Skip to main content

lance_namespace/
schema.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright The Lance Authors
3
4//! Schema conversion utilities for Lance Namespace.
5//!
6//! This module provides functions to convert between JsonArrow schema representations
7//! and Arrow schema types.
8
9use arrow::datatypes::{DataType, Field, Schema as ArrowSchema};
10use lance_core::{Error, Result};
11use lance_namespace_reqwest_client::models::{JsonArrowDataType, JsonArrowField, JsonArrowSchema};
12
13/// Convert Arrow Schema to JsonArrowSchema
14pub fn arrow_schema_to_json(arrow_schema: &ArrowSchema) -> Result<JsonArrowSchema> {
15    let fields: Result<Vec<JsonArrowField>> = arrow_schema
16        .fields()
17        .iter()
18        .map(|f| arrow_field_to_json(f.as_ref()))
19        .collect();
20
21    let metadata = if arrow_schema.metadata().is_empty() {
22        None
23    } else {
24        Some(arrow_schema.metadata().clone())
25    };
26
27    Ok(JsonArrowSchema {
28        fields: fields?,
29        metadata,
30    })
31}
32
33/// Convert Arrow Field to JsonArrowField
34fn arrow_field_to_json(arrow_field: &Field) -> Result<JsonArrowField> {
35    let data_type = arrow_type_to_json(arrow_field.data_type())?;
36
37    Ok(JsonArrowField {
38        name: arrow_field.name().clone(),
39        nullable: arrow_field.is_nullable(),
40        r#type: Box::new(data_type),
41        metadata: if arrow_field.metadata().is_empty() {
42            None
43        } else {
44            Some(arrow_field.metadata().clone())
45        },
46    })
47}
48
49/// Convert Arrow DataType to JsonArrowDataType
50fn arrow_type_to_json(data_type: &DataType) -> Result<JsonArrowDataType> {
51    match data_type {
52        // Primitive types
53        DataType::Null => Ok(JsonArrowDataType::new("null".to_string())),
54        DataType::Boolean => Ok(JsonArrowDataType::new("bool".to_string())),
55        DataType::Int8 => Ok(JsonArrowDataType::new("int8".to_string())),
56        DataType::UInt8 => Ok(JsonArrowDataType::new("uint8".to_string())),
57        DataType::Int16 => Ok(JsonArrowDataType::new("int16".to_string())),
58        DataType::UInt16 => Ok(JsonArrowDataType::new("uint16".to_string())),
59        DataType::Int32 => Ok(JsonArrowDataType::new("int32".to_string())),
60        DataType::UInt32 => Ok(JsonArrowDataType::new("uint32".to_string())),
61        DataType::Int64 => Ok(JsonArrowDataType::new("int64".to_string())),
62        DataType::UInt64 => Ok(JsonArrowDataType::new("uint64".to_string())),
63        DataType::Float16 => Ok(JsonArrowDataType::new("float16".to_string())),
64        DataType::Float32 => Ok(JsonArrowDataType::new("float32".to_string())),
65        DataType::Float64 => Ok(JsonArrowDataType::new("float64".to_string())),
66        DataType::Decimal32(precision, scale) => {
67            let mut dt = JsonArrowDataType::new("decimal32".to_string());
68            dt.length = Some(*precision as i64 * 1000 + *scale as i64); // Encode precision and scale
69            Ok(dt)
70        }
71        DataType::Decimal64(precision, scale) => {
72            let mut dt = JsonArrowDataType::new("decimal64".to_string());
73            dt.length = Some(*precision as i64 * 1000 + *scale as i64); // Encode precision and scale
74            Ok(dt)
75        }
76        DataType::Decimal128(precision, scale) => {
77            let mut dt = JsonArrowDataType::new("decimal128".to_string());
78            dt.length = Some(*precision as i64 * 1000 + *scale as i64); // Encode precision and scale
79            Ok(dt)
80        }
81        DataType::Decimal256(precision, scale) => {
82            let mut dt = JsonArrowDataType::new("decimal256".to_string());
83            dt.length = Some(*precision as i64 * 1000 + *scale as i64); // Encode precision and scale
84            Ok(dt)
85        }
86        DataType::Date32 => Ok(JsonArrowDataType::new("date32".to_string())),
87        DataType::Date64 => Ok(JsonArrowDataType::new("date64".to_string())),
88        DataType::Time32(_) => Ok(JsonArrowDataType::new("time32".to_string())),
89        DataType::Time64(_) => Ok(JsonArrowDataType::new("time64".to_string())),
90        DataType::Timestamp(_, _tz) => {
91            // TODO: We could encode timezone info if needed
92            Ok(JsonArrowDataType::new("timestamp".to_string()))
93        }
94        DataType::Duration(_) => Ok(JsonArrowDataType::new("duration".to_string())),
95        DataType::Interval(_) => Ok(JsonArrowDataType::new("interval".to_string())),
96
97        // String and Binary types
98        DataType::Utf8 => Ok(JsonArrowDataType::new("utf8".to_string())),
99        DataType::LargeUtf8 => Ok(JsonArrowDataType::new("large_utf8".to_string())),
100        DataType::Binary => Ok(JsonArrowDataType::new("binary".to_string())),
101        DataType::LargeBinary => Ok(JsonArrowDataType::new("large_binary".to_string())),
102        DataType::FixedSizeBinary(size) => {
103            let mut dt = JsonArrowDataType::new("fixed_size_binary".to_string());
104            dt.length = Some(*size as i64);
105            Ok(dt)
106        }
107
108        // Nested types
109        DataType::List(field) => {
110            let inner_type = arrow_type_to_json(field.data_type())?;
111            let inner_field = JsonArrowField {
112                name: field.name().clone(),
113                nullable: field.is_nullable(),
114                r#type: Box::new(inner_type),
115                metadata: if field.metadata().is_empty() {
116                    None
117                } else {
118                    Some(field.metadata().clone())
119                },
120            };
121            Ok(JsonArrowDataType {
122                r#type: "list".to_string(),
123                fields: Some(vec![inner_field]),
124                length: None,
125            })
126        }
127        DataType::LargeList(field) => {
128            let inner_type = arrow_type_to_json(field.data_type())?;
129            let inner_field = JsonArrowField {
130                name: field.name().clone(),
131                nullable: field.is_nullable(),
132                r#type: Box::new(inner_type),
133                metadata: if field.metadata().is_empty() {
134                    None
135                } else {
136                    Some(field.metadata().clone())
137                },
138            };
139            Ok(JsonArrowDataType {
140                r#type: "large_list".to_string(),
141                fields: Some(vec![inner_field]),
142                length: None,
143            })
144        }
145        DataType::FixedSizeList(field, size) => {
146            let inner_type = arrow_type_to_json(field.data_type())?;
147            let inner_field = JsonArrowField {
148                name: field.name().clone(),
149                nullable: field.is_nullable(),
150                r#type: Box::new(inner_type),
151                metadata: if field.metadata().is_empty() {
152                    None
153                } else {
154                    Some(field.metadata().clone())
155                },
156            };
157            Ok(JsonArrowDataType {
158                r#type: "fixed_size_list".to_string(),
159                fields: Some(vec![inner_field]),
160                length: Some(*size as i64),
161            })
162        }
163        DataType::Struct(fields) => {
164            let json_fields: Result<Vec<JsonArrowField>> = fields
165                .iter()
166                .map(|f| arrow_field_to_json(f.as_ref()))
167                .collect();
168            Ok(JsonArrowDataType {
169                r#type: "struct".to_string(),
170                fields: Some(json_fields?),
171                length: None,
172            })
173        }
174        DataType::Union(_, _) => {
175            // Union types are complex, for now we'll skip detailed conversion
176            Ok(JsonArrowDataType::new("union".to_string()))
177        }
178        DataType::Dictionary(_, value_type) => {
179            // For dictionary, return the value type
180            arrow_type_to_json(value_type)
181        }
182
183        DataType::Map(entries_field, keys_sorted) => {
184            if *keys_sorted {
185                return Err(Error::namespace(format!(
186                    "Map types with keys_sorted=true are not yet supported for JSON conversion: {:?}",
187                    data_type
188                )));
189            }
190            let inner_type = arrow_type_to_json(entries_field.data_type())?;
191            let inner_field = JsonArrowField {
192                name: entries_field.name().clone(),
193                nullable: entries_field.is_nullable(),
194                r#type: Box::new(inner_type),
195                metadata: if entries_field.metadata().is_empty() {
196                    None
197                } else {
198                    Some(entries_field.metadata().clone())
199                },
200            };
201            Ok(JsonArrowDataType {
202                r#type: "map".to_string(),
203                fields: Some(vec![inner_field]),
204                length: None,
205            })
206        }
207
208        // Unsupported types
209        DataType::RunEndEncoded(_, _) => Err(Error::namespace(format!(
210            "RunEndEncoded type is not yet supported for JSON conversion: {:?}",
211            data_type
212        ))),
213        DataType::ListView(_) | DataType::LargeListView(_) => Err(Error::namespace(format!(
214            "ListView types are not yet supported for JSON conversion: {:?}",
215            data_type
216        ))),
217        DataType::Utf8View | DataType::BinaryView => Err(Error::namespace(format!(
218            "View types are not yet supported for JSON conversion: {:?}",
219            data_type
220        ))),
221    }
222}
223
224/// Convert JsonArrowSchema to Arrow Schema
225pub fn convert_json_arrow_schema(json_schema: &JsonArrowSchema) -> Result<ArrowSchema> {
226    let fields: Result<Vec<Field>> = json_schema
227        .fields
228        .iter()
229        .map(convert_json_arrow_field)
230        .collect();
231
232    let metadata = json_schema.metadata.as_ref().cloned().unwrap_or_default();
233
234    Ok(ArrowSchema::new_with_metadata(fields?, metadata))
235}
236
237/// Convert JsonArrowField to Arrow Field
238pub fn convert_json_arrow_field(json_field: &JsonArrowField) -> Result<Field> {
239    let data_type = convert_json_arrow_type(&json_field.r#type)?;
240    let nullable = json_field.nullable;
241
242    let field = Field::new(&json_field.name, data_type, nullable);
243    Ok(match json_field.metadata.as_ref() {
244        Some(metadata) => field.with_metadata(metadata.clone()),
245        None => field,
246    })
247}
248
249/// Convert JsonArrowDataType to Arrow DataType
250pub fn convert_json_arrow_type(json_type: &JsonArrowDataType) -> Result<DataType> {
251    let type_name = json_type.r#type.to_lowercase();
252
253    match type_name.as_str() {
254        "null" => Ok(DataType::Null),
255        "bool" | "boolean" => Ok(DataType::Boolean),
256        "int8" => Ok(DataType::Int8),
257        "uint8" => Ok(DataType::UInt8),
258        "int16" => Ok(DataType::Int16),
259        "uint16" => Ok(DataType::UInt16),
260        "int32" => Ok(DataType::Int32),
261        "uint32" => Ok(DataType::UInt32),
262        "int64" => Ok(DataType::Int64),
263        "uint64" => Ok(DataType::UInt64),
264        "float32" => Ok(DataType::Float32),
265        "float64" => Ok(DataType::Float64),
266        "utf8" => Ok(DataType::Utf8),
267        "binary" => Ok(DataType::Binary),
268        _ => Err(Error::namespace(format!(
269            "Unsupported Arrow type: {}",
270            type_name
271        ))),
272    }
273}
274
275#[cfg(test)]
276mod tests {
277    use super::*;
278    use std::collections::HashMap;
279    use std::sync::Arc;
280
281    #[test]
282    fn test_extension_metadata_preserved_in_json_roundtrip() {
283        const ARROW_EXT_NAME_KEY: &str = "ARROW:extension:name";
284        const LANCE_JSON_EXT_NAME: &str = "lance.json";
285
286        let meta_field =
287            Field::new("meta", DataType::Binary, true).with_metadata(HashMap::from([(
288                ARROW_EXT_NAME_KEY.to_string(),
289                LANCE_JSON_EXT_NAME.to_string(),
290            )]));
291        let arrow_schema =
292            ArrowSchema::new(vec![Field::new("id", DataType::Int32, false), meta_field]);
293
294        let json_schema = arrow_schema_to_json(&arrow_schema).unwrap();
295        let meta_json_field = json_schema
296            .fields
297            .iter()
298            .find(|f| f.name == "meta")
299            .unwrap();
300        assert!(
301            meta_json_field
302                .metadata
303                .as_ref()
304                .unwrap()
305                .contains_key(ARROW_EXT_NAME_KEY)
306        );
307
308        let roundtrip = convert_json_arrow_schema(&json_schema).unwrap();
309        let meta_field = roundtrip.field_with_name("meta").unwrap();
310        assert_eq!(
311            meta_field.metadata().get(ARROW_EXT_NAME_KEY),
312            Some(&LANCE_JSON_EXT_NAME.to_string())
313        );
314    }
315
316    #[test]
317    fn test_convert_basic_types() {
318        // Test int32
319        let int_type = JsonArrowDataType::new("int32".to_string());
320        let result = convert_json_arrow_type(&int_type).unwrap();
321        assert_eq!(result, DataType::Int32);
322
323        // Test utf8
324        let string_type = JsonArrowDataType::new("utf8".to_string());
325        let result = convert_json_arrow_type(&string_type).unwrap();
326        assert_eq!(result, DataType::Utf8);
327
328        // Test float64
329        let float_type = JsonArrowDataType::new("float64".to_string());
330        let result = convert_json_arrow_type(&float_type).unwrap();
331        assert_eq!(result, DataType::Float64);
332
333        // Test binary
334        let binary_type = JsonArrowDataType::new("binary".to_string());
335        let result = convert_json_arrow_type(&binary_type).unwrap();
336        assert_eq!(result, DataType::Binary);
337    }
338
339    #[test]
340    fn test_convert_field() {
341        let int_type = JsonArrowDataType::new("int32".to_string());
342        let field = JsonArrowField {
343            name: "test_field".to_string(),
344            r#type: Box::new(int_type),
345            nullable: false,
346            metadata: None,
347        };
348
349        let result = convert_json_arrow_field(&field).unwrap();
350        assert_eq!(result.name(), "test_field");
351        assert_eq!(result.data_type(), &DataType::Int32);
352        assert!(!result.is_nullable());
353    }
354
355    #[test]
356    fn test_convert_schema() {
357        let int_type = JsonArrowDataType::new("int32".to_string());
358        let string_type = JsonArrowDataType::new("utf8".to_string());
359
360        let id_field = JsonArrowField {
361            name: "id".to_string(),
362            r#type: Box::new(int_type),
363            nullable: false,
364            metadata: None,
365        };
366
367        let name_field = JsonArrowField {
368            name: "name".to_string(),
369            r#type: Box::new(string_type),
370            nullable: true,
371            metadata: None,
372        };
373
374        let mut metadata = HashMap::new();
375        metadata.insert("key".to_string(), "value".to_string());
376
377        let schema = JsonArrowSchema {
378            fields: vec![id_field, name_field],
379            metadata: Some(metadata.clone()),
380        };
381
382        let result = convert_json_arrow_schema(&schema).unwrap();
383        assert_eq!(result.fields().len(), 2);
384        assert_eq!(result.field(0).name(), "id");
385        assert_eq!(result.field(1).name(), "name");
386        assert_eq!(result.metadata(), &metadata);
387    }
388
389    #[test]
390    fn test_unsupported_type() {
391        let unsupported_type = JsonArrowDataType::new("unsupported".to_string());
392        let result = convert_json_arrow_type(&unsupported_type);
393        assert!(result.is_err());
394        assert!(
395            result
396                .unwrap_err()
397                .to_string()
398                .contains("Unsupported Arrow type")
399        );
400    }
401
402    #[test]
403    fn test_list_type() {
404        use arrow::datatypes::Field;
405
406        let inner_field = Field::new("item", DataType::Int32, true);
407        let list_type = DataType::List(Arc::new(inner_field));
408
409        let result = arrow_type_to_json(&list_type).unwrap();
410        assert_eq!(result.r#type, "list");
411        assert!(result.fields.is_some());
412        let fields = result.fields.unwrap();
413        assert_eq!(fields.len(), 1);
414        assert_eq!(fields[0].name, "item");
415        assert_eq!(fields[0].r#type.r#type, "int32");
416    }
417
418    #[test]
419    fn test_struct_type() {
420        use arrow::datatypes::Field;
421
422        let fields = vec![
423            Field::new("id", DataType::Int64, false),
424            Field::new("name", DataType::Utf8, true),
425        ];
426        let struct_type = DataType::Struct(fields.into());
427
428        let result = arrow_type_to_json(&struct_type).unwrap();
429        assert_eq!(result.r#type, "struct");
430        assert!(result.fields.is_some());
431        let json_fields = result.fields.unwrap();
432        assert_eq!(json_fields.len(), 2);
433        assert_eq!(json_fields[0].name, "id");
434        assert_eq!(json_fields[0].r#type.r#type, "int64");
435        assert_eq!(json_fields[1].name, "name");
436        assert_eq!(json_fields[1].r#type.r#type, "utf8");
437    }
438
439    #[test]
440    fn test_fixed_size_list_type() {
441        use arrow::datatypes::Field;
442
443        let inner_field = Field::new("item", DataType::Float32, false);
444        let fixed_list_type = DataType::FixedSizeList(Arc::new(inner_field), 3);
445
446        let result = arrow_type_to_json(&fixed_list_type).unwrap();
447        assert_eq!(result.r#type, "fixed_size_list");
448        assert_eq!(result.length, Some(3));
449        assert!(result.fields.is_some());
450        let fields = result.fields.unwrap();
451        assert_eq!(fields.len(), 1);
452        assert_eq!(fields[0].r#type.r#type, "float32");
453    }
454
455    #[test]
456    fn test_nested_struct_with_list() {
457        use arrow::datatypes::Field;
458
459        let inner_list_field = Field::new("item", DataType::Utf8, true);
460        let list_type = DataType::List(Arc::new(inner_list_field));
461
462        let struct_fields = vec![
463            Field::new("id", DataType::Int32, false),
464            Field::new("tags", list_type, true),
465        ];
466        let struct_type = DataType::Struct(struct_fields.into());
467
468        let result = arrow_type_to_json(&struct_type).unwrap();
469        assert_eq!(result.r#type, "struct");
470        let json_fields = result.fields.unwrap();
471        assert_eq!(json_fields.len(), 2);
472        assert_eq!(json_fields[0].name, "id");
473        assert_eq!(json_fields[1].name, "tags");
474        assert_eq!(json_fields[1].r#type.r#type, "list");
475
476        // Check nested list structure
477        let list_fields = json_fields[1].r#type.fields.as_ref().unwrap();
478        assert_eq!(list_fields.len(), 1);
479        assert_eq!(list_fields[0].r#type.r#type, "utf8");
480    }
481
482    #[test]
483    fn test_map_type_supported() {
484        use arrow::datatypes::Field;
485
486        let key_field = Field::new("keys", DataType::Utf8, false);
487        let value_field = Field::new("values", DataType::Int32, true);
488        let map_type = DataType::Map(
489            Arc::new(Field::new(
490                "entries",
491                DataType::Struct(vec![key_field, value_field].into()),
492                false,
493            )),
494            false,
495        );
496
497        let result = arrow_type_to_json(&map_type);
498        assert!(result.is_ok());
499        let json_type = result.unwrap();
500        assert_eq!(json_type.r#type, "map");
501        assert!(json_type.fields.is_some());
502
503        let fields = json_type.fields.unwrap();
504        assert_eq!(fields.len(), 1);
505        assert_eq!(fields[0].name, "entries");
506        assert_eq!(fields[0].r#type.r#type, "struct");
507    }
508
509    #[test]
510    fn test_additional_types() {
511        // Test Date types
512        let date32 = arrow_type_to_json(&DataType::Date32).unwrap();
513        assert_eq!(date32.r#type, "date32");
514
515        let date64 = arrow_type_to_json(&DataType::Date64).unwrap();
516        assert_eq!(date64.r#type, "date64");
517
518        // Test FixedSizeBinary
519        let fixed_binary = arrow_type_to_json(&DataType::FixedSizeBinary(16)).unwrap();
520        assert_eq!(fixed_binary.r#type, "fixed_size_binary");
521        assert_eq!(fixed_binary.length, Some(16));
522
523        // Test Float16
524        let float16 = arrow_type_to_json(&DataType::Float16).unwrap();
525        assert_eq!(float16.r#type, "float16");
526    }
527}