arrow_schema/
field.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 crate::error::ArrowError;
19use std::cmp::Ordering;
20use std::collections::HashMap;
21use std::hash::{Hash, Hasher};
22use std::sync::Arc;
23
24use crate::datatype::DataType;
25#[cfg(feature = "canonical_extension_types")]
26use crate::extension::CanonicalExtensionType;
27use crate::schema::SchemaBuilder;
28use crate::{
29    Fields, UnionFields, UnionMode,
30    extension::{EXTENSION_TYPE_METADATA_KEY, EXTENSION_TYPE_NAME_KEY, ExtensionType},
31};
32
33/// A reference counted [`Field`]
34pub type FieldRef = Arc<Field>;
35
36/// Describes a single column in a [`Schema`](super::Schema).
37///
38/// A [`Schema`](super::Schema) is an ordered collection of
39/// [`Field`] objects. Fields contain:
40/// * `name`: the name of the field
41/// * `data_type`: the type of the field
42/// * `nullable`: if the field is nullable
43/// * `metadata`: a map of key-value pairs containing additional custom metadata
44///
45/// Arrow Extension types, are encoded in `Field`s metadata. See
46/// [`Self::try_extension_type`] to retrieve the [`ExtensionType`], if any.
47#[derive(Clone)]
48#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
49pub struct Field {
50    name: String,
51    data_type: DataType,
52    nullable: bool,
53    #[deprecated(
54        since = "54.0.0",
55        note = "The ability to preserve dictionary IDs will be removed. With it, all fields related to it."
56    )]
57    dict_id: i64,
58    dict_is_ordered: bool,
59    /// A map of key-value pairs containing additional custom meta data.
60    metadata: HashMap<String, String>,
61}
62
63impl std::fmt::Debug for Field {
64    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
65        #![expect(deprecated)] // Must still print dict_id, if set
66        let Self {
67            name,
68            data_type,
69            nullable,
70            dict_id,
71            dict_is_ordered,
72            metadata,
73        } = self;
74
75        let mut s = f.debug_struct("Field");
76
77        if name != "item" {
78            // Keep it short when debug-formatting `DataType::List`
79            s.field("name", name);
80        }
81
82        s.field("data_type", data_type);
83
84        if *nullable {
85            s.field("nullable", nullable);
86        }
87
88        if *dict_id != 0 {
89            s.field("dict_id", dict_id);
90        }
91
92        if *dict_is_ordered {
93            s.field("dict_is_ordered", dict_is_ordered);
94        }
95
96        if !metadata.is_empty() {
97            s.field("metadata", metadata);
98        }
99        s.finish()
100    }
101}
102
103// Auto-derive `PartialEq` traits will pull `dict_id` and `dict_is_ordered`
104// into comparison. However, these properties are only used in IPC context
105// for matching dictionary encoded data. They are not necessary to be same
106// to consider schema equality. For example, in C++ `Field` implementation,
107// it doesn't contain these dictionary properties too.
108impl PartialEq for Field {
109    fn eq(&self, other: &Self) -> bool {
110        self.name == other.name
111            && self.data_type == other.data_type
112            && self.nullable == other.nullable
113            && self.metadata == other.metadata
114    }
115}
116
117impl Eq for Field {}
118
119impl PartialOrd for Field {
120    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
121        Some(self.cmp(other))
122    }
123}
124
125impl Ord for Field {
126    fn cmp(&self, other: &Self) -> Ordering {
127        self.name
128            .cmp(other.name())
129            .then_with(|| self.data_type.cmp(other.data_type()))
130            .then_with(|| self.nullable.cmp(&other.nullable))
131            .then_with(|| {
132                // ensure deterministic key order
133                let mut keys: Vec<&String> =
134                    self.metadata.keys().chain(other.metadata.keys()).collect();
135                keys.sort();
136                for k in keys {
137                    match (self.metadata.get(k), other.metadata.get(k)) {
138                        (None, None) => {}
139                        (Some(_), None) => {
140                            return Ordering::Less;
141                        }
142                        (None, Some(_)) => {
143                            return Ordering::Greater;
144                        }
145                        (Some(v1), Some(v2)) => match v1.cmp(v2) {
146                            Ordering::Equal => {}
147                            other => {
148                                return other;
149                            }
150                        },
151                    }
152                }
153
154                Ordering::Equal
155            })
156    }
157}
158
159impl Hash for Field {
160    fn hash<H: Hasher>(&self, state: &mut H) {
161        self.name.hash(state);
162        self.data_type.hash(state);
163        self.nullable.hash(state);
164
165        // ensure deterministic key order
166        let mut keys: Vec<&String> = self.metadata.keys().collect();
167        keys.sort();
168        for k in keys {
169            k.hash(state);
170            self.metadata.get(k).expect("key valid").hash(state);
171        }
172    }
173}
174
175impl AsRef<Field> for Field {
176    fn as_ref(&self) -> &Field {
177        self
178    }
179}
180
181impl Field {
182    /// Default list member field name
183    pub const LIST_FIELD_DEFAULT_NAME: &'static str = "item";
184
185    /// Creates a new field with the given name, data type, and nullability
186    ///
187    /// # Example
188    /// ```
189    /// # use arrow_schema::{Field, DataType};
190    /// Field::new("field_name", DataType::Int32, true);
191    /// ```
192    pub fn new(name: impl Into<String>, data_type: DataType, nullable: bool) -> Self {
193        #[allow(deprecated)]
194        Field {
195            name: name.into(),
196            data_type,
197            nullable,
198            dict_id: 0,
199            dict_is_ordered: false,
200            metadata: HashMap::default(),
201        }
202    }
203
204    /// Creates a new `Field` suitable for [`DataType::List`] and
205    /// [`DataType::LargeList`]
206    ///
207    /// While not required, this method follows the convention of naming the
208    /// `Field` `"item"`.
209    ///
210    /// # Example
211    /// ```
212    /// # use arrow_schema::{Field, DataType};
213    /// assert_eq!(
214    ///   Field::new("item", DataType::Int32, true),
215    ///   Field::new_list_field(DataType::Int32, true)
216    /// );
217    /// ```
218    pub fn new_list_field(data_type: DataType, nullable: bool) -> Self {
219        Self::new(Self::LIST_FIELD_DEFAULT_NAME, data_type, nullable)
220    }
221
222    /// Creates a new field that has additional dictionary information
223    #[deprecated(
224        since = "54.0.0",
225        note = "The ability to preserve dictionary IDs will be removed. With the dict_id field disappearing this function signature will change by removing the dict_id parameter."
226    )]
227    pub fn new_dict(
228        name: impl Into<String>,
229        data_type: DataType,
230        nullable: bool,
231        dict_id: i64,
232        dict_is_ordered: bool,
233    ) -> Self {
234        #[allow(deprecated)]
235        Field {
236            name: name.into(),
237            data_type,
238            nullable,
239            dict_id,
240            dict_is_ordered,
241            metadata: HashMap::default(),
242        }
243    }
244
245    /// Create a new [`Field`] with [`DataType::Dictionary`]
246    ///
247    /// Use [`Self::new_dict`] for more advanced dictionary options
248    ///
249    /// # Panics
250    ///
251    /// Panics if [`!key.is_dictionary_key_type`][DataType::is_dictionary_key_type]
252    pub fn new_dictionary(
253        name: impl Into<String>,
254        key: DataType,
255        value: DataType,
256        nullable: bool,
257    ) -> Self {
258        assert!(
259            key.is_dictionary_key_type(),
260            "{key} is not a valid dictionary key"
261        );
262        let data_type = DataType::Dictionary(Box::new(key), Box::new(value));
263        Self::new(name, data_type, nullable)
264    }
265
266    /// Create a new [`Field`] with [`DataType::Struct`]
267    ///
268    /// - `name`: the name of the [`DataType::Struct`] field
269    /// - `fields`: the description of each struct element
270    /// - `nullable`: if the [`DataType::Struct`] array is nullable
271    pub fn new_struct(name: impl Into<String>, fields: impl Into<Fields>, nullable: bool) -> Self {
272        Self::new(name, DataType::Struct(fields.into()), nullable)
273    }
274
275    /// Create a new [`Field`] with [`DataType::List`]
276    ///
277    /// - `name`: the name of the [`DataType::List`] field
278    /// - `value`: the description of each list element
279    /// - `nullable`: if the [`DataType::List`] array is nullable
280    pub fn new_list(name: impl Into<String>, value: impl Into<FieldRef>, nullable: bool) -> Self {
281        Self::new(name, DataType::List(value.into()), nullable)
282    }
283
284    /// Create a new [`Field`] with [`DataType::LargeList`]
285    ///
286    /// - `name`: the name of the [`DataType::LargeList`] field
287    /// - `value`: the description of each list element
288    /// - `nullable`: if the [`DataType::LargeList`] array is nullable
289    pub fn new_large_list(
290        name: impl Into<String>,
291        value: impl Into<FieldRef>,
292        nullable: bool,
293    ) -> Self {
294        Self::new(name, DataType::LargeList(value.into()), nullable)
295    }
296
297    /// Create a new [`Field`] with [`DataType::FixedSizeList`]
298    ///
299    /// - `name`: the name of the [`DataType::FixedSizeList`] field
300    /// - `value`: the description of each list element
301    /// - `size`: the size of the fixed size list
302    /// - `nullable`: if the [`DataType::FixedSizeList`] array is nullable
303    pub fn new_fixed_size_list(
304        name: impl Into<String>,
305        value: impl Into<FieldRef>,
306        size: i32,
307        nullable: bool,
308    ) -> Self {
309        Self::new(name, DataType::FixedSizeList(value.into(), size), nullable)
310    }
311
312    /// Create a new [`Field`] with [`DataType::Map`]
313    ///
314    /// - `name`: the name of the [`DataType::Map`] field
315    /// - `entries`: the name of the inner [`DataType::Struct`] field
316    /// - `keys`: the map keys
317    /// - `values`: the map values
318    /// - `sorted`: if the [`DataType::Map`] array is sorted
319    /// - `nullable`: if the [`DataType::Map`] array is nullable
320    pub fn new_map(
321        name: impl Into<String>,
322        entries: impl Into<String>,
323        keys: impl Into<FieldRef>,
324        values: impl Into<FieldRef>,
325        sorted: bool,
326        nullable: bool,
327    ) -> Self {
328        let data_type = DataType::Map(
329            Arc::new(Field::new(
330                entries.into(),
331                DataType::Struct(Fields::from([keys.into(), values.into()])),
332                false, // The inner map field is always non-nullable (#1697),
333            )),
334            sorted,
335        );
336        Self::new(name, data_type, nullable)
337    }
338
339    /// Create a new [`Field`] with [`DataType::Union`]
340    ///
341    /// - `name`: the name of the [`DataType::Union`] field
342    /// - `type_ids`: the union type ids
343    /// - `fields`: the union fields
344    /// - `mode`: the union mode
345    ///
346    /// # Panics
347    ///
348    /// Panics if:
349    /// - any type ID is negative
350    /// - type IDs contain duplicates
351    /// - the number of type IDs does not equal the number of fields
352    pub fn new_union<S, F, T>(name: S, type_ids: T, fields: F, mode: UnionMode) -> Self
353    where
354        S: Into<String>,
355        F: IntoIterator,
356        F::Item: Into<FieldRef>,
357        T: IntoIterator<Item = i8>,
358    {
359        Self::new(
360            name,
361            DataType::Union(
362                UnionFields::try_new(type_ids, fields).expect("Invalid UnionField"),
363                mode,
364            ),
365            false, // Unions cannot be nullable
366        )
367    }
368
369    /// Sets the `Field`'s optional custom metadata.
370    #[inline]
371    pub fn set_metadata(&mut self, metadata: HashMap<String, String>) {
372        self.metadata = metadata;
373    }
374
375    /// Sets the metadata of this `Field` to be `metadata` and returns self
376    pub fn with_metadata(mut self, metadata: HashMap<String, String>) -> Self {
377        self.set_metadata(metadata);
378        self
379    }
380
381    /// Returns the immutable reference to the `Field`'s optional custom metadata.
382    #[inline]
383    pub const fn metadata(&self) -> &HashMap<String, String> {
384        &self.metadata
385    }
386
387    /// Returns a mutable reference to the `Field`'s optional custom metadata.
388    #[inline]
389    pub fn metadata_mut(&mut self) -> &mut HashMap<String, String> {
390        &mut self.metadata
391    }
392
393    /// Returns an immutable reference to the `Field`'s name.
394    #[inline]
395    pub const fn name(&self) -> &String {
396        &self.name
397    }
398
399    /// Set the name of this [`Field`]
400    #[inline]
401    pub fn set_name(&mut self, name: impl Into<String>) {
402        self.name = name.into();
403    }
404
405    /// Set the name of the [`Field`] and returns self.
406    ///
407    /// ```
408    /// # use arrow_schema::*;
409    /// let field = Field::new("c1", DataType::Int64, false)
410    ///    .with_name("c2");
411    ///
412    /// assert_eq!(field.name(), "c2");
413    /// ```
414    pub fn with_name(mut self, name: impl Into<String>) -> Self {
415        self.set_name(name);
416        self
417    }
418
419    /// Returns an immutable reference to the [`Field`]'s  [`DataType`].
420    #[inline]
421    pub const fn data_type(&self) -> &DataType {
422        &self.data_type
423    }
424
425    /// Set [`DataType`] of the [`Field`]
426    ///
427    /// ```
428    /// # use arrow_schema::*;
429    /// let mut field = Field::new("c1", DataType::Int64, false);
430    /// field.set_data_type(DataType::Utf8);
431    ///
432    /// assert_eq!(field.data_type(), &DataType::Utf8);
433    /// ```
434    #[inline]
435    pub fn set_data_type(&mut self, data_type: DataType) {
436        self.data_type = data_type;
437    }
438
439    /// Set [`DataType`] of the [`Field`] and returns self.
440    ///
441    /// ```
442    /// # use arrow_schema::*;
443    /// let field = Field::new("c1", DataType::Int64, false)
444    ///    .with_data_type(DataType::Utf8);
445    ///
446    /// assert_eq!(field.data_type(), &DataType::Utf8);
447    /// ```
448    pub fn with_data_type(mut self, data_type: DataType) -> Self {
449        self.set_data_type(data_type);
450        self
451    }
452
453    /// Returns the extension type name of this [`Field`], if set.
454    ///
455    /// This returns the value of [`EXTENSION_TYPE_NAME_KEY`], if set in
456    /// [`Field::metadata`]. If the key is missing, there is no extension type
457    /// name and this returns `None`.
458    ///
459    /// # Example
460    ///
461    /// ```
462    /// # use arrow_schema::{DataType, extension::EXTENSION_TYPE_NAME_KEY, Field};
463    ///
464    /// let field = Field::new("", DataType::Null, false);
465    /// assert_eq!(field.extension_type_name(), None);
466    ///
467    /// let field = Field::new("", DataType::Null, false).with_metadata(
468    ///    [(EXTENSION_TYPE_NAME_KEY.to_owned(), "example".to_owned())]
469    ///        .into_iter()
470    ///        .collect(),
471    /// );
472    /// assert_eq!(field.extension_type_name(), Some("example"));
473    /// ```
474    pub fn extension_type_name(&self) -> Option<&str> {
475        self.metadata()
476            .get(EXTENSION_TYPE_NAME_KEY)
477            .map(String::as_ref)
478    }
479
480    /// Returns the extension type metadata of this [`Field`], if set.
481    ///
482    /// This returns the value of [`EXTENSION_TYPE_METADATA_KEY`], if set in
483    /// [`Field::metadata`]. If the key is missing, there is no extension type
484    /// metadata and this returns `None`.
485    ///
486    /// # Example
487    ///
488    /// ```
489    /// # use arrow_schema::{DataType, extension::EXTENSION_TYPE_METADATA_KEY, Field};
490    ///
491    /// let field = Field::new("", DataType::Null, false);
492    /// assert_eq!(field.extension_type_metadata(), None);
493    ///
494    /// let field = Field::new("", DataType::Null, false).with_metadata(
495    ///    [(EXTENSION_TYPE_METADATA_KEY.to_owned(), "example".to_owned())]
496    ///        .into_iter()
497    ///        .collect(),
498    /// );
499    /// assert_eq!(field.extension_type_metadata(), Some("example"));
500    /// ```
501    pub fn extension_type_metadata(&self) -> Option<&str> {
502        self.metadata()
503            .get(EXTENSION_TYPE_METADATA_KEY)
504            .map(String::as_ref)
505    }
506
507    /// Returns an instance of the given [`ExtensionType`] of this [`Field`],
508    /// if set in the [`Field::metadata`].
509    ///
510    /// Note that using `try_extension_type` with an extension type that does
511    /// not match the name in the metadata will return an `ArrowError` which can
512    /// be slow due to string allocations. If you only want to check if a
513    /// [`Field`] has a specific [`ExtensionType`], see the example below.
514    ///
515    /// # Errors
516    ///
517    /// Returns an error if
518    /// - this field does not have the name of this extension type
519    ///   ([`ExtensionType::NAME`]) in the [`Field::metadata`] (mismatch or
520    ///   missing)
521    /// - the deserialization of the metadata
522    ///   ([`ExtensionType::deserialize_metadata`]) fails
523    /// - the construction of the extension type ([`ExtensionType::try_new`])
524    ///   fail (for example when the [`Field::data_type`] is not supported by
525    ///   the extension type ([`ExtensionType::supports_data_type`]))
526    ///
527    /// # Examples: Check and retrieve an extension type
528    /// You can use this to check if a [`Field`] has a specific
529    /// [`ExtensionType`] and retrieve it:
530    /// ```
531    /// # use arrow_schema::{DataType, Field, ArrowError};
532    /// # use arrow_schema::extension::ExtensionType;
533    /// # struct MyExtensionType;
534    /// # impl ExtensionType for MyExtensionType {
535    /// # const NAME: &'static str = "my_extension";
536    /// # type Metadata = String;
537    /// # fn supports_data_type(&self, data_type: &DataType) -> Result<(), ArrowError> { Ok(()) }
538    /// # fn try_new(data_type: &DataType, metadata: Self::Metadata) -> Result<Self, ArrowError> { Ok(Self) }
539    /// # fn serialize_metadata(&self) -> Option<String> { unimplemented!() }
540    /// # fn deserialize_metadata(s: Option<&str>) -> Result<Self::Metadata, ArrowError> { unimplemented!() }
541    /// # fn metadata(&self) -> &<Self as ExtensionType>::Metadata { todo!() }
542    /// # }
543    /// # fn get_field() -> Field { Field::new("field", DataType::Null, false) }
544    /// let field = get_field();
545    /// if let Ok(extension_type) = field.try_extension_type::<MyExtensionType>() {
546    ///   // do something with extension_type
547    /// }
548    /// ```
549    ///
550    /// # Example: Checking if a field has a specific extension type first
551    ///
552    /// Since `try_extension_type` returns an error, it is more
553    /// efficient to first check if the name matches before calling
554    /// `try_extension_type`:
555    /// ```
556    /// # use arrow_schema::{DataType, Field, ArrowError};
557    /// # use arrow_schema::extension::ExtensionType;
558    /// # struct MyExtensionType;
559    /// # impl ExtensionType for MyExtensionType {
560    /// # const NAME: &'static str = "my_extension";
561    /// # type Metadata = String;
562    /// # fn supports_data_type(&self, data_type: &DataType) -> Result<(), ArrowError> { Ok(()) }
563    /// # fn try_new(data_type: &DataType, metadata: Self::Metadata) -> Result<Self, ArrowError> { Ok(Self) }
564    /// # fn serialize_metadata(&self) -> Option<String> { unimplemented!() }
565    /// # fn deserialize_metadata(s: Option<&str>) -> Result<Self::Metadata, ArrowError> { unimplemented!() }
566    /// # fn metadata(&self) -> &<Self as ExtensionType>::Metadata { todo!() }
567    /// # }
568    /// # fn get_field() -> Field { Field::new("field", DataType::Null, false) }
569    /// let field = get_field();
570    /// // First check if the name matches before calling the potentially expensive `try_extension_type`
571    /// if field.extension_type_name() == Some(MyExtensionType::NAME) {
572    ///   if let Ok(extension_type) = field.try_extension_type::<MyExtensionType>() {
573    ///     // do something with extension_type
574    ///   }
575    /// }
576    /// ```
577    pub fn try_extension_type<E: ExtensionType>(&self) -> Result<E, ArrowError> {
578        E::try_new_from_field_metadata(self.data_type(), self.metadata())
579    }
580
581    /// Returns an instance of the given [`ExtensionType`] of this [`Field`],
582    /// panics if this [`Field`] does not have this extension type.
583    ///
584    /// # Panic
585    ///
586    /// This calls [`Field::try_extension_type`] and panics when it returns an
587    /// error.
588    pub fn extension_type<E: ExtensionType>(&self) -> E {
589        self.try_extension_type::<E>()
590            .unwrap_or_else(|e| panic!("{e}"))
591    }
592
593    /// Updates the metadata of this [`Field`] with the [`ExtensionType::NAME`]
594    /// and [`ExtensionType::metadata`] of the given [`ExtensionType`], if the
595    /// given extension type supports the [`Field::data_type`] of this field
596    /// ([`ExtensionType::supports_data_type`]).
597    ///
598    /// If the given extension type defines no metadata, a previously set
599    /// value of [`EXTENSION_TYPE_METADATA_KEY`] is cleared.
600    ///
601    /// # Error
602    ///
603    /// This functions returns an error if the data type of this field does not
604    /// match any of the supported storage types of the given extension type.
605    pub fn try_with_extension_type<E: ExtensionType>(
606        &mut self,
607        extension_type: E,
608    ) -> Result<(), ArrowError> {
609        // Make sure the data type of this field is supported
610        extension_type.supports_data_type(&self.data_type)?;
611
612        self.metadata
613            .insert(EXTENSION_TYPE_NAME_KEY.to_owned(), E::NAME.to_owned());
614        match extension_type.serialize_metadata() {
615            Some(metadata) => self
616                .metadata
617                .insert(EXTENSION_TYPE_METADATA_KEY.to_owned(), metadata),
618            // If this extension type has no metadata, we make sure to
619            // clear previously set metadata.
620            None => self.metadata.remove(EXTENSION_TYPE_METADATA_KEY),
621        };
622
623        Ok(())
624    }
625
626    /// Updates the metadata of this [`Field`] with the [`ExtensionType::NAME`]
627    /// and [`ExtensionType::metadata`] of the given [`ExtensionType`].
628    ///
629    /// # Panics
630    ///
631    /// This calls [`Field::try_with_extension_type`] and panics when it
632    /// returns an error.
633    pub fn with_extension_type<E: ExtensionType>(mut self, extension_type: E) -> Self {
634        self.try_with_extension_type(extension_type)
635            .unwrap_or_else(|e| panic!("{e}"));
636        self
637    }
638
639    /// Returns the [`CanonicalExtensionType`] of this [`Field`], if set.
640    ///
641    /// # Error
642    ///
643    /// Returns an error if
644    /// - this field does not have a canonical extension type (mismatch or missing)
645    /// - the canonical extension is not supported
646    /// - the construction of the extension type fails
647    #[cfg(feature = "canonical_extension_types")]
648    pub fn try_canonical_extension_type(&self) -> Result<CanonicalExtensionType, ArrowError> {
649        CanonicalExtensionType::try_from(self)
650    }
651
652    /// Indicates whether this [`Field`] supports null values.
653    ///
654    /// If true, the field *may* contain null values.
655    #[inline]
656    pub const fn is_nullable(&self) -> bool {
657        self.nullable
658    }
659
660    /// Set the `nullable` of this [`Field`].
661    ///
662    /// ```
663    /// # use arrow_schema::*;
664    /// let mut field = Field::new("c1", DataType::Int64, false);
665    /// field.set_nullable(true);
666    ///
667    /// assert_eq!(field.is_nullable(), true);
668    /// ```
669    #[inline]
670    pub fn set_nullable(&mut self, nullable: bool) {
671        self.nullable = nullable;
672    }
673
674    /// Set `nullable` of the [`Field`] and returns self.
675    ///
676    /// ```
677    /// # use arrow_schema::*;
678    /// let field = Field::new("c1", DataType::Int64, false)
679    ///    .with_nullable(true);
680    ///
681    /// assert_eq!(field.is_nullable(), true);
682    /// ```
683    pub fn with_nullable(mut self, nullable: bool) -> Self {
684        self.set_nullable(nullable);
685        self
686    }
687
688    /// Returns a (flattened) [`Vec`] containing all child [`Field`]s
689    /// within `self` contained within this field (including `self`)
690    pub(crate) fn fields(&self) -> Vec<&Field> {
691        let mut collected_fields = vec![self];
692        collected_fields.append(&mut Field::_fields(&self.data_type));
693
694        collected_fields
695    }
696
697    fn _fields(dt: &DataType) -> Vec<&Field> {
698        match dt {
699            DataType::Struct(fields) => fields.iter().flat_map(|f| f.fields()).collect(),
700            DataType::Union(fields, _) => fields.iter().flat_map(|(_, f)| f.fields()).collect(),
701            DataType::List(field)
702            | DataType::LargeList(field)
703            | DataType::ListView(field)
704            | DataType::LargeListView(field)
705            | DataType::FixedSizeList(field, _)
706            | DataType::Map(field, _) => field.fields(),
707            DataType::Dictionary(_, value_field) => Field::_fields(value_field.as_ref()),
708            DataType::RunEndEncoded(_, field) => field.fields(),
709            _ => vec![],
710        }
711    }
712
713    /// Returns a vector containing all (potentially nested) `Field` instances selected by the
714    /// dictionary ID they use
715    #[inline]
716    #[deprecated(
717        since = "54.0.0",
718        note = "The ability to preserve dictionary IDs will be removed. With it, all fields related to it."
719    )]
720    pub(crate) fn fields_with_dict_id(&self, id: i64) -> Vec<&Field> {
721        self.fields()
722            .into_iter()
723            .filter(|&field| {
724                #[allow(deprecated)]
725                let matching_dict_id = field.dict_id == id;
726                matches!(field.data_type(), DataType::Dictionary(_, _)) && matching_dict_id
727            })
728            .collect()
729    }
730
731    /// Returns the dictionary ID, if this is a dictionary type.
732    #[inline]
733    #[deprecated(
734        since = "54.0.0",
735        note = "The ability to preserve dictionary IDs will be removed. With it, all fields related to it."
736    )]
737    pub const fn dict_id(&self) -> Option<i64> {
738        match self.data_type {
739            #[allow(deprecated)]
740            DataType::Dictionary(_, _) => Some(self.dict_id),
741            _ => None,
742        }
743    }
744
745    /// Returns whether this `Field`'s dictionary is ordered, if this is a dictionary type.
746    ///
747    /// # Example
748    /// ```
749    /// # use arrow_schema::{DataType, Field};
750    /// // non dictionaries do not have a dict is ordered flat
751    /// let field = Field::new("c1", DataType::Int64, false);
752    /// assert_eq!(field.dict_is_ordered(), None);
753    /// // by default dictionary is not ordered
754    /// let field = Field::new("c1", DataType::Dictionary(Box::new(DataType::Int64), Box::new(DataType::Utf8)), false);
755    /// assert_eq!(field.dict_is_ordered(), Some(false));
756    /// let field = field.with_dict_is_ordered(true);
757    /// assert_eq!(field.dict_is_ordered(), Some(true));
758    /// ```
759    #[inline]
760    pub const fn dict_is_ordered(&self) -> Option<bool> {
761        match self.data_type {
762            DataType::Dictionary(_, _) => Some(self.dict_is_ordered),
763            _ => None,
764        }
765    }
766
767    /// Set the is ordered field for this `Field`, if it is a dictionary.
768    ///
769    /// Does nothing if this is not a dictionary type.
770    ///
771    /// See [`Field::dict_is_ordered`] for more information.
772    pub fn with_dict_is_ordered(mut self, dict_is_ordered: bool) -> Self {
773        if matches!(self.data_type, DataType::Dictionary(_, _)) {
774            self.dict_is_ordered = dict_is_ordered;
775        };
776        self
777    }
778
779    /// Merge this field into self if it is compatible.
780    ///
781    /// Struct fields are merged recursively.
782    ///
783    /// NOTE: `self` may be updated to a partial / unexpected state in case of merge failure.
784    ///
785    /// Example:
786    ///
787    /// ```
788    /// # use arrow_schema::*;
789    /// let mut field = Field::new("c1", DataType::Int64, false);
790    /// assert!(field.try_merge(&Field::new("c1", DataType::Int64, true)).is_ok());
791    /// assert!(field.is_nullable());
792    /// ```
793    pub fn try_merge(&mut self, from: &Field) -> Result<(), ArrowError> {
794        if from.dict_is_ordered != self.dict_is_ordered {
795            return Err(ArrowError::SchemaError(format!(
796                "Fail to merge schema field '{}' because from dict_is_ordered = {} does not match {}",
797                self.name, from.dict_is_ordered, self.dict_is_ordered
798            )));
799        }
800        // merge metadata
801        match (self.metadata().is_empty(), from.metadata().is_empty()) {
802            (false, false) => {
803                let mut merged = self.metadata().clone();
804                for (key, from_value) in from.metadata() {
805                    if let Some(self_value) = self.metadata.get(key) {
806                        if self_value != from_value {
807                            return Err(ArrowError::SchemaError(format!(
808                                "Fail to merge field '{}' due to conflicting metadata data value for key {}.
809                                    From value = {} does not match {}", self.name, key, from_value, self_value),
810                            ));
811                        }
812                    } else {
813                        merged.insert(key.clone(), from_value.clone());
814                    }
815                }
816                self.set_metadata(merged);
817            }
818            (true, false) => {
819                self.set_metadata(from.metadata().clone());
820            }
821            _ => {}
822        }
823        match &mut self.data_type {
824            DataType::Struct(nested_fields) => match &from.data_type {
825                DataType::Struct(from_nested_fields) => {
826                    let mut builder = SchemaBuilder::new();
827                    nested_fields
828                        .iter()
829                        .chain(from_nested_fields)
830                        .try_for_each(|f| builder.try_merge(f))?;
831                    *nested_fields = builder.finish().fields;
832                }
833                DataType::Null => {
834                    self.nullable = true;
835                }
836                _ => {
837                    return Err(ArrowError::SchemaError(format!(
838                        "Fail to merge schema field '{}' because the from data_type = {} is not DataType::Struct",
839                        self.name, from.data_type
840                    )));
841                }
842            },
843            DataType::Union(nested_fields, _) => match &from.data_type {
844                DataType::Union(from_nested_fields, _) => {
845                    nested_fields.try_merge(from_nested_fields)?
846                }
847                DataType::Null => {
848                    self.nullable = true;
849                }
850                _ => {
851                    return Err(ArrowError::SchemaError(format!(
852                        "Fail to merge schema field '{}' because the from data_type = {} is not DataType::Union",
853                        self.name, from.data_type
854                    )));
855                }
856            },
857            DataType::List(field) => match &from.data_type {
858                DataType::List(from_field) => {
859                    let mut f = (**field).clone();
860                    f.try_merge(from_field)?;
861                    (*field) = Arc::new(f);
862                }
863                DataType::Null => {
864                    self.nullable = true;
865                }
866                _ => {
867                    return Err(ArrowError::SchemaError(format!(
868                        "Fail to merge schema field '{}' because the from data_type = {} is not DataType::List",
869                        self.name, from.data_type
870                    )));
871                }
872            },
873            DataType::LargeList(field) => match &from.data_type {
874                DataType::LargeList(from_field) => {
875                    let mut f = (**field).clone();
876                    f.try_merge(from_field)?;
877                    (*field) = Arc::new(f);
878                }
879                DataType::Null => {
880                    self.nullable = true;
881                }
882                _ => {
883                    return Err(ArrowError::SchemaError(format!(
884                        "Fail to merge schema field '{}' because the from data_type = {} is not DataType::LargeList",
885                        self.name, from.data_type
886                    )));
887                }
888            },
889            DataType::Null => {
890                self.nullable = true;
891                self.data_type = from.data_type.clone();
892            }
893            DataType::Boolean
894            | DataType::Int8
895            | DataType::Int16
896            | DataType::Int32
897            | DataType::Int64
898            | DataType::UInt8
899            | DataType::UInt16
900            | DataType::UInt32
901            | DataType::UInt64
902            | DataType::Float16
903            | DataType::Float32
904            | DataType::Float64
905            | DataType::Timestamp(_, _)
906            | DataType::Date32
907            | DataType::Date64
908            | DataType::Time32(_)
909            | DataType::Time64(_)
910            | DataType::Duration(_)
911            | DataType::Binary
912            | DataType::LargeBinary
913            | DataType::BinaryView
914            | DataType::Interval(_)
915            | DataType::LargeListView(_)
916            | DataType::ListView(_)
917            | DataType::Map(_, _)
918            | DataType::Dictionary(_, _)
919            | DataType::RunEndEncoded(_, _)
920            | DataType::FixedSizeList(_, _)
921            | DataType::FixedSizeBinary(_)
922            | DataType::Utf8
923            | DataType::LargeUtf8
924            | DataType::Utf8View
925            | DataType::Decimal32(_, _)
926            | DataType::Decimal64(_, _)
927            | DataType::Decimal128(_, _)
928            | DataType::Decimal256(_, _) => {
929                if from.data_type == DataType::Null {
930                    self.nullable = true;
931                } else if self.data_type != from.data_type {
932                    return Err(ArrowError::SchemaError(format!(
933                        "Fail to merge schema field '{}' because the from data_type = {} does not equal {}",
934                        self.name, from.data_type, self.data_type
935                    )));
936                }
937            }
938        }
939        self.nullable |= from.nullable;
940
941        Ok(())
942    }
943
944    /// Check to see if `self` is a superset of `other` field. Superset is defined as:
945    ///
946    /// * if nullability doesn't match, self needs to be nullable
947    /// * self.metadata is a superset of other.metadata
948    /// * all other fields are equal
949    pub fn contains(&self, other: &Field) -> bool {
950        self.name == other.name
951        && self.data_type.contains(&other.data_type)
952        && self.dict_is_ordered == other.dict_is_ordered
953        // self need to be nullable or both of them are not nullable
954        && (self.nullable || !other.nullable)
955        // make sure self.metadata is a superset of other.metadata
956        && other.metadata.iter().all(|(k, v1)| {
957            self.metadata.get(k).map(|v2| v1 == v2).unwrap_or_default()
958        })
959    }
960
961    /// Return size of this instance in bytes.
962    ///
963    /// Includes the size of `Self`.
964    pub fn size(&self) -> usize {
965        std::mem::size_of_val(self) - std::mem::size_of_val(&self.data_type)
966            + self.data_type.size()
967            + self.name.capacity()
968            + (std::mem::size_of::<(String, String)>() * self.metadata.capacity())
969            + self
970                .metadata
971                .iter()
972                .map(|(k, v)| k.capacity() + v.capacity())
973                .sum::<usize>()
974    }
975}
976
977impl std::fmt::Display for Field {
978    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
979        #![expect(deprecated)] // Must still print dict_id, if set
980        let Self {
981            name,
982            data_type,
983            nullable,
984            dict_id,
985            dict_is_ordered,
986            metadata,
987        } = self;
988        let maybe_nullable = if *nullable { "nullable " } else { "" };
989        let metadata_str = if metadata.is_empty() {
990            String::new()
991        } else {
992            format!(", metadata: {metadata:?}")
993        };
994        let dict_id_str = if dict_id == &0 {
995            String::new()
996        } else {
997            format!(", dict_id: {dict_id}")
998        };
999        let dict_is_ordered_str = if *dict_is_ordered {
1000            ", dict_is_ordered"
1001        } else {
1002            ""
1003        };
1004        write!(
1005            f,
1006            "Field {{ {name:?}: {maybe_nullable}{data_type}{dict_id_str}{dict_is_ordered_str}{metadata_str} }}"
1007        )
1008    }
1009}
1010
1011#[cfg(test)]
1012mod test {
1013    use super::*;
1014    use std::collections::hash_map::DefaultHasher;
1015
1016    #[test]
1017    fn test_new_with_string() {
1018        // Fields should allow owned Strings to support reuse
1019        let s = "c1";
1020        Field::new(s, DataType::Int64, false);
1021    }
1022
1023    #[test]
1024    fn test_new_dict_with_string() {
1025        // Fields should allow owned Strings to support reuse
1026        let s = "c1";
1027        #[allow(deprecated)]
1028        Field::new_dict(s, DataType::Int64, false, 4, false);
1029    }
1030
1031    #[test]
1032    #[cfg_attr(miri, ignore)] // Can't handle the inlined strings of the assert_debug_snapshot macro
1033    fn test_debug_format_field() {
1034        // Make sure the `Debug` formatting of `Field` is readable and not too long
1035        insta::assert_debug_snapshot!(Field::new("item", DataType::UInt8, false), @r"
1036        Field {
1037            data_type: UInt8,
1038        }
1039        ");
1040        insta::assert_debug_snapshot!(Field::new("column", DataType::LargeUtf8, true), @r#"
1041        Field {
1042            name: "column",
1043            data_type: LargeUtf8,
1044            nullable: true,
1045        }
1046        "#);
1047    }
1048
1049    #[test]
1050    fn test_merge_incompatible_types() {
1051        let mut field = Field::new("c1", DataType::Int64, false);
1052        let result = field
1053            .try_merge(&Field::new("c1", DataType::Float32, true))
1054            .expect_err("should fail")
1055            .to_string();
1056        assert_eq!(
1057            "Schema error: Fail to merge schema field 'c1' because the from data_type = Float32 does not equal Int64",
1058            result
1059        );
1060    }
1061
1062    #[test]
1063    fn test_merge_with_null() {
1064        let mut field1 = Field::new("c1", DataType::Null, true);
1065        field1
1066            .try_merge(&Field::new("c1", DataType::Float32, false))
1067            .expect("should widen type to nullable float");
1068        assert_eq!(Field::new("c1", DataType::Float32, true), field1);
1069
1070        let mut field2 = Field::new("c2", DataType::Utf8, false);
1071        field2
1072            .try_merge(&Field::new("c2", DataType::Null, true))
1073            .expect("should widen type to nullable utf8");
1074        assert_eq!(Field::new("c2", DataType::Utf8, true), field2);
1075    }
1076
1077    #[test]
1078    fn test_merge_with_nested_null() {
1079        let mut struct1 = Field::new(
1080            "s1",
1081            DataType::Struct(Fields::from(vec![Field::new(
1082                "inner",
1083                DataType::Float32,
1084                false,
1085            )])),
1086            false,
1087        );
1088
1089        let struct2 = Field::new(
1090            "s2",
1091            DataType::Struct(Fields::from(vec![Field::new(
1092                "inner",
1093                DataType::Null,
1094                false,
1095            )])),
1096            true,
1097        );
1098
1099        struct1
1100            .try_merge(&struct2)
1101            .expect("should widen inner field's type to nullable float");
1102        assert_eq!(
1103            Field::new(
1104                "s1",
1105                DataType::Struct(Fields::from(vec![Field::new(
1106                    "inner",
1107                    DataType::Float32,
1108                    true,
1109                )])),
1110                true,
1111            ),
1112            struct1
1113        );
1114
1115        let mut list1 = Field::new(
1116            "l1",
1117            DataType::List(Field::new("inner", DataType::Float32, false).into()),
1118            false,
1119        );
1120
1121        let list2 = Field::new(
1122            "l2",
1123            DataType::List(Field::new("inner", DataType::Null, false).into()),
1124            true,
1125        );
1126
1127        list1
1128            .try_merge(&list2)
1129            .expect("should widen inner field's type to nullable float");
1130        assert_eq!(
1131            Field::new(
1132                "l1",
1133                DataType::List(Field::new("inner", DataType::Float32, true).into()),
1134                true,
1135            ),
1136            list1
1137        );
1138
1139        let mut large_list1 = Field::new(
1140            "ll1",
1141            DataType::LargeList(Field::new("inner", DataType::Float32, false).into()),
1142            false,
1143        );
1144
1145        let large_list2 = Field::new(
1146            "ll2",
1147            DataType::LargeList(Field::new("inner", DataType::Null, false).into()),
1148            true,
1149        );
1150
1151        large_list1
1152            .try_merge(&large_list2)
1153            .expect("should widen inner field's type to nullable float");
1154        assert_eq!(
1155            Field::new(
1156                "ll1",
1157                DataType::LargeList(Field::new("inner", DataType::Float32, true).into()),
1158                true,
1159            ),
1160            large_list1
1161        );
1162    }
1163
1164    #[test]
1165    fn test_fields_with_dict_id() {
1166        #[allow(deprecated)]
1167        let dict1 = Field::new_dict(
1168            "dict1",
1169            DataType::Dictionary(DataType::Utf8.into(), DataType::Int32.into()),
1170            false,
1171            10,
1172            false,
1173        );
1174        #[allow(deprecated)]
1175        let dict2 = Field::new_dict(
1176            "dict2",
1177            DataType::Dictionary(DataType::Int32.into(), DataType::Int8.into()),
1178            false,
1179            20,
1180            false,
1181        );
1182
1183        let field = Field::new(
1184            "struct<dict1, list[struct<dict2, list[struct<dict1]>]>",
1185            DataType::Struct(Fields::from(vec![
1186                dict1.clone(),
1187                Field::new(
1188                    "list[struct<dict1, list[struct<dict2>]>]",
1189                    DataType::List(Arc::new(Field::new(
1190                        "struct<dict1, list[struct<dict2>]>",
1191                        DataType::Struct(Fields::from(vec![
1192                            dict1.clone(),
1193                            Field::new(
1194                                "list[struct<dict2>]",
1195                                DataType::List(Arc::new(Field::new(
1196                                    "struct<dict2>",
1197                                    DataType::Struct(vec![dict2.clone()].into()),
1198                                    false,
1199                                ))),
1200                                false,
1201                            ),
1202                        ])),
1203                        false,
1204                    ))),
1205                    false,
1206                ),
1207            ])),
1208            false,
1209        );
1210
1211        #[allow(deprecated)]
1212        for field in field.fields_with_dict_id(10) {
1213            assert_eq!(dict1, *field);
1214        }
1215        #[allow(deprecated)]
1216        for field in field.fields_with_dict_id(20) {
1217            assert_eq!(dict2, *field);
1218        }
1219    }
1220
1221    fn get_field_hash(field: &Field) -> u64 {
1222        let mut s = DefaultHasher::new();
1223        field.hash(&mut s);
1224        s.finish()
1225    }
1226
1227    #[test]
1228    fn test_field_comparison_case() {
1229        // dictionary-encoding properties not used for field comparison
1230        #[allow(deprecated)]
1231        let dict1 = Field::new_dict(
1232            "dict1",
1233            DataType::Dictionary(DataType::Utf8.into(), DataType::Int32.into()),
1234            false,
1235            10,
1236            false,
1237        );
1238        #[allow(deprecated)]
1239        let dict2 = Field::new_dict(
1240            "dict1",
1241            DataType::Dictionary(DataType::Utf8.into(), DataType::Int32.into()),
1242            false,
1243            20,
1244            false,
1245        );
1246
1247        assert_eq!(dict1, dict2);
1248        assert_eq!(get_field_hash(&dict1), get_field_hash(&dict2));
1249
1250        #[allow(deprecated)]
1251        let dict1 = Field::new_dict(
1252            "dict0",
1253            DataType::Dictionary(DataType::Utf8.into(), DataType::Int32.into()),
1254            false,
1255            10,
1256            false,
1257        );
1258
1259        assert_ne!(dict1, dict2);
1260        assert_ne!(get_field_hash(&dict1), get_field_hash(&dict2));
1261    }
1262
1263    #[test]
1264    fn test_field_comparison_metadata() {
1265        let f1 = Field::new("x", DataType::Binary, false).with_metadata(HashMap::from([
1266            (String::from("k1"), String::from("v1")),
1267            (String::from("k2"), String::from("v2")),
1268        ]));
1269        let f2 = Field::new("x", DataType::Binary, false).with_metadata(HashMap::from([
1270            (String::from("k1"), String::from("v1")),
1271            (String::from("k3"), String::from("v3")),
1272        ]));
1273        let f3 = Field::new("x", DataType::Binary, false).with_metadata(HashMap::from([
1274            (String::from("k1"), String::from("v1")),
1275            (String::from("k3"), String::from("v4")),
1276        ]));
1277
1278        assert!(f1.cmp(&f2).is_lt());
1279        assert!(f2.cmp(&f3).is_lt());
1280        assert!(f1.cmp(&f3).is_lt());
1281    }
1282
1283    #[test]
1284    #[expect(clippy::needless_borrows_for_generic_args)] // intentional to exercise various references
1285    fn test_field_as_ref() {
1286        let field = || Field::new("x", DataType::Binary, false);
1287
1288        // AsRef can be used in a function accepting a field.
1289        // However, this case actually works a bit better when function takes `&Field`
1290        fn accept_ref(_: impl AsRef<Field>) {}
1291
1292        accept_ref(field());
1293        accept_ref(&field());
1294        accept_ref(&&field());
1295        accept_ref(Arc::new(field()));
1296        accept_ref(&Arc::new(field()));
1297        accept_ref(&&Arc::new(field()));
1298
1299        // AsRef can be used in a function accepting a collection of fields in any form,
1300        // such as &[Field], or &[Arc<Field>]
1301        fn accept_refs(_: impl IntoIterator<Item: AsRef<Field>>) {}
1302
1303        accept_refs(vec![field()]);
1304        accept_refs(vec![&field()]);
1305        accept_refs(vec![Arc::new(field())]);
1306        accept_refs(vec![&Arc::new(field())]);
1307        accept_refs(&vec![field()]);
1308        accept_refs(&vec![&field()]);
1309        accept_refs(&vec![Arc::new(field())]);
1310        accept_refs(&vec![&Arc::new(field())]);
1311    }
1312
1313    #[test]
1314    fn test_contains_reflexivity() {
1315        let mut field = Field::new("field1", DataType::Float16, false);
1316        field.set_metadata(HashMap::from([
1317            (String::from("k0"), String::from("v0")),
1318            (String::from("k1"), String::from("v1")),
1319        ]));
1320        assert!(field.contains(&field))
1321    }
1322
1323    #[test]
1324    fn test_contains_transitivity() {
1325        let child_field = Field::new("child1", DataType::Float16, false);
1326
1327        let mut field1 = Field::new(
1328            "field1",
1329            DataType::Struct(Fields::from(vec![child_field])),
1330            false,
1331        );
1332        field1.set_metadata(HashMap::from([(String::from("k1"), String::from("v1"))]));
1333
1334        let mut field2 = Field::new("field1", DataType::Struct(Fields::default()), true);
1335        field2.set_metadata(HashMap::from([(String::from("k2"), String::from("v2"))]));
1336        field2.try_merge(&field1).unwrap();
1337
1338        let mut field3 = Field::new("field1", DataType::Struct(Fields::default()), false);
1339        field3.set_metadata(HashMap::from([(String::from("k3"), String::from("v3"))]));
1340        field3.try_merge(&field2).unwrap();
1341
1342        assert!(field2.contains(&field1));
1343        assert!(field3.contains(&field2));
1344        assert!(field3.contains(&field1));
1345
1346        assert!(!field1.contains(&field2));
1347        assert!(!field1.contains(&field3));
1348        assert!(!field2.contains(&field3));
1349    }
1350
1351    #[test]
1352    fn test_contains_nullable() {
1353        let field1 = Field::new("field1", DataType::Boolean, true);
1354        let field2 = Field::new("field1", DataType::Boolean, false);
1355        assert!(field1.contains(&field2));
1356        assert!(!field2.contains(&field1));
1357    }
1358
1359    #[test]
1360    fn test_contains_must_have_same_fields() {
1361        let child_field1 = Field::new("child1", DataType::Float16, false);
1362        let child_field2 = Field::new("child2", DataType::Float16, false);
1363
1364        let field1 = Field::new(
1365            "field1",
1366            DataType::Struct(vec![child_field1.clone()].into()),
1367            true,
1368        );
1369        let field2 = Field::new(
1370            "field1",
1371            DataType::Struct(vec![child_field1, child_field2].into()),
1372            true,
1373        );
1374
1375        assert!(!field1.contains(&field2));
1376        assert!(!field2.contains(&field1));
1377
1378        // UnionFields with different type ID
1379        let field1 = Field::new(
1380            "field1",
1381            DataType::Union(
1382                UnionFields::try_new(
1383                    vec![1, 2],
1384                    vec![
1385                        Field::new("field1", DataType::UInt8, true),
1386                        Field::new("field3", DataType::Utf8, false),
1387                    ],
1388                )
1389                .unwrap(),
1390                UnionMode::Dense,
1391            ),
1392            true,
1393        );
1394        let field2 = Field::new(
1395            "field1",
1396            DataType::Union(
1397                UnionFields::try_new(
1398                    vec![1, 3],
1399                    vec![
1400                        Field::new("field1", DataType::UInt8, false),
1401                        Field::new("field3", DataType::Utf8, false),
1402                    ],
1403                )
1404                .unwrap(),
1405                UnionMode::Dense,
1406            ),
1407            true,
1408        );
1409        assert!(!field1.contains(&field2));
1410
1411        // UnionFields with same type ID
1412        let field1 = Field::new(
1413            "field1",
1414            DataType::Union(
1415                UnionFields::try_new(
1416                    vec![1, 2],
1417                    vec![
1418                        Field::new("field1", DataType::UInt8, true),
1419                        Field::new("field3", DataType::Utf8, false),
1420                    ],
1421                )
1422                .unwrap(),
1423                UnionMode::Dense,
1424            ),
1425            true,
1426        );
1427        let field2 = Field::new(
1428            "field1",
1429            DataType::Union(
1430                UnionFields::try_new(
1431                    vec![1, 2],
1432                    vec![
1433                        Field::new("field1", DataType::UInt8, false),
1434                        Field::new("field3", DataType::Utf8, false),
1435                    ],
1436                )
1437                .unwrap(),
1438                UnionMode::Dense,
1439            ),
1440            true,
1441        );
1442        assert!(field1.contains(&field2));
1443    }
1444
1445    #[cfg(feature = "serde")]
1446    fn assert_binary_serde_round_trip(field: Field) {
1447        let serialized = postcard::to_stdvec(&field).unwrap();
1448        let deserialized: Field = postcard::from_bytes(&serialized).unwrap();
1449        assert_eq!(field, deserialized)
1450    }
1451
1452    #[cfg(feature = "serde")]
1453    #[test]
1454    fn test_field_without_metadata_serde() {
1455        let field = Field::new("name", DataType::Boolean, true);
1456        assert_binary_serde_round_trip(field)
1457    }
1458
1459    #[cfg(feature = "serde")]
1460    #[test]
1461    fn test_field_with_empty_metadata_serde() {
1462        let field = Field::new("name", DataType::Boolean, false).with_metadata(HashMap::new());
1463
1464        assert_binary_serde_round_trip(field)
1465    }
1466
1467    #[cfg(feature = "serde")]
1468    #[test]
1469    fn test_field_with_nonempty_metadata_serde() {
1470        let mut metadata = HashMap::new();
1471        metadata.insert("hi".to_owned(), "".to_owned());
1472        let field = Field::new("name", DataType::Boolean, false).with_metadata(metadata);
1473
1474        assert_binary_serde_round_trip(field)
1475    }
1476
1477    #[test]
1478    fn test_merge_compound_with_null() {
1479        // Struct + Null
1480        let mut field = Field::new(
1481            "s",
1482            DataType::Struct(Fields::from(vec![Field::new("a", DataType::Int32, false)])),
1483            false,
1484        );
1485        field
1486            .try_merge(&Field::new("s", DataType::Null, true))
1487            .expect("Struct should merge with Null");
1488        assert!(field.is_nullable());
1489        assert!(matches!(field.data_type(), DataType::Struct(_)));
1490
1491        // List + Null
1492        let mut field = Field::new(
1493            "l",
1494            DataType::List(Field::new("item", DataType::Utf8, false).into()),
1495            false,
1496        );
1497        field
1498            .try_merge(&Field::new("l", DataType::Null, true))
1499            .expect("List should merge with Null");
1500        assert!(field.is_nullable());
1501        assert!(matches!(field.data_type(), DataType::List(_)));
1502
1503        // LargeList + Null
1504        let mut field = Field::new(
1505            "ll",
1506            DataType::LargeList(Field::new("item", DataType::Utf8, false).into()),
1507            false,
1508        );
1509        field
1510            .try_merge(&Field::new("ll", DataType::Null, true))
1511            .expect("LargeList should merge with Null");
1512        assert!(field.is_nullable());
1513        assert!(matches!(field.data_type(), DataType::LargeList(_)));
1514
1515        // Union + Null
1516        let mut field = Field::new(
1517            "u",
1518            DataType::Union(
1519                UnionFields::try_new(vec![0], vec![Field::new("f", DataType::Int32, false)])
1520                    .unwrap(),
1521                UnionMode::Dense,
1522            ),
1523            false,
1524        );
1525        field
1526            .try_merge(&Field::new("u", DataType::Null, true))
1527            .expect("Union should merge with Null");
1528        assert!(matches!(field.data_type(), DataType::Union(_, _)));
1529    }
1530}
arrow_schema/field.rs

arrow_schema/
field.rs
