arrow_schema/
fields.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::ops::Deref;
19use std::sync::Arc;
20
21use crate::{ArrowError, DataType, Field, FieldRef};
22
23/// A cheaply cloneable, owned slice of [`FieldRef`]
24///
25/// Similar to `Arc<Vec<FieldRef>>` or `Arc<[FieldRef]>`
26///
27/// Can be constructed in a number of ways
28///
29/// ```
30/// # use std::sync::Arc;
31/// # use arrow_schema::{DataType, Field, Fields, SchemaBuilder};
32/// // Can be constructed from Vec<Field>
33/// Fields::from(vec![Field::new("a", DataType::Boolean, false)]);
34/// // Can be constructed from Vec<FieldRef>
35/// Fields::from(vec![Arc::new(Field::new("a", DataType::Boolean, false))]);
36/// // Can be constructed from an iterator of Field
37/// std::iter::once(Field::new("a", DataType::Boolean, false)).collect::<Fields>();
38/// // Can be constructed from an iterator of FieldRef
39/// std::iter::once(Arc::new(Field::new("a", DataType::Boolean, false))).collect::<Fields>();
40/// ```
41///
42/// See [`SchemaBuilder`] for mutating or updating [`Fields`]
43///
44/// ```
45/// # use arrow_schema::{DataType, Field, SchemaBuilder};
46/// let mut builder = SchemaBuilder::new();
47/// builder.push(Field::new("a", DataType::Boolean, false));
48/// builder.push(Field::new("b", DataType::Boolean, false));
49/// let fields = builder.finish().fields;
50///
51/// let mut builder = SchemaBuilder::from(&fields);
52/// builder.remove(0);
53/// let new = builder.finish().fields;
54/// ```
55///
56/// [`SchemaBuilder`]: crate::SchemaBuilder
57#[derive(Clone, Eq, PartialEq, Ord, PartialOrd, Hash)]
58#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
59#[cfg_attr(feature = "serde", serde(transparent))]
60pub struct Fields(Arc<[FieldRef]>);
61
62impl std::fmt::Debug for Fields {
63    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
64        self.0.as_ref().fmt(f)
65    }
66}
67
68impl Fields {
69    /// Returns a new empty [`Fields`]
70    pub fn empty() -> Self {
71        Self(Arc::new([]))
72    }
73
74    /// Return size of this instance in bytes.
75    pub fn size(&self) -> usize {
76        self.iter()
77            .map(|field| field.size() + std::mem::size_of::<FieldRef>())
78            .sum()
79    }
80
81    /// Searches for a field by name, returning it along with its index if found
82    pub fn find(&self, name: &str) -> Option<(usize, &FieldRef)> {
83        self.0.iter().enumerate().find(|(_, b)| b.name() == name)
84    }
85
86    /// Check to see if `self` is a superset of `other`
87    ///
88    /// In particular returns true if both have the same number of fields, and [`Field::contains`]
89    /// for each field across self and other
90    ///
91    /// In other words, any record that conforms to `other` should also conform to `self`
92    pub fn contains(&self, other: &Fields) -> bool {
93        if Arc::ptr_eq(&self.0, &other.0) {
94            return true;
95        }
96        self.len() == other.len()
97            && self
98                .iter()
99                .zip(other.iter())
100                .all(|(a, b)| Arc::ptr_eq(a, b) || a.contains(b))
101    }
102
103    /// Returns a copy of this [`Fields`] containing only those [`FieldRef`] passing a predicate
104    ///
105    /// Performs a depth-first scan of [`Fields`] invoking `filter` for each [`FieldRef`]
106    /// containing no child [`FieldRef`], a leaf field, along with a count of the number
107    /// of such leaves encountered so far. Only [`FieldRef`] for which `filter`
108    /// returned `true` will be included in the result.
109    ///
110    /// This can therefore be used to select a subset of fields from nested types
111    /// such as [`DataType::Struct`] or [`DataType::List`].
112    ///
113    /// ```
114    /// # use arrow_schema::{DataType, Field, Fields};
115    /// let fields = Fields::from(vec![
116    ///     Field::new("a", DataType::Int32, true), // Leaf 0
117    ///     Field::new("b", DataType::Struct(Fields::from(vec![
118    ///         Field::new("c", DataType::Float32, false), // Leaf 1
119    ///         Field::new("d", DataType::Float64, false), // Leaf 2
120    ///         Field::new("e", DataType::Struct(Fields::from(vec![
121    ///             Field::new("f", DataType::Int32, false),   // Leaf 3
122    ///             Field::new("g", DataType::Float16, false), // Leaf 4
123    ///         ])), true),
124    ///     ])), false)
125    /// ]);
126    /// let filtered = fields.filter_leaves(|idx, _| [0, 2, 3, 4].contains(&idx));
127    /// let expected = Fields::from(vec![
128    ///     Field::new("a", DataType::Int32, true),
129    ///     Field::new("b", DataType::Struct(Fields::from(vec![
130    ///         Field::new("d", DataType::Float64, false),
131    ///         Field::new("e", DataType::Struct(Fields::from(vec![
132    ///             Field::new("f", DataType::Int32, false),
133    ///             Field::new("g", DataType::Float16, false),
134    ///         ])), true),
135    ///     ])), false)
136    /// ]);
137    /// assert_eq!(filtered, expected);
138    /// ```
139    pub fn filter_leaves<F: FnMut(usize, &FieldRef) -> bool>(&self, mut filter: F) -> Self {
140        self.try_filter_leaves(|idx, field| Ok(filter(idx, field)))
141            .unwrap()
142    }
143
144    /// Returns a copy of this [`Fields`] containing only those [`FieldRef`] passing a predicate
145    /// or an error if the predicate fails.
146    ///
147    /// See [`Fields::filter_leaves`] for more information.
148    pub fn try_filter_leaves<F: FnMut(usize, &FieldRef) -> Result<bool, ArrowError>>(
149        &self,
150        mut filter: F,
151    ) -> Result<Self, ArrowError> {
152        fn filter_field<F: FnMut(&FieldRef) -> Result<bool, ArrowError>>(
153            f: &FieldRef,
154            filter: &mut F,
155        ) -> Result<Option<FieldRef>, ArrowError> {
156            use DataType::*;
157
158            let v = match f.data_type() {
159                Dictionary(_, v) => v.as_ref(),       // Key must be integer
160                RunEndEncoded(_, v) => v.data_type(), // Run-ends must be integer
161                d => d,
162            };
163            let d = match v {
164                List(child) => {
165                    let fields = filter_field(child, filter)?;
166                    if let Some(fields) = fields {
167                        List(fields)
168                    } else {
169                        return Ok(None);
170                    }
171                }
172                LargeList(child) => {
173                    let fields = filter_field(child, filter)?;
174                    if let Some(fields) = fields {
175                        LargeList(fields)
176                    } else {
177                        return Ok(None);
178                    }
179                }
180                Map(child, ordered) => {
181                    let fields = filter_field(child, filter)?;
182                    if let Some(fields) = fields {
183                        Map(fields, *ordered)
184                    } else {
185                        return Ok(None);
186                    }
187                }
188                FixedSizeList(child, size) => {
189                    let fields = filter_field(child, filter)?;
190                    if let Some(fields) = fields {
191                        FixedSizeList(fields, *size)
192                    } else {
193                        return Ok(None);
194                    }
195                }
196                Struct(fields) => {
197                    let filtered: Result<Vec<_>, _> =
198                        fields.iter().map(|f| filter_field(f, filter)).collect();
199                    let filtered: Fields = filtered?
200                        .iter()
201                        .filter_map(|f| f.as_ref().cloned())
202                        .collect();
203
204                    if filtered.is_empty() {
205                        return Ok(None);
206                    }
207
208                    Struct(filtered)
209                }
210                Union(fields, mode) => {
211                    let filtered: Result<Vec<_>, _> = fields
212                        .iter()
213                        .map(|(id, f)| filter_field(f, filter).map(|f| f.map(|f| (id, f))))
214                        .collect();
215                    let filtered: UnionFields = filtered?
216                        .iter()
217                        .filter_map(|f| f.as_ref().cloned())
218                        .collect();
219
220                    if filtered.is_empty() {
221                        return Ok(None);
222                    }
223
224                    Union(filtered, *mode)
225                }
226                _ => {
227                    let filtered = filter(f)?;
228                    return Ok(filtered.then(|| f.clone()));
229                }
230            };
231            let d = match f.data_type() {
232                Dictionary(k, _) => Dictionary(k.clone(), Box::new(d)),
233                RunEndEncoded(v, f) => {
234                    RunEndEncoded(v.clone(), Arc::new(f.as_ref().clone().with_data_type(d)))
235                }
236                _ => d,
237            };
238            Ok(Some(Arc::new(f.as_ref().clone().with_data_type(d))))
239        }
240
241        let mut leaf_idx = 0;
242        let mut filter = |f: &FieldRef| {
243            let t = filter(leaf_idx, f)?;
244            leaf_idx += 1;
245            Ok(t)
246        };
247
248        let filtered: Result<Vec<_>, _> = self
249            .0
250            .iter()
251            .map(|f| filter_field(f, &mut filter))
252            .collect();
253        let filtered = filtered?
254            .iter()
255            .filter_map(|f| f.as_ref().cloned())
256            .collect();
257        Ok(filtered)
258    }
259}
260
261impl Default for Fields {
262    fn default() -> Self {
263        Self::empty()
264    }
265}
266
267impl FromIterator<Field> for Fields {
268    fn from_iter<T: IntoIterator<Item = Field>>(iter: T) -> Self {
269        iter.into_iter().map(Arc::new).collect()
270    }
271}
272
273impl FromIterator<FieldRef> for Fields {
274    fn from_iter<T: IntoIterator<Item = FieldRef>>(iter: T) -> Self {
275        Self(iter.into_iter().collect())
276    }
277}
278
279impl From<Vec<Field>> for Fields {
280    fn from(value: Vec<Field>) -> Self {
281        value.into_iter().collect()
282    }
283}
284
285impl From<Vec<FieldRef>> for Fields {
286    fn from(value: Vec<FieldRef>) -> Self {
287        Self(value.into())
288    }
289}
290
291impl From<&[FieldRef]> for Fields {
292    fn from(value: &[FieldRef]) -> Self {
293        Self(value.into())
294    }
295}
296
297impl<const N: usize> From<[FieldRef; N]> for Fields {
298    fn from(value: [FieldRef; N]) -> Self {
299        Self(Arc::new(value))
300    }
301}
302
303impl Deref for Fields {
304    type Target = [FieldRef];
305
306    fn deref(&self) -> &Self::Target {
307        self.0.as_ref()
308    }
309}
310
311impl<'a> IntoIterator for &'a Fields {
312    type Item = &'a FieldRef;
313    type IntoIter = std::slice::Iter<'a, FieldRef>;
314
315    fn into_iter(self) -> Self::IntoIter {
316        self.0.iter()
317    }
318}
319
320/// A cheaply cloneable, owned collection of [`FieldRef`] and their corresponding type ids
321#[derive(Clone, Eq, PartialEq, Ord, PartialOrd, Hash)]
322#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
323#[cfg_attr(feature = "serde", serde(transparent))]
324pub struct UnionFields(Arc<[(i8, FieldRef)]>);
325
326impl std::fmt::Debug for UnionFields {
327    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
328        self.0.as_ref().fmt(f)
329    }
330}
331
332/// Allows direct indexing into [`UnionFields`] to access fields by position.
333///
334/// # Panics
335///
336/// Panics if the index is out of bounds. Note that [`UnionFields`] supports
337/// a maximum of 128 fields, as type IDs are represented as `i8` values.
338///
339/// For a non-panicking alternative, use [`UnionFields::get`].
340impl std::ops::Index<usize> for UnionFields {
341    type Output = (i8, FieldRef);
342
343    fn index(&self, index: usize) -> &Self::Output {
344        &self.0[index]
345    }
346}
347
348impl UnionFields {
349    /// Create a new [`UnionFields`] with no fields
350    pub fn empty() -> Self {
351        Self(Arc::from([]))
352    }
353
354    /// Create a new [`UnionFields`] from a [`Fields`] and array of type_ids
355    ///
356    /// See <https://arrow.apache.org/docs/format/Columnar.html#union-layout>
357    ///
358    /// # Errors
359    ///
360    /// This function returns an error if:
361    /// - Any type_id appears more than once (duplicate type ids)
362    /// - The type_ids are duplicated
363    ///
364    /// # Examples
365    ///
366    /// ```
367    /// use arrow_schema::{DataType, Field, UnionFields};
368    /// // Create a new UnionFields with type id mapping
369    /// // 1 -> DataType::UInt8
370    /// // 3 -> DataType::Utf8
371    /// let result = UnionFields::try_new(
372    ///     vec![1, 3],
373    ///     vec![
374    ///         Field::new("field1", DataType::UInt8, false),
375    ///         Field::new("field3", DataType::Utf8, false),
376    ///     ],
377    /// );
378    /// assert!(result.is_ok());
379    ///
380    /// // This will fail due to duplicate type ids
381    /// let result = UnionFields::try_new(
382    ///     vec![1, 1],
383    ///     vec![
384    ///         Field::new("field1", DataType::UInt8, false),
385    ///         Field::new("field2", DataType::Utf8, false),
386    ///     ],
387    /// );
388    /// assert!(result.is_err());
389    /// ```
390    pub fn try_new<F, T>(type_ids: T, fields: F) -> Result<Self, ArrowError>
391    where
392        F: IntoIterator,
393        F::Item: Into<FieldRef>,
394        T: IntoIterator<Item = i8>,
395    {
396        let mut type_ids_iter = type_ids.into_iter();
397        let mut fields_iter = fields.into_iter().map(Into::into);
398
399        let mut seen_type_ids = 0u128;
400
401        let mut out = Vec::new();
402
403        loop {
404            match (type_ids_iter.next(), fields_iter.next()) {
405                (None, None) => return Ok(Self(out.into())),
406                (Some(type_id), Some(field)) => {
407                    // check type id is non-negative
408                    if type_id < 0 {
409                        return Err(ArrowError::InvalidArgumentError(format!(
410                            "type ids must be non-negative: {type_id}"
411                        )));
412                    }
413
414                    // check type id uniqueness
415                    let mask = 1_u128 << type_id;
416                    if (seen_type_ids & mask) != 0 {
417                        return Err(ArrowError::InvalidArgumentError(format!(
418                            "duplicate type id: {type_id}"
419                        )));
420                    }
421
422                    seen_type_ids |= mask;
423
424                    out.push((type_id, field));
425                }
426                (None, Some(_)) => {
427                    return Err(ArrowError::InvalidArgumentError(
428                        "fields iterator has more elements than type_ids iterator".to_string(),
429                    ));
430                }
431                (Some(_), None) => {
432                    return Err(ArrowError::InvalidArgumentError(
433                        "type_ids iterator has more elements than fields iterator".to_string(),
434                    ));
435                }
436            }
437        }
438    }
439
440    /// Create a new [`UnionFields`] from a collection of fields with automatically
441    /// assigned type IDs starting from 0.
442    ///
443    /// The type IDs are assigned in increasing order: 0, 1, 2, 3, etc.
444    ///
445    /// See <https://arrow.apache.org/docs/format/Columnar.html#union-layout>
446    ///
447    /// # Panics
448    ///
449    /// Panics if the number of fields exceeds 127 (the maximum value for i8 type IDs).
450    ///
451    /// If you want to avoid panics, use [`UnionFields::try_from_fields`] instead, which
452    /// returns a `Result`.
453    ///
454    /// # Examples
455    ///
456    /// ```
457    /// use arrow_schema::{DataType, Field, UnionFields};
458    /// // Create a new UnionFields with automatic type id assignment
459    /// // 0 -> DataType::UInt8
460    /// // 1 -> DataType::Utf8
461    /// let union_fields = UnionFields::from_fields(vec![
462    ///     Field::new("field1", DataType::UInt8, false),
463    ///     Field::new("field2", DataType::Utf8, false),
464    /// ]);
465    /// assert_eq!(union_fields.len(), 2);
466    /// ```
467    pub fn from_fields<F>(fields: F) -> Self
468    where
469        F: IntoIterator,
470        F::Item: Into<FieldRef>,
471    {
472        fields
473            .into_iter()
474            .enumerate()
475            .map(|(i, field)| {
476                let id = i8::try_from(i).expect("UnionFields cannot contain more than 128 fields");
477
478                (id, field.into())
479            })
480            .collect()
481    }
482
483    /// Create a new [`UnionFields`] from a collection of fields with automatically
484    /// assigned type IDs starting from 0.
485    ///
486    /// The type IDs are assigned in increasing order: 0, 1, 2, 3, etc.
487    ///
488    /// This is the non-panicking version of [`UnionFields::from_fields`].
489    ///
490    /// See <https://arrow.apache.org/docs/format/Columnar.html#union-layout>
491    ///
492    /// # Errors
493    ///
494    /// Returns an error if the number of fields exceeds 127 (the maximum value for i8 type IDs).
495    ///
496    /// # Examples
497    ///
498    /// ```
499    /// use arrow_schema::{DataType, Field, UnionFields};
500    /// // Create a new UnionFields with automatic type id assignment
501    /// // 0 -> DataType::UInt8
502    /// // 1 -> DataType::Utf8
503    /// let result = UnionFields::try_from_fields(vec![
504    ///     Field::new("field1", DataType::UInt8, false),
505    ///     Field::new("field2", DataType::Utf8, false),
506    /// ]);
507    /// assert!(result.is_ok());
508    /// assert_eq!(result.unwrap().len(), 2);
509    ///
510    /// // This will fail with too many fields
511    /// let many_fields: Vec<_> = (0..200)
512    ///     .map(|i| Field::new(format!("field{}", i), DataType::Int32, false))
513    ///     .collect();
514    /// let result = UnionFields::try_from_fields(many_fields);
515    /// assert!(result.is_err());
516    /// ```
517    pub fn try_from_fields<F>(fields: F) -> Result<Self, ArrowError>
518    where
519        F: IntoIterator,
520        F::Item: Into<FieldRef>,
521    {
522        let mut out = Vec::with_capacity(i8::MAX as usize + 1);
523
524        for (i, field) in fields.into_iter().enumerate() {
525            let id = i8::try_from(i).map_err(|_| {
526                ArrowError::InvalidArgumentError(
527                    "UnionFields cannot contain more than 128 fields".into(),
528                )
529            })?;
530
531            out.push((id, field.into()));
532        }
533
534        Ok(Self(out.into()))
535    }
536
537    /// Return size of this instance in bytes.
538    pub fn size(&self) -> usize {
539        self.iter()
540            .map(|(_, field)| field.size() + std::mem::size_of::<(i8, FieldRef)>())
541            .sum()
542    }
543
544    /// Returns the number of fields in this [`UnionFields`]
545    pub fn len(&self) -> usize {
546        self.0.len()
547    }
548
549    /// Returns `true` if this is empty
550    pub fn is_empty(&self) -> bool {
551        self.0.is_empty()
552    }
553
554    /// Returns an iterator over the fields and type ids in this [`UnionFields`]
555    pub fn iter(&self) -> impl Iterator<Item = (i8, &FieldRef)> + '_ {
556        self.0.iter().map(|(id, f)| (*id, f))
557    }
558
559    /// Returns a reference to the field at the given index, or `None` if out of bounds.
560    ///
561    /// This is a safe alternative to direct indexing via `[]`.
562    ///
563    /// # Example
564    ///
565    /// ```
566    /// use arrow_schema::{DataType, Field, UnionFields};
567    ///
568    /// let fields = UnionFields::try_new(
569    ///     vec![1, 3],
570    ///     vec![
571    ///         Field::new("field1", DataType::UInt8, false),
572    ///         Field::new("field3", DataType::Utf8, false),
573    ///     ],
574    /// ).unwrap();
575    ///
576    /// assert!(fields.get(0).is_some());
577    /// assert!(fields.get(1).is_some());
578    /// assert!(fields.get(2).is_none());
579    /// ```
580    pub fn get(&self, index: usize) -> Option<&(i8, FieldRef)> {
581        self.0.get(index)
582    }
583
584    /// Searches for a field by its type id, returning the type id and field reference if found.
585    /// Returns `None` if no field with the given type id exists.
586    pub fn find_by_type_id(&self, type_id: i8) -> Option<(i8, &FieldRef)> {
587        self.iter().find(|&(i, _)| i == type_id)
588    }
589
590    /// Searches for a field by value equality, returning its type id and reference if found.
591    /// Returns `None` if no matching field exists in this [`UnionFields`].
592    pub fn find_by_field(&self, field: &Field) -> Option<(i8, &FieldRef)> {
593        self.iter().find(|&(_, f)| f.as_ref() == field)
594    }
595
596    /// Merge this field into self if it is compatible.
597    ///
598    /// See [`Field::try_merge`]
599    pub(crate) fn try_merge(&mut self, other: &Self) -> Result<(), ArrowError> {
600        // TODO: This currently may produce duplicate type IDs (#3982)
601        let mut output: Vec<_> = self.iter().map(|(id, f)| (id, f.clone())).collect();
602        for (field_type_id, from_field) in other.iter() {
603            let mut is_new_field = true;
604            for (self_type_id, self_field) in output.iter_mut() {
605                if from_field == self_field {
606                    // If the nested fields in two unions are the same, they must have same
607                    // type id.
608                    if *self_type_id != field_type_id {
609                        return Err(ArrowError::SchemaError(format!(
610                            "Fail to merge schema field '{}' because the self_type_id = {} does not equal field_type_id = {}",
611                            self_field.name(),
612                            self_type_id,
613                            field_type_id
614                        )));
615                    }
616
617                    is_new_field = false;
618                    break;
619                }
620            }
621
622            if is_new_field {
623                output.push((field_type_id, from_field.clone()))
624            }
625        }
626        *self = output.into_iter().collect();
627        Ok(())
628    }
629}
630
631impl FromIterator<(i8, FieldRef)> for UnionFields {
632    fn from_iter<T: IntoIterator<Item = (i8, FieldRef)>>(iter: T) -> Self {
633        Self(iter.into_iter().collect())
634    }
635}
636
637#[cfg(test)]
638mod tests {
639    use super::*;
640    use crate::UnionMode;
641
642    #[test]
643    fn test_filter() {
644        let floats = Fields::from(vec![
645            Field::new("a", DataType::Float32, false),
646            Field::new("b", DataType::Float32, false),
647        ]);
648        let fields = Fields::from(vec![
649            Field::new("a", DataType::Int32, true),
650            Field::new("floats", DataType::Struct(floats.clone()), true),
651            Field::new("b", DataType::Int16, true),
652            Field::new(
653                "c",
654                DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8)),
655                false,
656            ),
657            Field::new(
658                "d",
659                DataType::Dictionary(
660                    Box::new(DataType::Int32),
661                    Box::new(DataType::Struct(floats.clone())),
662                ),
663                false,
664            ),
665            Field::new_list(
666                "e",
667                Field::new("floats", DataType::Struct(floats.clone()), true),
668                true,
669            ),
670            Field::new_fixed_size_list(
671                "f",
672                Field::new_list_field(DataType::Int32, false),
673                3,
674                false,
675            ),
676            Field::new_map(
677                "g",
678                Field::MAP_ENTRIES_FIELD_DEFAULT_NAME,
679                Field::new(
680                    Field::MAP_KEY_FIELD_DEFAULT_NAME,
681                    DataType::LargeUtf8,
682                    false,
683                ),
684                Field::new(Field::MAP_VALUE_FIELD_DEFAULT_NAME, DataType::Int32, true),
685                false,
686                false,
687            ),
688            Field::new(
689                "h",
690                DataType::Union(
691                    UnionFields::try_new(
692                        vec![1, 3],
693                        vec![
694                            Field::new("field1", DataType::UInt8, false),
695                            Field::new("field3", DataType::Utf8, false),
696                        ],
697                    )
698                    .unwrap(),
699                    UnionMode::Dense,
700                ),
701                true,
702            ),
703            Field::new(
704                "i",
705                DataType::RunEndEncoded(
706                    Arc::new(Field::new("run_ends", DataType::Int32, false)),
707                    Arc::new(Field::new("values", DataType::Struct(floats.clone()), true)),
708                ),
709                false,
710            ),
711        ]);
712
713        let floats_a = DataType::Struct(vec![floats[0].clone()].into());
714
715        let r = fields.filter_leaves(|idx, _| idx == 0 || idx == 1);
716        assert_eq!(r.len(), 2);
717        assert_eq!(r[0], fields[0]);
718        assert_eq!(r[1].data_type(), &floats_a);
719
720        let r = fields.filter_leaves(|_, f| f.name() == "a");
721        assert_eq!(r.len(), 5);
722        assert_eq!(r[0], fields[0]);
723        assert_eq!(r[1].data_type(), &floats_a);
724        assert_eq!(
725            r[2].data_type(),
726            &DataType::Dictionary(Box::new(DataType::Int32), Box::new(floats_a.clone()))
727        );
728        assert_eq!(
729            r[3].as_ref(),
730            &Field::new_list("e", Field::new("floats", floats_a.clone(), true), true)
731        );
732        assert_eq!(
733            r[4].as_ref(),
734            &Field::new(
735                "i",
736                DataType::RunEndEncoded(
737                    Arc::new(Field::new("run_ends", DataType::Int32, false)),
738                    Arc::new(Field::new("values", floats_a.clone(), true)),
739                ),
740                false,
741            )
742        );
743
744        let r = fields.filter_leaves(|_, f| f.name() == "floats");
745        assert_eq!(r.len(), 0);
746
747        let r = fields.filter_leaves(|idx, _| idx == 9);
748        assert_eq!(r.len(), 1);
749        assert_eq!(r[0], fields[6]);
750
751        let r = fields.filter_leaves(|idx, _| idx == 10 || idx == 11);
752        assert_eq!(r.len(), 1);
753        assert_eq!(r[0], fields[7]);
754
755        let union = DataType::Union(
756            UnionFields::try_new(vec![1], vec![Field::new("field1", DataType::UInt8, false)])
757                .unwrap(),
758            UnionMode::Dense,
759        );
760
761        let r = fields.filter_leaves(|idx, _| idx == 12);
762        assert_eq!(r.len(), 1);
763        assert_eq!(r[0].data_type(), &union);
764
765        let r = fields.filter_leaves(|idx, _| idx == 14 || idx == 15);
766        assert_eq!(r.len(), 1);
767        assert_eq!(r[0], fields[9]);
768
769        // Propagate error
770        let r = fields.try_filter_leaves(|_, _| Err(ArrowError::SchemaError("error".to_string())));
771        assert!(r.is_err());
772    }
773
774    #[test]
775    fn test_union_fields_try_new_valid() {
776        let res = UnionFields::try_new(
777            vec![1, 6, 7],
778            vec![
779                Field::new("f1", DataType::UInt8, false),
780                Field::new("f6", DataType::Utf8, false),
781                Field::new("f7", DataType::Int32, true),
782            ],
783        );
784        assert!(res.is_ok());
785        let union_fields = res.unwrap();
786        assert_eq!(union_fields.len(), 3);
787        assert_eq!(
788            union_fields.iter().map(|(id, _)| id).collect::<Vec<_>>(),
789            vec![1, 6, 7]
790        );
791    }
792
793    #[test]
794    fn test_union_fields_try_new_empty() {
795        let res = UnionFields::try_new(Vec::<i8>::new(), Vec::<Field>::new());
796        assert!(res.is_ok());
797        assert!(res.unwrap().is_empty());
798    }
799
800    #[test]
801    fn test_union_fields_try_new_duplicate_type_id() {
802        let res = UnionFields::try_new(
803            vec![1, 1],
804            vec![
805                Field::new("f1", DataType::UInt8, false),
806                Field::new("f2", DataType::Utf8, false),
807            ],
808        );
809        assert!(res.is_err());
810        assert!(
811            res.unwrap_err()
812                .to_string()
813                .contains("duplicate type id: 1")
814        );
815    }
816
817    #[test]
818    fn test_union_fields_try_new_duplicate_field() {
819        let field = Field::new("field", DataType::UInt8, false);
820        let res = UnionFields::try_new(vec![1, 2], vec![field.clone(), field]);
821        assert!(res.is_ok());
822    }
823
824    #[test]
825    fn test_union_fields_try_new_more_type_ids() {
826        let res = UnionFields::try_new(
827            vec![1, 2, 3],
828            vec![
829                Field::new("f1", DataType::UInt8, false),
830                Field::new("f2", DataType::Utf8, false),
831            ],
832        );
833        assert!(res.is_err());
834        assert!(
835            res.unwrap_err()
836                .to_string()
837                .contains("type_ids iterator has more elements")
838        );
839    }
840
841    #[test]
842    fn test_union_fields_try_new_more_fields() {
843        let res = UnionFields::try_new(
844            vec![1, 2],
845            vec![
846                Field::new("f1", DataType::UInt8, false),
847                Field::new("f2", DataType::Utf8, false),
848                Field::new("f3", DataType::Int32, true),
849            ],
850        );
851        assert!(res.is_err());
852        assert!(
853            res.unwrap_err()
854                .to_string()
855                .contains("fields iterator has more elements")
856        );
857    }
858
859    #[test]
860    fn test_union_fields_try_new_negative_type_ids() {
861        let res = UnionFields::try_new(
862            vec![-128, -1, 0, 127],
863            vec![
864                Field::new("field_min", DataType::UInt8, false),
865                Field::new("field_neg", DataType::Utf8, false),
866                Field::new("field_zero", DataType::Int32, true),
867                Field::new("field_max", DataType::Boolean, false),
868            ],
869        );
870        assert!(res.is_err());
871        assert!(
872            res.unwrap_err()
873                .to_string()
874                .contains("type ids must be non-negative")
875        )
876    }
877
878    #[test]
879    fn test_union_fields_try_new_complex_types() {
880        let res = UnionFields::try_new(
881            vec![0, 1, 2],
882            vec![
883                Field::new(
884                    "struct_field",
885                    DataType::Struct(Fields::from(vec![
886                        Field::new("a", DataType::Int32, false),
887                        Field::new("b", DataType::Utf8, true),
888                    ])),
889                    false,
890                ),
891                Field::new_list(
892                    "list_field",
893                    Field::new("item", DataType::Float64, true),
894                    true,
895                ),
896                Field::new(
897                    "dict_field",
898                    DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8)),
899                    false,
900                ),
901            ],
902        );
903        assert!(res.is_ok());
904        assert_eq!(res.unwrap().len(), 3);
905    }
906
907    #[test]
908    fn test_union_fields_try_new_single_field() {
909        let res = UnionFields::try_new(
910            vec![42],
911            vec![Field::new("only_field", DataType::Int64, false)],
912        );
913        assert!(res.is_ok());
914        let union_fields = res.unwrap();
915        assert_eq!(union_fields.len(), 1);
916        assert_eq!(union_fields.iter().next().unwrap().0, 42);
917    }
918
919    #[test]
920    fn test_union_fields_try_from_fields_empty() {
921        let res = UnionFields::try_from_fields(Vec::<Field>::new());
922        assert!(res.is_ok());
923        assert!(res.unwrap().is_empty());
924    }
925
926    #[test]
927    fn test_union_fields_try_from_fields_single() {
928        let res = UnionFields::try_from_fields(vec![Field::new("only", DataType::Int64, false)]);
929        assert!(res.is_ok());
930        let union_fields = res.unwrap();
931        assert_eq!(union_fields.len(), 1);
932        assert_eq!(union_fields.iter().next().unwrap().0, 0);
933    }
934
935    #[test]
936    fn test_union_fields_try_from_fields_too_many() {
937        let many_fields: Vec<_> = (0..200)
938            .map(|i| Field::new(format!("field{}", i), DataType::Int32, false))
939            .collect();
940        let res = UnionFields::try_from_fields(many_fields);
941        assert!(res.is_err());
942        assert!(
943            res.unwrap_err()
944                .to_string()
945                .contains("UnionFields cannot contain more than 128 fields")
946        );
947    }
948
949    #[test]
950    fn test_union_fields_try_from_fields_max_valid() {
951        let fields: Vec<_> = (0..=i8::MAX)
952            .map(|i| Field::new(format!("field{}", i), DataType::Int32, false))
953            .collect();
954        let res = UnionFields::try_from_fields(fields);
955        assert!(res.is_ok());
956        let union_fields = res.unwrap();
957        assert_eq!(union_fields.len(), 128);
958        assert_eq!(union_fields.iter().map(|(id, _)| id).min().unwrap(), 0);
959        assert_eq!(union_fields.iter().map(|(id, _)| id).max().unwrap(), 127);
960    }
961
962    #[test]
963    fn test_union_fields_try_from_fields_over_max() {
964        // 129 fields should fail
965        let fields: Vec<_> = (0..129)
966            .map(|i| Field::new(format!("field{}", i), DataType::Int32, false))
967            .collect();
968        let res = UnionFields::try_from_fields(fields);
969        assert!(res.is_err());
970    }
971
972    #[test]
973    fn test_union_fields_try_from_fields_complex_types() {
974        let res = UnionFields::try_from_fields(vec![
975            Field::new(
976                "struct_field",
977                DataType::Struct(Fields::from(vec![
978                    Field::new("a", DataType::Int32, false),
979                    Field::new("b", DataType::Utf8, true),
980                ])),
981                false,
982            ),
983            Field::new_list(
984                "list_field",
985                Field::new("item", DataType::Float64, true),
986                true,
987            ),
988            Field::new(
989                "dict_field",
990                DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8)),
991                false,
992            ),
993        ]);
994        assert!(res.is_ok());
995        assert_eq!(res.unwrap().len(), 3);
996    }
997}
arrow_schema/fields.rs

arrow_schema/
fields.rs
