1use std::fmt;
16use std::sync::Arc;
17
18use arrow::compute::cast as arrow_array_cast;
19use arrow::datatypes::{
20 DataType as ArrowDataType, IntervalUnit as ArrowIntervalUnit, TimeUnit as ArrowTimeUnit,
21};
22use arrow_schema::DECIMAL_DEFAULT_SCALE;
23use common_decimal::decimal128::DECIMAL128_MAX_PRECISION;
24use common_time::interval::IntervalUnit;
25use common_time::timestamp::TimeUnit;
26use enum_dispatch::enum_dispatch;
27use paste::paste;
28use serde::{Deserialize, Serialize};
29
30use crate::error::{self, Error, Result};
31use crate::type_id::LogicalTypeId;
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, JsonFormat, JsonType, ListType,
37 NullType, 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_vector(&self) -> bool {
232 matches!(self, ConcreteDataType::Vector(_))
233 }
234
235 pub fn numerics() -> Vec<ConcreteDataType> {
236 vec![
237 ConcreteDataType::int8_datatype(),
238 ConcreteDataType::int16_datatype(),
239 ConcreteDataType::int32_datatype(),
240 ConcreteDataType::int64_datatype(),
241 ConcreteDataType::uint8_datatype(),
242 ConcreteDataType::uint16_datatype(),
243 ConcreteDataType::uint32_datatype(),
244 ConcreteDataType::uint64_datatype(),
245 ConcreteDataType::float32_datatype(),
246 ConcreteDataType::float64_datatype(),
247 ]
248 }
249
250 pub fn unsigned_integers() -> Vec<ConcreteDataType> {
251 vec![
252 ConcreteDataType::uint8_datatype(),
253 ConcreteDataType::uint16_datatype(),
254 ConcreteDataType::uint32_datatype(),
255 ConcreteDataType::uint64_datatype(),
256 ]
257 }
258
259 pub fn timestamps() -> Vec<ConcreteDataType> {
260 vec![
261 ConcreteDataType::timestamp_second_datatype(),
262 ConcreteDataType::timestamp_millisecond_datatype(),
263 ConcreteDataType::timestamp_microsecond_datatype(),
264 ConcreteDataType::timestamp_nanosecond_datatype(),
265 ]
266 }
267
268 pub fn from_arrow_type(dt: &ArrowDataType) -> Self {
273 ConcreteDataType::try_from(dt).expect("Unimplemented type")
274 }
275
276 pub fn is_null(&self) -> bool {
277 matches!(self, ConcreteDataType::Null(NullType))
278 }
279
280 pub fn as_list(&self) -> Option<&ListType> {
282 match self {
283 ConcreteDataType::List(t) => Some(t),
284 _ => None,
285 }
286 }
287
288 pub fn as_struct(&self) -> Option<&StructType> {
289 match self {
290 ConcreteDataType::Struct(s) => Some(s),
291 _ => None,
292 }
293 }
294
295 pub fn as_timestamp(&self) -> Option<TimestampType> {
297 match self {
298 ConcreteDataType::Timestamp(t) => Some(*t),
299 _ => None,
300 }
301 }
302
303 pub fn numeric_precision(&self) -> Option<u8> {
305 match self {
306 ConcreteDataType::Int8(_) | ConcreteDataType::UInt8(_) => Some(3),
307 ConcreteDataType::Int16(_) | ConcreteDataType::UInt16(_) => Some(5),
308 ConcreteDataType::Int32(_) | ConcreteDataType::UInt32(_) => Some(10),
309 ConcreteDataType::Int64(_) => Some(19),
310 ConcreteDataType::UInt64(_) => Some(20),
311 ConcreteDataType::Float32(_) => Some(12),
312 ConcreteDataType::Float64(_) => Some(22),
313 ConcreteDataType::Decimal128(decimal_type) => Some(decimal_type.precision()),
314 _ => None,
315 }
316 }
317
318 pub fn numeric_scale(&self) -> Option<i8> {
320 match self {
321 ConcreteDataType::Int8(_)
322 | ConcreteDataType::UInt8(_)
323 | ConcreteDataType::Int16(_)
324 | ConcreteDataType::UInt16(_)
325 | ConcreteDataType::Int32(_)
326 | ConcreteDataType::UInt32(_)
327 | ConcreteDataType::Int64(_)
328 | ConcreteDataType::UInt64(_) => Some(0),
329 ConcreteDataType::Float32(_) | ConcreteDataType::Float64(_) => None,
330 ConcreteDataType::Decimal128(decimal_type) => Some(decimal_type.scale()),
331 _ => None,
332 }
333 }
334
335 pub fn as_time(&self) -> Option<TimeType> {
337 match self {
338 ConcreteDataType::Int64(_) => Some(TimeType::Millisecond(TimeMillisecondType)),
339 ConcreteDataType::Time(t) => Some(*t),
340 _ => None,
341 }
342 }
343
344 pub fn as_decimal128(&self) -> Option<Decimal128Type> {
345 match self {
346 ConcreteDataType::Decimal128(d) => Some(*d),
347 _ => None,
348 }
349 }
350
351 pub fn as_json(&self) -> Option<JsonType> {
352 match self {
353 ConcreteDataType::Json(j) => Some(j.clone()),
354 _ => None,
355 }
356 }
357
358 pub fn as_vector(&self) -> Option<VectorType> {
359 match self {
360 ConcreteDataType::Vector(v) => Some(*v),
361 _ => None,
362 }
363 }
364
365 pub fn can_arrow_type_cast_to(&self, to_type: &ConcreteDataType) -> bool {
367 let array = arrow_array::new_empty_array(&self.as_arrow_type());
368 arrow_array_cast(array.as_ref(), &to_type.as_arrow_type()).is_ok()
369 }
370
371 pub fn as_duration(&self) -> Option<DurationType> {
373 match self {
374 ConcreteDataType::Duration(d) => Some(*d),
375 _ => None,
376 }
377 }
378
379 pub fn postgres_datatype_name(&self) -> &'static str {
381 match self {
382 &ConcreteDataType::Null(_) => "UNKNOWN",
383 &ConcreteDataType::Boolean(_) => "BOOL",
384 &ConcreteDataType::Int8(_) | &ConcreteDataType::UInt8(_) => "CHAR",
385 &ConcreteDataType::Int16(_) | &ConcreteDataType::UInt16(_) => "INT2",
386 &ConcreteDataType::Int32(_) | &ConcreteDataType::UInt32(_) => "INT4",
387 &ConcreteDataType::Int64(_) | &ConcreteDataType::UInt64(_) => "INT8",
388 &ConcreteDataType::Float32(_) => "FLOAT4",
389 &ConcreteDataType::Float64(_) => "FLOAT8",
390 &ConcreteDataType::Binary(_) | &ConcreteDataType::Vector(_) => "BYTEA",
391 &ConcreteDataType::String(_) => "VARCHAR",
392 &ConcreteDataType::Date(_) => "DATE",
393 &ConcreteDataType::Timestamp(_) => "TIMESTAMP",
394 &ConcreteDataType::Time(_) => "TIME",
395 &ConcreteDataType::Interval(_) => "INTERVAL",
396 &ConcreteDataType::Decimal128(_) => "NUMERIC",
397 &ConcreteDataType::Json(_) => "JSON",
398 ConcreteDataType::List(list) => match list.item_type() {
399 &ConcreteDataType::Null(_) => "UNKNOWN",
400 &ConcreteDataType::Boolean(_) => "_BOOL",
401 &ConcreteDataType::Int8(_) | &ConcreteDataType::UInt8(_) => "_CHAR",
402 &ConcreteDataType::Int16(_) | &ConcreteDataType::UInt16(_) => "_INT2",
403 &ConcreteDataType::Int32(_) | &ConcreteDataType::UInt32(_) => "_INT4",
404 &ConcreteDataType::Int64(_) | &ConcreteDataType::UInt64(_) => "_INT8",
405 &ConcreteDataType::Float32(_) => "_FLOAT4",
406 &ConcreteDataType::Float64(_) => "_FLOAT8",
407 &ConcreteDataType::Binary(_) => "_BYTEA",
408 &ConcreteDataType::String(_) => "_VARCHAR",
409 &ConcreteDataType::Date(_) => "_DATE",
410 &ConcreteDataType::Timestamp(_) => "_TIMESTAMP",
411 &ConcreteDataType::Time(_) => "_TIME",
412 &ConcreteDataType::Interval(_) => "_INTERVAL",
413 &ConcreteDataType::Decimal128(_) => "_NUMERIC",
414 &ConcreteDataType::Json(_) => "_JSON",
415 &ConcreteDataType::Duration(_)
416 | &ConcreteDataType::Dictionary(_)
417 | &ConcreteDataType::Vector(_)
418 | &ConcreteDataType::List(_)
419 | &ConcreteDataType::Struct(_) => "UNKNOWN",
420 },
421 &ConcreteDataType::Duration(_)
422 | &ConcreteDataType::Dictionary(_)
423 | &ConcreteDataType::Struct(_) => "UNKNOWN",
424 }
425 }
426}
427
428impl From<&ConcreteDataType> for ConcreteDataType {
429 fn from(t: &ConcreteDataType) -> Self {
430 t.clone()
431 }
432}
433
434impl TryFrom<&ArrowDataType> for ConcreteDataType {
435 type Error = Error;
436
437 fn try_from(dt: &ArrowDataType) -> Result<ConcreteDataType> {
438 let concrete_type = match dt {
439 ArrowDataType::Null => Self::null_datatype(),
440 ArrowDataType::Boolean => Self::boolean_datatype(),
441 ArrowDataType::UInt8 => Self::uint8_datatype(),
442 ArrowDataType::UInt16 => Self::uint16_datatype(),
443 ArrowDataType::UInt32 => Self::uint32_datatype(),
444 ArrowDataType::UInt64 => Self::uint64_datatype(),
445 ArrowDataType::Int8 => Self::int8_datatype(),
446 ArrowDataType::Int16 => Self::int16_datatype(),
447 ArrowDataType::Int32 => Self::int32_datatype(),
448 ArrowDataType::Int64 => Self::int64_datatype(),
449 ArrowDataType::Float32 => Self::float32_datatype(),
450 ArrowDataType::Float64 => Self::float64_datatype(),
451 ArrowDataType::Date32 => Self::date_datatype(),
452 ArrowDataType::Timestamp(u, _) => ConcreteDataType::from_arrow_time_unit(u),
453 ArrowDataType::Interval(u) => ConcreteDataType::from_arrow_interval_unit(u),
454 ArrowDataType::Binary | ArrowDataType::LargeBinary | ArrowDataType::BinaryView => {
455 Self::binary_datatype()
456 }
457 ArrowDataType::Utf8 | ArrowDataType::Utf8View => Self::string_datatype(),
458 ArrowDataType::LargeUtf8 => Self::large_string_datatype(),
459 ArrowDataType::List(field) => Self::List(ListType::new(Arc::new(
460 ConcreteDataType::from_arrow_type(field.data_type()),
461 ))),
462 ArrowDataType::Dictionary(key_type, value_type) => {
463 let key_type = ConcreteDataType::from_arrow_type(key_type);
464 let value_type = ConcreteDataType::from_arrow_type(value_type);
465 Self::Dictionary(DictionaryType::new(key_type, value_type))
466 }
467 ArrowDataType::Time32(u) => ConcreteDataType::Time(TimeType::from_unit(u.into())),
468 ArrowDataType::Time64(u) => ConcreteDataType::Time(TimeType::from_unit(u.into())),
469 ArrowDataType::Duration(u) => {
470 ConcreteDataType::Duration(DurationType::from_unit(u.into()))
471 }
472 ArrowDataType::Decimal128(precision, scale) => {
473 ConcreteDataType::decimal128_datatype(*precision, *scale)
474 }
475 ArrowDataType::Struct(fields) => ConcreteDataType::Struct(fields.try_into()?),
476 ArrowDataType::Float16
477 | ArrowDataType::Date64
478 | ArrowDataType::FixedSizeBinary(_)
479 | ArrowDataType::ListView(_)
480 | ArrowDataType::FixedSizeList(_, _)
481 | ArrowDataType::LargeList(_)
482 | ArrowDataType::LargeListView(_)
483 | ArrowDataType::Union(_, _)
484 | ArrowDataType::Decimal256(_, _)
485 | ArrowDataType::Map(_, _)
486 | ArrowDataType::RunEndEncoded(_, _)
487 | ArrowDataType::Decimal32(_, _)
488 | ArrowDataType::Decimal64(_, _) => {
489 return error::UnsupportedArrowTypeSnafu {
490 arrow_type: dt.clone(),
491 }
492 .fail();
493 }
494 };
495
496 Ok(concrete_type)
497 }
498}
499
500macro_rules! impl_new_concrete_type_functions {
501 ($($Type: ident), +) => {
502 paste! {
503 impl ConcreteDataType {
504 $(
505 pub fn [<$Type:lower _datatype>]() -> ConcreteDataType {
506 ConcreteDataType::$Type([<$Type Type>]::default())
507 }
508 )+
509 }
510 }
511 }
512}
513
514impl_new_concrete_type_functions!(
515 Null, Boolean, UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64,
516 Binary, Date, String, Json
517);
518
519impl ConcreteDataType {
520 pub fn large_string_datatype() -> Self {
521 ConcreteDataType::String(StringType::large_utf8())
522 }
523
524 pub fn timestamp_second_datatype() -> Self {
525 ConcreteDataType::Timestamp(TimestampType::Second(TimestampSecondType))
526 }
527
528 pub fn timestamp_millisecond_datatype() -> Self {
529 ConcreteDataType::Timestamp(TimestampType::Millisecond(TimestampMillisecondType))
530 }
531
532 pub fn timestamp_microsecond_datatype() -> Self {
533 ConcreteDataType::Timestamp(TimestampType::Microsecond(TimestampMicrosecondType))
534 }
535
536 pub fn timestamp_nanosecond_datatype() -> Self {
537 ConcreteDataType::Timestamp(TimestampType::Nanosecond(TimestampNanosecondType))
538 }
539
540 pub fn time_datatype(unit: TimeUnit) -> Self {
542 ConcreteDataType::Time(TimeType::from_unit(unit))
543 }
544
545 pub fn time_second_datatype() -> Self {
547 Self::time_datatype(TimeUnit::Second)
548 }
549
550 pub fn time_millisecond_datatype() -> Self {
552 Self::time_datatype(TimeUnit::Millisecond)
553 }
554
555 pub fn time_microsecond_datatype() -> Self {
557 Self::time_datatype(TimeUnit::Microsecond)
558 }
559
560 pub fn time_nanosecond_datatype() -> Self {
562 Self::time_datatype(TimeUnit::Nanosecond)
563 }
564
565 pub fn duration_second_datatype() -> Self {
567 ConcreteDataType::Duration(DurationType::Second(DurationSecondType))
568 }
569
570 pub fn duration_millisecond_datatype() -> Self {
572 ConcreteDataType::Duration(DurationType::Millisecond(DurationMillisecondType))
573 }
574
575 pub fn duration_microsecond_datatype() -> Self {
577 ConcreteDataType::Duration(DurationType::Microsecond(DurationMicrosecondType))
578 }
579
580 pub fn duration_nanosecond_datatype() -> Self {
582 ConcreteDataType::Duration(DurationType::Nanosecond(DurationNanosecondType))
583 }
584
585 pub fn interval_month_day_nano_datatype() -> Self {
587 ConcreteDataType::Interval(IntervalType::MonthDayNano(IntervalMonthDayNanoType))
588 }
589
590 pub fn interval_year_month_datatype() -> Self {
592 ConcreteDataType::Interval(IntervalType::YearMonth(IntervalYearMonthType))
593 }
594
595 pub fn interval_day_time_datatype() -> Self {
597 ConcreteDataType::Interval(IntervalType::DayTime(IntervalDayTimeType))
598 }
599
600 pub fn timestamp_datatype(unit: TimeUnit) -> Self {
601 match unit {
602 TimeUnit::Second => Self::timestamp_second_datatype(),
603 TimeUnit::Millisecond => Self::timestamp_millisecond_datatype(),
604 TimeUnit::Microsecond => Self::timestamp_microsecond_datatype(),
605 TimeUnit::Nanosecond => Self::timestamp_nanosecond_datatype(),
606 }
607 }
608
609 pub fn from_arrow_time_unit(t: &ArrowTimeUnit) -> Self {
611 match t {
612 ArrowTimeUnit::Second => Self::timestamp_second_datatype(),
613 ArrowTimeUnit::Millisecond => Self::timestamp_millisecond_datatype(),
614 ArrowTimeUnit::Microsecond => Self::timestamp_microsecond_datatype(),
615 ArrowTimeUnit::Nanosecond => Self::timestamp_nanosecond_datatype(),
616 }
617 }
618
619 pub fn duration_datatype(unit: TimeUnit) -> Self {
620 match unit {
621 TimeUnit::Second => Self::duration_second_datatype(),
622 TimeUnit::Millisecond => Self::duration_millisecond_datatype(),
623 TimeUnit::Microsecond => Self::duration_microsecond_datatype(),
624 TimeUnit::Nanosecond => Self::duration_nanosecond_datatype(),
625 }
626 }
627
628 pub fn interval_datatype(unit: IntervalUnit) -> Self {
629 match unit {
630 IntervalUnit::YearMonth => Self::interval_year_month_datatype(),
631 IntervalUnit::DayTime => Self::interval_day_time_datatype(),
632 IntervalUnit::MonthDayNano => Self::interval_month_day_nano_datatype(),
633 }
634 }
635
636 pub fn from_arrow_interval_unit(u: &ArrowIntervalUnit) -> Self {
637 match u {
638 ArrowIntervalUnit::YearMonth => Self::interval_year_month_datatype(),
639 ArrowIntervalUnit::DayTime => Self::interval_day_time_datatype(),
640 ArrowIntervalUnit::MonthDayNano => Self::interval_month_day_nano_datatype(),
641 }
642 }
643
644 pub fn list_datatype(item_type: Arc<ConcreteDataType>) -> ConcreteDataType {
645 ConcreteDataType::List(ListType::new(item_type))
646 }
647
648 pub fn struct_datatype(fields: StructType) -> ConcreteDataType {
649 ConcreteDataType::Struct(fields)
650 }
651
652 pub fn dictionary_datatype(
653 key_type: ConcreteDataType,
654 value_type: ConcreteDataType,
655 ) -> ConcreteDataType {
656 ConcreteDataType::Dictionary(DictionaryType::new(key_type, value_type))
657 }
658
659 pub fn decimal128_datatype(precision: u8, scale: i8) -> ConcreteDataType {
660 ConcreteDataType::Decimal128(Decimal128Type::new(precision, scale))
661 }
662
663 pub fn decimal128_default_datatype() -> ConcreteDataType {
664 Self::decimal128_datatype(DECIMAL128_MAX_PRECISION, DECIMAL_DEFAULT_SCALE)
665 }
666
667 pub fn vector_datatype(dim: u32) -> ConcreteDataType {
668 ConcreteDataType::Vector(VectorType::new(dim))
669 }
670
671 pub fn vector_default_datatype() -> ConcreteDataType {
672 Self::vector_datatype(0)
673 }
674
675 pub fn json_native_datatype(inner_type: ConcreteDataType) -> ConcreteDataType {
676 ConcreteDataType::Json(JsonType::new(JsonFormat::Native(Box::new(inner_type))))
677 }
678}
679
680#[enum_dispatch::enum_dispatch]
682pub trait DataType: std::fmt::Debug + Send + Sync {
683 fn name(&self) -> String;
685
686 fn logical_type_id(&self) -> LogicalTypeId;
688
689 fn default_value(&self) -> Value;
691
692 fn as_arrow_type(&self) -> ArrowDataType;
694
695 fn create_mutable_vector(&self, capacity: usize) -> Box<dyn MutableVector>;
697
698 fn try_cast(&self, from: Value) -> Option<Value>;
701}
702
703pub type DataTypeRef = Arc<dyn DataType>;
704
705#[cfg(test)]
706mod tests {
707 use arrow::datatypes::Field;
708
709 use super::*;
710
711 #[test]
712 fn test_concrete_type_as_datatype_trait() {
713 let concrete_type = ConcreteDataType::boolean_datatype();
714
715 assert_eq!("Boolean", concrete_type.to_string());
716 assert_eq!(Value::Boolean(false), concrete_type.default_value());
717 assert_eq!(LogicalTypeId::Boolean, concrete_type.logical_type_id());
718 assert_eq!(ArrowDataType::Boolean, concrete_type.as_arrow_type());
719 }
720
721 #[test]
722 fn test_from_arrow_type() {
723 assert!(matches!(
724 ConcreteDataType::from_arrow_type(&ArrowDataType::Null),
725 ConcreteDataType::Null(_)
726 ));
727 assert!(matches!(
728 ConcreteDataType::from_arrow_type(&ArrowDataType::Boolean),
729 ConcreteDataType::Boolean(_)
730 ));
731 assert!(matches!(
732 ConcreteDataType::from_arrow_type(&ArrowDataType::Binary),
733 ConcreteDataType::Binary(_)
734 ));
735 assert!(matches!(
736 ConcreteDataType::from_arrow_type(&ArrowDataType::LargeBinary),
737 ConcreteDataType::Binary(_)
738 ));
739 assert!(matches!(
740 ConcreteDataType::from_arrow_type(&ArrowDataType::Int8),
741 ConcreteDataType::Int8(_)
742 ));
743 assert!(matches!(
744 ConcreteDataType::from_arrow_type(&ArrowDataType::Int16),
745 ConcreteDataType::Int16(_)
746 ));
747 assert!(matches!(
748 ConcreteDataType::from_arrow_type(&ArrowDataType::Int32),
749 ConcreteDataType::Int32(_)
750 ));
751 assert!(matches!(
752 ConcreteDataType::from_arrow_type(&ArrowDataType::Int64),
753 ConcreteDataType::Int64(_)
754 ));
755 assert!(matches!(
756 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt8),
757 ConcreteDataType::UInt8(_)
758 ));
759 assert!(matches!(
760 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt16),
761 ConcreteDataType::UInt16(_)
762 ));
763 assert!(matches!(
764 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt32),
765 ConcreteDataType::UInt32(_)
766 ));
767 assert!(matches!(
768 ConcreteDataType::from_arrow_type(&ArrowDataType::UInt64),
769 ConcreteDataType::UInt64(_)
770 ));
771 assert!(matches!(
772 ConcreteDataType::from_arrow_type(&ArrowDataType::Float32),
773 ConcreteDataType::Float32(_)
774 ));
775 assert!(matches!(
776 ConcreteDataType::from_arrow_type(&ArrowDataType::Float64),
777 ConcreteDataType::Float64(_)
778 ));
779 assert!(matches!(
780 ConcreteDataType::from_arrow_type(&ArrowDataType::Utf8),
781 ConcreteDataType::String(_)
782 ));
783 let large_string_type = ConcreteDataType::from_arrow_type(&ArrowDataType::LargeUtf8);
785 assert!(matches!(large_string_type, ConcreteDataType::String(_)));
786 if let ConcreteDataType::String(string_type) = &large_string_type {
787 assert!(string_type.is_large());
788 } else {
789 panic!("Expected a String type");
790 }
791 assert_eq!(
792 ConcreteDataType::from_arrow_type(&ArrowDataType::List(Arc::new(Field::new(
793 "item",
794 ArrowDataType::Int32,
795 true,
796 )))),
797 ConcreteDataType::List(ListType::new(Arc::new(ConcreteDataType::int32_datatype())))
798 );
799 assert!(matches!(
800 ConcreteDataType::from_arrow_type(&ArrowDataType::Date32),
801 ConcreteDataType::Date(_)
802 ));
803 }
804
805 #[test]
806 fn test_large_utf8_round_trip() {
807 let large_utf8_arrow = ArrowDataType::LargeUtf8;
809 let concrete_type = ConcreteDataType::from_arrow_type(&large_utf8_arrow);
810 let back_to_arrow = concrete_type.as_arrow_type();
811
812 assert!(matches!(concrete_type, ConcreteDataType::String(_)));
813 assert_eq!(large_utf8_arrow, back_to_arrow);
815
816 let utf8_concrete = ConcreteDataType::from_arrow_type(&ArrowDataType::Utf8);
818 let large_utf8_concrete = ConcreteDataType::from_arrow_type(&ArrowDataType::LargeUtf8);
819
820 assert!(matches!(utf8_concrete, ConcreteDataType::String(_)));
821 assert!(matches!(large_utf8_concrete, ConcreteDataType::String(_)));
822
823 if let (ConcreteDataType::String(utf8_type), ConcreteDataType::String(large_type)) =
825 (&utf8_concrete, &large_utf8_concrete)
826 {
827 assert!(!utf8_type.is_large());
828 assert!(large_type.is_large());
829 } else {
830 panic!("Expected both to be String types");
831 }
832
833 assert_ne!(utf8_concrete, large_utf8_concrete);
835 }
836
837 #[test]
838 fn test_from_arrow_timestamp() {
839 assert_eq!(
840 ConcreteDataType::timestamp_millisecond_datatype(),
841 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Millisecond)
842 );
843 assert_eq!(
844 ConcreteDataType::timestamp_microsecond_datatype(),
845 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Microsecond)
846 );
847 assert_eq!(
848 ConcreteDataType::timestamp_nanosecond_datatype(),
849 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Nanosecond)
850 );
851 assert_eq!(
852 ConcreteDataType::timestamp_second_datatype(),
853 ConcreteDataType::from_arrow_time_unit(&ArrowTimeUnit::Second)
854 );
855 }
856
857 #[test]
858 fn test_is_null() {
859 assert!(ConcreteDataType::null_datatype().is_null());
860 assert!(!ConcreteDataType::int32_datatype().is_null());
861 }
862
863 #[test]
864 fn test_is_float() {
865 assert!(!ConcreteDataType::int32_datatype().is_float());
866 assert!(ConcreteDataType::float32_datatype().is_float());
867 assert!(ConcreteDataType::float64_datatype().is_float());
868 }
869
870 #[test]
871 fn test_is_boolean() {
872 assert!(!ConcreteDataType::int32_datatype().is_boolean());
873 assert!(!ConcreteDataType::float32_datatype().is_boolean());
874 assert!(ConcreteDataType::boolean_datatype().is_boolean());
875 }
876
877 #[test]
878 fn test_is_decimal() {
879 assert!(!ConcreteDataType::int32_datatype().is_decimal());
880 assert!(!ConcreteDataType::float32_datatype().is_decimal());
881 assert!(ConcreteDataType::decimal128_datatype(10, 2).is_decimal());
882 assert!(ConcreteDataType::decimal128_datatype(18, 6).is_decimal());
883 }
884
885 #[test]
886 fn test_is_stringifiable() {
887 assert!(!ConcreteDataType::int32_datatype().is_stringifiable());
888 assert!(!ConcreteDataType::float32_datatype().is_stringifiable());
889 assert!(ConcreteDataType::string_datatype().is_stringifiable());
890 assert!(ConcreteDataType::binary_datatype().is_stringifiable());
891 assert!(ConcreteDataType::date_datatype().is_stringifiable());
892 assert!(ConcreteDataType::timestamp_second_datatype().is_stringifiable());
893 assert!(ConcreteDataType::timestamp_millisecond_datatype().is_stringifiable());
894 assert!(ConcreteDataType::timestamp_microsecond_datatype().is_stringifiable());
895 assert!(ConcreteDataType::timestamp_nanosecond_datatype().is_stringifiable());
896 assert!(ConcreteDataType::time_second_datatype().is_stringifiable());
897 assert!(ConcreteDataType::time_millisecond_datatype().is_stringifiable());
898 assert!(ConcreteDataType::time_microsecond_datatype().is_stringifiable());
899 assert!(ConcreteDataType::time_nanosecond_datatype().is_stringifiable());
900
901 assert!(ConcreteDataType::interval_year_month_datatype().is_stringifiable());
902 assert!(ConcreteDataType::interval_day_time_datatype().is_stringifiable());
903 assert!(ConcreteDataType::interval_month_day_nano_datatype().is_stringifiable());
904
905 assert!(ConcreteDataType::duration_second_datatype().is_stringifiable());
906 assert!(ConcreteDataType::duration_millisecond_datatype().is_stringifiable());
907 assert!(ConcreteDataType::duration_microsecond_datatype().is_stringifiable());
908 assert!(ConcreteDataType::duration_nanosecond_datatype().is_stringifiable());
909 assert!(ConcreteDataType::decimal128_datatype(10, 2).is_stringifiable());
910 assert!(ConcreteDataType::vector_default_datatype().is_stringifiable());
911 }
912
913 #[test]
914 fn test_is_signed() {
915 assert!(ConcreteDataType::int8_datatype().is_signed());
916 assert!(ConcreteDataType::int16_datatype().is_signed());
917 assert!(ConcreteDataType::int32_datatype().is_signed());
918 assert!(ConcreteDataType::int64_datatype().is_signed());
919 assert!(ConcreteDataType::date_datatype().is_signed());
920 assert!(ConcreteDataType::timestamp_second_datatype().is_signed());
921 assert!(ConcreteDataType::timestamp_millisecond_datatype().is_signed());
922 assert!(ConcreteDataType::timestamp_microsecond_datatype().is_signed());
923 assert!(ConcreteDataType::timestamp_nanosecond_datatype().is_signed());
924 assert!(ConcreteDataType::time_second_datatype().is_signed());
925 assert!(ConcreteDataType::time_millisecond_datatype().is_signed());
926 assert!(ConcreteDataType::time_microsecond_datatype().is_signed());
927 assert!(ConcreteDataType::time_nanosecond_datatype().is_signed());
928 assert!(ConcreteDataType::interval_year_month_datatype().is_signed());
929 assert!(ConcreteDataType::interval_day_time_datatype().is_signed());
930 assert!(ConcreteDataType::interval_month_day_nano_datatype().is_signed());
931 assert!(ConcreteDataType::duration_second_datatype().is_signed());
932 assert!(ConcreteDataType::duration_millisecond_datatype().is_signed());
933 assert!(ConcreteDataType::duration_microsecond_datatype().is_signed());
934 assert!(ConcreteDataType::duration_nanosecond_datatype().is_signed());
935
936 assert!(!ConcreteDataType::uint8_datatype().is_signed());
937 assert!(!ConcreteDataType::uint16_datatype().is_signed());
938 assert!(!ConcreteDataType::uint32_datatype().is_signed());
939 assert!(!ConcreteDataType::uint64_datatype().is_signed());
940
941 assert!(!ConcreteDataType::float32_datatype().is_signed());
942 assert!(!ConcreteDataType::float64_datatype().is_signed());
943
944 assert!(ConcreteDataType::decimal128_datatype(10, 2).is_signed());
945 }
946
947 #[test]
948 fn test_is_unsigned() {
949 assert!(!ConcreteDataType::int8_datatype().is_unsigned());
950 assert!(!ConcreteDataType::int16_datatype().is_unsigned());
951 assert!(!ConcreteDataType::int32_datatype().is_unsigned());
952 assert!(!ConcreteDataType::int64_datatype().is_unsigned());
953 assert!(!ConcreteDataType::date_datatype().is_unsigned());
954 assert!(!ConcreteDataType::timestamp_second_datatype().is_unsigned());
955 assert!(!ConcreteDataType::timestamp_millisecond_datatype().is_unsigned());
956 assert!(!ConcreteDataType::timestamp_microsecond_datatype().is_unsigned());
957 assert!(!ConcreteDataType::timestamp_nanosecond_datatype().is_unsigned());
958 assert!(!ConcreteDataType::time_second_datatype().is_unsigned());
959 assert!(!ConcreteDataType::time_millisecond_datatype().is_unsigned());
960 assert!(!ConcreteDataType::time_microsecond_datatype().is_unsigned());
961 assert!(!ConcreteDataType::time_nanosecond_datatype().is_unsigned());
962 assert!(!ConcreteDataType::interval_year_month_datatype().is_unsigned());
963 assert!(!ConcreteDataType::interval_day_time_datatype().is_unsigned());
964 assert!(!ConcreteDataType::interval_month_day_nano_datatype().is_unsigned());
965 assert!(!ConcreteDataType::duration_second_datatype().is_unsigned());
966 assert!(!ConcreteDataType::duration_millisecond_datatype().is_unsigned());
967 assert!(!ConcreteDataType::duration_microsecond_datatype().is_unsigned());
968 assert!(!ConcreteDataType::duration_nanosecond_datatype().is_unsigned());
969 assert!(!ConcreteDataType::decimal128_datatype(10, 2).is_unsigned());
970
971 assert!(ConcreteDataType::uint8_datatype().is_unsigned());
972 assert!(ConcreteDataType::uint16_datatype().is_unsigned());
973 assert!(ConcreteDataType::uint32_datatype().is_unsigned());
974 assert!(ConcreteDataType::uint64_datatype().is_unsigned());
975
976 assert!(!ConcreteDataType::float32_datatype().is_unsigned());
977 assert!(!ConcreteDataType::float64_datatype().is_unsigned());
978 }
979
980 #[test]
981 fn test_numerics() {
982 let nums = ConcreteDataType::numerics();
983 assert_eq!(10, nums.len());
984 }
985
986 #[test]
987 fn test_as_list() {
988 let list_type =
989 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype()));
990 assert_eq!(
991 ListType::new(Arc::new(ConcreteDataType::int32_datatype())),
992 *list_type.as_list().unwrap()
993 );
994 assert!(ConcreteDataType::int32_datatype().as_list().is_none());
995 }
996
997 #[test]
998 fn test_display_concrete_data_type() {
999 assert_eq!(ConcreteDataType::null_datatype().to_string(), "Null");
1000 assert_eq!(ConcreteDataType::boolean_datatype().to_string(), "Boolean");
1001 assert_eq!(ConcreteDataType::binary_datatype().to_string(), "Binary");
1002 assert_eq!(ConcreteDataType::int8_datatype().to_string(), "Int8");
1003 assert_eq!(ConcreteDataType::int16_datatype().to_string(), "Int16");
1004 assert_eq!(ConcreteDataType::int32_datatype().to_string(), "Int32");
1005 assert_eq!(ConcreteDataType::int64_datatype().to_string(), "Int64");
1006 assert_eq!(ConcreteDataType::uint8_datatype().to_string(), "UInt8");
1007 assert_eq!(ConcreteDataType::uint16_datatype().to_string(), "UInt16");
1008 assert_eq!(ConcreteDataType::uint32_datatype().to_string(), "UInt32");
1009 assert_eq!(ConcreteDataType::uint64_datatype().to_string(), "UInt64");
1010 assert_eq!(ConcreteDataType::float32_datatype().to_string(), "Float32");
1011 assert_eq!(ConcreteDataType::float64_datatype().to_string(), "Float64");
1012 assert_eq!(ConcreteDataType::string_datatype().to_string(), "String");
1013 assert_eq!(ConcreteDataType::date_datatype().to_string(), "Date");
1014 assert_eq!(
1015 ConcreteDataType::timestamp_millisecond_datatype().to_string(),
1016 "TimestampMillisecond"
1017 );
1018 assert_eq!(
1019 ConcreteDataType::time_millisecond_datatype().to_string(),
1020 "TimeMillisecond"
1021 );
1022 assert_eq!(
1023 ConcreteDataType::interval_month_day_nano_datatype().to_string(),
1024 "IntervalMonthDayNano"
1025 );
1026 assert_eq!(
1027 ConcreteDataType::duration_second_datatype().to_string(),
1028 "DurationSecond"
1029 );
1030 assert_eq!(
1031 ConcreteDataType::decimal128_datatype(10, 2).to_string(),
1032 "Decimal(10, 2)"
1033 );
1034 assert_eq!(
1036 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype()))
1037 .to_string(),
1038 "List<Int32>"
1039 );
1040 assert_eq!(
1041 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::Dictionary(
1042 DictionaryType::new(
1043 ConcreteDataType::int32_datatype(),
1044 ConcreteDataType::string_datatype()
1045 )
1046 )))
1047 .to_string(),
1048 "List<Dictionary<Int32, String>>"
1049 );
1050 assert_eq!(
1051 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::list_datatype(Arc::new(
1052 ConcreteDataType::list_datatype(Arc::new(ConcreteDataType::int32_datatype()))
1053 ))))
1054 .to_string(),
1055 "List<List<List<Int32>>>"
1056 );
1057 assert_eq!(
1058 ConcreteDataType::dictionary_datatype(
1059 ConcreteDataType::int32_datatype(),
1060 ConcreteDataType::string_datatype()
1061 )
1062 .to_string(),
1063 "Dictionary<Int32, String>"
1064 );
1065 assert_eq!(
1066 ConcreteDataType::vector_datatype(3).to_string(),
1067 "Vector(3)"
1068 );
1069 }
1070}