parquet 0.5.12 → 0.6.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/Cargo.lock +295 -98
- data/Cargo.toml +1 -1
- data/Gemfile +1 -0
- data/README.md +94 -3
- data/ext/parquet/Cargo.toml +8 -5
- data/ext/parquet/src/adapter_ffi.rs +156 -0
- data/ext/parquet/src/lib.rs +13 -21
- data/ext/parquet-core/Cargo.toml +23 -0
- data/ext/parquet-core/src/arrow_conversion.rs +1133 -0
- data/ext/parquet-core/src/error.rs +163 -0
- data/ext/parquet-core/src/lib.rs +60 -0
- data/ext/parquet-core/src/reader.rs +263 -0
- data/ext/parquet-core/src/schema.rs +283 -0
- data/ext/parquet-core/src/test_utils.rs +308 -0
- data/ext/parquet-core/src/traits/mod.rs +5 -0
- data/ext/parquet-core/src/traits/schema.rs +151 -0
- data/ext/parquet-core/src/value.rs +209 -0
- data/ext/parquet-core/src/writer.rs +839 -0
- data/ext/parquet-core/tests/arrow_conversion_tests.rs +423 -0
- data/ext/parquet-core/tests/binary_data.rs +437 -0
- data/ext/parquet-core/tests/column_projection.rs +557 -0
- data/ext/parquet-core/tests/complex_types.rs +821 -0
- data/ext/parquet-core/tests/compression_tests.rs +434 -0
- data/ext/parquet-core/tests/concurrent_access.rs +430 -0
- data/ext/parquet-core/tests/decimal_tests.rs +488 -0
- data/ext/parquet-core/tests/edge_cases_corner_cases.rs +322 -0
- data/ext/parquet-core/tests/error_handling_comprehensive_tests.rs +547 -0
- data/ext/parquet-core/tests/null_handling_tests.rs +430 -0
- data/ext/parquet-core/tests/performance_memory.rs +181 -0
- data/ext/parquet-core/tests/primitive_types.rs +547 -0
- data/ext/parquet-core/tests/real_world_patterns.rs +777 -0
- data/ext/parquet-core/tests/roundtrip_correctness.rs +279 -0
- data/ext/parquet-core/tests/schema_comprehensive_tests.rs +534 -0
- data/ext/parquet-core/tests/temporal_tests.rs +518 -0
- data/ext/parquet-core/tests/test_helpers.rs +132 -0
- data/ext/parquet-core/tests/writer_tests.rs +545 -0
- data/ext/parquet-ruby-adapter/Cargo.toml +22 -0
- data/ext/parquet-ruby-adapter/build.rs +5 -0
- data/ext/parquet-ruby-adapter/examples/try_into_value_demo.rs +98 -0
- data/ext/parquet-ruby-adapter/src/batch_manager.rs +116 -0
- data/ext/parquet-ruby-adapter/src/chunk_reader.rs +237 -0
- data/ext/parquet-ruby-adapter/src/converter.rs +1685 -0
- data/ext/parquet-ruby-adapter/src/error.rs +148 -0
- data/ext/{parquet/src/ruby_reader.rs → parquet-ruby-adapter/src/io.rs} +190 -56
- data/ext/parquet-ruby-adapter/src/lib.rs +90 -0
- data/ext/parquet-ruby-adapter/src/logger.rs +64 -0
- data/ext/parquet-ruby-adapter/src/metadata.rs +427 -0
- data/ext/parquet-ruby-adapter/src/reader.rs +317 -0
- data/ext/parquet-ruby-adapter/src/schema.rs +810 -0
- data/ext/parquet-ruby-adapter/src/string_cache.rs +106 -0
- data/ext/parquet-ruby-adapter/src/try_into_value.rs +91 -0
- data/ext/parquet-ruby-adapter/src/types.rs +94 -0
- data/ext/parquet-ruby-adapter/src/utils.rs +186 -0
- data/ext/parquet-ruby-adapter/src/writer.rs +435 -0
- data/lib/parquet/schema.rb +19 -0
- data/lib/parquet/version.rb +1 -1
- metadata +50 -24
- data/ext/parquet/src/enumerator.rs +0 -68
- data/ext/parquet/src/header_cache.rs +0 -99
- data/ext/parquet/src/logger.rs +0 -171
- data/ext/parquet/src/reader/common.rs +0 -111
- data/ext/parquet/src/reader/mod.rs +0 -211
- data/ext/parquet/src/reader/parquet_column_reader.rs +0 -44
- data/ext/parquet/src/reader/parquet_row_reader.rs +0 -43
- data/ext/parquet/src/reader/unified/mod.rs +0 -363
- data/ext/parquet/src/types/core_types.rs +0 -120
- data/ext/parquet/src/types/mod.rs +0 -100
- data/ext/parquet/src/types/parquet_value.rs +0 -1275
- data/ext/parquet/src/types/record_types.rs +0 -603
- data/ext/parquet/src/types/schema_converter.rs +0 -290
- data/ext/parquet/src/types/schema_node.rs +0 -424
- data/ext/parquet/src/types/timestamp.rs +0 -285
- data/ext/parquet/src/types/type_conversion.rs +0 -1949
- data/ext/parquet/src/types/writer_types.rs +0 -329
- data/ext/parquet/src/utils.rs +0 -184
- data/ext/parquet/src/writer/mod.rs +0 -505
- data/ext/parquet/src/writer/write_columns.rs +0 -238
- data/ext/parquet/src/writer/write_rows.rs +0 -488
@@ -0,0 +1,821 @@
|
|
1
|
+
use bytes::Bytes;
|
2
|
+
use indexmap::IndexMap;
|
3
|
+
use num::BigInt;
|
4
|
+
use ordered_float::OrderedFloat;
|
5
|
+
use parquet_core::*;
|
6
|
+
use std::sync::Arc;
|
7
|
+
|
8
|
+
#[test]
|
9
|
+
fn test_write_and_read_lists() {
|
10
|
+
// Create schema with list fields
|
11
|
+
let schema = SchemaBuilder::new()
|
12
|
+
.with_root(SchemaNode::Struct {
|
13
|
+
name: "root".to_string(),
|
14
|
+
nullable: false,
|
15
|
+
fields: vec![
|
16
|
+
SchemaNode::Primitive {
|
17
|
+
name: "id".to_string(),
|
18
|
+
primitive_type: PrimitiveType::Int32,
|
19
|
+
nullable: false,
|
20
|
+
format: None,
|
21
|
+
},
|
22
|
+
SchemaNode::List {
|
23
|
+
name: "tags".to_string(),
|
24
|
+
nullable: true,
|
25
|
+
item: Box::new(SchemaNode::Primitive {
|
26
|
+
name: "tag".to_string(),
|
27
|
+
primitive_type: PrimitiveType::String,
|
28
|
+
nullable: false,
|
29
|
+
format: None,
|
30
|
+
}),
|
31
|
+
},
|
32
|
+
SchemaNode::List {
|
33
|
+
name: "scores".to_string(),
|
34
|
+
nullable: false,
|
35
|
+
item: Box::new(SchemaNode::Primitive {
|
36
|
+
name: "score".to_string(),
|
37
|
+
primitive_type: PrimitiveType::Float64,
|
38
|
+
nullable: true,
|
39
|
+
format: None,
|
40
|
+
}),
|
41
|
+
},
|
42
|
+
],
|
43
|
+
})
|
44
|
+
.build()
|
45
|
+
.unwrap();
|
46
|
+
|
47
|
+
// Create test data with various list scenarios
|
48
|
+
let rows = vec![
|
49
|
+
// Row with populated lists
|
50
|
+
vec![
|
51
|
+
ParquetValue::Int32(1),
|
52
|
+
ParquetValue::List(vec![
|
53
|
+
ParquetValue::String(Arc::from("rust")),
|
54
|
+
ParquetValue::String(Arc::from("parquet")),
|
55
|
+
ParquetValue::String(Arc::from("ffi")),
|
56
|
+
]),
|
57
|
+
ParquetValue::List(vec![
|
58
|
+
ParquetValue::Float64(OrderedFloat(95.5)),
|
59
|
+
ParquetValue::Float64(OrderedFloat(87.3)),
|
60
|
+
ParquetValue::Null,
|
61
|
+
]),
|
62
|
+
],
|
63
|
+
// Row with empty lists
|
64
|
+
vec![
|
65
|
+
ParquetValue::Int32(2),
|
66
|
+
ParquetValue::List(vec![]),
|
67
|
+
ParquetValue::List(vec![]),
|
68
|
+
],
|
69
|
+
// Row with null list
|
70
|
+
vec![
|
71
|
+
ParquetValue::Int32(3),
|
72
|
+
ParquetValue::Null,
|
73
|
+
ParquetValue::List(vec![ParquetValue::Float64(OrderedFloat(100.0))]),
|
74
|
+
],
|
75
|
+
// Row with single-element lists
|
76
|
+
vec![
|
77
|
+
ParquetValue::Int32(4),
|
78
|
+
ParquetValue::List(vec![ParquetValue::String(Arc::from("single"))]),
|
79
|
+
ParquetValue::List(vec![ParquetValue::Float64(OrderedFloat(42.0))]),
|
80
|
+
],
|
81
|
+
];
|
82
|
+
|
83
|
+
// Write to buffer
|
84
|
+
let mut buffer = Vec::new();
|
85
|
+
{
|
86
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
87
|
+
writer.write_rows(rows.clone()).unwrap();
|
88
|
+
writer.close().unwrap();
|
89
|
+
}
|
90
|
+
|
91
|
+
// Read back and verify
|
92
|
+
let bytes = Bytes::from(buffer);
|
93
|
+
let reader = Reader::new(bytes);
|
94
|
+
|
95
|
+
let read_rows: Vec<_> = reader
|
96
|
+
.read_rows()
|
97
|
+
.unwrap()
|
98
|
+
.collect::<Result<Vec<_>>>()
|
99
|
+
.unwrap();
|
100
|
+
|
101
|
+
assert_eq!(rows.len(), read_rows.len());
|
102
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
103
|
+
assert_eq!(expected, actual);
|
104
|
+
}
|
105
|
+
}
|
106
|
+
|
107
|
+
#[test]
|
108
|
+
fn test_write_and_read_maps() {
|
109
|
+
// Create schema with map field
|
110
|
+
let schema = SchemaBuilder::new()
|
111
|
+
.with_root(SchemaNode::Struct {
|
112
|
+
name: "root".to_string(),
|
113
|
+
nullable: false,
|
114
|
+
fields: vec![
|
115
|
+
SchemaNode::Primitive {
|
116
|
+
name: "user_id".to_string(),
|
117
|
+
primitive_type: PrimitiveType::Int64,
|
118
|
+
nullable: false,
|
119
|
+
format: None,
|
120
|
+
},
|
121
|
+
SchemaNode::Map {
|
122
|
+
name: "attributes".to_string(),
|
123
|
+
nullable: true,
|
124
|
+
key: Box::new(SchemaNode::Primitive {
|
125
|
+
name: "key".to_string(),
|
126
|
+
primitive_type: PrimitiveType::String,
|
127
|
+
nullable: false,
|
128
|
+
format: None,
|
129
|
+
}),
|
130
|
+
value: Box::new(SchemaNode::Primitive {
|
131
|
+
name: "value".to_string(),
|
132
|
+
primitive_type: PrimitiveType::String,
|
133
|
+
nullable: true,
|
134
|
+
format: None,
|
135
|
+
}),
|
136
|
+
},
|
137
|
+
],
|
138
|
+
})
|
139
|
+
.build()
|
140
|
+
.unwrap();
|
141
|
+
|
142
|
+
let rows = vec![
|
143
|
+
// Row with populated map
|
144
|
+
vec![
|
145
|
+
ParquetValue::Int64(1001),
|
146
|
+
ParquetValue::Map(vec![
|
147
|
+
(
|
148
|
+
ParquetValue::String(Arc::from("name")),
|
149
|
+
ParquetValue::String(Arc::from("Alice")),
|
150
|
+
),
|
151
|
+
(
|
152
|
+
ParquetValue::String(Arc::from("role")),
|
153
|
+
ParquetValue::String(Arc::from("admin")),
|
154
|
+
),
|
155
|
+
(
|
156
|
+
ParquetValue::String(Arc::from("department")),
|
157
|
+
ParquetValue::Null,
|
158
|
+
),
|
159
|
+
]),
|
160
|
+
],
|
161
|
+
// Row with empty map
|
162
|
+
vec![ParquetValue::Int64(1002), ParquetValue::Map(vec![])],
|
163
|
+
// Row with null map
|
164
|
+
vec![ParquetValue::Int64(1003), ParquetValue::Null],
|
165
|
+
// Row with single-entry map
|
166
|
+
vec![
|
167
|
+
ParquetValue::Int64(1004),
|
168
|
+
ParquetValue::Map(vec![(
|
169
|
+
ParquetValue::String(Arc::from("status")),
|
170
|
+
ParquetValue::String(Arc::from("active")),
|
171
|
+
)]),
|
172
|
+
],
|
173
|
+
];
|
174
|
+
|
175
|
+
// Write to buffer
|
176
|
+
let mut buffer = Vec::new();
|
177
|
+
{
|
178
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
179
|
+
writer.write_rows(rows.clone()).unwrap();
|
180
|
+
writer.close().unwrap();
|
181
|
+
}
|
182
|
+
|
183
|
+
// Read back and verify
|
184
|
+
let bytes = Bytes::from(buffer);
|
185
|
+
let reader = Reader::new(bytes);
|
186
|
+
|
187
|
+
let read_rows: Vec<_> = reader
|
188
|
+
.read_rows()
|
189
|
+
.unwrap()
|
190
|
+
.collect::<Result<Vec<_>>>()
|
191
|
+
.unwrap();
|
192
|
+
|
193
|
+
assert_eq!(rows.len(), read_rows.len());
|
194
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
195
|
+
assert_eq!(expected, actual);
|
196
|
+
}
|
197
|
+
}
|
198
|
+
|
199
|
+
#[test]
|
200
|
+
fn test_write_and_read_nested_structs() {
|
201
|
+
// Create schema with nested structs
|
202
|
+
let schema = SchemaBuilder::new()
|
203
|
+
.with_root(SchemaNode::Struct {
|
204
|
+
name: "root".to_string(),
|
205
|
+
nullable: false,
|
206
|
+
fields: vec![
|
207
|
+
SchemaNode::Primitive {
|
208
|
+
name: "id".to_string(),
|
209
|
+
primitive_type: PrimitiveType::Int32,
|
210
|
+
nullable: false,
|
211
|
+
format: None,
|
212
|
+
},
|
213
|
+
SchemaNode::Struct {
|
214
|
+
name: "address".to_string(),
|
215
|
+
nullable: true,
|
216
|
+
fields: vec![
|
217
|
+
SchemaNode::Primitive {
|
218
|
+
name: "street".to_string(),
|
219
|
+
primitive_type: PrimitiveType::String,
|
220
|
+
nullable: true, // Changed to nullable
|
221
|
+
format: None,
|
222
|
+
},
|
223
|
+
SchemaNode::Primitive {
|
224
|
+
name: "city".to_string(),
|
225
|
+
primitive_type: PrimitiveType::String,
|
226
|
+
nullable: true, // Changed to nullable
|
227
|
+
format: None,
|
228
|
+
},
|
229
|
+
SchemaNode::Struct {
|
230
|
+
name: "coordinates".to_string(),
|
231
|
+
nullable: true,
|
232
|
+
fields: vec![
|
233
|
+
SchemaNode::Primitive {
|
234
|
+
name: "latitude".to_string(),
|
235
|
+
primitive_type: PrimitiveType::Float64,
|
236
|
+
nullable: true, // Changed to nullable
|
237
|
+
format: None,
|
238
|
+
},
|
239
|
+
SchemaNode::Primitive {
|
240
|
+
name: "longitude".to_string(),
|
241
|
+
primitive_type: PrimitiveType::Float64,
|
242
|
+
nullable: true, // Changed to nullable
|
243
|
+
format: None,
|
244
|
+
},
|
245
|
+
],
|
246
|
+
},
|
247
|
+
],
|
248
|
+
},
|
249
|
+
],
|
250
|
+
})
|
251
|
+
.build()
|
252
|
+
.unwrap();
|
253
|
+
|
254
|
+
let rows = vec![
|
255
|
+
// Row with fully populated nested struct
|
256
|
+
vec![
|
257
|
+
ParquetValue::Int32(1),
|
258
|
+
ParquetValue::Record({
|
259
|
+
let mut map = IndexMap::new();
|
260
|
+
map.insert(
|
261
|
+
Arc::from("street"),
|
262
|
+
ParquetValue::String(Arc::from("123 Main St")),
|
263
|
+
);
|
264
|
+
map.insert(
|
265
|
+
Arc::from("city"),
|
266
|
+
ParquetValue::String(Arc::from("Seattle")),
|
267
|
+
);
|
268
|
+
map.insert(
|
269
|
+
Arc::from("coordinates"),
|
270
|
+
ParquetValue::Record({
|
271
|
+
let mut coords = IndexMap::new();
|
272
|
+
coords.insert(
|
273
|
+
Arc::from("latitude"),
|
274
|
+
ParquetValue::Float64(OrderedFloat(47.6062)),
|
275
|
+
);
|
276
|
+
coords.insert(
|
277
|
+
Arc::from("longitude"),
|
278
|
+
ParquetValue::Float64(OrderedFloat(-122.3321)),
|
279
|
+
);
|
280
|
+
coords
|
281
|
+
}),
|
282
|
+
);
|
283
|
+
map
|
284
|
+
}),
|
285
|
+
],
|
286
|
+
// Row with null nested struct
|
287
|
+
vec![ParquetValue::Int32(2), ParquetValue::Null],
|
288
|
+
// Row with struct containing all required fields
|
289
|
+
vec![
|
290
|
+
ParquetValue::Int32(3),
|
291
|
+
ParquetValue::Record({
|
292
|
+
let mut map = IndexMap::new();
|
293
|
+
map.insert(
|
294
|
+
Arc::from("street"),
|
295
|
+
ParquetValue::String(Arc::from("456 Oak Ave")),
|
296
|
+
);
|
297
|
+
map.insert(
|
298
|
+
Arc::from("city"),
|
299
|
+
ParquetValue::String(Arc::from("Portland")),
|
300
|
+
);
|
301
|
+
// Now we can use null since fields are nullable
|
302
|
+
map.insert(Arc::from("coordinates"), ParquetValue::Null);
|
303
|
+
map
|
304
|
+
}),
|
305
|
+
],
|
306
|
+
];
|
307
|
+
|
308
|
+
// Write to buffer
|
309
|
+
let mut buffer = Vec::new();
|
310
|
+
{
|
311
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
312
|
+
writer.write_rows(rows.clone()).unwrap();
|
313
|
+
writer.close().unwrap();
|
314
|
+
}
|
315
|
+
|
316
|
+
// Read back and verify
|
317
|
+
let bytes = Bytes::from(buffer);
|
318
|
+
let reader = Reader::new(bytes);
|
319
|
+
|
320
|
+
let read_rows: Vec<_> = reader
|
321
|
+
.read_rows()
|
322
|
+
.unwrap()
|
323
|
+
.collect::<Result<Vec<_>>>()
|
324
|
+
.unwrap();
|
325
|
+
|
326
|
+
assert_eq!(rows.len(), read_rows.len());
|
327
|
+
|
328
|
+
// Check first row (fully populated)
|
329
|
+
assert_eq!(rows[0], read_rows[0]);
|
330
|
+
|
331
|
+
// Check second row - when we write a null struct, it reads back as null
|
332
|
+
assert_eq!(read_rows[1][0], ParquetValue::Int32(2)); // ID should match
|
333
|
+
match &read_rows[1][1] {
|
334
|
+
ParquetValue::Null => {
|
335
|
+
// Correct - null structs read back as null
|
336
|
+
}
|
337
|
+
_ => panic!("Expected second row address to be Null"),
|
338
|
+
}
|
339
|
+
|
340
|
+
// Check third row - same issue with null coordinates
|
341
|
+
match &read_rows[2][1] {
|
342
|
+
ParquetValue::Record(record) => {
|
343
|
+
// Check the string fields match
|
344
|
+
assert_eq!(
|
345
|
+
record.get("street"),
|
346
|
+
Some(&ParquetValue::String(Arc::from("456 Oak Ave")))
|
347
|
+
);
|
348
|
+
assert_eq!(
|
349
|
+
record.get("city"),
|
350
|
+
Some(&ParquetValue::String(Arc::from("Portland")))
|
351
|
+
);
|
352
|
+
// Verify coordinates is null (not a struct with null fields)
|
353
|
+
match record.get("coordinates") {
|
354
|
+
Some(ParquetValue::Null) => {
|
355
|
+
// Correct - null nested struct reads back as null
|
356
|
+
}
|
357
|
+
_ => panic!("Expected coordinates to be Null"),
|
358
|
+
}
|
359
|
+
}
|
360
|
+
_ => panic!("Expected third row address to be a Record"),
|
361
|
+
}
|
362
|
+
}
|
363
|
+
|
364
|
+
#[test]
|
365
|
+
fn test_complex_list_of_structs() {
|
366
|
+
// Create schema with list of structs
|
367
|
+
let schema = SchemaBuilder::new()
|
368
|
+
.with_root(SchemaNode::Struct {
|
369
|
+
name: "root".to_string(),
|
370
|
+
nullable: false,
|
371
|
+
fields: vec![
|
372
|
+
SchemaNode::Primitive {
|
373
|
+
name: "order_id".to_string(),
|
374
|
+
primitive_type: PrimitiveType::Int64,
|
375
|
+
nullable: false,
|
376
|
+
format: None,
|
377
|
+
},
|
378
|
+
SchemaNode::List {
|
379
|
+
name: "items".to_string(),
|
380
|
+
nullable: false,
|
381
|
+
item: Box::new(SchemaNode::Struct {
|
382
|
+
name: "item".to_string(),
|
383
|
+
nullable: false,
|
384
|
+
fields: vec![
|
385
|
+
SchemaNode::Primitive {
|
386
|
+
name: "product_id".to_string(),
|
387
|
+
primitive_type: PrimitiveType::Int32,
|
388
|
+
nullable: false,
|
389
|
+
format: None,
|
390
|
+
},
|
391
|
+
SchemaNode::Primitive {
|
392
|
+
name: "quantity".to_string(),
|
393
|
+
primitive_type: PrimitiveType::Int32,
|
394
|
+
nullable: false,
|
395
|
+
format: None,
|
396
|
+
},
|
397
|
+
SchemaNode::Primitive {
|
398
|
+
name: "price".to_string(),
|
399
|
+
primitive_type: PrimitiveType::Decimal128(10, 2),
|
400
|
+
nullable: false,
|
401
|
+
format: None,
|
402
|
+
},
|
403
|
+
],
|
404
|
+
}),
|
405
|
+
},
|
406
|
+
],
|
407
|
+
})
|
408
|
+
.build()
|
409
|
+
.unwrap();
|
410
|
+
|
411
|
+
let rows = vec![
|
412
|
+
// Order with multiple items
|
413
|
+
vec![
|
414
|
+
ParquetValue::Int64(100001),
|
415
|
+
ParquetValue::List(vec![
|
416
|
+
ParquetValue::Record({
|
417
|
+
let mut item = IndexMap::new();
|
418
|
+
item.insert(Arc::from("product_id"), ParquetValue::Int32(1));
|
419
|
+
item.insert(Arc::from("quantity"), ParquetValue::Int32(2));
|
420
|
+
item.insert(Arc::from("price"), ParquetValue::Decimal128(1999, 2));
|
421
|
+
item
|
422
|
+
}),
|
423
|
+
ParquetValue::Record({
|
424
|
+
let mut item = IndexMap::new();
|
425
|
+
item.insert(Arc::from("product_id"), ParquetValue::Int32(2));
|
426
|
+
item.insert(Arc::from("quantity"), ParquetValue::Int32(1));
|
427
|
+
item.insert(Arc::from("price"), ParquetValue::Decimal128(4995, 2));
|
428
|
+
item
|
429
|
+
}),
|
430
|
+
]),
|
431
|
+
],
|
432
|
+
// Order with single item
|
433
|
+
vec![
|
434
|
+
ParquetValue::Int64(100002),
|
435
|
+
ParquetValue::List(vec![ParquetValue::Record({
|
436
|
+
let mut item = IndexMap::new();
|
437
|
+
item.insert(Arc::from("product_id"), ParquetValue::Int32(3));
|
438
|
+
item.insert(Arc::from("quantity"), ParquetValue::Int32(5));
|
439
|
+
item.insert(Arc::from("price"), ParquetValue::Decimal128(999, 2));
|
440
|
+
item
|
441
|
+
})]),
|
442
|
+
],
|
443
|
+
// Order with no items
|
444
|
+
vec![ParquetValue::Int64(100003), ParquetValue::List(vec![])],
|
445
|
+
];
|
446
|
+
|
447
|
+
// Write to buffer
|
448
|
+
let mut buffer = Vec::new();
|
449
|
+
{
|
450
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
451
|
+
writer.write_rows(rows.clone()).unwrap();
|
452
|
+
writer.close().unwrap();
|
453
|
+
}
|
454
|
+
|
455
|
+
// Read back and verify
|
456
|
+
let bytes = Bytes::from(buffer);
|
457
|
+
let reader = Reader::new(bytes);
|
458
|
+
|
459
|
+
let read_rows: Vec<_> = reader
|
460
|
+
.read_rows()
|
461
|
+
.unwrap()
|
462
|
+
.collect::<Result<Vec<_>>>()
|
463
|
+
.unwrap();
|
464
|
+
|
465
|
+
assert_eq!(rows.len(), read_rows.len());
|
466
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
467
|
+
assert_eq!(expected, actual);
|
468
|
+
}
|
469
|
+
}
|
470
|
+
|
471
|
+
#[test]
|
472
|
+
fn test_map_with_complex_values() {
|
473
|
+
// Create schema with map containing struct values
|
474
|
+
let schema = SchemaBuilder::new()
|
475
|
+
.with_root(SchemaNode::Struct {
|
476
|
+
name: "root".to_string(),
|
477
|
+
nullable: false,
|
478
|
+
fields: vec![
|
479
|
+
SchemaNode::Primitive {
|
480
|
+
name: "session_id".to_string(),
|
481
|
+
primitive_type: PrimitiveType::String,
|
482
|
+
nullable: false,
|
483
|
+
format: None,
|
484
|
+
},
|
485
|
+
SchemaNode::Map {
|
486
|
+
name: "metrics".to_string(),
|
487
|
+
nullable: false,
|
488
|
+
key: Box::new(SchemaNode::Primitive {
|
489
|
+
name: "metric_name".to_string(),
|
490
|
+
primitive_type: PrimitiveType::String,
|
491
|
+
nullable: false,
|
492
|
+
format: None,
|
493
|
+
}),
|
494
|
+
value: Box::new(SchemaNode::Struct {
|
495
|
+
name: "metric_data".to_string(),
|
496
|
+
nullable: false,
|
497
|
+
fields: vec![
|
498
|
+
SchemaNode::Primitive {
|
499
|
+
name: "value".to_string(),
|
500
|
+
primitive_type: PrimitiveType::Float64,
|
501
|
+
nullable: false,
|
502
|
+
format: None,
|
503
|
+
},
|
504
|
+
SchemaNode::Primitive {
|
505
|
+
name: "unit".to_string(),
|
506
|
+
primitive_type: PrimitiveType::String,
|
507
|
+
nullable: false,
|
508
|
+
format: None,
|
509
|
+
},
|
510
|
+
SchemaNode::Primitive {
|
511
|
+
name: "timestamp".to_string(),
|
512
|
+
primitive_type: PrimitiveType::TimestampMillis(None),
|
513
|
+
nullable: false,
|
514
|
+
format: None,
|
515
|
+
},
|
516
|
+
],
|
517
|
+
}),
|
518
|
+
},
|
519
|
+
],
|
520
|
+
})
|
521
|
+
.build()
|
522
|
+
.unwrap();
|
523
|
+
|
524
|
+
let rows = vec![vec![
|
525
|
+
ParquetValue::String(Arc::from("session-123")),
|
526
|
+
ParquetValue::Map(vec![
|
527
|
+
(
|
528
|
+
ParquetValue::String(Arc::from("cpu_usage")),
|
529
|
+
ParquetValue::Record({
|
530
|
+
let mut data = IndexMap::new();
|
531
|
+
data.insert(
|
532
|
+
Arc::from("value"),
|
533
|
+
ParquetValue::Float64(OrderedFloat(85.5)),
|
534
|
+
);
|
535
|
+
data.insert(
|
536
|
+
Arc::from("unit"),
|
537
|
+
ParquetValue::String(Arc::from("percent")),
|
538
|
+
);
|
539
|
+
data.insert(
|
540
|
+
Arc::from("timestamp"),
|
541
|
+
ParquetValue::TimestampMillis(1640000000000, None),
|
542
|
+
);
|
543
|
+
data
|
544
|
+
}),
|
545
|
+
),
|
546
|
+
(
|
547
|
+
ParquetValue::String(Arc::from("memory_usage")),
|
548
|
+
ParquetValue::Record({
|
549
|
+
let mut data = IndexMap::new();
|
550
|
+
data.insert(
|
551
|
+
Arc::from("value"),
|
552
|
+
ParquetValue::Float64(OrderedFloat(1024.0)),
|
553
|
+
);
|
554
|
+
data.insert(Arc::from("unit"), ParquetValue::String(Arc::from("MB")));
|
555
|
+
data.insert(
|
556
|
+
Arc::from("timestamp"),
|
557
|
+
ParquetValue::TimestampMillis(1640000001000, None),
|
558
|
+
);
|
559
|
+
data
|
560
|
+
}),
|
561
|
+
),
|
562
|
+
]),
|
563
|
+
]];
|
564
|
+
|
565
|
+
// Write to buffer
|
566
|
+
let mut buffer = Vec::new();
|
567
|
+
{
|
568
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
569
|
+
writer.write_rows(rows.clone()).unwrap();
|
570
|
+
writer.close().unwrap();
|
571
|
+
}
|
572
|
+
|
573
|
+
// Read back and verify
|
574
|
+
let bytes = Bytes::from(buffer);
|
575
|
+
let reader = Reader::new(bytes);
|
576
|
+
|
577
|
+
let read_rows: Vec<_> = reader
|
578
|
+
.read_rows()
|
579
|
+
.unwrap()
|
580
|
+
.collect::<Result<Vec<_>>>()
|
581
|
+
.unwrap();
|
582
|
+
|
583
|
+
assert_eq!(rows.len(), read_rows.len());
|
584
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
585
|
+
assert_eq!(expected, actual);
|
586
|
+
}
|
587
|
+
}
|
588
|
+
|
589
|
+
#[test]
|
590
|
+
fn test_deeply_nested_structures() {
|
591
|
+
// Create a deeply nested schema: struct -> list -> struct -> list
|
592
|
+
// (avoiding map with struct values which is not supported)
|
593
|
+
let schema = SchemaBuilder::new()
|
594
|
+
.with_root(SchemaNode::Struct {
|
595
|
+
name: "root".to_string(),
|
596
|
+
nullable: false,
|
597
|
+
fields: vec![
|
598
|
+
SchemaNode::Primitive {
|
599
|
+
name: "doc_id".to_string(),
|
600
|
+
primitive_type: PrimitiveType::String,
|
601
|
+
nullable: false,
|
602
|
+
format: None,
|
603
|
+
},
|
604
|
+
SchemaNode::Struct {
|
605
|
+
name: "content".to_string(),
|
606
|
+
nullable: false,
|
607
|
+
fields: vec![SchemaNode::List {
|
608
|
+
name: "sections".to_string(),
|
609
|
+
nullable: false,
|
610
|
+
item: Box::new(SchemaNode::Struct {
|
611
|
+
name: "section".to_string(),
|
612
|
+
nullable: false,
|
613
|
+
fields: vec![
|
614
|
+
SchemaNode::Primitive {
|
615
|
+
name: "title".to_string(),
|
616
|
+
primitive_type: PrimitiveType::String,
|
617
|
+
nullable: false,
|
618
|
+
format: None,
|
619
|
+
},
|
620
|
+
SchemaNode::List {
|
621
|
+
name: "paragraphs".to_string(),
|
622
|
+
nullable: false,
|
623
|
+
item: Box::new(SchemaNode::Struct {
|
624
|
+
name: "paragraph".to_string(),
|
625
|
+
nullable: false,
|
626
|
+
fields: vec![
|
627
|
+
SchemaNode::Primitive {
|
628
|
+
name: "text".to_string(),
|
629
|
+
primitive_type: PrimitiveType::String,
|
630
|
+
nullable: false,
|
631
|
+
format: None,
|
632
|
+
},
|
633
|
+
SchemaNode::Primitive {
|
634
|
+
name: "score".to_string(),
|
635
|
+
primitive_type: PrimitiveType::Float32,
|
636
|
+
nullable: true,
|
637
|
+
format: None,
|
638
|
+
},
|
639
|
+
],
|
640
|
+
}),
|
641
|
+
},
|
642
|
+
],
|
643
|
+
}),
|
644
|
+
}],
|
645
|
+
},
|
646
|
+
],
|
647
|
+
})
|
648
|
+
.build()
|
649
|
+
.unwrap();
|
650
|
+
|
651
|
+
let rows = vec![vec![
|
652
|
+
ParquetValue::String(Arc::from("doc-001")),
|
653
|
+
ParquetValue::Record({
|
654
|
+
let mut content = IndexMap::new();
|
655
|
+
content.insert(
|
656
|
+
Arc::from("sections"),
|
657
|
+
ParquetValue::List(vec![
|
658
|
+
ParquetValue::Record({
|
659
|
+
let mut section = IndexMap::new();
|
660
|
+
section.insert(
|
661
|
+
Arc::from("title"),
|
662
|
+
ParquetValue::String(Arc::from("Introduction")),
|
663
|
+
);
|
664
|
+
section.insert(
|
665
|
+
Arc::from("paragraphs"),
|
666
|
+
ParquetValue::List(vec![
|
667
|
+
ParquetValue::Record({
|
668
|
+
let mut para = IndexMap::new();
|
669
|
+
para.insert(
|
670
|
+
Arc::from("text"),
|
671
|
+
ParquetValue::String(Arc::from(
|
672
|
+
"Welcome to this document.",
|
673
|
+
)),
|
674
|
+
);
|
675
|
+
para.insert(
|
676
|
+
Arc::from("score"),
|
677
|
+
ParquetValue::Float32(OrderedFloat(0.95)),
|
678
|
+
);
|
679
|
+
para
|
680
|
+
}),
|
681
|
+
ParquetValue::Record({
|
682
|
+
let mut para = IndexMap::new();
|
683
|
+
para.insert(
|
684
|
+
Arc::from("text"),
|
685
|
+
ParquetValue::String(Arc::from(
|
686
|
+
"This is the second paragraph.",
|
687
|
+
)),
|
688
|
+
);
|
689
|
+
para.insert(Arc::from("score"), ParquetValue::Null);
|
690
|
+
para
|
691
|
+
}),
|
692
|
+
]),
|
693
|
+
);
|
694
|
+
section
|
695
|
+
}),
|
696
|
+
ParquetValue::Record({
|
697
|
+
let mut section = IndexMap::new();
|
698
|
+
section.insert(
|
699
|
+
Arc::from("title"),
|
700
|
+
ParquetValue::String(Arc::from("Conclusion")),
|
701
|
+
);
|
702
|
+
section.insert(
|
703
|
+
Arc::from("paragraphs"),
|
704
|
+
ParquetValue::List(vec![ParquetValue::Record({
|
705
|
+
let mut para = IndexMap::new();
|
706
|
+
para.insert(
|
707
|
+
Arc::from("text"),
|
708
|
+
ParquetValue::String(Arc::from("In summary...")),
|
709
|
+
);
|
710
|
+
para.insert(
|
711
|
+
Arc::from("score"),
|
712
|
+
ParquetValue::Float32(OrderedFloat(0.88)),
|
713
|
+
);
|
714
|
+
para
|
715
|
+
})]),
|
716
|
+
);
|
717
|
+
section
|
718
|
+
}),
|
719
|
+
]),
|
720
|
+
);
|
721
|
+
content
|
722
|
+
}),
|
723
|
+
]];
|
724
|
+
|
725
|
+
// Write to buffer
|
726
|
+
let mut buffer = Vec::new();
|
727
|
+
{
|
728
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
729
|
+
writer.write_rows(rows.clone()).unwrap();
|
730
|
+
writer.close().unwrap();
|
731
|
+
}
|
732
|
+
|
733
|
+
// Read back and verify
|
734
|
+
let bytes = Bytes::from(buffer);
|
735
|
+
let reader = Reader::new(bytes);
|
736
|
+
|
737
|
+
let read_rows: Vec<_> = reader
|
738
|
+
.read_rows()
|
739
|
+
.unwrap()
|
740
|
+
.collect::<Result<Vec<_>>>()
|
741
|
+
.unwrap();
|
742
|
+
|
743
|
+
assert_eq!(rows.len(), read_rows.len());
|
744
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
745
|
+
assert_eq!(expected, actual);
|
746
|
+
}
|
747
|
+
}
|
748
|
+
|
749
|
+
#[test]
|
750
|
+
fn test_decimal256_complex_type() {
|
751
|
+
// Test Decimal256 within complex structures
|
752
|
+
let schema = SchemaBuilder::new()
|
753
|
+
.with_root(SchemaNode::Struct {
|
754
|
+
name: "root".to_string(),
|
755
|
+
nullable: false,
|
756
|
+
fields: vec![
|
757
|
+
SchemaNode::Primitive {
|
758
|
+
name: "id".to_string(),
|
759
|
+
primitive_type: PrimitiveType::Int32,
|
760
|
+
nullable: false,
|
761
|
+
format: None,
|
762
|
+
},
|
763
|
+
SchemaNode::List {
|
764
|
+
name: "large_values".to_string(),
|
765
|
+
nullable: false,
|
766
|
+
item: Box::new(SchemaNode::Primitive {
|
767
|
+
name: "value".to_string(),
|
768
|
+
primitive_type: PrimitiveType::Decimal256(50, 10),
|
769
|
+
nullable: true,
|
770
|
+
format: None,
|
771
|
+
}),
|
772
|
+
},
|
773
|
+
],
|
774
|
+
})
|
775
|
+
.build()
|
776
|
+
.unwrap();
|
777
|
+
|
778
|
+
let rows = vec![
|
779
|
+
vec![
|
780
|
+
ParquetValue::Int32(1),
|
781
|
+
ParquetValue::List(vec![
|
782
|
+
ParquetValue::Decimal256(
|
783
|
+
BigInt::parse_bytes(b"123456789012345678901234567890", 10).unwrap(),
|
784
|
+
10,
|
785
|
+
),
|
786
|
+
ParquetValue::Decimal256(
|
787
|
+
BigInt::parse_bytes(b"-987654321098765432109876543210", 10).unwrap(),
|
788
|
+
10,
|
789
|
+
),
|
790
|
+
ParquetValue::Null,
|
791
|
+
]),
|
792
|
+
],
|
793
|
+
vec![
|
794
|
+
ParquetValue::Int32(2),
|
795
|
+
ParquetValue::List(vec![ParquetValue::Decimal256(BigInt::from(0), 10)]),
|
796
|
+
],
|
797
|
+
];
|
798
|
+
|
799
|
+
// Write to buffer
|
800
|
+
let mut buffer = Vec::new();
|
801
|
+
{
|
802
|
+
let mut writer = Writer::new(&mut buffer, schema.clone()).unwrap();
|
803
|
+
writer.write_rows(rows.clone()).unwrap();
|
804
|
+
writer.close().unwrap();
|
805
|
+
}
|
806
|
+
|
807
|
+
// Read back and verify
|
808
|
+
let bytes = Bytes::from(buffer);
|
809
|
+
let reader = Reader::new(bytes);
|
810
|
+
|
811
|
+
let read_rows: Vec<_> = reader
|
812
|
+
.read_rows()
|
813
|
+
.unwrap()
|
814
|
+
.collect::<Result<Vec<_>>>()
|
815
|
+
.unwrap();
|
816
|
+
|
817
|
+
assert_eq!(rows.len(), read_rows.len());
|
818
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
819
|
+
assert_eq!(expected, actual);
|
820
|
+
}
|
821
|
+
}
|