arrow_schema/datatype.rs
1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements. See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership. The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License. You may obtain a copy of the License at
8//
9// http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied. See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use std::str::FromStr;
19use std::sync::Arc;
20
21use crate::{ArrowError, Field, FieldRef, Fields, UnionFields};
22
23/// Datatypes supported by this implementation of Apache Arrow.
24///
25/// The variants of this enum include primitive fixed size types as well as
26/// parametric or nested types. See [`Schema.fbs`] for Arrow's specification.
27///
28/// # Examples
29///
30/// Primitive types
31/// ```
32/// # use arrow_schema::DataType;
33/// // create a new 32-bit signed integer
34/// let data_type = DataType::Int32;
35/// ```
36///
37/// Nested Types
38/// ```
39/// # use arrow_schema::{DataType, Field};
40/// # use std::sync::Arc;
41/// // create a new list of 32-bit signed integers directly
42/// let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true)));
43/// // Create the same list type with constructor
44/// let list_data_type2 = DataType::new_list(DataType::Int32, true);
45/// assert_eq!(list_data_type, list_data_type2);
46/// ```
47///
48/// Dictionary Types
49/// ```
50/// # use arrow_schema::{DataType};
51/// // String Dictionary (key type Int32 and value type Utf8)
52/// let data_type = DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8));
53/// ```
54///
55/// Timestamp Types
56/// ```
57/// # use arrow_schema::{DataType, TimeUnit};
58/// // timestamp with millisecond precision without timezone specified
59/// let data_type = DataType::Timestamp(TimeUnit::Millisecond, None);
60/// // timestamp with nanosecond precision in UTC timezone
61/// let data_type = DataType::Timestamp(TimeUnit::Nanosecond, Some("UTC".into()));
62///```
63///
64/// # Display and FromStr
65///
66/// The `Display` and `FromStr` implementations for `DataType` are
67/// human-readable, parseable, and reversible.
68///
69/// ```
70/// # use arrow_schema::DataType;
71/// let data_type = DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8));
72/// let data_type_string = data_type.to_string();
73/// assert_eq!(data_type_string, "Dictionary(Int32, Utf8)");
74/// // display can be parsed back into the original type
75/// let parsed_data_type: DataType = data_type.to_string().parse().unwrap();
76/// assert_eq!(data_type, parsed_data_type);
77/// ```
78///
79/// # Nested Support
80/// Currently, the Rust implementation supports the following nested types:
81/// - `List<T>`
82/// - `LargeList<T>`
83/// - `FixedSizeList<T>`
84/// - `Struct<T, U, V, ...>`
85/// - `Union<T, U, V, ...>`
86/// - `Map<K, V>`
87///
88/// Nested types can themselves be nested within other arrays.
89/// For more information on these types please see
90/// [the physical memory layout of Apache Arrow]
91///
92/// [`Schema.fbs`]: https://github.com/apache/arrow/blob/main/format/Schema.fbs
93/// [the physical memory layout of Apache Arrow]: https://arrow.apache.org/docs/format/Columnar.html#physical-memory-layout
94#[derive(Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
95#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
96pub enum DataType {
97 /// Null type
98 Null,
99 /// A boolean datatype representing the values `true` and `false`.
100 Boolean,
101 /// A signed 8-bit integer.
102 Int8,
103 /// A signed 16-bit integer.
104 Int16,
105 /// A signed 32-bit integer.
106 Int32,
107 /// A signed 64-bit integer.
108 Int64,
109 /// An unsigned 8-bit integer.
110 UInt8,
111 /// An unsigned 16-bit integer.
112 UInt16,
113 /// An unsigned 32-bit integer.
114 UInt32,
115 /// An unsigned 64-bit integer.
116 UInt64,
117 /// A 16-bit floating point number.
118 Float16,
119 /// A 32-bit floating point number.
120 Float32,
121 /// A 64-bit floating point number.
122 Float64,
123 /// A timestamp with an optional timezone.
124 ///
125 /// Time is measured as a Unix epoch, counting the seconds from
126 /// 00:00:00.000 on 1 January 1970, excluding leap seconds,
127 /// as a signed 64-bit integer.
128 ///
129 /// The time zone is a string indicating the name of a time zone, one of:
130 ///
131 /// * As used in the Olson time zone database (the "tz database" or
132 /// "tzdata"), such as "America/New_York"
133 /// * An absolute time zone offset of the form +XX:XX or -XX:XX, such as +07:30
134 ///
135 /// Timestamps with a non-empty timezone
136 /// ------------------------------------
137 ///
138 /// If a Timestamp column has a non-empty timezone value, its epoch is
139 /// 1970-01-01 00:00:00 (January 1st 1970, midnight) in the *UTC* timezone
140 /// (the Unix epoch), regardless of the Timestamp's own timezone.
141 ///
142 /// Therefore, timestamp values with a non-empty timezone correspond to
143 /// physical points in time together with some additional information about
144 /// how the data was obtained and/or how to display it (the timezone).
145 ///
146 /// For example, the timestamp value 0 with the timezone string "Europe/Paris"
147 /// corresponds to "January 1st 1970, 00h00" in the UTC timezone, but the
148 /// application may prefer to display it as "January 1st 1970, 01h00" in
149 /// the Europe/Paris timezone (which is the same physical point in time).
150 ///
151 /// One consequence is that timestamp values with a non-empty timezone
152 /// can be compared and ordered directly, since they all share the same
153 /// well-known point of reference (the Unix epoch).
154 ///
155 /// Timestamps with an unset / empty timezone
156 /// -----------------------------------------
157 ///
158 /// If a Timestamp column has no timezone value, its epoch is
159 /// 1970-01-01 00:00:00 (January 1st 1970, midnight) in an *unknown* timezone.
160 ///
161 /// Therefore, timestamp values without a timezone cannot be meaningfully
162 /// interpreted as physical points in time, but only as calendar / clock
163 /// indications ("wall clock time") in an unspecified timezone.
164 ///
165 /// For example, the timestamp value 0 with an empty timezone string
166 /// corresponds to "January 1st 1970, 00h00" in an unknown timezone: there
167 /// is not enough information to interpret it as a well-defined physical
168 /// point in time.
169 ///
170 /// One consequence is that timestamp values without a timezone cannot
171 /// be reliably compared or ordered, since they may have different points of
172 /// reference. In particular, it is *not* possible to interpret an unset
173 /// or empty timezone as the same as "UTC".
174 ///
175 /// Conversion between timezones
176 /// ----------------------------
177 ///
178 /// If a Timestamp column has a non-empty timezone, changing the timezone
179 /// to a different non-empty value is a metadata-only operation:
180 /// the timestamp values need not change as their point of reference remains
181 /// the same (the Unix epoch).
182 ///
183 /// However, if a Timestamp column has no timezone value, changing it to a
184 /// non-empty value requires to think about the desired semantics.
185 /// One possibility is to assume that the original timestamp values are
186 /// relative to the epoch of the timezone being set; timestamp values should
187 /// then adjusted to the Unix epoch (for example, changing the timezone from
188 /// empty to "Europe/Paris" would require converting the timestamp values
189 /// from "Europe/Paris" to "UTC", which seems counter-intuitive but is
190 /// nevertheless correct).
191 ///
192 /// ```
193 /// # use arrow_schema::{DataType, TimeUnit};
194 /// DataType::Timestamp(TimeUnit::Second, None);
195 /// DataType::Timestamp(TimeUnit::Second, Some("literal".into()));
196 /// DataType::Timestamp(TimeUnit::Second, Some("string".to_string().into()));
197 /// ```
198 ///
199 /// # Timezone representation
200 /// ----------------------------
201 /// It is possible to use either the timezone string representation, such as "UTC", or the absolute time zone offset "+00:00".
202 /// For timezones with fixed offsets, such as "UTC" or "JST", the offset representation is recommended, as it is more explicit and less ambiguous.
203 ///
204 /// Most arrow-rs functionalities use the absolute offset representation,
205 /// such as [`PrimitiveArray::with_timezone_utc`] that applies a
206 /// UTC timezone to timestamp arrays.
207 ///
208 /// [`PrimitiveArray::with_timezone_utc`]: https://docs.rs/arrow/latest/arrow/array/struct.PrimitiveArray.html#method.with_timezone_utc
209 ///
210 /// Timezone string parsing
211 /// -----------------------
212 /// When feature `chrono-tz` is not enabled, allowed timezone strings are fixed offsets of the form "+09:00", "-09" or "+0930".
213 ///
214 /// When feature `chrono-tz` is enabled, additional strings supported by [chrono_tz](https://docs.rs/chrono-tz/latest/chrono_tz/)
215 /// are also allowed, which include [IANA database](https://en.wikipedia.org/wiki/List_of_tz_database_time_zones)
216 /// timezones.
217 Timestamp(TimeUnit, Option<Arc<str>>),
218 /// A signed 32-bit date representing the elapsed time since UNIX epoch (1970-01-01)
219 /// in days.
220 Date32,
221 /// A signed 64-bit date representing the elapsed time since UNIX epoch (1970-01-01)
222 /// in milliseconds.
223 ///
224 /// # Valid Ranges
225 ///
226 /// According to the Arrow specification ([Schema.fbs]), values of Date64
227 /// are treated as the number of *days*, in milliseconds, since the UNIX
228 /// epoch. Therefore, values of this type must be evenly divisible by
229 /// `86_400_000`, the number of milliseconds in a standard day.
230 ///
231 /// It is not valid to store milliseconds that do not represent an exact
232 /// day. The reason for this restriction is compatibility with other
233 /// language's native libraries (specifically Java), which historically
234 /// lacked a dedicated date type and only supported timestamps.
235 ///
236 /// # Validation
237 ///
238 /// This library does not validate or enforce that Date64 values are evenly
239 /// divisible by `86_400_000` for performance and usability reasons. Date64
240 /// values are treated similarly to `Timestamp(TimeUnit::Millisecond,
241 /// None)`: values will be displayed with a time of day if the value does
242 /// not represent an exact day, and arithmetic will be done at the
243 /// millisecond granularity.
244 ///
245 /// # Recommendation
246 ///
247 /// Users should prefer [`Date32`] to cleanly represent the number
248 /// of days, or one of the Timestamp variants to include time as part of the
249 /// representation, depending on their use case.
250 ///
251 /// # Further Reading
252 ///
253 /// For more details, see [#5288](https://github.com/apache/arrow-rs/issues/5288).
254 ///
255 /// [`Date32`]: Self::Date32
256 /// [Schema.fbs]: https://github.com/apache/arrow/blob/main/format/Schema.fbs
257 Date64,
258 /// A signed 32-bit time representing the elapsed time since midnight in the unit of `TimeUnit`.
259 /// Must be either seconds or milliseconds.
260 Time32(TimeUnit),
261 /// A signed 64-bit time representing the elapsed time since midnight in the unit of `TimeUnit`.
262 /// Must be either microseconds or nanoseconds.
263 Time64(TimeUnit),
264 /// Measure of elapsed time in either seconds, milliseconds, microseconds or nanoseconds.
265 Duration(TimeUnit),
266 /// A "calendar" interval which models types that don't necessarily
267 /// have a precise duration without the context of a base timestamp (e.g.
268 /// days can differ in length during day light savings time transitions).
269 Interval(IntervalUnit),
270 /// Opaque binary data of variable length.
271 ///
272 /// A single Binary array can store up to [`i32::MAX`] bytes
273 /// of binary data in total.
274 Binary,
275 /// Opaque binary data of fixed size.
276 ///
277 /// Enum parameter specifies the number of bytes per value, defined by the
278 /// [`byteWidth` field] in the Arrow Spec
279 ///
280 /// [`byteWidth` field]: https://github.com/apache/arrow/blob/2a89d03bbefd620b42126b8e00f8ae57e99cd638/format/Schema.fbs#L211
281 FixedSizeBinary(i32),
282 /// Opaque binary data of variable length and 64-bit offsets.
283 ///
284 /// A single LargeBinary array can store up to [`i64::MAX`] bytes
285 /// of binary data in total.
286 LargeBinary,
287 /// Opaque binary data of variable length.
288 ///
289 /// Logically the same as [`Binary`], but the internal representation uses a view
290 /// struct that contains the string length and either the string's entire data
291 /// inline (for small strings) or an inlined prefix, an index of another buffer,
292 /// and an offset pointing to a slice in that buffer (for non-small strings).
293 ///
294 /// [`Binary`]: Self::Binary
295 BinaryView,
296 /// A variable-length string in Unicode with UTF-8 encoding.
297 ///
298 /// A single Utf8 array can store up to [`i32::MAX`] bytes
299 /// of string data in total.
300 Utf8,
301 /// A variable-length string in Unicode with UFT-8 encoding and 64-bit offsets.
302 ///
303 /// A single LargeUtf8 array can store up to [`i64::MAX`] bytes
304 /// of string data in total.
305 LargeUtf8,
306 /// A variable-length string in Unicode with UTF-8 encoding
307 ///
308 /// Logically the same as [`Utf8`], but the internal representation uses a view
309 /// struct that contains the string length and either the string's entire data
310 /// inline (for small strings) or an inlined prefix, an index of another buffer,
311 /// and an offset pointing to a slice in that buffer (for non-small strings).
312 ///
313 /// [`Utf8`]: Self::Utf8
314 Utf8View,
315 /// A list of some logical data type with variable length.
316 ///
317 /// A single List array can store up to [`i32::MAX`] elements in total.
318 List(FieldRef),
319 /// A list of some logical data type with variable length.
320 ///
321 /// Logically the same as [`List`], but the internal representation differs in how child
322 /// data is referenced, allowing flexibility in how data is layed out.
323 ///
324 /// [`List`]: Self::List
325 ListView(FieldRef),
326 /// A list of some logical data type with fixed length.
327 FixedSizeList(FieldRef, i32),
328 /// A list of some logical data type with variable length and 64-bit offsets.
329 ///
330 /// A single LargeList array can store up to [`i64::MAX`] elements in total.
331 LargeList(FieldRef),
332 /// A list of some logical data type with variable length and 64-bit offsets.
333 ///
334 /// Logically the same as [`LargeList`], but the internal representation differs in how child
335 /// data is referenced, allowing flexibility in how data is layed out.
336 ///
337 /// [`LargeList`]: Self::LargeList
338 LargeListView(FieldRef),
339 /// A nested datatype that contains a number of sub-fields.
340 Struct(Fields),
341 /// A nested datatype that can represent slots of differing types. Components:
342 ///
343 /// 1. [`UnionFields`]
344 /// 2. The type of union (Sparse or Dense)
345 Union(UnionFields, UnionMode),
346 /// A dictionary encoded array (`key_type`, `value_type`), where
347 /// each array element is an index of `key_type` into an
348 /// associated dictionary of `value_type`.
349 ///
350 /// Dictionary arrays are used to store columns of `value_type`
351 /// that contain many repeated values using less memory, but with
352 /// a higher CPU overhead for some operations.
353 ///
354 /// This type mostly used to represent low cardinality string
355 /// arrays or a limited set of primitive types as integers.
356 Dictionary(Box<DataType>, Box<DataType>),
357 /// Exact 32-bit width decimal value with precision and scale
358 ///
359 /// * precision is the maximum number of digits in the unscaled value
360 /// * scale controls the position of the decimal point
361 ///
362 /// The represented value is the unscaled integer multiplied by 10^{-scale}.
363 /// For example, the unscaled value 12345 with precision 5 and scale 2
364 /// represents 123.45.
365 ///
366 /// Scale can also be negative. For example, the unscaled value 12 with
367 /// precision 2 and scale -3 represents 12000.
368 Decimal32(u8, i8),
369 /// Exact 64-bit width decimal value with precision and scale
370 ///
371 /// * precision is the maximum number of digits in the unscaled value
372 /// * scale controls the position of the decimal point
373 ///
374 /// The represented value is the unscaled integer multiplied by 10^{-scale}.
375 /// For example, the unscaled value 12345 with precision 5 and scale 2
376 /// represents 123.45.
377 ///
378 /// Scale can also be negative. For example, the unscaled value 12 with
379 /// precision 2 and scale -3 represents 12000.
380 Decimal64(u8, i8),
381 /// Exact 128-bit width decimal value with precision and scale
382 ///
383 /// * precision is the maximum number of digits in the unscaled value
384 /// * scale controls the position of the decimal point
385 ///
386 /// The represented value is the unscaled integer multiplied by 10^{-scale}.
387 /// For example, the unscaled value 12345 with precision 5 and scale 2
388 /// represents 123.45.
389 ///
390 /// Scale can also be negative. For example, the unscaled value 12 with
391 /// precision 2 and scale -3 represents 12000.
392 Decimal128(u8, i8),
393 /// Exact 256-bit width decimal value with precision and scale
394 ///
395 /// * precision is the maximum number of digits in the unscaled value
396 /// * scale controls the position of the decimal point
397 ///
398 /// The represented value is the unscaled integer multiplied by 10^{-scale}.
399 /// For example, the unscaled value 12345 with precision 5 and scale 2
400 /// represents 123.45.
401 ///
402 /// Scale can also be negative. For example, the unscaled value 12 with
403 /// precision 2 and scale -3 represents 12000.
404 Decimal256(u8, i8),
405 /// A Map is a logical nested type that is represented as
406 ///
407 /// `List<entries: Struct<key: K, value: V>>`
408 ///
409 /// The keys and values are each respectively contiguous.
410 /// The key and value types are not constrained, but keys should be
411 /// hashable and unique.
412 /// Whether the keys are sorted can be set in the `bool` after the `Field`.
413 ///
414 /// In a field with Map type, the field has a child Struct field, which then
415 /// has two children: key type and the second the value type. The names of the
416 /// child fields may be respectively "entries", "key", and "value", but this is
417 /// not enforced.
418 Map(FieldRef, bool),
419 /// A run-end encoding (REE) is a variation of run-length encoding (RLE). These
420 /// encodings are well-suited for representing data containing sequences of the
421 /// same value, called runs. Each run is represented as a value and an integer giving
422 /// the index in the array where the run ends.
423 ///
424 /// A run-end encoded array has no buffers by itself, but has two child arrays. The
425 /// first child array, called the run ends array, holds either 16, 32, or 64-bit
426 /// signed integers. The actual values of each run are held in the second child array.
427 ///
428 /// These child arrays are prescribed the standard names of "run_ends" and "values"
429 /// respectively.
430 RunEndEncoded(FieldRef, FieldRef),
431}
432
433/// An absolute length of time in seconds, milliseconds, microseconds or nanoseconds.
434#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
435#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
436pub enum TimeUnit {
437 /// Time in seconds.
438 Second,
439 /// Time in milliseconds.
440 Millisecond,
441 /// Time in microseconds.
442 Microsecond,
443 /// Time in nanoseconds.
444 Nanosecond,
445}
446
447impl std::fmt::Display for TimeUnit {
448 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
449 match self {
450 TimeUnit::Second => write!(f, "s"),
451 TimeUnit::Millisecond => write!(f, "ms"),
452 TimeUnit::Microsecond => write!(f, "µs"),
453 TimeUnit::Nanosecond => write!(f, "ns"),
454 }
455 }
456}
457
458/// YEAR_MONTH, DAY_TIME, MONTH_DAY_NANO interval in SQL style.
459#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
460#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
461pub enum IntervalUnit {
462 /// Indicates the number of elapsed whole months, stored as 4-byte integers.
463 YearMonth,
464 /// Indicates the number of elapsed days and milliseconds,
465 /// stored as 2 contiguous 32-bit integers (days, milliseconds) (8-bytes in total).
466 DayTime,
467 /// A triple of the number of elapsed months, days, and nanoseconds.
468 /// The values are stored contiguously in 16 byte blocks. Months and
469 /// days are encoded as 32 bit integers and nanoseconds is encoded as a
470 /// 64 bit integer. All integers are signed. Each field is independent
471 /// (e.g. there is no constraint that nanoseconds have the same sign
472 /// as days or that the quantity of nanoseconds represents less
473 /// than a day's worth of time).
474 MonthDayNano,
475}
476
477/// Sparse or Dense union layouts
478#[derive(Debug, Clone, PartialEq, Eq, Hash, PartialOrd, Ord, Copy)]
479#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
480pub enum UnionMode {
481 /// Sparse union layout
482 Sparse,
483 /// Dense union layout
484 Dense,
485}
486
487/// Parses `str` into a `DataType`.
488///
489/// This is the reverse of [`DataType`]'s `Display`
490/// impl, and maintains the invariant that
491/// `DataType::try_from(&data_type.to_string()).unwrap() == data_type`
492///
493/// # Example
494/// ```
495/// use arrow_schema::DataType;
496///
497/// let data_type: DataType = "Int32".parse().unwrap();
498/// assert_eq!(data_type, DataType::Int32);
499/// ```
500impl FromStr for DataType {
501 type Err = ArrowError;
502
503 fn from_str(s: &str) -> Result<Self, Self::Err> {
504 crate::datatype_parse::parse_data_type(s)
505 }
506}
507
508impl TryFrom<&str> for DataType {
509 type Error = ArrowError;
510
511 fn try_from(value: &str) -> Result<Self, Self::Error> {
512 value.parse()
513 }
514}
515
516impl DataType {
517 /// Returns true if the type is primitive: (numeric, temporal).
518 #[inline]
519 pub fn is_primitive(&self) -> bool {
520 self.is_numeric() || self.is_temporal()
521 }
522
523 /// Returns true if this type is numeric: (UInt*, Int*, Float*, Decimal*).
524 #[inline]
525 pub fn is_numeric(&self) -> bool {
526 use DataType::*;
527 matches!(
528 self,
529 UInt8
530 | UInt16
531 | UInt32
532 | UInt64
533 | Int8
534 | Int16
535 | Int32
536 | Int64
537 | Float16
538 | Float32
539 | Float64
540 | Decimal32(_, _)
541 | Decimal64(_, _)
542 | Decimal128(_, _)
543 | Decimal256(_, _)
544 )
545 }
546
547 /// Returns true if this type is temporal: (Date*, Time*, Duration, or Interval).
548 #[inline]
549 pub fn is_temporal(&self) -> bool {
550 use DataType::*;
551 matches!(
552 self,
553 Date32 | Date64 | Timestamp(_, _) | Time32(_) | Time64(_) | Duration(_) | Interval(_)
554 )
555 }
556
557 /// Returns true if this type is floating: (Float*).
558 #[inline]
559 pub fn is_floating(&self) -> bool {
560 use DataType::*;
561 matches!(self, Float16 | Float32 | Float64)
562 }
563
564 /// Returns true if this type is integer: (Int*, UInt*).
565 #[inline]
566 pub fn is_integer(&self) -> bool {
567 self.is_signed_integer() || self.is_unsigned_integer()
568 }
569
570 /// Returns true if this type is signed integer: (Int*).
571 #[inline]
572 pub fn is_signed_integer(&self) -> bool {
573 use DataType::*;
574 matches!(self, Int8 | Int16 | Int32 | Int64)
575 }
576
577 /// Returns true if this type is unsigned integer: (UInt*).
578 #[inline]
579 pub fn is_unsigned_integer(&self) -> bool {
580 use DataType::*;
581 matches!(self, UInt8 | UInt16 | UInt32 | UInt64)
582 }
583
584 /// Returns true if this type is decimal: (Decimal*).
585 #[inline]
586 pub fn is_decimal(&self) -> bool {
587 use DataType::*;
588 matches!(
589 self,
590 Decimal32(..) | Decimal64(..) | Decimal128(..) | Decimal256(..)
591 )
592 }
593
594 /// Returns true if this type is valid as a dictionary key
595 #[inline]
596 pub fn is_dictionary_key_type(&self) -> bool {
597 self.is_integer()
598 }
599
600 /// Returns true if this type is valid for run-ends array in RunArray
601 #[inline]
602 pub fn is_run_ends_type(&self) -> bool {
603 use DataType::*;
604 matches!(self, Int16 | Int32 | Int64)
605 }
606
607 /// Returns true if this type is nested (List, FixedSizeList, LargeList, ListView. LargeListView, Struct, Union,
608 /// or Map), or a dictionary of a nested type
609 #[inline]
610 pub fn is_nested(&self) -> bool {
611 use DataType::*;
612 match self {
613 Dictionary(_, v) => DataType::is_nested(v.as_ref()),
614 RunEndEncoded(_, v) => DataType::is_nested(v.data_type()),
615 List(_)
616 | FixedSizeList(_, _)
617 | LargeList(_)
618 | ListView(_)
619 | LargeListView(_)
620 | Struct(_)
621 | Union(_, _)
622 | Map(_, _) => true,
623 _ => false,
624 }
625 }
626
627 /// Returns true if this type is DataType::Null.
628 #[inline]
629 pub fn is_null(&self) -> bool {
630 use DataType::*;
631 matches!(self, Null)
632 }
633
634 /// Returns true if this type is a String type
635 #[inline]
636 pub fn is_string(&self) -> bool {
637 use DataType::*;
638 matches!(self, Utf8 | LargeUtf8 | Utf8View)
639 }
640
641 /// Returns true if this type is a List type.
642 ///
643 /// List types include List, LargeList, FixedSizeList, ListView, and LargeListView.
644 #[inline]
645 pub fn is_list(&self) -> bool {
646 use DataType::*;
647 matches!(
648 self,
649 List(_) | LargeList(_) | FixedSizeList(_, _) | ListView(_) | LargeListView(_)
650 )
651 }
652
653 /// Returns true if this type is a Binary type.
654 ///
655 /// Binary types include Binary, LargeBinary, FixedSizeBinary and BinaryView.
656 #[inline]
657 pub fn is_binary(&self) -> bool {
658 use DataType::*;
659 matches!(self, Binary | LargeBinary | FixedSizeBinary(_) | BinaryView)
660 }
661
662 /// Compares the datatype with another, ignoring nested field names
663 /// and metadata.
664 pub fn equals_datatype(&self, other: &DataType) -> bool {
665 match (&self, other) {
666 (DataType::List(a), DataType::List(b))
667 | (DataType::LargeList(a), DataType::LargeList(b))
668 | (DataType::ListView(a), DataType::ListView(b))
669 | (DataType::LargeListView(a), DataType::LargeListView(b)) => {
670 a.is_nullable() == b.is_nullable() && a.data_type().equals_datatype(b.data_type())
671 }
672 (DataType::FixedSizeList(a, a_size), DataType::FixedSizeList(b, b_size)) => {
673 a_size == b_size
674 && a.is_nullable() == b.is_nullable()
675 && a.data_type().equals_datatype(b.data_type())
676 }
677 (DataType::Struct(a), DataType::Struct(b)) => {
678 a.len() == b.len()
679 && a.iter().zip(b).all(|(a, b)| {
680 a.is_nullable() == b.is_nullable()
681 && a.data_type().equals_datatype(b.data_type())
682 })
683 }
684 (DataType::Map(a_field, a_is_sorted), DataType::Map(b_field, b_is_sorted)) => {
685 a_field.is_nullable() == b_field.is_nullable()
686 && a_field.data_type().equals_datatype(b_field.data_type())
687 && a_is_sorted == b_is_sorted
688 }
689 (DataType::Dictionary(a_key, a_value), DataType::Dictionary(b_key, b_value)) => {
690 a_key.equals_datatype(b_key) && a_value.equals_datatype(b_value)
691 }
692 (
693 DataType::RunEndEncoded(a_run_ends, a_values),
694 DataType::RunEndEncoded(b_run_ends, b_values),
695 ) => {
696 a_run_ends.is_nullable() == b_run_ends.is_nullable()
697 && a_run_ends
698 .data_type()
699 .equals_datatype(b_run_ends.data_type())
700 && a_values.is_nullable() == b_values.is_nullable()
701 && a_values.data_type().equals_datatype(b_values.data_type())
702 }
703 (
704 DataType::Union(a_union_fields, a_union_mode),
705 DataType::Union(b_union_fields, b_union_mode),
706 ) => {
707 a_union_mode == b_union_mode
708 && a_union_fields.len() == b_union_fields.len()
709 && a_union_fields.iter().all(|a| {
710 b_union_fields.iter().any(|b| {
711 a.0 == b.0
712 && a.1.is_nullable() == b.1.is_nullable()
713 && a.1.data_type().equals_datatype(b.1.data_type())
714 })
715 })
716 }
717 _ => self == other,
718 }
719 }
720
721 /// Returns the byte width of this type if it is a primitive type
722 ///
723 /// Returns `None` if not a primitive type
724 #[inline]
725 pub fn primitive_width(&self) -> Option<usize> {
726 match self {
727 DataType::Null => None,
728 DataType::Boolean => None,
729 DataType::Int8 | DataType::UInt8 => Some(1),
730 DataType::Int16 | DataType::UInt16 | DataType::Float16 => Some(2),
731 DataType::Int32 | DataType::UInt32 | DataType::Float32 => Some(4),
732 DataType::Int64 | DataType::UInt64 | DataType::Float64 => Some(8),
733 DataType::Timestamp(_, _) => Some(8),
734 DataType::Date32 | DataType::Time32(_) => Some(4),
735 DataType::Date64 | DataType::Time64(_) => Some(8),
736 DataType::Duration(_) => Some(8),
737 DataType::Interval(IntervalUnit::YearMonth) => Some(4),
738 DataType::Interval(IntervalUnit::DayTime) => Some(8),
739 DataType::Interval(IntervalUnit::MonthDayNano) => Some(16),
740 DataType::Decimal32(_, _) => Some(4),
741 DataType::Decimal64(_, _) => Some(8),
742 DataType::Decimal128(_, _) => Some(16),
743 DataType::Decimal256(_, _) => Some(32),
744 DataType::Utf8 | DataType::LargeUtf8 | DataType::Utf8View => None,
745 DataType::Binary | DataType::LargeBinary | DataType::BinaryView => None,
746 DataType::FixedSizeBinary(_) => None,
747 DataType::List(_)
748 | DataType::ListView(_)
749 | DataType::LargeList(_)
750 | DataType::LargeListView(_)
751 | DataType::Map(_, _) => None,
752 DataType::FixedSizeList(_, _) => None,
753 DataType::Struct(_) => None,
754 DataType::Union(_, _) => None,
755 DataType::Dictionary(_, _) => None,
756 DataType::RunEndEncoded(_, _) => None,
757 }
758 }
759
760 /// Return size of this instance in bytes.
761 ///
762 /// Includes the size of `Self`.
763 pub fn size(&self) -> usize {
764 std::mem::size_of_val(self)
765 + match self {
766 DataType::Null
767 | DataType::Boolean
768 | DataType::Int8
769 | DataType::Int16
770 | DataType::Int32
771 | DataType::Int64
772 | DataType::UInt8
773 | DataType::UInt16
774 | DataType::UInt32
775 | DataType::UInt64
776 | DataType::Float16
777 | DataType::Float32
778 | DataType::Float64
779 | DataType::Date32
780 | DataType::Date64
781 | DataType::Time32(_)
782 | DataType::Time64(_)
783 | DataType::Duration(_)
784 | DataType::Interval(_)
785 | DataType::Binary
786 | DataType::FixedSizeBinary(_)
787 | DataType::LargeBinary
788 | DataType::BinaryView
789 | DataType::Utf8
790 | DataType::LargeUtf8
791 | DataType::Utf8View
792 | DataType::Decimal32(_, _)
793 | DataType::Decimal64(_, _)
794 | DataType::Decimal128(_, _)
795 | DataType::Decimal256(_, _) => 0,
796 DataType::Timestamp(_, s) => s.as_ref().map(|s| s.len()).unwrap_or_default(),
797 DataType::List(field)
798 | DataType::ListView(field)
799 | DataType::FixedSizeList(field, _)
800 | DataType::LargeList(field)
801 | DataType::LargeListView(field)
802 | DataType::Map(field, _) => field.size(),
803 DataType::Struct(fields) => fields.size(),
804 DataType::Union(fields, _) => fields.size(),
805 DataType::Dictionary(dt1, dt2) => dt1.size() + dt2.size(),
806 DataType::RunEndEncoded(run_ends, values) => {
807 run_ends.size() - std::mem::size_of_val(run_ends) + values.size()
808 - std::mem::size_of_val(values)
809 }
810 }
811 }
812
813 /// Check to see if `self` is a superset of `other`
814 ///
815 /// If DataType is a nested type, then it will check to see if the nested type is a superset of the other nested type
816 /// else it will check to see if the DataType is equal to the other DataType
817 pub fn contains(&self, other: &DataType) -> bool {
818 match (self, other) {
819 (DataType::List(f1), DataType::List(f2))
820 | (DataType::LargeList(f1), DataType::LargeList(f2))
821 | (DataType::ListView(f1), DataType::ListView(f2))
822 | (DataType::LargeListView(f1), DataType::LargeListView(f2)) => f1.contains(f2),
823 (DataType::FixedSizeList(f1, s1), DataType::FixedSizeList(f2, s2)) => {
824 s1 == s2 && f1.contains(f2)
825 }
826 (DataType::Map(f1, s1), DataType::Map(f2, s2)) => s1 == s2 && f1.contains(f2),
827 (DataType::Struct(f1), DataType::Struct(f2)) => f1.contains(f2),
828 (DataType::Union(f1, s1), DataType::Union(f2, s2)) => {
829 s1 == s2
830 && f1
831 .iter()
832 .all(|f1| f2.iter().any(|f2| f1.0 == f2.0 && f1.1.contains(f2.1)))
833 }
834 (DataType::Dictionary(k1, v1), DataType::Dictionary(k2, v2)) => {
835 k1.contains(k2) && v1.contains(v2)
836 }
837 _ => self == other,
838 }
839 }
840
841 /// Create a [`DataType::List`] with elements of the specified type
842 /// and nullability, and conventionally named inner [`Field`] (`"item"`).
843 ///
844 /// To specify field level metadata, construct the inner [`Field`]
845 /// directly via [`Field::new`] or [`Field::new_list_field`].
846 pub fn new_list(data_type: DataType, nullable: bool) -> Self {
847 DataType::List(Arc::new(Field::new_list_field(data_type, nullable)))
848 }
849
850 /// Create a [`DataType::LargeList`] with elements of the specified type
851 /// and nullability, and conventionally named inner [`Field`] (`"item"`).
852 ///
853 /// To specify field level metadata, construct the inner [`Field`]
854 /// directly via [`Field::new`] or [`Field::new_list_field`].
855 pub fn new_large_list(data_type: DataType, nullable: bool) -> Self {
856 DataType::LargeList(Arc::new(Field::new_list_field(data_type, nullable)))
857 }
858
859 /// Create a [`DataType::FixedSizeList`] with elements of the specified type, size
860 /// and nullability, and conventionally named inner [`Field`] (`"item"`).
861 ///
862 /// To specify field level metadata, construct the inner [`Field`]
863 /// directly via [`Field::new`] or [`Field::new_list_field`].
864 pub fn new_fixed_size_list(data_type: DataType, size: i32, nullable: bool) -> Self {
865 DataType::FixedSizeList(Arc::new(Field::new_list_field(data_type, nullable)), size)
866 }
867}
868
869/// The maximum precision for [DataType::Decimal32] values
870pub const DECIMAL32_MAX_PRECISION: u8 = 9;
871
872/// The maximum scale for [DataType::Decimal32] values
873pub const DECIMAL32_MAX_SCALE: i8 = 9;
874
875/// The maximum precision for [DataType::Decimal64] values
876pub const DECIMAL64_MAX_PRECISION: u8 = 18;
877
878/// The maximum scale for [DataType::Decimal64] values
879pub const DECIMAL64_MAX_SCALE: i8 = 18;
880
881/// The maximum precision for [DataType::Decimal128] values
882pub const DECIMAL128_MAX_PRECISION: u8 = 38;
883
884/// The maximum scale for [DataType::Decimal128] values
885pub const DECIMAL128_MAX_SCALE: i8 = 38;
886
887/// The maximum precision for [DataType::Decimal256] values
888pub const DECIMAL256_MAX_PRECISION: u8 = 76;
889
890/// The maximum scale for [DataType::Decimal256] values
891pub const DECIMAL256_MAX_SCALE: i8 = 76;
892
893/// The default scale for [DataType::Decimal32] values
894pub const DECIMAL32_DEFAULT_SCALE: i8 = 2;
895
896/// The default scale for [DataType::Decimal64] values
897pub const DECIMAL64_DEFAULT_SCALE: i8 = 6;
898
899/// The default scale for [DataType::Decimal128] and [DataType::Decimal256]
900/// values
901pub const DECIMAL_DEFAULT_SCALE: i8 = 10;
902
903#[cfg(test)]
904mod tests {
905 use super::*;
906
907 #[test]
908 #[cfg(feature = "serde")]
909 fn serde_struct_type() {
910 use std::collections::HashMap;
911
912 let kv_array = [("k".to_string(), "v".to_string())];
913 let field_metadata: HashMap<String, String> = kv_array.iter().cloned().collect();
914
915 // Non-empty map: should be converted as JSON obj { ... }
916 let first_name =
917 Field::new("first_name", DataType::Utf8, false).with_metadata(field_metadata);
918
919 // Empty map: should be omitted.
920 let last_name =
921 Field::new("last_name", DataType::Utf8, false).with_metadata(HashMap::default());
922
923 let person = DataType::Struct(Fields::from(vec![
924 first_name,
925 last_name,
926 Field::new(
927 "address",
928 DataType::Struct(Fields::from(vec![
929 Field::new("street", DataType::Utf8, false),
930 Field::new("zip", DataType::UInt16, false),
931 ])),
932 false,
933 ),
934 ]));
935
936 let serialized = serde_json::to_string(&person).unwrap();
937
938 // NOTE that this is testing the default (derived) serialization format, not the
939 // JSON format specified in metadata.md
940
941 assert_eq!(
942 "{\"Struct\":[\
943 {\"name\":\"first_name\",\"data_type\":\"Utf8\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{\"k\":\"v\"}},\
944 {\"name\":\"last_name\",\"data_type\":\"Utf8\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}},\
945 {\"name\":\"address\",\"data_type\":{\"Struct\":\
946 [{\"name\":\"street\",\"data_type\":\"Utf8\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}},\
947 {\"name\":\"zip\",\"data_type\":\"UInt16\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}}\
948 ]},\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}}]}",
949 serialized
950 );
951
952 let deserialized = serde_json::from_str(&serialized).unwrap();
953
954 assert_eq!(person, deserialized);
955 }
956
957 #[test]
958 fn test_list_datatype_equality() {
959 // tests that list type equality is checked while ignoring list names
960 let list_a = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true)));
961 let list_b = DataType::List(Arc::new(Field::new("array", DataType::Int32, true)));
962 let list_c = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, false)));
963 let list_d = DataType::List(Arc::new(Field::new_list_field(DataType::UInt32, true)));
964 assert!(list_a.equals_datatype(&list_b));
965 assert!(!list_a.equals_datatype(&list_c));
966 assert!(!list_b.equals_datatype(&list_c));
967 assert!(!list_a.equals_datatype(&list_d));
968
969 let list_e =
970 DataType::FixedSizeList(Arc::new(Field::new_list_field(list_a.clone(), false)), 3);
971 let list_f =
972 DataType::FixedSizeList(Arc::new(Field::new("array", list_b.clone(), false)), 3);
973 let list_g = DataType::FixedSizeList(
974 Arc::new(Field::new_list_field(DataType::FixedSizeBinary(3), true)),
975 3,
976 );
977 assert!(list_e.equals_datatype(&list_f));
978 assert!(!list_e.equals_datatype(&list_g));
979 assert!(!list_f.equals_datatype(&list_g));
980
981 let list_h = DataType::Struct(Fields::from(vec![Field::new("f1", list_e, true)]));
982 let list_i = DataType::Struct(Fields::from(vec![Field::new("f1", list_f.clone(), true)]));
983 let list_j = DataType::Struct(Fields::from(vec![Field::new("f1", list_f.clone(), false)]));
984 let list_k = DataType::Struct(Fields::from(vec![
985 Field::new("f1", list_f.clone(), false),
986 Field::new("f2", list_g.clone(), false),
987 Field::new("f3", DataType::Utf8, true),
988 ]));
989 let list_l = DataType::Struct(Fields::from(vec![
990 Field::new("ff1", list_f.clone(), false),
991 Field::new("ff2", list_g.clone(), false),
992 Field::new("ff3", DataType::LargeUtf8, true),
993 ]));
994 let list_m = DataType::Struct(Fields::from(vec![
995 Field::new("ff1", list_f, false),
996 Field::new("ff2", list_g, false),
997 Field::new("ff3", DataType::Utf8, true),
998 ]));
999 assert!(list_h.equals_datatype(&list_i));
1000 assert!(!list_h.equals_datatype(&list_j));
1001 assert!(!list_k.equals_datatype(&list_l));
1002 assert!(list_k.equals_datatype(&list_m));
1003
1004 let list_n = DataType::Map(Arc::new(Field::new("f1", list_a.clone(), true)), true);
1005 let list_o = DataType::Map(Arc::new(Field::new("f2", list_b.clone(), true)), true);
1006 let list_p = DataType::Map(Arc::new(Field::new("f2", list_b.clone(), true)), false);
1007 let list_q = DataType::Map(Arc::new(Field::new("f2", list_c.clone(), true)), true);
1008 let list_r = DataType::Map(Arc::new(Field::new("f1", list_a.clone(), false)), true);
1009
1010 assert!(list_n.equals_datatype(&list_o));
1011 assert!(!list_n.equals_datatype(&list_p));
1012 assert!(!list_n.equals_datatype(&list_q));
1013 assert!(!list_n.equals_datatype(&list_r));
1014
1015 let list_s = DataType::Dictionary(Box::new(DataType::UInt8), Box::new(list_a));
1016 let list_t = DataType::Dictionary(Box::new(DataType::UInt8), Box::new(list_b.clone()));
1017 let list_u = DataType::Dictionary(Box::new(DataType::Int8), Box::new(list_b));
1018 let list_v = DataType::Dictionary(Box::new(DataType::UInt8), Box::new(list_c));
1019
1020 assert!(list_s.equals_datatype(&list_t));
1021 assert!(!list_s.equals_datatype(&list_u));
1022 assert!(!list_s.equals_datatype(&list_v));
1023
1024 let union_a = DataType::Union(
1025 UnionFields::try_new(
1026 vec![1, 2],
1027 vec![
1028 Field::new("f1", DataType::Utf8, false),
1029 Field::new("f2", DataType::UInt8, false),
1030 ],
1031 )
1032 .unwrap(),
1033 UnionMode::Sparse,
1034 );
1035 let union_b = DataType::Union(
1036 UnionFields::try_new(
1037 vec![1, 2],
1038 vec![
1039 Field::new("ff1", DataType::Utf8, false),
1040 Field::new("ff2", DataType::UInt8, false),
1041 ],
1042 )
1043 .unwrap(),
1044 UnionMode::Sparse,
1045 );
1046 let union_c = DataType::Union(
1047 UnionFields::try_new(
1048 vec![2, 1],
1049 vec![
1050 Field::new("fff2", DataType::UInt8, false),
1051 Field::new("fff1", DataType::Utf8, false),
1052 ],
1053 )
1054 .unwrap(),
1055 UnionMode::Sparse,
1056 );
1057 let union_d = DataType::Union(
1058 UnionFields::try_new(
1059 vec![2, 1],
1060 vec![
1061 Field::new("fff1", DataType::Int8, false),
1062 Field::new("fff2", DataType::UInt8, false),
1063 ],
1064 )
1065 .unwrap(),
1066 UnionMode::Sparse,
1067 );
1068 let union_e = DataType::Union(
1069 UnionFields::try_new(
1070 vec![1, 2],
1071 vec![
1072 Field::new("f1", DataType::Utf8, true),
1073 Field::new("f2", DataType::UInt8, false),
1074 ],
1075 )
1076 .unwrap(),
1077 UnionMode::Sparse,
1078 );
1079
1080 assert!(union_a.equals_datatype(&union_b));
1081 assert!(union_a.equals_datatype(&union_c));
1082 assert!(!union_a.equals_datatype(&union_d));
1083 assert!(!union_a.equals_datatype(&union_e));
1084
1085 let list_w = DataType::RunEndEncoded(
1086 Arc::new(Field::new("f1", DataType::Int64, true)),
1087 Arc::new(Field::new("f2", DataType::Utf8, true)),
1088 );
1089 let list_x = DataType::RunEndEncoded(
1090 Arc::new(Field::new("ff1", DataType::Int64, true)),
1091 Arc::new(Field::new("ff2", DataType::Utf8, true)),
1092 );
1093 let list_y = DataType::RunEndEncoded(
1094 Arc::new(Field::new("ff1", DataType::UInt16, true)),
1095 Arc::new(Field::new("ff2", DataType::Utf8, true)),
1096 );
1097 let list_z = DataType::RunEndEncoded(
1098 Arc::new(Field::new("f1", DataType::Int64, false)),
1099 Arc::new(Field::new("f2", DataType::Utf8, true)),
1100 );
1101
1102 assert!(list_w.equals_datatype(&list_x));
1103 assert!(!list_w.equals_datatype(&list_y));
1104 assert!(!list_w.equals_datatype(&list_z));
1105 }
1106
1107 #[test]
1108 fn create_struct_type() {
1109 let _person = DataType::Struct(Fields::from(vec![
1110 Field::new("first_name", DataType::Utf8, false),
1111 Field::new("last_name", DataType::Utf8, false),
1112 Field::new(
1113 "address",
1114 DataType::Struct(Fields::from(vec![
1115 Field::new("street", DataType::Utf8, false),
1116 Field::new("zip", DataType::UInt16, false),
1117 ])),
1118 false,
1119 ),
1120 ]));
1121 }
1122
1123 #[test]
1124 fn test_nested() {
1125 let list = DataType::List(Arc::new(Field::new("foo", DataType::Utf8, true)));
1126 let list_view = DataType::ListView(Arc::new(Field::new("foo", DataType::Utf8, true)));
1127 let large_list_view =
1128 DataType::LargeListView(Arc::new(Field::new("foo", DataType::Utf8, true)));
1129
1130 assert!(!DataType::is_nested(&DataType::Boolean));
1131 assert!(!DataType::is_nested(&DataType::Int32));
1132 assert!(!DataType::is_nested(&DataType::Utf8));
1133 assert!(DataType::is_nested(&list));
1134 assert!(DataType::is_nested(&list_view));
1135 assert!(DataType::is_nested(&large_list_view));
1136
1137 assert!(!DataType::is_nested(&DataType::Dictionary(
1138 Box::new(DataType::Int32),
1139 Box::new(DataType::Boolean)
1140 )));
1141 assert!(!DataType::is_nested(&DataType::Dictionary(
1142 Box::new(DataType::Int32),
1143 Box::new(DataType::Int64)
1144 )));
1145 assert!(!DataType::is_nested(&DataType::Dictionary(
1146 Box::new(DataType::Int32),
1147 Box::new(DataType::LargeUtf8)
1148 )));
1149 assert!(DataType::is_nested(&DataType::Dictionary(
1150 Box::new(DataType::Int32),
1151 Box::new(list)
1152 )));
1153 }
1154
1155 #[test]
1156 fn test_integer() {
1157 // is_integer
1158 assert!(DataType::is_integer(&DataType::Int32));
1159 assert!(DataType::is_integer(&DataType::UInt64));
1160 assert!(!DataType::is_integer(&DataType::Float16));
1161
1162 // is_signed_integer
1163 assert!(DataType::is_signed_integer(&DataType::Int32));
1164 assert!(!DataType::is_signed_integer(&DataType::UInt64));
1165 assert!(!DataType::is_signed_integer(&DataType::Float16));
1166
1167 // is_unsigned_integer
1168 assert!(!DataType::is_unsigned_integer(&DataType::Int32));
1169 assert!(DataType::is_unsigned_integer(&DataType::UInt64));
1170 assert!(!DataType::is_unsigned_integer(&DataType::Float16));
1171
1172 // is_dictionary_key_type
1173 assert!(DataType::is_dictionary_key_type(&DataType::Int32));
1174 assert!(DataType::is_dictionary_key_type(&DataType::UInt64));
1175 assert!(!DataType::is_dictionary_key_type(&DataType::Float16));
1176 }
1177
1178 #[test]
1179 fn test_string() {
1180 assert!(DataType::is_string(&DataType::Utf8));
1181 assert!(DataType::is_string(&DataType::LargeUtf8));
1182 assert!(DataType::is_string(&DataType::Utf8View));
1183 assert!(!DataType::is_string(&DataType::Int32));
1184 }
1185
1186 #[test]
1187 fn test_floating() {
1188 assert!(DataType::is_floating(&DataType::Float16));
1189 assert!(!DataType::is_floating(&DataType::Int32));
1190 }
1191
1192 #[test]
1193 fn test_decimal() {
1194 assert!(DataType::is_decimal(&DataType::Decimal32(4, 2)));
1195 assert!(DataType::is_decimal(&DataType::Decimal64(4, 2)));
1196 assert!(DataType::is_decimal(&DataType::Decimal128(4, 2)));
1197 assert!(DataType::is_decimal(&DataType::Decimal256(4, 2)));
1198 assert!(!DataType::is_decimal(&DataType::Float16));
1199 }
1200
1201 #[test]
1202 fn test_datatype_is_null() {
1203 assert!(DataType::is_null(&DataType::Null));
1204 assert!(!DataType::is_null(&DataType::Int32));
1205 }
1206
1207 #[test]
1208 fn test_is_list() {
1209 assert!(DataType::is_list(&DataType::new_list(
1210 DataType::Int16,
1211 true
1212 )));
1213 assert!(DataType::is_list(&DataType::new_large_list(
1214 DataType::Int16,
1215 true
1216 )));
1217 assert!(DataType::is_list(&DataType::new_fixed_size_list(
1218 DataType::Int16,
1219 5,
1220 true
1221 )));
1222 assert!(DataType::is_list(&DataType::ListView(Arc::new(
1223 Field::new("f", DataType::Int16, true)
1224 ))));
1225 assert!(DataType::is_list(&DataType::LargeListView(Arc::new(
1226 Field::new("f", DataType::Int16, true)
1227 ))));
1228 assert!(!DataType::is_list(&DataType::Binary));
1229 }
1230
1231 #[test]
1232 fn test_is_binary() {
1233 assert!(DataType::is_binary(&DataType::Binary));
1234 assert!(DataType::is_binary(&DataType::LargeBinary));
1235 assert!(DataType::is_binary(&DataType::BinaryView));
1236 assert!(!DataType::is_list(&DataType::Utf8View));
1237 }
1238
1239 #[test]
1240 fn size_should_not_regress() {
1241 assert_eq!(std::mem::size_of::<DataType>(), 24);
1242 }
1243
1244 #[test]
1245 #[should_panic(expected = "duplicate type id: 1")]
1246 fn test_union_with_duplicated_type_id() {
1247 let type_ids = vec![1, 1];
1248 let _union = DataType::Union(
1249 UnionFields::try_new(
1250 type_ids,
1251 vec![
1252 Field::new("f1", DataType::Int32, false),
1253 Field::new("f2", DataType::Utf8, false),
1254 ],
1255 )
1256 .unwrap(),
1257 UnionMode::Dense,
1258 );
1259 }
1260
1261 #[test]
1262 fn test_try_from_str() {
1263 let data_type: DataType = "Int32".try_into().unwrap();
1264 assert_eq!(data_type, DataType::Int32);
1265 }
1266
1267 #[test]
1268 fn test_from_str() {
1269 let data_type: DataType = "UInt64".parse().unwrap();
1270 assert_eq!(data_type, DataType::UInt64);
1271 }
1272
1273 #[test]
1274 #[cfg_attr(miri, ignore)] // Can't handle the inlined strings of the assert_debug_snapshot macro
1275 fn test_debug_format_field() {
1276 // Make sure the `Debug` formatting of `DataType` is readable and not too long
1277 insta::assert_debug_snapshot!(DataType::new_list(DataType::Int8, false), @r"
1278 List(
1279 Field {
1280 data_type: Int8,
1281 },
1282 )
1283 ");
1284 }
1285}