arrow_integration_test/
lib.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
18//! Partial support for the [Apache Arrow JSON test data format](https://github.com/apache/arrow/blob/master/docs/source/format/Integration.rst#json-test-data-format)
19//!
20//! These utilities define structs that read the integration JSON format for integration testing purposes.
21//!
22//! This is not a canonical format, but provides a human-readable way of verifying language implementations
23//!
24//! <div class="warning">
25//!
26//! This crate is **only intended for integration testing the
27//! [Arrow project](https://github.com/apache/arrow-rs)**. It is not [intended for usage outside of
28//! this context](https://github.com/apache/arrow-rs/issues/8684#issuecomment-3433193158).
29//!
30//! </div>
31
32#![doc(
33    html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
34    html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
35)]
36#![cfg_attr(docsrs, feature(doc_cfg))]
37#![warn(missing_docs)]
38use arrow_buffer::{IntervalDayTime, IntervalMonthDayNano, ScalarBuffer};
39use hex::decode;
40use num_bigint::BigInt;
41use num_traits::Signed;
42use serde::{Deserialize, Serialize};
43use serde_json::{Map as SJMap, Value};
44use std::collections::HashMap;
45use std::sync::Arc;
46
47use arrow::array::*;
48use arrow::buffer::{Buffer, MutableBuffer};
49use arrow::datatypes::*;
50use arrow::error::{ArrowError, Result};
51use arrow::util::bit_util;
52
53mod datatype;
54mod field;
55mod schema;
56
57pub use datatype::*;
58pub use field::*;
59pub use schema::*;
60
61/// A struct that represents an Arrow file with a schema and record batches
62///
63/// See <https://github.com/apache/arrow/blob/master/docs/source/format/Integration.rst#json-test-data-format>
64#[derive(Deserialize, Serialize, Debug)]
65pub struct ArrowJson {
66    /// The Arrow schema for JSON file
67    pub schema: ArrowJsonSchema,
68    /// The `RecordBatch`es in the JSON file
69    pub batches: Vec<ArrowJsonBatch>,
70    /// The dictionaries in the JSON file
71    #[serde(skip_serializing_if = "Option::is_none")]
72    pub dictionaries: Option<Vec<ArrowJsonDictionaryBatch>>,
73}
74
75/// A struct that partially reads the Arrow JSON schema.
76///
77/// Fields are left as JSON `Value` as they vary by `DataType`
78#[derive(Deserialize, Serialize, Debug)]
79pub struct ArrowJsonSchema {
80    /// An array of JSON fields
81    pub fields: Vec<ArrowJsonField>,
82    /// An array of metadata key-value pairs
83    #[serde(skip_serializing_if = "Option::is_none")]
84    pub metadata: Option<Vec<HashMap<String, String>>>,
85}
86
87/// Fields are left as JSON `Value` as they vary by `DataType`
88#[derive(Deserialize, Serialize, Debug)]
89pub struct ArrowJsonField {
90    /// The name of the field
91    pub name: String,
92    /// The data type of the field,
93    /// can be any valid JSON value
94    #[serde(rename = "type")]
95    pub field_type: Value,
96    /// Whether the field is nullable
97    pub nullable: bool,
98    /// The children fields
99    pub children: Vec<ArrowJsonField>,
100    /// The dictionary for the field
101    #[serde(skip_serializing_if = "Option::is_none")]
102    pub dictionary: Option<ArrowJsonFieldDictionary>,
103    /// The metadata for the field, if any
104    #[serde(skip_serializing_if = "Option::is_none")]
105    pub metadata: Option<Value>,
106}
107
108impl From<&FieldRef> for ArrowJsonField {
109    fn from(value: &FieldRef) -> Self {
110        Self::from(value.as_ref())
111    }
112}
113
114impl From<&Field> for ArrowJsonField {
115    fn from(field: &Field) -> Self {
116        let metadata_value = match field.metadata().is_empty() {
117            false => {
118                let mut array = Vec::new();
119                for (k, v) in field.metadata() {
120                    let mut kv_map = SJMap::new();
121                    kv_map.insert(k.clone(), Value::String(v.clone()));
122                    array.push(Value::Object(kv_map));
123                }
124                if !array.is_empty() {
125                    Some(Value::Array(array))
126                } else {
127                    None
128                }
129            }
130            _ => None,
131        };
132
133        Self {
134            name: field.name().clone(),
135            field_type: data_type_to_json(field.data_type()),
136            nullable: field.is_nullable(),
137            children: vec![],
138            dictionary: None, // TODO: not enough info
139            metadata: metadata_value,
140        }
141    }
142}
143
144/// Represents a dictionary-encoded field in the Arrow JSON format
145#[derive(Deserialize, Serialize, Debug)]
146pub struct ArrowJsonFieldDictionary {
147    /// A unique identifier for the dictionary
148    pub id: i64,
149    /// The type of the dictionary index
150    #[serde(rename = "indexType")]
151    pub index_type: DictionaryIndexType,
152    /// Whether the dictionary is ordered
153    #[serde(rename = "isOrdered")]
154    pub is_ordered: bool,
155}
156
157/// Type of an index for a dictionary-encoded field in the Arrow JSON format
158#[derive(Deserialize, Serialize, Debug)]
159pub struct DictionaryIndexType {
160    /// The name of the dictionary index type
161    pub name: String,
162    /// Whether the dictionary index type is signed
163    #[serde(rename = "isSigned")]
164    pub is_signed: bool,
165    /// The bit width of the dictionary index type
166    #[serde(rename = "bitWidth")]
167    pub bit_width: i64,
168}
169
170/// A struct that partially reads the Arrow JSON record batch
171#[derive(Deserialize, Serialize, Debug, Clone)]
172pub struct ArrowJsonBatch {
173    count: usize,
174    /// The columns in the record batch
175    pub columns: Vec<ArrowJsonColumn>,
176}
177
178/// A struct that partially reads the Arrow JSON dictionary batch
179#[derive(Deserialize, Serialize, Debug, Clone)]
180pub struct ArrowJsonDictionaryBatch {
181    /// The unique identifier for the dictionary
182    pub id: i64,
183    /// The data for the dictionary
184    pub data: ArrowJsonBatch,
185}
186
187/// A struct that partially reads the Arrow JSON column/array
188#[derive(Deserialize, Serialize, Clone, Debug)]
189pub struct ArrowJsonColumn {
190    name: String,
191    /// The number of elements in the column
192    pub count: usize,
193    /// The validity bitmap to determine null values
194    #[serde(rename = "VALIDITY")]
195    pub validity: Option<Vec<u8>>,
196    /// The data values in the column
197    #[serde(rename = "DATA")]
198    pub data: Option<Vec<Value>>,
199    /// The offsets for variable-sized data types
200    #[serde(rename = "OFFSET")]
201    pub offset: Option<Vec<Value>>, // leaving as Value as 64-bit offsets are strings
202    /// The type id for union types
203    #[serde(rename = "TYPE_ID")]
204    pub type_id: Option<Vec<i8>>,
205    /// The sizes for ListView/LargeListView types
206    #[serde(rename = "SIZE")]
207    pub size: Option<Vec<Value>>,
208    /// The views for BinaryView/Utf8View types
209    #[serde(rename = "VIEWS")]
210    pub views: Option<Vec<Value>>,
211    /// The variadic data buffers for BinaryView/Utf8View types
212    #[serde(rename = "VARIADIC_DATA_BUFFERS")]
213    pub variadic_data_buffers: Option<Vec<String>>,
214    /// The children columns for nested types
215    pub children: Option<Vec<ArrowJsonColumn>>,
216}
217
218impl ArrowJson {
219    /// Compare the Arrow JSON with a record batch reader
220    pub fn equals_reader(&self, reader: &mut dyn RecordBatchReader) -> Result<bool> {
221        if !self.schema.equals_schema(&reader.schema()) {
222            return Ok(false);
223        }
224
225        for json_batch in self.get_record_batches()? {
226            let batch = reader.next();
227            match batch {
228                Some(Ok(batch)) => {
229                    if json_batch != batch {
230                        println!("json: {json_batch:?}");
231                        println!("batch: {batch:?}");
232                        return Ok(false);
233                    }
234                }
235                Some(Err(e)) => return Err(e),
236                None => return Ok(false),
237            }
238        }
239
240        Ok(true)
241    }
242
243    /// Convert the stored dictionaries to `Vec[RecordBatch]`
244    pub fn get_record_batches(&self) -> Result<Vec<RecordBatch>> {
245        let schema = self.schema.to_arrow_schema()?;
246
247        let mut dictionaries = HashMap::new();
248        self.dictionaries.iter().for_each(|dict_batches| {
249            dict_batches.iter().for_each(|d| {
250                dictionaries.insert(d.id, d.clone());
251            });
252        });
253
254        let batches: Result<Vec<_>> = self
255            .batches
256            .iter()
257            .map(|col| record_batch_from_json(&schema, col.clone(), Some(&dictionaries)))
258            .collect();
259
260        batches
261    }
262}
263
264impl ArrowJsonSchema {
265    /// Compare the Arrow JSON schema with the Arrow `Schema`
266    fn equals_schema(&self, schema: &Schema) -> bool {
267        let field_len = self.fields.len();
268        if field_len != schema.fields().len() {
269            return false;
270        }
271        for i in 0..field_len {
272            let json_field = &self.fields[i];
273            let field = schema.field(i);
274            if !json_field.equals_field(field) {
275                return false;
276            }
277        }
278        true
279    }
280
281    fn to_arrow_schema(&self) -> Result<Schema> {
282        let arrow_fields: Result<Vec<_>> = self
283            .fields
284            .iter()
285            .map(|field| field.to_arrow_field())
286            .collect();
287
288        if let Some(metadatas) = &self.metadata {
289            let mut metadata: HashMap<String, String> = HashMap::new();
290
291            metadatas.iter().for_each(|pair| {
292                let key = pair.get("key").unwrap();
293                let value = pair.get("value").unwrap();
294                metadata.insert(key.clone(), value.clone());
295            });
296
297            Ok(Schema::new_with_metadata(arrow_fields?, metadata))
298        } else {
299            Ok(Schema::new(arrow_fields?))
300        }
301    }
302}
303
304impl ArrowJsonField {
305    /// Compare the Arrow JSON field with the Arrow `Field`
306    fn equals_field(&self, field: &Field) -> bool {
307        // convert to a field
308        match self.to_arrow_field() {
309            Ok(self_field) => {
310                assert_eq!(&self_field, field, "Arrow fields not the same");
311                true
312            }
313            Err(e) => {
314                eprintln!("Encountered error while converting JSON field to Arrow field: {e:?}");
315                false
316            }
317        }
318    }
319
320    /// Convert to an Arrow Field
321    /// TODO: convert to use an Into
322    fn to_arrow_field(&self) -> Result<Field> {
323        // a bit regressive, but we have to convert the field to JSON in order to convert it
324        let field =
325            serde_json::to_value(self).map_err(|error| ArrowError::JsonError(error.to_string()))?;
326        field_from_json(&field)
327    }
328}
329
330/// Generates a [`RecordBatch`] from an Arrow JSON batch, given a schema
331pub fn record_batch_from_json(
332    schema: &Schema,
333    json_batch: ArrowJsonBatch,
334    json_dictionaries: Option<&HashMap<i64, ArrowJsonDictionaryBatch>>,
335) -> Result<RecordBatch> {
336    let mut columns = vec![];
337
338    for (field, json_col) in schema.fields().iter().zip(json_batch.columns) {
339        let col = array_from_json(field, json_col, json_dictionaries)?;
340        columns.push(col);
341    }
342
343    RecordBatch::try_new(Arc::new(schema.clone()), columns)
344}
345
346/// Construct an Arrow array from a partially typed JSON column
347pub fn array_from_json(
348    field: &Field,
349    json_col: ArrowJsonColumn,
350    dictionaries: Option<&HashMap<i64, ArrowJsonDictionaryBatch>>,
351) -> Result<ArrayRef> {
352    match field.data_type() {
353        DataType::Null => Ok(Arc::new(NullArray::new(json_col.count))),
354        DataType::Boolean => {
355            let mut b = BooleanBuilder::with_capacity(json_col.count);
356            for (is_valid, value) in json_col
357                .validity
358                .as_ref()
359                .unwrap()
360                .iter()
361                .zip(json_col.data.unwrap())
362            {
363                match is_valid {
364                    1 => b.append_value(value.as_bool().unwrap()),
365                    _ => b.append_null(),
366                }
367            }
368            Ok(Arc::new(b.finish()))
369        }
370        DataType::Int8 => {
371            let mut b = Int8Builder::with_capacity(json_col.count);
372            for (is_valid, value) in json_col
373                .validity
374                .as_ref()
375                .unwrap()
376                .iter()
377                .zip(json_col.data.unwrap())
378            {
379                match is_valid {
380                    1 => b.append_value(value.as_i64().ok_or_else(|| {
381                        ArrowError::JsonError(format!("Unable to get {value:?} as int64"))
382                    })? as i8),
383                    _ => b.append_null(),
384                }
385            }
386            Ok(Arc::new(b.finish()))
387        }
388        DataType::Int16 => {
389            let mut b = Int16Builder::with_capacity(json_col.count);
390            for (is_valid, value) in json_col
391                .validity
392                .as_ref()
393                .unwrap()
394                .iter()
395                .zip(json_col.data.unwrap())
396            {
397                match is_valid {
398                    1 => b.append_value(value.as_i64().unwrap() as i16),
399                    _ => b.append_null(),
400                }
401            }
402            Ok(Arc::new(b.finish()))
403        }
404        DataType::Int32 | DataType::Date32 | DataType::Time32(_) => {
405            let mut b = Int32Builder::with_capacity(json_col.count);
406            for (is_valid, value) in json_col
407                .validity
408                .as_ref()
409                .unwrap()
410                .iter()
411                .zip(json_col.data.unwrap())
412            {
413                match is_valid {
414                    1 => b.append_value(value.as_i64().unwrap() as i32),
415                    _ => b.append_null(),
416                }
417            }
418            let array = Arc::new(b.finish()) as ArrayRef;
419            arrow::compute::cast(&array, field.data_type())
420        }
421        DataType::Interval(IntervalUnit::YearMonth) => {
422            let mut b = IntervalYearMonthBuilder::with_capacity(json_col.count);
423            for (is_valid, value) in json_col
424                .validity
425                .as_ref()
426                .unwrap()
427                .iter()
428                .zip(json_col.data.unwrap())
429            {
430                match is_valid {
431                    1 => b.append_value(value.as_i64().unwrap() as i32),
432                    _ => b.append_null(),
433                }
434            }
435            Ok(Arc::new(b.finish()))
436        }
437        DataType::Int64
438        | DataType::Date64
439        | DataType::Time64(_)
440        | DataType::Timestamp(_, _)
441        | DataType::Duration(_) => {
442            let mut b = Int64Builder::with_capacity(json_col.count);
443            for (is_valid, value) in json_col
444                .validity
445                .as_ref()
446                .unwrap()
447                .iter()
448                .zip(json_col.data.unwrap())
449            {
450                match is_valid {
451                    1 => b.append_value(match value {
452                        Value::Number(n) => n.as_i64().unwrap(),
453                        Value::String(s) => s.parse().expect("Unable to parse string as i64"),
454                        _ => panic!("Unable to parse {value:?} as number"),
455                    }),
456                    _ => b.append_null(),
457                }
458            }
459            let array = Arc::new(b.finish()) as ArrayRef;
460            arrow::compute::cast(&array, field.data_type())
461        }
462        DataType::Interval(IntervalUnit::DayTime) => {
463            let mut b = IntervalDayTimeBuilder::with_capacity(json_col.count);
464            for (is_valid, value) in json_col
465                .validity
466                .as_ref()
467                .unwrap()
468                .iter()
469                .zip(json_col.data.unwrap())
470            {
471                match is_valid {
472                    1 => b.append_value(match value {
473                        Value::Object(ref map)
474                            if map.contains_key("days") && map.contains_key("milliseconds") =>
475                        {
476                            match field.data_type() {
477                                DataType::Interval(IntervalUnit::DayTime) => {
478                                    let days = map.get("days").unwrap();
479                                    let milliseconds = map.get("milliseconds").unwrap();
480
481                                    match (days, milliseconds) {
482                                        (Value::Number(d), Value::Number(m)) => {
483                                            let days = d.as_i64().unwrap() as _;
484                                            let millis = m.as_i64().unwrap() as _;
485                                            IntervalDayTime::new(days, millis)
486                                        }
487                                        _ => {
488                                            panic!("Unable to parse {value:?} as interval daytime")
489                                        }
490                                    }
491                                }
492                                _ => panic!("Unable to parse {value:?} as interval daytime"),
493                            }
494                        }
495                        _ => panic!("Unable to parse {value:?} as number"),
496                    }),
497                    _ => b.append_null(),
498                }
499            }
500            Ok(Arc::new(b.finish()))
501        }
502        DataType::UInt8 => {
503            let mut b = UInt8Builder::with_capacity(json_col.count);
504            for (is_valid, value) in json_col
505                .validity
506                .as_ref()
507                .unwrap()
508                .iter()
509                .zip(json_col.data.unwrap())
510            {
511                match is_valid {
512                    1 => b.append_value(value.as_u64().unwrap() as u8),
513                    _ => b.append_null(),
514                }
515            }
516            Ok(Arc::new(b.finish()))
517        }
518        DataType::UInt16 => {
519            let mut b = UInt16Builder::with_capacity(json_col.count);
520            for (is_valid, value) in json_col
521                .validity
522                .as_ref()
523                .unwrap()
524                .iter()
525                .zip(json_col.data.unwrap())
526            {
527                match is_valid {
528                    1 => b.append_value(value.as_u64().unwrap() as u16),
529                    _ => b.append_null(),
530                }
531            }
532            Ok(Arc::new(b.finish()))
533        }
534        DataType::UInt32 => {
535            let mut b = UInt32Builder::with_capacity(json_col.count);
536            for (is_valid, value) in json_col
537                .validity
538                .as_ref()
539                .unwrap()
540                .iter()
541                .zip(json_col.data.unwrap())
542            {
543                match is_valid {
544                    1 => b.append_value(value.as_u64().unwrap() as u32),
545                    _ => b.append_null(),
546                }
547            }
548            Ok(Arc::new(b.finish()))
549        }
550        DataType::UInt64 => {
551            let mut b = UInt64Builder::with_capacity(json_col.count);
552            for (is_valid, value) in json_col
553                .validity
554                .as_ref()
555                .unwrap()
556                .iter()
557                .zip(json_col.data.unwrap())
558            {
559                match is_valid {
560                    1 => {
561                        if value.is_string() {
562                            b.append_value(
563                                value
564                                    .as_str()
565                                    .unwrap()
566                                    .parse()
567                                    .expect("Unable to parse string as u64"),
568                            )
569                        } else if value.is_number() {
570                            b.append_value(value.as_u64().expect("Unable to read number as u64"))
571                        } else {
572                            panic!("Unable to parse value {value:?} as u64")
573                        }
574                    }
575                    _ => b.append_null(),
576                }
577            }
578            Ok(Arc::new(b.finish()))
579        }
580        DataType::Interval(IntervalUnit::MonthDayNano) => {
581            let mut b = IntervalMonthDayNanoBuilder::with_capacity(json_col.count);
582            for (is_valid, value) in json_col
583                .validity
584                .as_ref()
585                .unwrap()
586                .iter()
587                .zip(json_col.data.unwrap())
588            {
589                match is_valid {
590                    1 => b.append_value(match value {
591                        Value::Object(v) => {
592                            let months = v.get("months").unwrap();
593                            let days = v.get("days").unwrap();
594                            let nanoseconds = v.get("nanoseconds").unwrap();
595                            match (months, days, nanoseconds) {
596                                (
597                                    Value::Number(months),
598                                    Value::Number(days),
599                                    Value::Number(nanoseconds),
600                                ) => {
601                                    let months = months.as_i64().unwrap() as i32;
602                                    let days = days.as_i64().unwrap() as i32;
603                                    let nanoseconds = nanoseconds.as_i64().unwrap();
604                                    IntervalMonthDayNano::new(months, days, nanoseconds)
605                                }
606                                (_, _, _) => {
607                                    panic!("Unable to parse {v:?} as MonthDayNano")
608                                }
609                            }
610                        }
611                        _ => panic!("Unable to parse {value:?} as MonthDayNano"),
612                    }),
613                    _ => b.append_null(),
614                }
615            }
616            Ok(Arc::new(b.finish()))
617        }
618        DataType::Float32 => {
619            let mut b = Float32Builder::with_capacity(json_col.count);
620            for (is_valid, value) in json_col
621                .validity
622                .as_ref()
623                .unwrap()
624                .iter()
625                .zip(json_col.data.unwrap())
626            {
627                match is_valid {
628                    1 => b.append_value(value.as_f64().unwrap() as f32),
629                    _ => b.append_null(),
630                }
631            }
632            Ok(Arc::new(b.finish()))
633        }
634        DataType::Float64 => {
635            let mut b = Float64Builder::with_capacity(json_col.count);
636            for (is_valid, value) in json_col
637                .validity
638                .as_ref()
639                .unwrap()
640                .iter()
641                .zip(json_col.data.unwrap())
642            {
643                match is_valid {
644                    1 => b.append_value(value.as_f64().unwrap()),
645                    _ => b.append_null(),
646                }
647            }
648            Ok(Arc::new(b.finish()))
649        }
650        DataType::Binary => {
651            let mut b = BinaryBuilder::with_capacity(json_col.count, 1024);
652            for (is_valid, value) in json_col
653                .validity
654                .as_ref()
655                .unwrap()
656                .iter()
657                .zip(json_col.data.unwrap())
658            {
659                match is_valid {
660                    1 => {
661                        let v = decode(value.as_str().unwrap()).unwrap();
662                        b.append_value(&v)
663                    }
664                    _ => b.append_null(),
665                }
666            }
667            Ok(Arc::new(b.finish()))
668        }
669        DataType::LargeBinary => {
670            let mut b = LargeBinaryBuilder::with_capacity(json_col.count, 1024);
671            for (is_valid, value) in json_col
672                .validity
673                .as_ref()
674                .unwrap()
675                .iter()
676                .zip(json_col.data.unwrap())
677            {
678                match is_valid {
679                    1 => {
680                        let v = decode(value.as_str().unwrap()).unwrap();
681                        b.append_value(&v)
682                    }
683                    _ => b.append_null(),
684                }
685            }
686            Ok(Arc::new(b.finish()))
687        }
688        DataType::Utf8 => {
689            let mut b = StringBuilder::with_capacity(json_col.count, 1024);
690            for (is_valid, value) in json_col
691                .validity
692                .as_ref()
693                .unwrap()
694                .iter()
695                .zip(json_col.data.unwrap())
696            {
697                match is_valid {
698                    1 => b.append_value(value.as_str().unwrap()),
699                    _ => b.append_null(),
700                }
701            }
702            Ok(Arc::new(b.finish()))
703        }
704        DataType::LargeUtf8 => {
705            let mut b = LargeStringBuilder::with_capacity(json_col.count, 1024);
706            for (is_valid, value) in json_col
707                .validity
708                .as_ref()
709                .unwrap()
710                .iter()
711                .zip(json_col.data.unwrap())
712            {
713                match is_valid {
714                    1 => b.append_value(value.as_str().unwrap()),
715                    _ => b.append_null(),
716                }
717            }
718            Ok(Arc::new(b.finish()))
719        }
720        DataType::FixedSizeBinary(len) => {
721            let mut b = FixedSizeBinaryBuilder::with_capacity(json_col.count, *len);
722            for (is_valid, value) in json_col
723                .validity
724                .as_ref()
725                .unwrap()
726                .iter()
727                .zip(json_col.data.unwrap())
728            {
729                match is_valid {
730                    1 => {
731                        let v = hex::decode(value.as_str().unwrap()).unwrap();
732                        b.append_value(&v)?
733                    }
734                    _ => b.append_null(),
735                }
736            }
737            Ok(Arc::new(b.finish()))
738        }
739        DataType::List(child_field) => {
740            let null_buf = create_null_buf(&json_col);
741            let children = json_col.children.clone().unwrap();
742            let child_array = array_from_json(child_field, children[0].clone(), dictionaries)?;
743            let offsets: Vec<i32> = json_col
744                .offset
745                .unwrap()
746                .iter()
747                .map(|v| v.as_i64().unwrap() as i32)
748                .collect();
749            let list_data = ArrayData::builder(field.data_type().clone())
750                .len(json_col.count)
751                .offset(0)
752                .add_buffer(Buffer::from(offsets.to_byte_slice()))
753                .add_child_data(child_array.into_data())
754                .null_bit_buffer(Some(null_buf))
755                .build()
756                .unwrap();
757            Ok(Arc::new(ListArray::from(list_data)))
758        }
759        DataType::LargeList(child_field) => {
760            let null_buf = create_null_buf(&json_col);
761            let children = json_col.children.clone().unwrap();
762            let child_array = array_from_json(child_field, children[0].clone(), dictionaries)?;
763            let offsets: Vec<i64> = json_col
764                .offset
765                .unwrap()
766                .iter()
767                .map(|v| match v {
768                    Value::Number(n) => n.as_i64().unwrap(),
769                    Value::String(s) => s.parse::<i64>().unwrap(),
770                    _ => panic!("64-bit offset must be either string or number"),
771                })
772                .collect();
773            let list_data = ArrayData::builder(field.data_type().clone())
774                .len(json_col.count)
775                .offset(0)
776                .add_buffer(Buffer::from(offsets.to_byte_slice()))
777                .add_child_data(child_array.into_data())
778                .null_bit_buffer(Some(null_buf))
779                .build()
780                .unwrap();
781            Ok(Arc::new(LargeListArray::from(list_data)))
782        }
783        DataType::ListView(child_field) => {
784            let null_buf = create_null_buf(&json_col);
785            let children = json_col.children.clone().unwrap();
786            let child_array = array_from_json(child_field, children[0].clone(), dictionaries)?;
787            let offsets: Vec<i32> = json_col
788                .offset
789                .unwrap()
790                .iter()
791                .map(|v| v.as_i64().unwrap() as i32)
792                .collect();
793            let sizes: Vec<i32> = json_col
794                .size
795                .unwrap()
796                .iter()
797                .map(|v| v.as_i64().unwrap() as i32)
798                .collect();
799            let list_data = ArrayData::builder(field.data_type().clone())
800                .len(json_col.count)
801                .add_buffer(Buffer::from(offsets.to_byte_slice()))
802                .add_buffer(Buffer::from(sizes.to_byte_slice()))
803                .add_child_data(child_array.into_data())
804                .null_bit_buffer(Some(null_buf))
805                .build()
806                .unwrap();
807            Ok(Arc::new(ListViewArray::from(list_data)))
808        }
809        DataType::LargeListView(child_field) => {
810            let null_buf = create_null_buf(&json_col);
811            let children = json_col.children.clone().unwrap();
812            let child_array = array_from_json(child_field, children[0].clone(), dictionaries)?;
813            let offsets: Vec<i64> = json_col
814                .offset
815                .unwrap()
816                .iter()
817                .map(|v| match v {
818                    Value::Number(n) => n.as_i64().unwrap(),
819                    Value::String(s) => s.parse::<i64>().unwrap(),
820                    _ => panic!("64-bit offset must be either string or number"),
821                })
822                .collect();
823            let sizes: Vec<i64> = json_col
824                .size
825                .unwrap()
826                .iter()
827                .map(|v| match v {
828                    Value::Number(n) => n.as_i64().unwrap(),
829                    Value::String(s) => s.parse::<i64>().unwrap(),
830                    _ => panic!("64-bit size must be either string or number"),
831                })
832                .collect();
833            let list_data = ArrayData::builder(field.data_type().clone())
834                .len(json_col.count)
835                .add_buffer(Buffer::from(offsets.to_byte_slice()))
836                .add_buffer(Buffer::from(sizes.to_byte_slice()))
837                .add_child_data(child_array.into_data())
838                .null_bit_buffer(Some(null_buf))
839                .build()
840                .unwrap();
841            Ok(Arc::new(LargeListViewArray::from(list_data)))
842        }
843        DataType::FixedSizeList(child_field, _) => {
844            let children = json_col.children.clone().unwrap();
845            let child_array = array_from_json(child_field, children[0].clone(), dictionaries)?;
846            let null_buf = create_null_buf(&json_col);
847            let list_data = ArrayData::builder(field.data_type().clone())
848                .len(json_col.count)
849                .add_child_data(child_array.into_data())
850                .null_bit_buffer(Some(null_buf))
851                .build()
852                .unwrap();
853            Ok(Arc::new(FixedSizeListArray::from(list_data)))
854        }
855        DataType::Struct(fields) => {
856            // construct struct with null data
857            let null_buf = create_null_buf(&json_col);
858            let mut array_data = ArrayData::builder(field.data_type().clone())
859                .len(json_col.count)
860                .null_bit_buffer(Some(null_buf));
861
862            for (field, col) in fields.iter().zip(json_col.children.unwrap()) {
863                let array = array_from_json(field, col, dictionaries)?;
864                array_data = array_data.add_child_data(array.into_data());
865            }
866
867            let array = StructArray::from(array_data.build().unwrap());
868            Ok(Arc::new(array))
869        }
870        DataType::Dictionary(key_type, value_type) => {
871            #[expect(deprecated)]
872            let dict_id = field.dict_id().ok_or_else(|| {
873                ArrowError::JsonError(format!("Unable to find dict_id for field {field}"))
874            })?;
875            // find dictionary
876            let dictionary = dictionaries
877                .ok_or_else(|| {
878                    ArrowError::JsonError(format!(
879                        "Unable to find any dictionaries for field {field}"
880                    ))
881                })?
882                .get(&dict_id);
883            match dictionary {
884                Some(dictionary) => dictionary_array_from_json(
885                    field,
886                    json_col,
887                    key_type,
888                    value_type,
889                    dictionary,
890                    dictionaries,
891                ),
892                None => Err(ArrowError::JsonError(format!(
893                    "Unable to find dictionary for field {field}"
894                ))),
895            }
896        }
897        DataType::Decimal32(precision, scale) => {
898            let mut b = Decimal32Builder::with_capacity(json_col.count);
899            for (is_valid, value) in json_col
900                .validity
901                .as_ref()
902                .unwrap()
903                .iter()
904                .zip(json_col.data.unwrap())
905            {
906                match is_valid {
907                    1 => b.append_value(value.as_str().unwrap().parse::<i32>().unwrap()),
908                    _ => b.append_null(),
909                }
910            }
911            Ok(Arc::new(
912                b.finish().with_precision_and_scale(*precision, *scale)?,
913            ))
914        }
915        DataType::Decimal64(precision, scale) => {
916            let mut b = Decimal64Builder::with_capacity(json_col.count);
917            for (is_valid, value) in json_col
918                .validity
919                .as_ref()
920                .unwrap()
921                .iter()
922                .zip(json_col.data.unwrap())
923            {
924                match is_valid {
925                    1 => b.append_value(value.as_str().unwrap().parse::<i64>().unwrap()),
926                    _ => b.append_null(),
927                }
928            }
929            Ok(Arc::new(
930                b.finish().with_precision_and_scale(*precision, *scale)?,
931            ))
932        }
933        DataType::Decimal128(precision, scale) => {
934            let mut b = Decimal128Builder::with_capacity(json_col.count);
935            for (is_valid, value) in json_col
936                .validity
937                .as_ref()
938                .unwrap()
939                .iter()
940                .zip(json_col.data.unwrap())
941            {
942                match is_valid {
943                    1 => b.append_value(value.as_str().unwrap().parse::<i128>().unwrap()),
944                    _ => b.append_null(),
945                }
946            }
947            Ok(Arc::new(
948                b.finish().with_precision_and_scale(*precision, *scale)?,
949            ))
950        }
951        DataType::Decimal256(precision, scale) => {
952            let mut b = Decimal256Builder::with_capacity(json_col.count);
953            for (is_valid, value) in json_col
954                .validity
955                .as_ref()
956                .unwrap()
957                .iter()
958                .zip(json_col.data.unwrap())
959            {
960                match is_valid {
961                    1 => {
962                        let str = value.as_str().unwrap();
963                        let integer = BigInt::parse_bytes(str.as_bytes(), 10).unwrap();
964                        let integer_bytes = integer.to_signed_bytes_le();
965                        // Sign-extend the minimal-length two's-complement
966                        // encoding to the full 32 bytes: 0x00 fill for
967                        // non-negative values, 0xFF for negative ones.
968                        let mut bytes = if integer.is_negative() {
969                            [255_u8; 32]
970                        } else {
971                            [0_u8; 32]
972                        };
973                        bytes[0..integer_bytes.len()].copy_from_slice(integer_bytes.as_slice());
974                        b.append_value(i256::from_le_bytes(bytes));
975                    }
976                    _ => b.append_null(),
977                }
978            }
979            Ok(Arc::new(
980                b.finish().with_precision_and_scale(*precision, *scale)?,
981            ))
982        }
983        DataType::Map(child_field, _) => {
984            let null_buf = create_null_buf(&json_col);
985            let children = json_col.children.clone().unwrap();
986            let child_array = array_from_json(child_field, children[0].clone(), dictionaries)?;
987            let offsets: Vec<i32> = json_col
988                .offset
989                .unwrap()
990                .iter()
991                .map(|v| v.as_i64().unwrap() as i32)
992                .collect();
993            let array_data = ArrayData::builder(field.data_type().clone())
994                .len(json_col.count)
995                .add_buffer(Buffer::from(offsets.to_byte_slice()))
996                .add_child_data(child_array.into_data())
997                .null_bit_buffer(Some(null_buf))
998                .build()
999                .unwrap();
1000
1001            let array = MapArray::from(array_data);
1002            Ok(Arc::new(array))
1003        }
1004        DataType::Union(fields, _) => {
1005            let Some(type_ids) = json_col.type_id else {
1006                return Err(ArrowError::JsonError(
1007                    "Cannot find expected type_id in json column".to_string(),
1008                ));
1009            };
1010
1011            let offset: Option<ScalarBuffer<i32>> = json_col
1012                .offset
1013                .map(|offsets| offsets.iter().map(|v| v.as_i64().unwrap() as i32).collect());
1014
1015            let mut children = Vec::with_capacity(fields.len());
1016            for ((_, field), col) in fields.iter().zip(json_col.children.unwrap()) {
1017                let array = array_from_json(field, col, dictionaries)?;
1018                children.push(array);
1019            }
1020
1021            let array =
1022                UnionArray::try_new(fields.clone(), type_ids.into(), offset, children).unwrap();
1023            Ok(Arc::new(array))
1024        }
1025        DataType::Utf8View => {
1026            let views = json_col.views.ok_or_else(|| {
1027                ArrowError::JsonError("Utf8View requires VIEWS field".to_string())
1028            })?;
1029            let variadic_buffers = json_col.variadic_data_buffers.unwrap_or_default();
1030            let validity = json_col.validity.as_ref();
1031
1032            let mut builder = StringViewBuilder::new();
1033            for (i, view) in views.iter().enumerate() {
1034                let is_valid = validity.map_or(1, |v| v[i]);
1035                if is_valid == 0 {
1036                    builder.append_null();
1037                } else {
1038                    let view_obj = view.as_object().unwrap();
1039                    let size = view_obj["SIZE"].as_u64().unwrap() as usize;
1040                    // Check for INLINED key presence - inlined if SIZE <= 12
1041                    if let Some(inlined) = view_obj.get("INLINED") {
1042                        builder.append_value(inlined.as_str().unwrap());
1043                    } else {
1044                        // Reference to variadic buffer
1045                        let buffer_index = view_obj["BUFFER_INDEX"].as_u64().unwrap() as usize;
1046                        let offset = view_obj["OFFSET"].as_u64().unwrap() as usize;
1047                        let buffer_data = hex::decode(&variadic_buffers[buffer_index]).unwrap();
1048                        let s = std::str::from_utf8(&buffer_data[offset..offset + size]).unwrap();
1049                        builder.append_value(s);
1050                    }
1051                }
1052            }
1053            Ok(Arc::new(builder.finish()))
1054        }
1055        DataType::BinaryView => {
1056            let views = json_col.views.ok_or_else(|| {
1057                ArrowError::JsonError("BinaryView requires VIEWS field".to_string())
1058            })?;
1059            let variadic_buffers = json_col.variadic_data_buffers.unwrap_or_default();
1060            let validity = json_col.validity.as_ref();
1061
1062            let mut builder = BinaryViewBuilder::new();
1063            for (i, view) in views.iter().enumerate() {
1064                let is_valid = validity.map_or(1, |v| v[i]);
1065                if is_valid == 0 {
1066                    builder.append_null();
1067                } else {
1068                    let view_obj = view.as_object().unwrap();
1069                    let size = view_obj["SIZE"].as_u64().unwrap() as usize;
1070                    // Check for INLINED key presence - inlined if SIZE <= 12
1071                    if let Some(inlined) = view_obj.get("INLINED") {
1072                        let data = hex::decode(inlined.as_str().unwrap()).unwrap();
1073                        builder.append_value(&data);
1074                    } else {
1075                        // Reference to variadic buffer
1076                        let buffer_index = view_obj["BUFFER_INDEX"].as_u64().unwrap() as usize;
1077                        let offset = view_obj["OFFSET"].as_u64().unwrap() as usize;
1078                        let buffer_data = hex::decode(&variadic_buffers[buffer_index]).unwrap();
1079                        builder.append_value(&buffer_data[offset..offset + size]);
1080                    }
1081                }
1082            }
1083            Ok(Arc::new(builder.finish()))
1084        }
1085        DataType::RunEndEncoded(run_ends_field, values_field) => {
1086            let children = json_col.children.clone().unwrap();
1087            if children.len() != 2 {
1088                return Err(ArrowError::JsonError(
1089                    "RunEndEncoded requires exactly 2 children".to_string(),
1090                ));
1091            }
1092            let run_ends_array =
1093                array_from_json(run_ends_field, children[0].clone(), dictionaries)?;
1094            let values_array = array_from_json(values_field, children[1].clone(), dictionaries)?;
1095
1096            let run_array_data = ArrayData::builder(field.data_type().clone())
1097                .len(json_col.count)
1098                .add_child_data(run_ends_array.into_data())
1099                .add_child_data(values_array.into_data())
1100                .build()
1101                .unwrap();
1102
1103            Ok(make_array(run_array_data))
1104        }
1105        t => Err(ArrowError::JsonError(format!(
1106            "data type {t} not supported"
1107        ))),
1108    }
1109}
1110
1111/// Construct a [`DictionaryArray`] from a partially typed JSON column
1112pub fn dictionary_array_from_json(
1113    field: &Field,
1114    json_col: ArrowJsonColumn,
1115    dict_key: &DataType,
1116    dict_value: &DataType,
1117    dictionary: &ArrowJsonDictionaryBatch,
1118    dictionaries: Option<&HashMap<i64, ArrowJsonDictionaryBatch>>,
1119) -> Result<ArrayRef> {
1120    match dict_key {
1121        DataType::Int8
1122        | DataType::Int16
1123        | DataType::Int32
1124        | DataType::Int64
1125        | DataType::UInt8
1126        | DataType::UInt16
1127        | DataType::UInt32
1128        | DataType::UInt64 => {
1129            let null_buf = create_null_buf(&json_col);
1130
1131            // build the key data into a buffer, then construct values separately
1132            #[expect(deprecated)]
1133            let key_field = Field::new_dict(
1134                "key",
1135                dict_key.clone(),
1136                field.is_nullable(),
1137                #[expect(deprecated)]
1138                field
1139                    .dict_id()
1140                    .expect("Dictionary fields must have a dict_id value"),
1141                field
1142                    .dict_is_ordered()
1143                    .expect("Dictionary fields must have a dict_is_ordered value"),
1144            );
1145            let keys = array_from_json(&key_field, json_col, None)?;
1146            // note: not enough info on nullability of dictionary
1147            let value_field = Field::new("value", dict_value.clone(), true);
1148            let values = array_from_json(
1149                &value_field,
1150                dictionary.data.columns[0].clone(),
1151                dictionaries,
1152            )?;
1153
1154            // convert key and value to dictionary data
1155            let dict_data = ArrayData::builder(field.data_type().clone())
1156                .len(keys.len())
1157                .add_buffer(keys.to_data().buffers()[0].clone())
1158                .null_bit_buffer(Some(null_buf))
1159                .add_child_data(values.into_data())
1160                .build()
1161                .unwrap();
1162
1163            let array = match dict_key {
1164                DataType::Int8 => Arc::new(Int8DictionaryArray::from(dict_data)) as ArrayRef,
1165                DataType::Int16 => Arc::new(Int16DictionaryArray::from(dict_data)),
1166                DataType::Int32 => Arc::new(Int32DictionaryArray::from(dict_data)),
1167                DataType::Int64 => Arc::new(Int64DictionaryArray::from(dict_data)),
1168                DataType::UInt8 => Arc::new(UInt8DictionaryArray::from(dict_data)),
1169                DataType::UInt16 => Arc::new(UInt16DictionaryArray::from(dict_data)),
1170                DataType::UInt32 => Arc::new(UInt32DictionaryArray::from(dict_data)),
1171                DataType::UInt64 => Arc::new(UInt64DictionaryArray::from(dict_data)),
1172                _ => unreachable!(),
1173            };
1174            Ok(array)
1175        }
1176        _ => Err(ArrowError::JsonError(format!(
1177            "Dictionary key type {dict_key:?} not supported"
1178        ))),
1179    }
1180}
1181
1182/// A helper to create a null buffer from a `Vec<bool>`
1183fn create_null_buf(json_col: &ArrowJsonColumn) -> Buffer {
1184    let num_bytes = bit_util::ceil(json_col.count, 8);
1185    let mut null_buf = MutableBuffer::new(num_bytes).with_bitset(num_bytes, false);
1186    json_col
1187        .validity
1188        .clone()
1189        .unwrap()
1190        .iter()
1191        .enumerate()
1192        .for_each(|(i, v)| {
1193            let null_slice = null_buf.as_slice_mut();
1194            if *v != 0 {
1195                bit_util::set_bit(null_slice, i);
1196            }
1197        });
1198    null_buf.into()
1199}
1200
1201impl ArrowJsonBatch {
1202    /// Convert a [`RecordBatch`] to an [`ArrowJsonBatch`]
1203    ///
1204    /// <div class="warning">
1205    ///
1206    /// This function is **deliberately incomplete**! As noted in the crate-level documentation,
1207    /// this crate is only intended for use within the Arrow project itself.
1208    ///
1209    /// Right now, this function only supports `DataType::Int8` columns. Other data types will lead
1210    /// to an empty `ArrowJsonColumn`.
1211    ///
1212    /// </div>
1213    pub fn from_batch(batch: &RecordBatch) -> ArrowJsonBatch {
1214        let mut json_batch = ArrowJsonBatch {
1215            count: batch.num_rows(),
1216            columns: Vec::with_capacity(batch.num_columns()),
1217        };
1218
1219        for (col, field) in batch.columns().iter().zip(batch.schema().fields.iter()) {
1220            let json_col = match field.data_type() {
1221                DataType::Int8 => {
1222                    let col = col.as_any().downcast_ref::<Int8Array>().unwrap();
1223
1224                    let mut validity: Vec<u8> = Vec::with_capacity(col.len());
1225                    let mut data: Vec<Value> = Vec::with_capacity(col.len());
1226
1227                    for i in 0..col.len() {
1228                        if col.is_null(i) {
1229                            validity.push(1);
1230                            data.push(0i8.into());
1231                        } else {
1232                            validity.push(0);
1233                            data.push(col.value(i).into());
1234                        }
1235                    }
1236
1237                    ArrowJsonColumn {
1238                        name: field.name().clone(),
1239                        count: col.len(),
1240                        validity: Some(validity),
1241                        data: Some(data),
1242                        offset: None,
1243                        type_id: None,
1244                        size: None,
1245                        views: None,
1246                        variadic_data_buffers: None,
1247                        children: None,
1248                    }
1249                }
1250                _ => ArrowJsonColumn {
1251                    name: field.name().clone(),
1252                    count: col.len(),
1253                    validity: None,
1254                    data: None,
1255                    offset: None,
1256                    type_id: None,
1257                    size: None,
1258                    views: None,
1259                    variadic_data_buffers: None,
1260                    children: None,
1261                },
1262            };
1263
1264            json_batch.columns.push(json_col);
1265        }
1266
1267        json_batch
1268    }
1269}
1270
1271#[cfg(test)]
1272mod tests {
1273    use super::*;
1274
1275    #[test]
1276    fn test_decimal256_from_json() {
1277        let field = Field::new("c", DataType::Decimal256(76, 0), false);
1278        let col: ArrowJsonColumn = serde_json::from_str(
1279            r#"{
1280                "name": "c",
1281                "count": 5,
1282                "VALIDITY": [1, 1, 1, 1, 1],
1283                "DATA": [
1284                    "0",
1285                    "1",
1286                    "-1",
1287                    "123456789012345678901234567890",
1288                    "-123456789012345678901234567890"
1289                ]
1290            }"#,
1291        )
1292        .unwrap();
1293        let arr = array_from_json(&field, col, None).unwrap();
1294        let arr = arr.as_any().downcast_ref::<Decimal256Array>().unwrap();
1295        assert_eq!(arr.value(0), i256::ZERO);
1296        assert_eq!(arr.value(1), i256::from_i128(1));
1297        assert_eq!(arr.value(2), i256::from_i128(-1));
1298        let big = i256::from_string("123456789012345678901234567890").unwrap();
1299        assert_eq!(arr.value(3), big);
1300        assert_eq!(arr.value(4), i256::ZERO - big);
1301    }
1302
1303    #[test]
1304    fn test_schema_equality() {
1305        let json = r#"
1306        {
1307            "fields": [
1308                {
1309                    "name": "c1",
1310                    "type": {"name": "int", "isSigned": true, "bitWidth": 32},
1311                    "nullable": true,
1312                    "children": []
1313                },
1314                {
1315                    "name": "c2",
1316                    "type": {"name": "floatingpoint", "precision": "DOUBLE"},
1317                    "nullable": true,
1318                    "children": []
1319                },
1320                {
1321                    "name": "c3",
1322                    "type": {"name": "utf8"},
1323                    "nullable": true,
1324                    "children": []
1325                },
1326                {
1327                    "name": "c4",
1328                    "type": {
1329                        "name": "list"
1330                    },
1331                    "nullable": true,
1332                    "children": [
1333                        {
1334                            "name": "custom_item",
1335                            "type": {
1336                                "name": "int",
1337                                "isSigned": true,
1338                                "bitWidth": 32
1339                            },
1340                            "nullable": false,
1341                            "children": []
1342                        }
1343                    ]
1344                }
1345            ]
1346        }"#;
1347        let json_schema: ArrowJsonSchema = serde_json::from_str(json).unwrap();
1348        let schema = Schema::new(vec![
1349            Field::new("c1", DataType::Int32, true),
1350            Field::new("c2", DataType::Float64, true),
1351            Field::new("c3", DataType::Utf8, true),
1352            Field::new(
1353                "c4",
1354                DataType::List(Arc::new(Field::new("custom_item", DataType::Int32, false))),
1355                true,
1356            ),
1357        ]);
1358        assert!(json_schema.equals_schema(&schema));
1359    }
1360
1361    #[test]
1362    fn test_arrow_data_equality() {
1363        let secs_tz = Some("Europe/Budapest".into());
1364        let millis_tz = Some("America/New_York".into());
1365        let micros_tz = Some("UTC".into());
1366        let nanos_tz = Some("Africa/Johannesburg".into());
1367
1368        let schema = Schema::new(vec![
1369            Field::new("bools-with-metadata-map", DataType::Boolean, true)
1370                .with_metadata([("k", "v")]),
1371            Field::new("bools-with-metadata-vec", DataType::Boolean, true)
1372                .with_metadata([("k2", "v2")]),
1373            Field::new("bools", DataType::Boolean, true),
1374            Field::new("int8s", DataType::Int8, true),
1375            Field::new("int16s", DataType::Int16, true),
1376            Field::new("int32s", DataType::Int32, true),
1377            Field::new("int64s", DataType::Int64, true),
1378            Field::new("uint8s", DataType::UInt8, true),
1379            Field::new("uint16s", DataType::UInt16, true),
1380            Field::new("uint32s", DataType::UInt32, true),
1381            Field::new("uint64s", DataType::UInt64, true),
1382            Field::new("float32s", DataType::Float32, true),
1383            Field::new("float64s", DataType::Float64, true),
1384            Field::new("date_days", DataType::Date32, true),
1385            Field::new("date_millis", DataType::Date64, true),
1386            Field::new("time_secs", DataType::Time32(TimeUnit::Second), true),
1387            Field::new("time_millis", DataType::Time32(TimeUnit::Millisecond), true),
1388            Field::new("time_micros", DataType::Time64(TimeUnit::Microsecond), true),
1389            Field::new("time_nanos", DataType::Time64(TimeUnit::Nanosecond), true),
1390            Field::new("ts_secs", DataType::Timestamp(TimeUnit::Second, None), true),
1391            Field::new(
1392                "ts_millis",
1393                DataType::Timestamp(TimeUnit::Millisecond, None),
1394                true,
1395            ),
1396            Field::new(
1397                "ts_micros",
1398                DataType::Timestamp(TimeUnit::Microsecond, None),
1399                true,
1400            ),
1401            Field::new(
1402                "ts_nanos",
1403                DataType::Timestamp(TimeUnit::Nanosecond, None),
1404                true,
1405            ),
1406            Field::new(
1407                "ts_secs_tz",
1408                DataType::Timestamp(TimeUnit::Second, secs_tz.clone()),
1409                true,
1410            ),
1411            Field::new(
1412                "ts_millis_tz",
1413                DataType::Timestamp(TimeUnit::Millisecond, millis_tz.clone()),
1414                true,
1415            ),
1416            Field::new(
1417                "ts_micros_tz",
1418                DataType::Timestamp(TimeUnit::Microsecond, micros_tz.clone()),
1419                true,
1420            ),
1421            Field::new(
1422                "ts_nanos_tz",
1423                DataType::Timestamp(TimeUnit::Nanosecond, nanos_tz.clone()),
1424                true,
1425            ),
1426            Field::new("utf8s", DataType::Utf8, true),
1427            Field::new(
1428                "lists",
1429                DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))),
1430                true,
1431            ),
1432            Field::new(
1433                "structs",
1434                DataType::Struct(Fields::from(vec![
1435                    Field::new("int32s", DataType::Int32, true),
1436                    Field::new("utf8s", DataType::Utf8, true),
1437                ])),
1438                true,
1439            ),
1440            Field::new("utf8views", DataType::Utf8View, true),
1441            Field::new("binaryviews", DataType::BinaryView, true),
1442            Field::new(
1443                "listviews",
1444                DataType::ListView(Arc::new(Field::new_list_field(DataType::Int32, true))),
1445                true,
1446            ),
1447            Field::new(
1448                "largelistviews",
1449                DataType::LargeListView(Arc::new(Field::new_list_field(DataType::Int32, true))),
1450                true,
1451            ),
1452            Field::new(
1453                "runendencoded",
1454                DataType::RunEndEncoded(
1455                    Arc::new(Field::new("run_ends", DataType::Int16, false)),
1456                    Arc::new(Field::new("values", DataType::Int32, true)),
1457                ),
1458                true,
1459            ),
1460        ]);
1461
1462        let bools_with_metadata_map = BooleanArray::from(vec![Some(true), None, Some(false)]);
1463        let bools_with_metadata_vec = BooleanArray::from(vec![Some(true), None, Some(false)]);
1464        let bools = BooleanArray::from(vec![Some(true), None, Some(false)]);
1465        let int8s = Int8Array::from(vec![Some(1), None, Some(3)]);
1466        let int16s = Int16Array::from(vec![Some(1), None, Some(3)]);
1467        let int32s = Int32Array::from(vec![Some(1), None, Some(3)]);
1468        let int64s = Int64Array::from(vec![Some(1), None, Some(3)]);
1469        let uint8s = UInt8Array::from(vec![Some(1), None, Some(3)]);
1470        let uint16s = UInt16Array::from(vec![Some(1), None, Some(3)]);
1471        let uint32s = UInt32Array::from(vec![Some(1), None, Some(3)]);
1472        let uint64s = UInt64Array::from(vec![Some(1), None, Some(3)]);
1473        let float32s = Float32Array::from(vec![Some(1.0), None, Some(3.0)]);
1474        let float64s = Float64Array::from(vec![Some(1.0), None, Some(3.0)]);
1475        let date_days = Date32Array::from(vec![Some(1196848), None, None]);
1476        let date_millis = Date64Array::from(vec![
1477            Some(167903550396207),
1478            Some(29923997007884),
1479            Some(30612271819236),
1480        ]);
1481        let time_secs = Time32SecondArray::from(vec![Some(27974), Some(78592), Some(43207)]);
1482        let time_millis =
1483            Time32MillisecondArray::from(vec![Some(6613125), Some(74667230), Some(52260079)]);
1484        let time_micros = Time64MicrosecondArray::from(vec![Some(62522958593), None, None]);
1485        let time_nanos =
1486            Time64NanosecondArray::from(vec![Some(73380123595985), None, Some(16584393546415)]);
1487        let ts_secs = TimestampSecondArray::from(vec![None, Some(193438817552), None]);
1488        let ts_millis =
1489            TimestampMillisecondArray::from(vec![None, Some(38606916383008), Some(58113709376587)]);
1490        let ts_micros = TimestampMicrosecondArray::from(vec![None, None, None]);
1491        let ts_nanos = TimestampNanosecondArray::from(vec![None, None, Some(-6473623571954960143)]);
1492        let ts_secs_tz = TimestampSecondArray::from(vec![None, Some(193438817552), None])
1493            .with_timezone_opt(secs_tz);
1494        let ts_millis_tz =
1495            TimestampMillisecondArray::from(vec![None, Some(38606916383008), Some(58113709376587)])
1496                .with_timezone_opt(millis_tz);
1497        let ts_micros_tz =
1498            TimestampMicrosecondArray::from(vec![None, None, None]).with_timezone_opt(micros_tz);
1499        let ts_nanos_tz =
1500            TimestampNanosecondArray::from(vec![None, None, Some(-6473623571954960143)])
1501                .with_timezone_opt(nanos_tz);
1502        let utf8s = StringArray::from(vec![Some("aa"), None, Some("bbb")]);
1503
1504        let value_data = Int32Array::from(vec![None, Some(2), None, None]);
1505        let value_offsets = Buffer::from_slice_ref([0, 3, 4, 4]);
1506        let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true)));
1507        let list_data = ArrayData::builder(list_data_type)
1508            .len(3)
1509            .add_buffer(value_offsets)
1510            .add_child_data(value_data.into_data())
1511            .null_bit_buffer(Some(Buffer::from([0b00000011])))
1512            .build()
1513            .unwrap();
1514        let lists = ListArray::from(list_data);
1515
1516        let structs_int32s = Int32Array::from(vec![None, Some(-2), None]);
1517        let structs_utf8s = StringArray::from(vec![None, None, Some("aaaaaa")]);
1518        let struct_data_type = DataType::Struct(Fields::from(vec![
1519            Field::new("int32s", DataType::Int32, true),
1520            Field::new("utf8s", DataType::Utf8, true),
1521        ]));
1522        let struct_data = ArrayData::builder(struct_data_type)
1523            .len(3)
1524            .add_child_data(structs_int32s.into_data())
1525            .add_child_data(structs_utf8s.into_data())
1526            .null_bit_buffer(Some(Buffer::from([0b00000011])))
1527            .build()
1528            .unwrap();
1529        let structs = StructArray::from(struct_data);
1530
1531        let utf8views =
1532            StringViewArray::from(vec![Some("hello"), None, Some("this is not inlined")]);
1533        let binaryviews = BinaryViewArray::from_iter(vec![
1534            Some(b"\xf3\x4d".as_slice()),
1535            Some(b"\x00\x01\x02\x03\x04\x05\x06\x07\x08\x09\x0a\x0b\x0c\x0d\x0e\x0f".as_slice()),
1536            None,
1537        ]);
1538
1539        let listview_value_data = Int32Array::from(vec![Some(1), Some(2), Some(3), None, Some(5)]);
1540        let listview_offsets = Buffer::from_slice_ref([0i32, 2, 2]);
1541        let listview_sizes = Buffer::from_slice_ref([2i32, 0, 3]);
1542        let listview_data_type =
1543            DataType::ListView(Arc::new(Field::new_list_field(DataType::Int32, true)));
1544        let listview_data = ArrayData::builder(listview_data_type)
1545            .len(3)
1546            .add_buffer(listview_offsets)
1547            .add_buffer(listview_sizes)
1548            .add_child_data(listview_value_data.into_data())
1549            .null_bit_buffer(Some(Buffer::from([0b00000101])))
1550            .build()
1551            .unwrap();
1552        let listviews = ListViewArray::from(listview_data);
1553
1554        let largelistview_value_data = Int32Array::from(vec![Some(10), None, Some(30)]);
1555        let largelistview_offsets = Buffer::from_slice_ref([0i64, 2, 3]);
1556        let largelistview_sizes = Buffer::from_slice_ref([2i64, 1, 0]);
1557        let largelistview_data_type =
1558            DataType::LargeListView(Arc::new(Field::new_list_field(DataType::Int32, true)));
1559        let largelistview_data = ArrayData::builder(largelistview_data_type)
1560            .len(3)
1561            .add_buffer(largelistview_offsets)
1562            .add_buffer(largelistview_sizes)
1563            .add_child_data(largelistview_value_data.into_data())
1564            .null_bit_buffer(Some(Buffer::from([0b00000011])))
1565            .build()
1566            .unwrap();
1567        let largelistviews = LargeListViewArray::from(largelistview_data);
1568
1569        let ree_run_ends = Int16Array::from(vec![2, 3]);
1570        let ree_values = Int32Array::from(vec![Some(100), None]);
1571        let ree_data_type = DataType::RunEndEncoded(
1572            Arc::new(Field::new("run_ends", DataType::Int16, false)),
1573            Arc::new(Field::new("values", DataType::Int32, true)),
1574        );
1575        let ree_data = ArrayData::builder(ree_data_type)
1576            .len(3)
1577            .add_child_data(ree_run_ends.into_data())
1578            .add_child_data(ree_values.into_data())
1579            .build()
1580            .unwrap();
1581        let runendencoded = RunArray::<Int16Type>::from(ree_data);
1582
1583        let record_batch = RecordBatch::try_new(
1584            Arc::new(schema.clone()),
1585            vec![
1586                Arc::new(bools_with_metadata_map),
1587                Arc::new(bools_with_metadata_vec),
1588                Arc::new(bools),
1589                Arc::new(int8s),
1590                Arc::new(int16s),
1591                Arc::new(int32s),
1592                Arc::new(int64s),
1593                Arc::new(uint8s),
1594                Arc::new(uint16s),
1595                Arc::new(uint32s),
1596                Arc::new(uint64s),
1597                Arc::new(float32s),
1598                Arc::new(float64s),
1599                Arc::new(date_days),
1600                Arc::new(date_millis),
1601                Arc::new(time_secs),
1602                Arc::new(time_millis),
1603                Arc::new(time_micros),
1604                Arc::new(time_nanos),
1605                Arc::new(ts_secs),
1606                Arc::new(ts_millis),
1607                Arc::new(ts_micros),
1608                Arc::new(ts_nanos),
1609                Arc::new(ts_secs_tz),
1610                Arc::new(ts_millis_tz),
1611                Arc::new(ts_micros_tz),
1612                Arc::new(ts_nanos_tz),
1613                Arc::new(utf8s),
1614                Arc::new(lists),
1615                Arc::new(structs),
1616                Arc::new(utf8views),
1617                Arc::new(binaryviews),
1618                Arc::new(listviews),
1619                Arc::new(largelistviews),
1620                Arc::new(runendencoded),
1621            ],
1622        )
1623        .unwrap();
1624        let json = std::fs::read_to_string("data/integration.json").unwrap();
1625        let arrow_json: ArrowJson = serde_json::from_str(&json).unwrap();
1626        // test schemas
1627        assert!(arrow_json.schema.equals_schema(&schema));
1628        // test record batch
1629        assert_eq!(arrow_json.get_record_batches().unwrap()[0], record_batch);
1630    }
1631}
arrow_integration_test/lib.rs

arrow_integration_test/
lib.rs
