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 +1,488 @@
|
|
1
|
+
use bytes::Bytes;
|
2
|
+
use num::BigInt;
|
3
|
+
use parquet_core::*;
|
4
|
+
use std::sync::Arc;
|
5
|
+
|
6
|
+
#[test]
|
7
|
+
fn test_decimal128_precision_scale_combinations() {
|
8
|
+
// Test various precision and scale combinations for Decimal128
|
9
|
+
let schema = SchemaBuilder::new()
|
10
|
+
.with_root(SchemaNode::Struct {
|
11
|
+
name: "root".to_string(),
|
12
|
+
nullable: false,
|
13
|
+
fields: vec![
|
14
|
+
SchemaNode::Primitive {
|
15
|
+
name: "dec_5_2".to_string(),
|
16
|
+
primitive_type: PrimitiveType::Decimal128(5, 2),
|
17
|
+
nullable: false,
|
18
|
+
format: None,
|
19
|
+
},
|
20
|
+
SchemaNode::Primitive {
|
21
|
+
name: "dec_9_2".to_string(),
|
22
|
+
primitive_type: PrimitiveType::Decimal128(9, 2),
|
23
|
+
nullable: false,
|
24
|
+
format: None,
|
25
|
+
},
|
26
|
+
SchemaNode::Primitive {
|
27
|
+
name: "dec_18_0".to_string(),
|
28
|
+
primitive_type: PrimitiveType::Decimal128(18, 0),
|
29
|
+
nullable: false,
|
30
|
+
format: None,
|
31
|
+
},
|
32
|
+
SchemaNode::Primitive {
|
33
|
+
name: "dec_38_0".to_string(),
|
34
|
+
primitive_type: PrimitiveType::Decimal128(38, 0),
|
35
|
+
nullable: false,
|
36
|
+
format: None,
|
37
|
+
},
|
38
|
+
SchemaNode::Primitive {
|
39
|
+
name: "dec_38_10".to_string(),
|
40
|
+
primitive_type: PrimitiveType::Decimal128(38, 10),
|
41
|
+
nullable: false,
|
42
|
+
format: None,
|
43
|
+
},
|
44
|
+
SchemaNode::Primitive {
|
45
|
+
name: "dec_38_38".to_string(),
|
46
|
+
primitive_type: PrimitiveType::Decimal128(38, 38),
|
47
|
+
nullable: false,
|
48
|
+
format: None,
|
49
|
+
},
|
50
|
+
],
|
51
|
+
})
|
52
|
+
.build()
|
53
|
+
.unwrap();
|
54
|
+
|
55
|
+
let test_cases = vec![
|
56
|
+
// Maximum positive values for each precision
|
57
|
+
vec![
|
58
|
+
ParquetValue::Decimal128(99999, 2), // 999.99 (max for 5,2)
|
59
|
+
ParquetValue::Decimal128(999_999_999, 2), // 9999999.99
|
60
|
+
ParquetValue::Decimal128(999999999999999999, 0), // 18 digits
|
61
|
+
ParquetValue::Decimal128(99999999999999999999999999999999999999_i128, 0), // Max 38 digits
|
62
|
+
ParquetValue::Decimal128(99_999_999_999_999_999_999_999_999_999_999_999_999i128, 10), // 38 digits, 10 scale
|
63
|
+
ParquetValue::Decimal128(99999999999999999999999999999999999999_i128, 38), // 0.99999... (38 9s after decimal)
|
64
|
+
],
|
65
|
+
// Maximum negative values
|
66
|
+
vec![
|
67
|
+
ParquetValue::Decimal128(-99999, 2), // -999.99 (min for 5,2)
|
68
|
+
ParquetValue::Decimal128(-999_999_999, 2), // -9999999.99
|
69
|
+
ParquetValue::Decimal128(-999999999999999999, 0), // -18 digits
|
70
|
+
ParquetValue::Decimal128(-99999999999999999999999999999999999999_i128, 0), // Min 38 digits
|
71
|
+
ParquetValue::Decimal128(-99_999_999_999_999_999_999_999_999_999_999_999_999i128, 10), // -38 digits, 10 scale
|
72
|
+
ParquetValue::Decimal128(-99999999999999999999999999999999999999_i128, 38), // -0.99999...
|
73
|
+
],
|
74
|
+
// Zero values
|
75
|
+
vec![
|
76
|
+
ParquetValue::Decimal128(0, 2), // 0.00
|
77
|
+
ParquetValue::Decimal128(0, 2),
|
78
|
+
ParquetValue::Decimal128(0, 0),
|
79
|
+
ParquetValue::Decimal128(0, 0), // 0
|
80
|
+
ParquetValue::Decimal128(0, 10),
|
81
|
+
ParquetValue::Decimal128(0, 38), // 0.00000... (38 zeros)
|
82
|
+
],
|
83
|
+
// Small positive values
|
84
|
+
vec![
|
85
|
+
ParquetValue::Decimal128(1, 2), // 0.01 (smallest positive for 5,2)
|
86
|
+
ParquetValue::Decimal128(1, 2), // 0.01
|
87
|
+
ParquetValue::Decimal128(1, 0), // 1
|
88
|
+
ParquetValue::Decimal128(1, 0), // 1
|
89
|
+
ParquetValue::Decimal128(1, 10), // 0.0000000001
|
90
|
+
ParquetValue::Decimal128(1, 38), // 0.00000...01 (37 zeros then 1)
|
91
|
+
],
|
92
|
+
// Maximum value for i128
|
93
|
+
vec![
|
94
|
+
ParquetValue::Decimal128(99999, 2),
|
95
|
+
ParquetValue::Decimal128(999_999_999, 2),
|
96
|
+
ParquetValue::Decimal128(i64::MAX as i128, 0),
|
97
|
+
ParquetValue::Decimal128(i128::MAX, 0),
|
98
|
+
ParquetValue::Decimal128(12345678901234567890123456789_i128, 10),
|
99
|
+
ParquetValue::Decimal128(1, 38),
|
100
|
+
],
|
101
|
+
// Minimum value for i128
|
102
|
+
vec![
|
103
|
+
ParquetValue::Decimal128(-99999, 2),
|
104
|
+
ParquetValue::Decimal128(-999_999_999, 2),
|
105
|
+
ParquetValue::Decimal128(i64::MIN as i128, 0),
|
106
|
+
ParquetValue::Decimal128(i128::MIN, 0),
|
107
|
+
ParquetValue::Decimal128(-12345678901234567890123456789_i128, 10),
|
108
|
+
ParquetValue::Decimal128(-1, 38),
|
109
|
+
],
|
110
|
+
];
|
111
|
+
|
112
|
+
let mut buffer = Vec::new();
|
113
|
+
{
|
114
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
115
|
+
writer.write_rows(test_cases.clone()).unwrap();
|
116
|
+
writer.close().unwrap();
|
117
|
+
}
|
118
|
+
|
119
|
+
// Read back and verify
|
120
|
+
let bytes = Bytes::from(buffer);
|
121
|
+
let reader = Reader::new(bytes);
|
122
|
+
|
123
|
+
let read_rows: Vec<_> = reader
|
124
|
+
.read_rows()
|
125
|
+
.unwrap()
|
126
|
+
.collect::<Result<Vec<_>>>()
|
127
|
+
.unwrap();
|
128
|
+
|
129
|
+
assert_eq!(read_rows.len(), test_cases.len());
|
130
|
+
for (expected, actual) in test_cases.iter().zip(read_rows.iter()) {
|
131
|
+
assert_eq!(expected, actual);
|
132
|
+
}
|
133
|
+
}
|
134
|
+
|
135
|
+
#[test]
|
136
|
+
fn test_decimal256_large_values() {
|
137
|
+
// Test very large Decimal256 values
|
138
|
+
let schema = SchemaBuilder::new()
|
139
|
+
.with_root(SchemaNode::Struct {
|
140
|
+
name: "root".to_string(),
|
141
|
+
nullable: false,
|
142
|
+
fields: vec![
|
143
|
+
SchemaNode::Primitive {
|
144
|
+
name: "dec256_76_0".to_string(),
|
145
|
+
primitive_type: PrimitiveType::Decimal256(76, 0),
|
146
|
+
nullable: false,
|
147
|
+
format: None,
|
148
|
+
},
|
149
|
+
SchemaNode::Primitive {
|
150
|
+
name: "dec256_76_38".to_string(),
|
151
|
+
primitive_type: PrimitiveType::Decimal256(76, 38),
|
152
|
+
nullable: false,
|
153
|
+
format: None,
|
154
|
+
},
|
155
|
+
],
|
156
|
+
})
|
157
|
+
.build()
|
158
|
+
.unwrap();
|
159
|
+
|
160
|
+
// Test values that require Decimal256
|
161
|
+
let large_positive = BigInt::parse_bytes(
|
162
|
+
b"9999999999999999999999999999999999999999999999999999999999999999999999999",
|
163
|
+
10,
|
164
|
+
)
|
165
|
+
.unwrap();
|
166
|
+
let large_negative = -large_positive.clone();
|
167
|
+
let medium_value = BigInt::parse_bytes(
|
168
|
+
b"123456789012345678901234567890123456789012345678901234567890",
|
169
|
+
10,
|
170
|
+
)
|
171
|
+
.unwrap();
|
172
|
+
|
173
|
+
let test_cases = vec![
|
174
|
+
vec![
|
175
|
+
ParquetValue::Decimal256(large_positive.clone(), 0),
|
176
|
+
ParquetValue::Decimal256(medium_value.clone(), 38),
|
177
|
+
],
|
178
|
+
vec![
|
179
|
+
ParquetValue::Decimal256(large_negative.clone(), 0),
|
180
|
+
ParquetValue::Decimal256(-medium_value.clone(), 38),
|
181
|
+
],
|
182
|
+
vec![
|
183
|
+
ParquetValue::Decimal256(BigInt::from(0), 0),
|
184
|
+
ParquetValue::Decimal256(BigInt::from(1), 38),
|
185
|
+
],
|
186
|
+
];
|
187
|
+
|
188
|
+
let mut buffer = Vec::new();
|
189
|
+
{
|
190
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
191
|
+
writer.write_rows(test_cases.clone()).unwrap();
|
192
|
+
writer.close().unwrap();
|
193
|
+
}
|
194
|
+
|
195
|
+
// Read back and verify
|
196
|
+
let bytes = Bytes::from(buffer);
|
197
|
+
let reader = Reader::new(bytes);
|
198
|
+
|
199
|
+
let read_rows: Vec<_> = reader
|
200
|
+
.read_rows()
|
201
|
+
.unwrap()
|
202
|
+
.collect::<Result<Vec<_>>>()
|
203
|
+
.unwrap();
|
204
|
+
|
205
|
+
assert_eq!(read_rows.len(), test_cases.len());
|
206
|
+
for (expected, actual) in test_cases.iter().zip(read_rows.iter()) {
|
207
|
+
assert_eq!(expected, actual);
|
208
|
+
}
|
209
|
+
}
|
210
|
+
|
211
|
+
#[test]
|
212
|
+
fn test_decimal_null_handling() {
|
213
|
+
// Test nullable decimal fields
|
214
|
+
let schema = SchemaBuilder::new()
|
215
|
+
.with_root(SchemaNode::Struct {
|
216
|
+
name: "root".to_string(),
|
217
|
+
nullable: false,
|
218
|
+
fields: vec![
|
219
|
+
SchemaNode::Primitive {
|
220
|
+
name: "nullable_dec128".to_string(),
|
221
|
+
primitive_type: PrimitiveType::Decimal128(18, 4),
|
222
|
+
nullable: true,
|
223
|
+
format: None,
|
224
|
+
},
|
225
|
+
SchemaNode::Primitive {
|
226
|
+
name: "nullable_dec256".to_string(),
|
227
|
+
primitive_type: PrimitiveType::Decimal256(50, 10),
|
228
|
+
nullable: true,
|
229
|
+
format: None,
|
230
|
+
},
|
231
|
+
],
|
232
|
+
})
|
233
|
+
.build()
|
234
|
+
.unwrap();
|
235
|
+
|
236
|
+
let rows = vec![
|
237
|
+
vec![
|
238
|
+
ParquetValue::Decimal128(123456789012345678, 4),
|
239
|
+
ParquetValue::Decimal256(BigInt::from(987654321), 10),
|
240
|
+
],
|
241
|
+
vec![
|
242
|
+
ParquetValue::Null,
|
243
|
+
ParquetValue::Decimal256(BigInt::from(0), 10),
|
244
|
+
],
|
245
|
+
vec![
|
246
|
+
ParquetValue::Decimal128(-999999999999999999, 4),
|
247
|
+
ParquetValue::Null,
|
248
|
+
],
|
249
|
+
vec![ParquetValue::Null, ParquetValue::Null],
|
250
|
+
];
|
251
|
+
|
252
|
+
let mut buffer = Vec::new();
|
253
|
+
{
|
254
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
255
|
+
writer.write_rows(rows.clone()).unwrap();
|
256
|
+
writer.close().unwrap();
|
257
|
+
}
|
258
|
+
|
259
|
+
// Read back and verify
|
260
|
+
let bytes = Bytes::from(buffer);
|
261
|
+
let reader = Reader::new(bytes);
|
262
|
+
|
263
|
+
let read_rows: Vec<_> = reader
|
264
|
+
.read_rows()
|
265
|
+
.unwrap()
|
266
|
+
.collect::<Result<Vec<_>>>()
|
267
|
+
.unwrap();
|
268
|
+
|
269
|
+
assert_eq!(read_rows.len(), rows.len());
|
270
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
271
|
+
assert_eq!(expected, actual);
|
272
|
+
}
|
273
|
+
}
|
274
|
+
|
275
|
+
#[test]
|
276
|
+
fn test_decimal_in_complex_types() {
|
277
|
+
use indexmap::IndexMap;
|
278
|
+
|
279
|
+
// Test decimals within lists, maps, and structs
|
280
|
+
let schema = SchemaBuilder::new()
|
281
|
+
.with_root(SchemaNode::Struct {
|
282
|
+
name: "root".to_string(),
|
283
|
+
nullable: false,
|
284
|
+
fields: vec![
|
285
|
+
SchemaNode::List {
|
286
|
+
name: "decimal_list".to_string(),
|
287
|
+
nullable: false,
|
288
|
+
item: Box::new(SchemaNode::Primitive {
|
289
|
+
name: "item".to_string(),
|
290
|
+
primitive_type: PrimitiveType::Decimal128(10, 2),
|
291
|
+
nullable: false,
|
292
|
+
format: None,
|
293
|
+
}),
|
294
|
+
},
|
295
|
+
SchemaNode::Map {
|
296
|
+
name: "decimal_map".to_string(),
|
297
|
+
nullable: false,
|
298
|
+
key: Box::new(SchemaNode::Primitive {
|
299
|
+
name: "key".to_string(),
|
300
|
+
primitive_type: PrimitiveType::String,
|
301
|
+
nullable: false,
|
302
|
+
format: None,
|
303
|
+
}),
|
304
|
+
value: Box::new(SchemaNode::Primitive {
|
305
|
+
name: "value".to_string(),
|
306
|
+
primitive_type: PrimitiveType::Decimal256(40, 10),
|
307
|
+
nullable: true,
|
308
|
+
format: None,
|
309
|
+
}),
|
310
|
+
},
|
311
|
+
SchemaNode::Struct {
|
312
|
+
name: "decimal_struct".to_string(),
|
313
|
+
nullable: true,
|
314
|
+
fields: vec![
|
315
|
+
SchemaNode::Primitive {
|
316
|
+
name: "amount".to_string(),
|
317
|
+
primitive_type: PrimitiveType::Decimal128(18, 2),
|
318
|
+
nullable: false,
|
319
|
+
format: None,
|
320
|
+
},
|
321
|
+
SchemaNode::Primitive {
|
322
|
+
name: "large_amount".to_string(),
|
323
|
+
primitive_type: PrimitiveType::Decimal256(50, 6),
|
324
|
+
nullable: true,
|
325
|
+
format: None,
|
326
|
+
},
|
327
|
+
],
|
328
|
+
},
|
329
|
+
],
|
330
|
+
})
|
331
|
+
.build()
|
332
|
+
.unwrap();
|
333
|
+
|
334
|
+
let rows = vec![
|
335
|
+
vec![
|
336
|
+
// List of decimals
|
337
|
+
ParquetValue::List(vec![
|
338
|
+
ParquetValue::Decimal128(100, 2), // 1.00
|
339
|
+
ParquetValue::Decimal128(250, 2), // 2.50
|
340
|
+
ParquetValue::Decimal128(-375, 2), // -3.75
|
341
|
+
]),
|
342
|
+
// Map with decimal values
|
343
|
+
ParquetValue::Map(vec![
|
344
|
+
(
|
345
|
+
ParquetValue::String(Arc::from("total")),
|
346
|
+
ParquetValue::Decimal256(BigInt::from(1234567890123456789_i64), 10),
|
347
|
+
),
|
348
|
+
(
|
349
|
+
ParquetValue::String(Arc::from("discount")),
|
350
|
+
ParquetValue::Null,
|
351
|
+
),
|
352
|
+
(
|
353
|
+
ParquetValue::String(Arc::from("tax")),
|
354
|
+
ParquetValue::Decimal256(BigInt::from(98765432109876543_i64), 10),
|
355
|
+
),
|
356
|
+
]),
|
357
|
+
// Struct with decimals
|
358
|
+
ParquetValue::Record({
|
359
|
+
let mut map = IndexMap::new();
|
360
|
+
map.insert(
|
361
|
+
Arc::from("amount"),
|
362
|
+
ParquetValue::Decimal128(999999999999999999, 2),
|
363
|
+
);
|
364
|
+
map.insert(
|
365
|
+
Arc::from("large_amount"),
|
366
|
+
ParquetValue::Decimal256(
|
367
|
+
BigInt::parse_bytes(b"12345678901234567890123456789012345678901234", 10)
|
368
|
+
.unwrap(),
|
369
|
+
6,
|
370
|
+
),
|
371
|
+
);
|
372
|
+
map
|
373
|
+
}),
|
374
|
+
],
|
375
|
+
vec![
|
376
|
+
// Empty list
|
377
|
+
ParquetValue::List(vec![]),
|
378
|
+
// Empty map
|
379
|
+
ParquetValue::Map(vec![]),
|
380
|
+
// Null struct
|
381
|
+
ParquetValue::Null,
|
382
|
+
],
|
383
|
+
];
|
384
|
+
|
385
|
+
let mut buffer = Vec::new();
|
386
|
+
{
|
387
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
388
|
+
writer.write_rows(rows.clone()).unwrap();
|
389
|
+
writer.close().unwrap();
|
390
|
+
}
|
391
|
+
|
392
|
+
// Read back and verify
|
393
|
+
let bytes = Bytes::from(buffer);
|
394
|
+
let reader = Reader::new(bytes);
|
395
|
+
|
396
|
+
let read_rows: Vec<_> = reader
|
397
|
+
.read_rows()
|
398
|
+
.unwrap()
|
399
|
+
.collect::<Result<Vec<_>>>()
|
400
|
+
.unwrap();
|
401
|
+
|
402
|
+
assert_eq!(read_rows.len(), rows.len());
|
403
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
404
|
+
assert_eq!(expected, actual);
|
405
|
+
}
|
406
|
+
}
|
407
|
+
|
408
|
+
#[test]
|
409
|
+
fn test_decimal_precision_edge_cases() {
|
410
|
+
// Test decimal values at the edge of precision limits
|
411
|
+
let schema = SchemaBuilder::new()
|
412
|
+
.with_root(SchemaNode::Struct {
|
413
|
+
name: "root".to_string(),
|
414
|
+
nullable: false,
|
415
|
+
fields: vec![
|
416
|
+
SchemaNode::Primitive {
|
417
|
+
name: "dec_9_2".to_string(),
|
418
|
+
primitive_type: PrimitiveType::Decimal128(9, 2),
|
419
|
+
nullable: false,
|
420
|
+
format: None,
|
421
|
+
},
|
422
|
+
SchemaNode::Primitive {
|
423
|
+
name: "dec_18_0".to_string(),
|
424
|
+
primitive_type: PrimitiveType::Decimal128(18, 0),
|
425
|
+
nullable: false,
|
426
|
+
format: None,
|
427
|
+
},
|
428
|
+
SchemaNode::Primitive {
|
429
|
+
name: "dec_38_10".to_string(),
|
430
|
+
primitive_type: PrimitiveType::Decimal128(38, 10),
|
431
|
+
nullable: false,
|
432
|
+
format: None,
|
433
|
+
},
|
434
|
+
],
|
435
|
+
})
|
436
|
+
.build()
|
437
|
+
.unwrap();
|
438
|
+
|
439
|
+
let decimal_values = vec![
|
440
|
+
// Maximum positive values for each precision
|
441
|
+
vec![
|
442
|
+
ParquetValue::Decimal128(999_999_999, 2), // 9999999.99
|
443
|
+
ParquetValue::Decimal128(999999999999999999, 0), // 18 digits
|
444
|
+
ParquetValue::Decimal128(99_999_999_999_999_999_999_999_999_999_999_999_999i128, 10), // 38 digits, 10 scale
|
445
|
+
],
|
446
|
+
// Maximum negative values
|
447
|
+
vec![
|
448
|
+
ParquetValue::Decimal128(-999_999_999, 2), // -9999999.99
|
449
|
+
ParquetValue::Decimal128(-999999999999999999, 0), // -18 digits
|
450
|
+
ParquetValue::Decimal128(-99_999_999_999_999_999_999_999_999_999_999_999_999i128, 10), // -38 digits, 10 scale
|
451
|
+
],
|
452
|
+
// Zero values
|
453
|
+
vec![
|
454
|
+
ParquetValue::Decimal128(0, 2),
|
455
|
+
ParquetValue::Decimal128(0, 0),
|
456
|
+
ParquetValue::Decimal128(0, 10),
|
457
|
+
],
|
458
|
+
// Small values
|
459
|
+
vec![
|
460
|
+
ParquetValue::Decimal128(1, 2), // 0.01
|
461
|
+
ParquetValue::Decimal128(1, 0), // 1
|
462
|
+
ParquetValue::Decimal128(1, 10), // 0.0000000001
|
463
|
+
],
|
464
|
+
];
|
465
|
+
|
466
|
+
let mut buffer = Vec::new();
|
467
|
+
{
|
468
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
469
|
+
writer.write_rows(decimal_values.clone()).unwrap();
|
470
|
+
writer.close().unwrap();
|
471
|
+
}
|
472
|
+
|
473
|
+
// Read back and verify
|
474
|
+
let bytes = Bytes::from(buffer);
|
475
|
+
let reader = Reader::new(bytes);
|
476
|
+
|
477
|
+
let read_rows: Vec<_> = reader
|
478
|
+
.read_rows()
|
479
|
+
.unwrap()
|
480
|
+
.collect::<Result<Vec<_>>>()
|
481
|
+
.unwrap();
|
482
|
+
|
483
|
+
assert_eq!(read_rows.len(), decimal_values.len());
|
484
|
+
|
485
|
+
for (expected, actual) in decimal_values.iter().zip(read_rows.iter()) {
|
486
|
+
assert_eq!(expected, actual);
|
487
|
+
}
|
488
|
+
}
|