond_datatype().is_unsigned());
1029 assert!(!ConcreteDataType::time_nanosecond_datatype().is_unsigned());
1030 assert!(!ConcreteDataType::interval_year_month_datatype().is_unsigned());
1031 assert!(!ConcreteDataType::interval_day_time_datatype().is_unsigned());
1032 assert!(!ConcreteDataType::interval_month_day_nano_datatype().is_unsigned());
1033 assert!(!ConcreteDataType::duration_second_datatype().is_unsigned());
1034 assert!(!ConcreteDataType::duration_millisecond_datatype().is_unsigned());
1035 assert!(!ConcreteDataType::duration_microsecond_datatype().is_unsigned());
1036 assert!(!ConcreteDataType::duration_nanosecond_datatype().is_unsigned());
1037 assert!(!ConcreteDataType::decimal128_datatype(10, 2).is_unsigned());
1038
1039 assert!(ConcreteDataType::uint8_datatype().is_unsigned());
1040 assert!(ConcreteDataType::uint16_datatype().is_unsigned());
1041 assert!(ConcreteDataType::uint32_datatype().is_unsigned());
1042 assert!(ConcreteDataType::uint64_datatype().is_unsigned());
1043
1044 assert!(!ConcreteDataType::float32_datatype().is_unsigned());
1045 assert!(!ConcreteDataType::float64_datatype().is_unsigned());
1046 }
1047
1048 #[test]
1049 fn test_numerics() {
1050 let nums = ConcreteDataType::numerics();
1051 assert_eq!(10, nums.len());
1052 }
1053
1054 #[test]
1055 fn test_as_list() {
1056 let list_type =
1057 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype()));
1058 assert_eq!(
1059 ListType::new(Arc::new(ConcreteDataType::int32_datatype())),
1060 *list_type.as_list().unwrap()
1061 );
1062 assert!(ConcreteDataType::int32_datatype().as_list().is_none());
1063 }
1064
1065 #[test]
1066 fn test_display_concrete_data_type() {
1067 assert_eq!(ConcreteDataType::null_datatype().to_string(), "Null");
1068 assert_eq!(ConcreteDataType::boolean_datatype().to_string(), "Boolean");
1069 assert_eq!(ConcreteDataType::binary_datatype().to_string(), "Binary");
1070 assert_eq!(ConcreteDataType::int8_datatype().to_string(), "Int8");
1071 assert_eq!(ConcreteDataType::int16_datatype().to_string(), "Int16");
1072 assert_eq!(ConcreteDataType::int32_datatype().to_string(), "Int32");
1073 assert_eq!(ConcreteDataType::int64_datatype().to_string(), "Int64");
1074 assert_eq!(ConcreteDataType::uint8_datatype().to_string(), "UInt8");
1075 assert_eq!(ConcreteDataType::uint16_datatype().to_string(), "UInt16");
1076 assert_eq!(ConcreteDataType::uint32_datatype().to_string(), "UInt32");
1077 assert_eq!(ConcreteDataType::uint64_datatype().to_string(), "UInt64");
1078 assert_eq!(ConcreteDataType::float32_datatype().to_string(), "Float32");
1079 assert_eq!(ConcreteDataType::float64_datatype().to_string(), "Float64");
1080 assert_eq!(ConcreteDataType::string_datatype().to_string(), "String");
1081 assert_eq!(ConcreteDataType::date_datatype().to_string(), "Date");
1082 assert_eq!(
1083 ConcreteDataType::timestamp_millisecond_datatype().to_string(),
1084 "TimestampMillisecond"
1085 );
1086 assert_eq!(
1087 ConcreteDataType::time_millisecond_datatype().to_string(),
1088 "TimeMillisecond"
1089 );
1090 assert_eq!(
1091 ConcreteDataType::interval_month_day_nano_datatype().to_string(),
1092 "IntervalMonthDayNano"
1093 );
1094 assert_eq!(
1095 ConcreteDataType::duration_second_datatype().to_string(),
1096 "DurationSecond"
1097 );
1098 assert_eq!(
1099 ConcreteDataType::decimal128_datatype(10, 2).to_string(),
1100 "Decimal(10, 2)"
1101 );
1102 assert_eq!(
1104 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype()))
1105 .to_string(),
1106 "List<Int32>"
1107 );
1108 assert_eq!(
1109 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::Dictionary(
1110 DictionaryType::new(
1111 ConcreteDataType::int32_datatype(),
1112 ConcreteDataType::string_datatype()
1113 )
1114 )))
1115 .to_string(),
1116 "List<Dictionary<Int32, String>>"
1117 );
1118 assert_eq!(
1119 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::list_datatype(Arc::new(
1120 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype()))
1121 ))))
1122 .to_string(),
1123 "List<List<List<Int32>>>"
1124 );
1125 assert_eq!(
1126 ConcreteDataType::dictionary_datatype(
1127 ConcreteDataType::int32_datatype(),
1128 ConcreteDataType::string_datatype()
1129 )
1130 .to_string(),
1131 "Dictionary<Int32, String>"
1132 );
1133 assert_eq!(
1134 ConcreteDataType::vector_datatype(3).to_string(),
1135 "Vector(3)"
1136 );
1137 }
1138}