parquet-tyfoom 0.8.0
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 +7 -0
- data/Cargo.lock +1854 -0
- data/Cargo.toml +3 -0
- data/Gemfile +21 -0
- data/LICENSE +21 -0
- data/README.md +428 -0
- data/Rakefile +43 -0
- data/ext/parquet/Cargo.toml +39 -0
- data/ext/parquet/build.rs +5 -0
- data/ext/parquet/extconf.rb +4 -0
- data/ext/parquet/src/adapter_ffi.rs +297 -0
- data/ext/parquet/src/allocator.rs +13 -0
- data/ext/parquet/src/lib.rs +24 -0
- data/ext/parquet-core/Cargo.toml +24 -0
- data/ext/parquet-core/src/arrow_conversion.rs +1243 -0
- data/ext/parquet-core/src/error.rs +189 -0
- data/ext/parquet-core/src/lib.rs +60 -0
- data/ext/parquet-core/src/reader.rs +368 -0
- data/ext/parquet-core/src/schema.rs +452 -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 +190 -0
- data/ext/parquet-core/src/value.rs +220 -0
- data/ext/parquet-core/src/writer.rs +1241 -0
- data/ext/parquet-core/tests/arrow_conversion_tests.rs +484 -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 +431 -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 +540 -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/review_regressions.rs +787 -0
- data/ext/parquet-core/tests/roundtrip_correctness.rs +279 -0
- data/ext/parquet-core/tests/schema_comprehensive_tests.rs +542 -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 +24 -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/chunk_reader.rs +237 -0
- data/ext/parquet-ruby-adapter/src/converter.rs +1734 -0
- data/ext/parquet-ruby-adapter/src/error.rs +141 -0
- data/ext/parquet-ruby-adapter/src/io.rs +432 -0
- data/ext/parquet-ruby-adapter/src/lib.rs +91 -0
- data/ext/parquet-ruby-adapter/src/logger.rs +67 -0
- data/ext/parquet-ruby-adapter/src/metadata.rs +529 -0
- data/ext/parquet-ruby-adapter/src/reader.rs +339 -0
- data/ext/parquet-ruby-adapter/src/schema.rs +884 -0
- data/ext/parquet-ruby-adapter/src/string_cache.rs +115 -0
- data/ext/parquet-ruby-adapter/src/string_cache_test.rs +122 -0
- data/ext/parquet-ruby-adapter/src/string_storage.rs +632 -0
- data/ext/parquet-ruby-adapter/src/try_into_value.rs +91 -0
- data/ext/parquet-ruby-adapter/src/types.rs +98 -0
- data/ext/parquet-ruby-adapter/src/utils.rs +280 -0
- data/ext/parquet-ruby-adapter/src/writer.rs +625 -0
- data/lib/parquet/schema.rb +262 -0
- data/lib/parquet/version.rb +3 -0
- data/lib/parquet.rb +11 -0
- data/lib/parquet.rbi +181 -0
- metadata +165 -0
|
@@ -0,0 +1,547 @@
|
|
|
1
|
+
use bytes::Bytes;
|
|
2
|
+
use ordered_float::OrderedFloat;
|
|
3
|
+
use parquet_core::*;
|
|
4
|
+
use triomphe::Arc;
|
|
5
|
+
|
|
6
|
+
// =============================================================================
|
|
7
|
+
// Boolean Type Tests
|
|
8
|
+
// =============================================================================
|
|
9
|
+
|
|
10
|
+
#[test]
|
|
11
|
+
fn test_boolean_values() {
|
|
12
|
+
let schema = SchemaBuilder::new()
|
|
13
|
+
.with_root(SchemaNode::Struct {
|
|
14
|
+
name: "root".to_string(),
|
|
15
|
+
nullable: false,
|
|
16
|
+
fields: vec![
|
|
17
|
+
SchemaNode::Primitive {
|
|
18
|
+
name: "bool_field".to_string(),
|
|
19
|
+
primitive_type: PrimitiveType::Boolean,
|
|
20
|
+
nullable: false,
|
|
21
|
+
format: None,
|
|
22
|
+
},
|
|
23
|
+
SchemaNode::Primitive {
|
|
24
|
+
name: "nullable_bool".to_string(),
|
|
25
|
+
primitive_type: PrimitiveType::Boolean,
|
|
26
|
+
nullable: true,
|
|
27
|
+
format: None,
|
|
28
|
+
},
|
|
29
|
+
],
|
|
30
|
+
})
|
|
31
|
+
.build()
|
|
32
|
+
.unwrap();
|
|
33
|
+
|
|
34
|
+
let rows = vec![
|
|
35
|
+
vec![ParquetValue::Boolean(true), ParquetValue::Boolean(true)],
|
|
36
|
+
vec![ParquetValue::Boolean(false), ParquetValue::Boolean(false)],
|
|
37
|
+
vec![ParquetValue::Boolean(true), ParquetValue::Null],
|
|
38
|
+
vec![ParquetValue::Boolean(false), ParquetValue::Null],
|
|
39
|
+
// Many repeated values to test encoding efficiency
|
|
40
|
+
vec![ParquetValue::Boolean(true), ParquetValue::Boolean(true)],
|
|
41
|
+
vec![ParquetValue::Boolean(true), ParquetValue::Boolean(true)],
|
|
42
|
+
vec![ParquetValue::Boolean(true), ParquetValue::Boolean(true)],
|
|
43
|
+
vec![ParquetValue::Boolean(false), ParquetValue::Boolean(false)],
|
|
44
|
+
vec![ParquetValue::Boolean(false), ParquetValue::Boolean(false)],
|
|
45
|
+
];
|
|
46
|
+
|
|
47
|
+
let mut buffer = Vec::new();
|
|
48
|
+
{
|
|
49
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
|
50
|
+
writer.write_rows(rows.clone()).unwrap();
|
|
51
|
+
writer.close().unwrap();
|
|
52
|
+
}
|
|
53
|
+
|
|
54
|
+
// Read back and verify
|
|
55
|
+
let bytes = Bytes::from(buffer);
|
|
56
|
+
let reader = Reader::new(bytes);
|
|
57
|
+
|
|
58
|
+
let read_rows: Vec<_> = reader
|
|
59
|
+
.read_rows()
|
|
60
|
+
.unwrap()
|
|
61
|
+
.collect::<Result<Vec<_>>>()
|
|
62
|
+
.unwrap();
|
|
63
|
+
|
|
64
|
+
assert_eq!(rows.len(), read_rows.len());
|
|
65
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
|
66
|
+
assert_eq!(expected, actual);
|
|
67
|
+
}
|
|
68
|
+
}
|
|
69
|
+
|
|
70
|
+
#[test]
|
|
71
|
+
fn test_boolean_in_complex_types() {
|
|
72
|
+
let schema = SchemaBuilder::new()
|
|
73
|
+
.with_root(SchemaNode::Struct {
|
|
74
|
+
name: "root".to_string(),
|
|
75
|
+
nullable: false,
|
|
76
|
+
fields: vec![
|
|
77
|
+
SchemaNode::List {
|
|
78
|
+
name: "bool_list".to_string(),
|
|
79
|
+
nullable: false,
|
|
80
|
+
item: Box::new(SchemaNode::Primitive {
|
|
81
|
+
name: "item".to_string(),
|
|
82
|
+
primitive_type: PrimitiveType::Boolean,
|
|
83
|
+
nullable: false,
|
|
84
|
+
format: None,
|
|
85
|
+
}),
|
|
86
|
+
},
|
|
87
|
+
SchemaNode::Map {
|
|
88
|
+
name: "bool_map".to_string(),
|
|
89
|
+
nullable: false,
|
|
90
|
+
key: Box::new(SchemaNode::Primitive {
|
|
91
|
+
name: "key".to_string(),
|
|
92
|
+
primitive_type: PrimitiveType::String,
|
|
93
|
+
nullable: false,
|
|
94
|
+
format: None,
|
|
95
|
+
}),
|
|
96
|
+
value: Box::new(SchemaNode::Primitive {
|
|
97
|
+
name: "value".to_string(),
|
|
98
|
+
primitive_type: PrimitiveType::Boolean,
|
|
99
|
+
nullable: true,
|
|
100
|
+
format: None,
|
|
101
|
+
}),
|
|
102
|
+
},
|
|
103
|
+
],
|
|
104
|
+
})
|
|
105
|
+
.build()
|
|
106
|
+
.unwrap();
|
|
107
|
+
|
|
108
|
+
let rows = vec![
|
|
109
|
+
vec![
|
|
110
|
+
ParquetValue::List(vec![
|
|
111
|
+
ParquetValue::Boolean(true),
|
|
112
|
+
ParquetValue::Boolean(false),
|
|
113
|
+
ParquetValue::Boolean(true),
|
|
114
|
+
]),
|
|
115
|
+
ParquetValue::Map(vec![
|
|
116
|
+
(
|
|
117
|
+
ParquetValue::String(Arc::from("enabled")),
|
|
118
|
+
ParquetValue::Boolean(true),
|
|
119
|
+
),
|
|
120
|
+
(
|
|
121
|
+
ParquetValue::String(Arc::from("disabled")),
|
|
122
|
+
ParquetValue::Boolean(false),
|
|
123
|
+
),
|
|
124
|
+
(
|
|
125
|
+
ParquetValue::String(Arc::from("unknown")),
|
|
126
|
+
ParquetValue::Null,
|
|
127
|
+
),
|
|
128
|
+
]),
|
|
129
|
+
],
|
|
130
|
+
vec![ParquetValue::List(vec![]), ParquetValue::Map(vec![])],
|
|
131
|
+
];
|
|
132
|
+
|
|
133
|
+
let mut buffer = Vec::new();
|
|
134
|
+
{
|
|
135
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
|
136
|
+
writer.write_rows(rows.clone()).unwrap();
|
|
137
|
+
writer.close().unwrap();
|
|
138
|
+
}
|
|
139
|
+
|
|
140
|
+
// Read back and verify
|
|
141
|
+
let bytes = Bytes::from(buffer);
|
|
142
|
+
let reader = Reader::new(bytes);
|
|
143
|
+
|
|
144
|
+
let read_rows: Vec<_> = reader
|
|
145
|
+
.read_rows()
|
|
146
|
+
.unwrap()
|
|
147
|
+
.collect::<Result<Vec<_>>>()
|
|
148
|
+
.unwrap();
|
|
149
|
+
|
|
150
|
+
assert_eq!(rows.len(), read_rows.len());
|
|
151
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
|
152
|
+
assert_eq!(expected, actual);
|
|
153
|
+
}
|
|
154
|
+
}
|
|
155
|
+
|
|
156
|
+
// =============================================================================
|
|
157
|
+
// String Type Tests
|
|
158
|
+
// =============================================================================
|
|
159
|
+
|
|
160
|
+
#[test]
|
|
161
|
+
fn test_string_roundtrip() {
|
|
162
|
+
let schema = SchemaBuilder::new()
|
|
163
|
+
.with_root(SchemaNode::Struct {
|
|
164
|
+
name: "root".to_string(),
|
|
165
|
+
nullable: false,
|
|
166
|
+
fields: vec![SchemaNode::Primitive {
|
|
167
|
+
name: "text".to_string(),
|
|
168
|
+
primitive_type: PrimitiveType::String,
|
|
169
|
+
nullable: false,
|
|
170
|
+
format: None,
|
|
171
|
+
}],
|
|
172
|
+
})
|
|
173
|
+
.build()
|
|
174
|
+
.unwrap();
|
|
175
|
+
|
|
176
|
+
let repeated_str = "Repeated".repeat(100);
|
|
177
|
+
let long_str = "x".repeat(10_000);
|
|
178
|
+
|
|
179
|
+
let test_strings = vec![
|
|
180
|
+
"", // Empty string
|
|
181
|
+
"Hello, World!", // ASCII
|
|
182
|
+
"Hello, 世界! 🦀", // Unicode with emoji
|
|
183
|
+
"Line1\nLine2\rLine3\r\nLine4", // Various line endings
|
|
184
|
+
"\t\t\tTabbed\t\t\t", // Tabs
|
|
185
|
+
"Special chars: !@#$%^&*()_+-=[]{}|;':\",./<>?",
|
|
186
|
+
repeated_str.as_str(), // Long repeated string
|
|
187
|
+
long_str.as_str(), // Very long string
|
|
188
|
+
];
|
|
189
|
+
|
|
190
|
+
let rows: Vec<Vec<ParquetValue>> = test_strings
|
|
191
|
+
.into_iter()
|
|
192
|
+
.map(|s| vec![ParquetValue::String(Arc::from(s))])
|
|
193
|
+
.collect();
|
|
194
|
+
|
|
195
|
+
let mut buffer = Vec::new();
|
|
196
|
+
{
|
|
197
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
|
198
|
+
writer.write_rows(rows.clone()).unwrap();
|
|
199
|
+
writer.close().unwrap();
|
|
200
|
+
}
|
|
201
|
+
|
|
202
|
+
// Read back and verify
|
|
203
|
+
let bytes = Bytes::from(buffer);
|
|
204
|
+
let reader = Reader::new(bytes);
|
|
205
|
+
|
|
206
|
+
let read_rows: Vec<_> = reader
|
|
207
|
+
.read_rows()
|
|
208
|
+
.unwrap()
|
|
209
|
+
.collect::<Result<Vec<_>>>()
|
|
210
|
+
.unwrap();
|
|
211
|
+
|
|
212
|
+
assert_eq!(rows.len(), read_rows.len());
|
|
213
|
+
for (expected, actual) in rows.iter().zip(read_rows.iter()) {
|
|
214
|
+
assert_eq!(expected, actual);
|
|
215
|
+
}
|
|
216
|
+
}
|
|
217
|
+
|
|
218
|
+
// =============================================================================
|
|
219
|
+
// Numeric Type Tests
|
|
220
|
+
// =============================================================================
|
|
221
|
+
|
|
222
|
+
#[test]
|
|
223
|
+
fn test_float_special_values() {
|
|
224
|
+
// Test NaN, Infinity, and -Infinity handling
|
|
225
|
+
let schema = SchemaBuilder::new()
|
|
226
|
+
.with_root(SchemaNode::Struct {
|
|
227
|
+
name: "root".to_string(),
|
|
228
|
+
nullable: false,
|
|
229
|
+
fields: vec![
|
|
230
|
+
SchemaNode::Primitive {
|
|
231
|
+
name: "float32_val".to_string(),
|
|
232
|
+
primitive_type: PrimitiveType::Float32,
|
|
233
|
+
nullable: false,
|
|
234
|
+
format: None,
|
|
235
|
+
},
|
|
236
|
+
SchemaNode::Primitive {
|
|
237
|
+
name: "float64_val".to_string(),
|
|
238
|
+
primitive_type: PrimitiveType::Float64,
|
|
239
|
+
nullable: false,
|
|
240
|
+
format: None,
|
|
241
|
+
},
|
|
242
|
+
],
|
|
243
|
+
})
|
|
244
|
+
.build()
|
|
245
|
+
.unwrap();
|
|
246
|
+
|
|
247
|
+
let special_values = vec![
|
|
248
|
+
// Normal values
|
|
249
|
+
vec![
|
|
250
|
+
ParquetValue::Float32(OrderedFloat(1.23f32)),
|
|
251
|
+
ParquetValue::Float64(OrderedFloat(4.56f64)),
|
|
252
|
+
],
|
|
253
|
+
// Positive infinity
|
|
254
|
+
vec![
|
|
255
|
+
ParquetValue::Float32(OrderedFloat(f32::INFINITY)),
|
|
256
|
+
ParquetValue::Float64(OrderedFloat(f64::INFINITY)),
|
|
257
|
+
],
|
|
258
|
+
// Negative infinity
|
|
259
|
+
vec![
|
|
260
|
+
ParquetValue::Float32(OrderedFloat(f32::NEG_INFINITY)),
|
|
261
|
+
ParquetValue::Float64(OrderedFloat(f64::NEG_INFINITY)),
|
|
262
|
+
],
|
|
263
|
+
// NaN values
|
|
264
|
+
vec![
|
|
265
|
+
ParquetValue::Float32(OrderedFloat(f32::NAN)),
|
|
266
|
+
ParquetValue::Float64(OrderedFloat(f64::NAN)),
|
|
267
|
+
],
|
|
268
|
+
// Zero values
|
|
269
|
+
vec![
|
|
270
|
+
ParquetValue::Float32(OrderedFloat(0.0f32)),
|
|
271
|
+
ParquetValue::Float64(OrderedFloat(0.0f64)),
|
|
272
|
+
],
|
|
273
|
+
// Negative zero
|
|
274
|
+
vec![
|
|
275
|
+
ParquetValue::Float32(OrderedFloat(-0.0f32)),
|
|
276
|
+
ParquetValue::Float64(OrderedFloat(-0.0f64)),
|
|
277
|
+
],
|
|
278
|
+
// Very small values
|
|
279
|
+
vec![
|
|
280
|
+
ParquetValue::Float32(OrderedFloat(f32::MIN_POSITIVE)),
|
|
281
|
+
ParquetValue::Float64(OrderedFloat(f64::MIN_POSITIVE)),
|
|
282
|
+
],
|
|
283
|
+
// Very large values
|
|
284
|
+
vec![
|
|
285
|
+
ParquetValue::Float32(OrderedFloat(f32::MAX)),
|
|
286
|
+
ParquetValue::Float64(OrderedFloat(f64::MAX)),
|
|
287
|
+
],
|
|
288
|
+
];
|
|
289
|
+
|
|
290
|
+
let mut buffer = Vec::new();
|
|
291
|
+
{
|
|
292
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
|
293
|
+
writer.write_rows(special_values.clone()).unwrap();
|
|
294
|
+
writer.close().unwrap();
|
|
295
|
+
}
|
|
296
|
+
|
|
297
|
+
// Read back and verify
|
|
298
|
+
let bytes = Bytes::from(buffer);
|
|
299
|
+
let reader = Reader::new(bytes);
|
|
300
|
+
|
|
301
|
+
let read_rows: Vec<_> = reader
|
|
302
|
+
.read_rows()
|
|
303
|
+
.unwrap()
|
|
304
|
+
.collect::<Result<Vec<_>>>()
|
|
305
|
+
.unwrap();
|
|
306
|
+
|
|
307
|
+
assert_eq!(read_rows.len(), special_values.len());
|
|
308
|
+
|
|
309
|
+
// Verify special values are preserved
|
|
310
|
+
for (expected, actual) in special_values.iter().zip(read_rows.iter()) {
|
|
311
|
+
for (exp_val, act_val) in expected.iter().zip(actual.iter()) {
|
|
312
|
+
match (exp_val, act_val) {
|
|
313
|
+
(
|
|
314
|
+
ParquetValue::Float32(OrderedFloat(e)),
|
|
315
|
+
ParquetValue::Float32(OrderedFloat(a)),
|
|
316
|
+
) => {
|
|
317
|
+
if e.is_nan() {
|
|
318
|
+
assert!(a.is_nan());
|
|
319
|
+
} else {
|
|
320
|
+
assert_eq!(e, a);
|
|
321
|
+
}
|
|
322
|
+
}
|
|
323
|
+
(
|
|
324
|
+
ParquetValue::Float64(OrderedFloat(e)),
|
|
325
|
+
ParquetValue::Float64(OrderedFloat(a)),
|
|
326
|
+
) => {
|
|
327
|
+
if e.is_nan() {
|
|
328
|
+
assert!(a.is_nan());
|
|
329
|
+
} else {
|
|
330
|
+
assert_eq!(e, a);
|
|
331
|
+
}
|
|
332
|
+
}
|
|
333
|
+
_ => panic!("Type mismatch"),
|
|
334
|
+
}
|
|
335
|
+
}
|
|
336
|
+
}
|
|
337
|
+
}
|
|
338
|
+
|
|
339
|
+
// Macro to generate integer boundary tests for each type
|
|
340
|
+
macro_rules! test_integer_boundaries {
|
|
341
|
+
($test_name:ident, $type_name:expr, $primitive_type:expr, $rust_type:ty, $parquet_variant:ident, $test_values:expr) => {
|
|
342
|
+
#[test]
|
|
343
|
+
fn $test_name() {
|
|
344
|
+
let schema = SchemaBuilder::new()
|
|
345
|
+
.with_root(SchemaNode::Struct {
|
|
346
|
+
name: "root".to_string(),
|
|
347
|
+
nullable: false,
|
|
348
|
+
fields: vec![SchemaNode::Primitive {
|
|
349
|
+
name: $type_name.to_string(),
|
|
350
|
+
primitive_type: $primitive_type,
|
|
351
|
+
nullable: false,
|
|
352
|
+
format: None,
|
|
353
|
+
}],
|
|
354
|
+
})
|
|
355
|
+
.build()
|
|
356
|
+
.unwrap();
|
|
357
|
+
|
|
358
|
+
let boundary_values: Vec<Vec<ParquetValue>> = $test_values
|
|
359
|
+
.into_iter()
|
|
360
|
+
.map(|v| vec![ParquetValue::$parquet_variant(v)])
|
|
361
|
+
.collect();
|
|
362
|
+
|
|
363
|
+
let mut buffer = Vec::new();
|
|
364
|
+
{
|
|
365
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
|
366
|
+
writer.write_rows(boundary_values.clone()).unwrap();
|
|
367
|
+
writer.close().unwrap();
|
|
368
|
+
}
|
|
369
|
+
|
|
370
|
+
// Read back and verify
|
|
371
|
+
let bytes = Bytes::from(buffer);
|
|
372
|
+
let reader = Reader::new(bytes);
|
|
373
|
+
|
|
374
|
+
let read_rows: Vec<_> = reader
|
|
375
|
+
.read_rows()
|
|
376
|
+
.unwrap()
|
|
377
|
+
.collect::<Result<Vec<_>>>()
|
|
378
|
+
.unwrap();
|
|
379
|
+
|
|
380
|
+
assert_eq!(read_rows.len(), boundary_values.len());
|
|
381
|
+
for (expected, actual) in boundary_values.iter().zip(read_rows.iter()) {
|
|
382
|
+
assert_eq!(expected, actual);
|
|
383
|
+
}
|
|
384
|
+
}
|
|
385
|
+
};
|
|
386
|
+
}
|
|
387
|
+
|
|
388
|
+
// Generate tests for all integer types
|
|
389
|
+
test_integer_boundaries!(
|
|
390
|
+
test_int8_boundaries,
|
|
391
|
+
"int8",
|
|
392
|
+
PrimitiveType::Int8,
|
|
393
|
+
i8,
|
|
394
|
+
Int8,
|
|
395
|
+
vec![i8::MIN, i8::MAX, 0, -1, 42, -42]
|
|
396
|
+
);
|
|
397
|
+
|
|
398
|
+
test_integer_boundaries!(
|
|
399
|
+
test_int16_boundaries,
|
|
400
|
+
"int16",
|
|
401
|
+
PrimitiveType::Int16,
|
|
402
|
+
i16,
|
|
403
|
+
Int16,
|
|
404
|
+
vec![i16::MIN, i16::MAX, 0, -1, 1000, -1000]
|
|
405
|
+
);
|
|
406
|
+
|
|
407
|
+
test_integer_boundaries!(
|
|
408
|
+
test_int32_boundaries,
|
|
409
|
+
"int32",
|
|
410
|
+
PrimitiveType::Int32,
|
|
411
|
+
i32,
|
|
412
|
+
Int32,
|
|
413
|
+
vec![i32::MIN, i32::MAX, 0, -1, 1_000_000, -1_000_000]
|
|
414
|
+
);
|
|
415
|
+
|
|
416
|
+
test_integer_boundaries!(
|
|
417
|
+
test_int64_boundaries,
|
|
418
|
+
"int64",
|
|
419
|
+
PrimitiveType::Int64,
|
|
420
|
+
i64,
|
|
421
|
+
Int64,
|
|
422
|
+
vec![
|
|
423
|
+
i64::MIN,
|
|
424
|
+
i64::MAX,
|
|
425
|
+
0,
|
|
426
|
+
-1,
|
|
427
|
+
1_000_000_000_000,
|
|
428
|
+
-1_000_000_000_000
|
|
429
|
+
]
|
|
430
|
+
);
|
|
431
|
+
|
|
432
|
+
test_integer_boundaries!(
|
|
433
|
+
test_uint8_boundaries,
|
|
434
|
+
"uint8",
|
|
435
|
+
PrimitiveType::UInt8,
|
|
436
|
+
u8,
|
|
437
|
+
UInt8,
|
|
438
|
+
vec![u8::MIN, u8::MAX, 0, 1, 128, 255]
|
|
439
|
+
);
|
|
440
|
+
|
|
441
|
+
test_integer_boundaries!(
|
|
442
|
+
test_uint16_boundaries,
|
|
443
|
+
"uint16",
|
|
444
|
+
PrimitiveType::UInt16,
|
|
445
|
+
u16,
|
|
446
|
+
UInt16,
|
|
447
|
+
vec![u16::MIN, u16::MAX, 0, 1, 32768, 65535]
|
|
448
|
+
);
|
|
449
|
+
|
|
450
|
+
test_integer_boundaries!(
|
|
451
|
+
test_uint32_boundaries,
|
|
452
|
+
"uint32",
|
|
453
|
+
PrimitiveType::UInt32,
|
|
454
|
+
u32,
|
|
455
|
+
UInt32,
|
|
456
|
+
vec![u32::MIN, u32::MAX, 0, 1, 2_147_483_648, 4_294_967_295]
|
|
457
|
+
);
|
|
458
|
+
|
|
459
|
+
test_integer_boundaries!(
|
|
460
|
+
test_uint64_boundaries,
|
|
461
|
+
"uint64",
|
|
462
|
+
PrimitiveType::UInt64,
|
|
463
|
+
u64,
|
|
464
|
+
UInt64,
|
|
465
|
+
vec![
|
|
466
|
+
u64::MIN,
|
|
467
|
+
u64::MAX,
|
|
468
|
+
0,
|
|
469
|
+
1,
|
|
470
|
+
9_223_372_036_854_775_808,
|
|
471
|
+
18_446_744_073_709_551_615
|
|
472
|
+
]
|
|
473
|
+
);
|
|
474
|
+
|
|
475
|
+
#[test]
|
|
476
|
+
fn test_mixed_numeric_types() {
|
|
477
|
+
// Test writing different numeric types and reading them back
|
|
478
|
+
let schema = SchemaBuilder::new()
|
|
479
|
+
.with_root(SchemaNode::Struct {
|
|
480
|
+
name: "root".to_string(),
|
|
481
|
+
nullable: false,
|
|
482
|
+
fields: vec![
|
|
483
|
+
SchemaNode::Primitive {
|
|
484
|
+
name: "as_int8".to_string(),
|
|
485
|
+
primitive_type: PrimitiveType::Int8,
|
|
486
|
+
nullable: false,
|
|
487
|
+
format: None,
|
|
488
|
+
},
|
|
489
|
+
SchemaNode::Primitive {
|
|
490
|
+
name: "as_int32".to_string(),
|
|
491
|
+
primitive_type: PrimitiveType::Int32,
|
|
492
|
+
nullable: false,
|
|
493
|
+
format: None,
|
|
494
|
+
},
|
|
495
|
+
SchemaNode::Primitive {
|
|
496
|
+
name: "as_float".to_string(),
|
|
497
|
+
primitive_type: PrimitiveType::Float32,
|
|
498
|
+
nullable: false,
|
|
499
|
+
format: None,
|
|
500
|
+
},
|
|
501
|
+
],
|
|
502
|
+
})
|
|
503
|
+
.build()
|
|
504
|
+
.unwrap();
|
|
505
|
+
|
|
506
|
+
// Test values that fit in all types
|
|
507
|
+
let test_values = vec![
|
|
508
|
+
vec![
|
|
509
|
+
ParquetValue::Int8(42),
|
|
510
|
+
ParquetValue::Int32(42),
|
|
511
|
+
ParquetValue::Float32(OrderedFloat(42.0)),
|
|
512
|
+
],
|
|
513
|
+
vec![
|
|
514
|
+
ParquetValue::Int8(-50),
|
|
515
|
+
ParquetValue::Int32(-50),
|
|
516
|
+
ParquetValue::Float32(OrderedFloat(-50.0)),
|
|
517
|
+
],
|
|
518
|
+
vec![
|
|
519
|
+
ParquetValue::Int8(0),
|
|
520
|
+
ParquetValue::Int32(0),
|
|
521
|
+
ParquetValue::Float32(OrderedFloat(0.0)),
|
|
522
|
+
],
|
|
523
|
+
];
|
|
524
|
+
|
|
525
|
+
let mut buffer = Vec::new();
|
|
526
|
+
{
|
|
527
|
+
let mut writer = Writer::new(&mut buffer, schema).unwrap();
|
|
528
|
+
writer.write_rows(test_values.clone()).unwrap();
|
|
529
|
+
writer.close().unwrap();
|
|
530
|
+
}
|
|
531
|
+
|
|
532
|
+
// Read back and verify
|
|
533
|
+
let bytes = Bytes::from(buffer);
|
|
534
|
+
let reader = Reader::new(bytes);
|
|
535
|
+
|
|
536
|
+
let read_rows: Vec<_> = reader
|
|
537
|
+
.read_rows()
|
|
538
|
+
.unwrap()
|
|
539
|
+
.collect::<Result<Vec<_>>>()
|
|
540
|
+
.unwrap();
|
|
541
|
+
|
|
542
|
+
assert_eq!(read_rows.len(), test_values.len());
|
|
543
|
+
|
|
544
|
+
for (expected, actual) in test_values.iter().zip(read_rows.iter()) {
|
|
545
|
+
assert_eq!(expected, actual);
|
|
546
|
+
}
|
|
547
|
+
}
|