parquet 0.2.12-arm64-darwin → 0.2.15-arm64-darwin
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/Gemfile +5 -3
- data/Rakefile +16 -0
- data/lib/parquet/3.2/parquet.bundle +0 -0
- data/lib/parquet/3.3/parquet.bundle +0 -0
- data/lib/parquet/3.4/parquet.bundle +0 -0
- data/lib/parquet/version.rb +1 -1
- data/lib/parquet.rb +6 -1
- metadata +14 -45
- data/Cargo.lock +0 -1449
- data/Cargo.toml +0 -3
- data/ext/parquet/Cargo.toml +0 -28
- data/ext/parquet/extconf.rb +0 -4
- data/ext/parquet/src/allocator.rs +0 -13
- data/ext/parquet/src/enumerator.rs +0 -52
- data/ext/parquet/src/header_cache.rs +0 -100
- data/ext/parquet/src/lib.rs +0 -29
- data/ext/parquet/src/reader/mod.rs +0 -44
- data/ext/parquet/src/reader/parquet_column_reader.rs +0 -214
- data/ext/parquet/src/reader/parquet_row_reader.rs +0 -157
- data/ext/parquet/src/ruby_integration.rs +0 -77
- data/ext/parquet/src/ruby_reader.rs +0 -171
- data/ext/parquet/src/types/core_types.rs +0 -75
- data/ext/parquet/src/types/mod.rs +0 -30
- data/ext/parquet/src/types/parquet_value.rs +0 -462
- data/ext/parquet/src/types/record_types.rs +0 -204
- data/ext/parquet/src/types/timestamp.rs +0 -85
- data/ext/parquet/src/types/type_conversion.rs +0 -809
- data/ext/parquet/src/types/writer_types.rs +0 -283
- data/ext/parquet/src/utils.rs +0 -148
- data/ext/parquet/src/writer/mod.rs +0 -575
@@ -1,462 +0,0 @@
|
|
1
|
-
use crate::{impl_date_conversion, impl_timestamp_array_conversion, impl_timestamp_conversion};
|
2
|
-
|
3
|
-
use super::*;
|
4
|
-
|
5
|
-
#[derive(Debug, Clone)]
|
6
|
-
pub enum ParquetValue {
|
7
|
-
Int8(i8),
|
8
|
-
Int16(i16),
|
9
|
-
Int32(i32),
|
10
|
-
Int64(i64),
|
11
|
-
UInt8(u8),
|
12
|
-
UInt16(u16),
|
13
|
-
UInt32(u32),
|
14
|
-
UInt64(u64),
|
15
|
-
Float16(f32), // f16 converted to f32
|
16
|
-
Float32(f32),
|
17
|
-
Float64(f64),
|
18
|
-
Boolean(bool),
|
19
|
-
String(String),
|
20
|
-
Bytes(Vec<u8>),
|
21
|
-
Date32(i32),
|
22
|
-
Date64(i64),
|
23
|
-
TimestampSecond(i64, Option<Arc<str>>),
|
24
|
-
TimestampMillis(i64, Option<Arc<str>>),
|
25
|
-
TimestampMicros(i64, Option<Arc<str>>),
|
26
|
-
TimestampNanos(i64, Option<Arc<str>>),
|
27
|
-
List(Vec<ParquetValue>),
|
28
|
-
Map(HashMap<ParquetValue, ParquetValue>),
|
29
|
-
Null,
|
30
|
-
}
|
31
|
-
|
32
|
-
impl PartialEq for ParquetValue {
|
33
|
-
fn eq(&self, other: &Self) -> bool {
|
34
|
-
match (self, other) {
|
35
|
-
(ParquetValue::Int8(a), ParquetValue::Int8(b)) => a == b,
|
36
|
-
(ParquetValue::Int16(a), ParquetValue::Int16(b)) => a == b,
|
37
|
-
(ParquetValue::Int32(a), ParquetValue::Int32(b)) => a == b,
|
38
|
-
(ParquetValue::Int64(a), ParquetValue::Int64(b)) => a == b,
|
39
|
-
(ParquetValue::UInt8(a), ParquetValue::UInt8(b)) => a == b,
|
40
|
-
(ParquetValue::UInt16(a), ParquetValue::UInt16(b)) => a == b,
|
41
|
-
(ParquetValue::UInt32(a), ParquetValue::UInt32(b)) => a == b,
|
42
|
-
(ParquetValue::UInt64(a), ParquetValue::UInt64(b)) => a == b,
|
43
|
-
(ParquetValue::Float16(a), ParquetValue::Float16(b)) => a == b,
|
44
|
-
(ParquetValue::Float32(a), ParquetValue::Float32(b)) => a == b,
|
45
|
-
(ParquetValue::Float64(a), ParquetValue::Float64(b)) => a == b,
|
46
|
-
(ParquetValue::Boolean(a), ParquetValue::Boolean(b)) => a == b,
|
47
|
-
(ParquetValue::String(a), ParquetValue::String(b)) => a == b,
|
48
|
-
(ParquetValue::Bytes(a), ParquetValue::Bytes(b)) => a == b,
|
49
|
-
(ParquetValue::Date32(a), ParquetValue::Date32(b)) => a == b,
|
50
|
-
(ParquetValue::Date64(a), ParquetValue::Date64(b)) => a == b,
|
51
|
-
(ParquetValue::TimestampSecond(a, _), ParquetValue::TimestampSecond(b, _)) => a == b,
|
52
|
-
(ParquetValue::TimestampMillis(a, _), ParquetValue::TimestampMillis(b, _)) => a == b,
|
53
|
-
(ParquetValue::TimestampMicros(a, _), ParquetValue::TimestampMicros(b, _)) => a == b,
|
54
|
-
(ParquetValue::TimestampNanos(a, _), ParquetValue::TimestampNanos(b, _)) => a == b,
|
55
|
-
(ParquetValue::List(a), ParquetValue::List(b)) => a == b,
|
56
|
-
(ParquetValue::Null, ParquetValue::Null) => true,
|
57
|
-
_ => false,
|
58
|
-
}
|
59
|
-
}
|
60
|
-
}
|
61
|
-
|
62
|
-
impl Eq for ParquetValue {}
|
63
|
-
|
64
|
-
impl std::hash::Hash for ParquetValue {
|
65
|
-
fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
|
66
|
-
match self {
|
67
|
-
ParquetValue::Int8(i) => i.hash(state),
|
68
|
-
ParquetValue::Int16(i) => i.hash(state),
|
69
|
-
ParquetValue::Int32(i) => i.hash(state),
|
70
|
-
ParquetValue::Int64(i) => i.hash(state),
|
71
|
-
ParquetValue::UInt8(i) => i.hash(state),
|
72
|
-
ParquetValue::UInt16(i) => i.hash(state),
|
73
|
-
ParquetValue::UInt32(i) => i.hash(state),
|
74
|
-
ParquetValue::UInt64(i) => i.hash(state),
|
75
|
-
ParquetValue::Float16(f) => f.to_bits().hash(state),
|
76
|
-
ParquetValue::Float32(f) => f.to_bits().hash(state),
|
77
|
-
ParquetValue::Float64(f) => f.to_bits().hash(state),
|
78
|
-
ParquetValue::Boolean(b) => b.hash(state),
|
79
|
-
ParquetValue::String(s) => s.hash(state),
|
80
|
-
ParquetValue::Bytes(b) => b.hash(state),
|
81
|
-
ParquetValue::Date32(d) => d.hash(state),
|
82
|
-
ParquetValue::Date64(d) => d.hash(state),
|
83
|
-
ParquetValue::TimestampSecond(ts, tz) => {
|
84
|
-
ts.hash(state);
|
85
|
-
tz.hash(state);
|
86
|
-
}
|
87
|
-
ParquetValue::TimestampMillis(ts, tz) => {
|
88
|
-
ts.hash(state);
|
89
|
-
tz.hash(state);
|
90
|
-
}
|
91
|
-
ParquetValue::TimestampMicros(ts, tz) => {
|
92
|
-
ts.hash(state);
|
93
|
-
tz.hash(state);
|
94
|
-
}
|
95
|
-
ParquetValue::TimestampNanos(ts, tz) => {
|
96
|
-
ts.hash(state);
|
97
|
-
tz.hash(state);
|
98
|
-
}
|
99
|
-
ParquetValue::List(l) => l.hash(state),
|
100
|
-
ParquetValue::Map(_m) => panic!("Map is not hashable"),
|
101
|
-
ParquetValue::Null => 0_i32.hash(state),
|
102
|
-
}
|
103
|
-
}
|
104
|
-
}
|
105
|
-
|
106
|
-
impl IntoValue for ParquetValue {
|
107
|
-
fn into_value_with(self, handle: &Ruby) -> Value {
|
108
|
-
match self {
|
109
|
-
ParquetValue::Int8(i) => i.into_value_with(handle),
|
110
|
-
ParquetValue::Int16(i) => i.into_value_with(handle),
|
111
|
-
ParquetValue::Int32(i) => i.into_value_with(handle),
|
112
|
-
ParquetValue::Int64(i) => i.into_value_with(handle),
|
113
|
-
ParquetValue::UInt8(i) => i.into_value_with(handle),
|
114
|
-
ParquetValue::UInt16(i) => i.into_value_with(handle),
|
115
|
-
ParquetValue::UInt32(i) => i.into_value_with(handle),
|
116
|
-
ParquetValue::UInt64(i) => i.into_value_with(handle),
|
117
|
-
ParquetValue::Float16(f) => f.into_value_with(handle),
|
118
|
-
ParquetValue::Float32(f) => f.into_value_with(handle),
|
119
|
-
ParquetValue::Float64(f) => f.into_value_with(handle),
|
120
|
-
ParquetValue::Boolean(b) => b.into_value_with(handle),
|
121
|
-
ParquetValue::String(s) => s.into_value_with(handle),
|
122
|
-
ParquetValue::Bytes(b) => handle.str_from_slice(&b).as_value(),
|
123
|
-
ParquetValue::Date32(d) => impl_date_conversion!(d, handle),
|
124
|
-
ParquetValue::Date64(d) => impl_date_conversion!(d, handle),
|
125
|
-
timestamp @ ParquetValue::TimestampSecond(_, _) => {
|
126
|
-
impl_timestamp_conversion!(timestamp, TimestampSecond, handle)
|
127
|
-
}
|
128
|
-
timestamp @ ParquetValue::TimestampMillis(_, _) => {
|
129
|
-
impl_timestamp_conversion!(timestamp, TimestampMillis, handle)
|
130
|
-
}
|
131
|
-
timestamp @ ParquetValue::TimestampMicros(_, _) => {
|
132
|
-
impl_timestamp_conversion!(timestamp, TimestampMicros, handle)
|
133
|
-
}
|
134
|
-
timestamp @ ParquetValue::TimestampNanos(_, _) => {
|
135
|
-
impl_timestamp_conversion!(timestamp, TimestampNanos, handle)
|
136
|
-
}
|
137
|
-
ParquetValue::List(l) => {
|
138
|
-
let ary = handle.ary_new_capa(l.len());
|
139
|
-
l.into_iter()
|
140
|
-
.try_for_each(|v| ary.push(v.into_value_with(handle)))
|
141
|
-
.unwrap();
|
142
|
-
ary.into_value_with(handle)
|
143
|
-
}
|
144
|
-
ParquetValue::Map(m) => {
|
145
|
-
let hash = handle.hash_new_capa(m.len());
|
146
|
-
m.into_iter()
|
147
|
-
.try_for_each(|(k, v)| {
|
148
|
-
hash.aset(k.into_value_with(handle), v.into_value_with(handle))
|
149
|
-
})
|
150
|
-
.unwrap();
|
151
|
-
hash.into_value_with(handle)
|
152
|
-
}
|
153
|
-
ParquetValue::Null => handle.qnil().as_value(),
|
154
|
-
}
|
155
|
-
}
|
156
|
-
}
|
157
|
-
|
158
|
-
impl ParquetValue {
|
159
|
-
pub fn from_value(value: Value, type_: &ParquetSchemaType) -> Result<Self, MagnusError> {
|
160
|
-
if value.is_nil() {
|
161
|
-
return Ok(ParquetValue::Null);
|
162
|
-
}
|
163
|
-
|
164
|
-
match type_ {
|
165
|
-
ParquetSchemaType::Int8 => {
|
166
|
-
let v = NumericConverter::<i8>::convert_with_string_fallback(value)?;
|
167
|
-
Ok(ParquetValue::Int8(v))
|
168
|
-
}
|
169
|
-
ParquetSchemaType::Int16 => {
|
170
|
-
let v = NumericConverter::<i16>::convert_with_string_fallback(value)?;
|
171
|
-
Ok(ParquetValue::Int16(v))
|
172
|
-
}
|
173
|
-
ParquetSchemaType::Int32 => {
|
174
|
-
let v = NumericConverter::<i32>::convert_with_string_fallback(value)?;
|
175
|
-
Ok(ParquetValue::Int32(v))
|
176
|
-
}
|
177
|
-
ParquetSchemaType::Int64 => {
|
178
|
-
let v = NumericConverter::<i64>::convert_with_string_fallback(value)?;
|
179
|
-
Ok(ParquetValue::Int64(v))
|
180
|
-
}
|
181
|
-
ParquetSchemaType::UInt8 => {
|
182
|
-
let v = NumericConverter::<u8>::convert_with_string_fallback(value)?;
|
183
|
-
Ok(ParquetValue::UInt8(v))
|
184
|
-
}
|
185
|
-
ParquetSchemaType::UInt16 => {
|
186
|
-
let v = NumericConverter::<u16>::convert_with_string_fallback(value)?;
|
187
|
-
Ok(ParquetValue::UInt16(v))
|
188
|
-
}
|
189
|
-
ParquetSchemaType::UInt32 => {
|
190
|
-
let v = NumericConverter::<u32>::convert_with_string_fallback(value)?;
|
191
|
-
Ok(ParquetValue::UInt32(v))
|
192
|
-
}
|
193
|
-
ParquetSchemaType::UInt64 => {
|
194
|
-
let v = NumericConverter::<u64>::convert_with_string_fallback(value)?;
|
195
|
-
Ok(ParquetValue::UInt64(v))
|
196
|
-
}
|
197
|
-
ParquetSchemaType::Float => {
|
198
|
-
let v = NumericConverter::<f32>::convert_with_string_fallback(value)?;
|
199
|
-
Ok(ParquetValue::Float32(v))
|
200
|
-
}
|
201
|
-
ParquetSchemaType::Double => {
|
202
|
-
let v = NumericConverter::<f64>::convert_with_string_fallback(value)?;
|
203
|
-
Ok(ParquetValue::Float64(v))
|
204
|
-
}
|
205
|
-
ParquetSchemaType::String => {
|
206
|
-
let v = String::try_convert(value)?;
|
207
|
-
Ok(ParquetValue::String(v))
|
208
|
-
}
|
209
|
-
ParquetSchemaType::Binary => {
|
210
|
-
let v = convert_to_binary(value)?;
|
211
|
-
Ok(ParquetValue::Bytes(v))
|
212
|
-
}
|
213
|
-
ParquetSchemaType::Boolean => {
|
214
|
-
let v = convert_to_boolean(value)?;
|
215
|
-
Ok(ParquetValue::Boolean(v))
|
216
|
-
}
|
217
|
-
ParquetSchemaType::Date32 => {
|
218
|
-
let v = convert_to_date32(value, None)?;
|
219
|
-
Ok(ParquetValue::Date32(v))
|
220
|
-
}
|
221
|
-
ParquetSchemaType::TimestampMillis => {
|
222
|
-
let v = convert_to_timestamp_millis(value, None)?;
|
223
|
-
Ok(ParquetValue::TimestampMillis(v, None))
|
224
|
-
}
|
225
|
-
ParquetSchemaType::TimestampMicros => {
|
226
|
-
let v = convert_to_timestamp_micros(value, None)?;
|
227
|
-
Ok(ParquetValue::TimestampMicros(v, None))
|
228
|
-
}
|
229
|
-
ParquetSchemaType::List(_) | ParquetSchemaType::Map(_) => Err(MagnusError::new(
|
230
|
-
magnus::exception::type_error(),
|
231
|
-
"Nested lists and maps are not supported",
|
232
|
-
)),
|
233
|
-
}
|
234
|
-
}
|
235
|
-
}
|
236
|
-
|
237
|
-
#[derive(Debug)]
|
238
|
-
pub struct ParquetValueVec(Vec<ParquetValue>);
|
239
|
-
|
240
|
-
impl ParquetValueVec {
|
241
|
-
pub fn into_inner(self) -> Vec<ParquetValue> {
|
242
|
-
self.0
|
243
|
-
}
|
244
|
-
}
|
245
|
-
|
246
|
-
impl IntoIterator for ParquetValueVec {
|
247
|
-
type Item = ParquetValue;
|
248
|
-
type IntoIter = std::vec::IntoIter<ParquetValue>;
|
249
|
-
|
250
|
-
fn into_iter(self) -> Self::IntoIter {
|
251
|
-
self.0.into_iter()
|
252
|
-
}
|
253
|
-
}
|
254
|
-
|
255
|
-
impl std::cmp::PartialEq for ParquetValueVec {
|
256
|
-
fn eq(&self, other: &Self) -> bool {
|
257
|
-
self.0 == other.0
|
258
|
-
}
|
259
|
-
}
|
260
|
-
|
261
|
-
impl std::cmp::Eq for ParquetValueVec {}
|
262
|
-
|
263
|
-
impl TryFrom<Arc<dyn Array>> for ParquetValueVec {
|
264
|
-
type Error = String;
|
265
|
-
|
266
|
-
fn try_from(column: Arc<dyn Array>) -> Result<Self, Self::Error> {
|
267
|
-
ParquetValueVec::try_from(&*column)
|
268
|
-
}
|
269
|
-
}
|
270
|
-
|
271
|
-
macro_rules! impl_numeric_array_conversion {
|
272
|
-
($column:expr, $array_type:ty, $variant:ident) => {{
|
273
|
-
let array = downcast_array::<$array_type>($column);
|
274
|
-
if array.is_nullable() {
|
275
|
-
array
|
276
|
-
.values()
|
277
|
-
.iter()
|
278
|
-
.enumerate()
|
279
|
-
.map(|(i, x)| {
|
280
|
-
if array.is_null(i) {
|
281
|
-
ParquetValue::Null
|
282
|
-
} else {
|
283
|
-
ParquetValue::$variant(*x)
|
284
|
-
}
|
285
|
-
})
|
286
|
-
.collect()
|
287
|
-
} else {
|
288
|
-
array
|
289
|
-
.values()
|
290
|
-
.iter()
|
291
|
-
.map(|x| ParquetValue::$variant(*x))
|
292
|
-
.collect()
|
293
|
-
}
|
294
|
-
}};
|
295
|
-
}
|
296
|
-
macro_rules! impl_boolean_array_conversion {
|
297
|
-
($column:expr, $array_type:ty, $variant:ident) => {{
|
298
|
-
let array = downcast_array::<$array_type>($column);
|
299
|
-
if array.is_nullable() {
|
300
|
-
array
|
301
|
-
.values()
|
302
|
-
.iter()
|
303
|
-
.enumerate()
|
304
|
-
.map(|(i, x)| {
|
305
|
-
if array.is_null(i) {
|
306
|
-
ParquetValue::Null
|
307
|
-
} else {
|
308
|
-
ParquetValue::$variant(x)
|
309
|
-
}
|
310
|
-
})
|
311
|
-
.collect()
|
312
|
-
} else {
|
313
|
-
array
|
314
|
-
.values()
|
315
|
-
.iter()
|
316
|
-
.map(|x| ParquetValue::$variant(x))
|
317
|
-
.collect()
|
318
|
-
}
|
319
|
-
}};
|
320
|
-
}
|
321
|
-
|
322
|
-
impl TryFrom<&dyn Array> for ParquetValueVec {
|
323
|
-
type Error = String;
|
324
|
-
|
325
|
-
fn try_from(column: &dyn Array) -> Result<Self, Self::Error> {
|
326
|
-
let tmp_vec = match column.data_type() {
|
327
|
-
DataType::Boolean => impl_boolean_array_conversion!(column, BooleanArray, Boolean),
|
328
|
-
DataType::Int8 => impl_numeric_array_conversion!(column, Int8Array, Int8),
|
329
|
-
DataType::Int16 => impl_numeric_array_conversion!(column, Int16Array, Int16),
|
330
|
-
DataType::Int32 => impl_numeric_array_conversion!(column, Int32Array, Int32),
|
331
|
-
DataType::Int64 => impl_numeric_array_conversion!(column, Int64Array, Int64),
|
332
|
-
DataType::UInt8 => impl_numeric_array_conversion!(column, UInt8Array, UInt8),
|
333
|
-
DataType::UInt16 => impl_numeric_array_conversion!(column, UInt16Array, UInt16),
|
334
|
-
DataType::UInt32 => impl_numeric_array_conversion!(column, UInt32Array, UInt32),
|
335
|
-
DataType::UInt64 => impl_numeric_array_conversion!(column, UInt64Array, UInt64),
|
336
|
-
DataType::Float32 => impl_numeric_array_conversion!(column, Float32Array, Float32),
|
337
|
-
DataType::Float64 => impl_numeric_array_conversion!(column, Float64Array, Float64),
|
338
|
-
DataType::Date32 => impl_numeric_array_conversion!(column, Date32Array, Date32),
|
339
|
-
DataType::Date64 => impl_numeric_array_conversion!(column, Date64Array, Date64),
|
340
|
-
DataType::Timestamp(TimeUnit::Second, tz) => {
|
341
|
-
impl_timestamp_array_conversion!(column, TimestampSecondArray, TimestampSecond, tz)
|
342
|
-
}
|
343
|
-
DataType::Timestamp(TimeUnit::Millisecond, tz) => {
|
344
|
-
impl_timestamp_array_conversion!(
|
345
|
-
column,
|
346
|
-
TimestampMillisecondArray,
|
347
|
-
TimestampMillis,
|
348
|
-
tz
|
349
|
-
)
|
350
|
-
}
|
351
|
-
DataType::Timestamp(TimeUnit::Microsecond, tz) => {
|
352
|
-
impl_timestamp_array_conversion!(
|
353
|
-
column,
|
354
|
-
TimestampMicrosecondArray,
|
355
|
-
TimestampMicros,
|
356
|
-
tz
|
357
|
-
)
|
358
|
-
}
|
359
|
-
DataType::Timestamp(TimeUnit::Nanosecond, tz) => {
|
360
|
-
impl_timestamp_array_conversion!(
|
361
|
-
column,
|
362
|
-
TimestampNanosecondArray,
|
363
|
-
TimestampNanos,
|
364
|
-
tz
|
365
|
-
)
|
366
|
-
}
|
367
|
-
DataType::Float16 => {
|
368
|
-
let array = downcast_array::<Float16Array>(column);
|
369
|
-
if array.is_nullable() {
|
370
|
-
array
|
371
|
-
.values()
|
372
|
-
.iter()
|
373
|
-
.enumerate()
|
374
|
-
.map(|(i, x)| {
|
375
|
-
if array.is_null(i) {
|
376
|
-
ParquetValue::Null
|
377
|
-
} else {
|
378
|
-
ParquetValue::Float16(f32::from(*x))
|
379
|
-
}
|
380
|
-
})
|
381
|
-
.collect()
|
382
|
-
} else {
|
383
|
-
array
|
384
|
-
.values()
|
385
|
-
.iter()
|
386
|
-
.map(|x| ParquetValue::Float16(f32::from(*x)))
|
387
|
-
.collect()
|
388
|
-
}
|
389
|
-
}
|
390
|
-
DataType::Utf8 => {
|
391
|
-
let array = downcast_array::<StringArray>(column);
|
392
|
-
array
|
393
|
-
.iter()
|
394
|
-
.map(|opt_x| match opt_x {
|
395
|
-
Some(x) => ParquetValue::String(x.to_string()),
|
396
|
-
None => ParquetValue::Null,
|
397
|
-
})
|
398
|
-
.collect()
|
399
|
-
}
|
400
|
-
DataType::Binary => {
|
401
|
-
let array = downcast_array::<BinaryArray>(column);
|
402
|
-
array
|
403
|
-
.iter()
|
404
|
-
.map(|opt_x| match opt_x {
|
405
|
-
Some(x) => ParquetValue::Bytes(x.to_vec()),
|
406
|
-
None => ParquetValue::Null,
|
407
|
-
})
|
408
|
-
.collect()
|
409
|
-
}
|
410
|
-
DataType::List(_field) => {
|
411
|
-
let list_array = downcast_array::<ListArray>(column);
|
412
|
-
list_array
|
413
|
-
.iter()
|
414
|
-
.map(|x| match x {
|
415
|
-
Some(values) => match ParquetValueVec::try_from(values) {
|
416
|
-
Ok(vec) => ParquetValue::List(vec.into_inner()),
|
417
|
-
Err(e) => {
|
418
|
-
panic!("Error converting list array to ParquetValueVec: {}", e)
|
419
|
-
}
|
420
|
-
},
|
421
|
-
None => ParquetValue::Null,
|
422
|
-
})
|
423
|
-
.collect()
|
424
|
-
}
|
425
|
-
DataType::Struct(_) => {
|
426
|
-
let struct_array = downcast_array::<StructArray>(column);
|
427
|
-
let mut values = Vec::with_capacity(struct_array.len());
|
428
|
-
for i in 0..struct_array.len() {
|
429
|
-
if struct_array.is_null(i) {
|
430
|
-
values.push(ParquetValue::Null);
|
431
|
-
continue;
|
432
|
-
}
|
433
|
-
|
434
|
-
let mut map = std::collections::HashMap::new();
|
435
|
-
for (field_idx, field) in struct_array.fields().iter().enumerate() {
|
436
|
-
let column = struct_array.column(field_idx);
|
437
|
-
let field_values = match ParquetValueVec::try_from(column.slice(i, 1)) {
|
438
|
-
Ok(vec) => vec.into_inner(),
|
439
|
-
Err(e) => {
|
440
|
-
panic!("Error converting struct field to ParquetValueVec: {}", e)
|
441
|
-
}
|
442
|
-
};
|
443
|
-
map.insert(
|
444
|
-
ParquetValue::String(field.name().to_string()),
|
445
|
-
field_values.into_iter().next().unwrap(),
|
446
|
-
);
|
447
|
-
}
|
448
|
-
values.push(ParquetValue::Map(map));
|
449
|
-
}
|
450
|
-
values
|
451
|
-
}
|
452
|
-
DataType::Null => {
|
453
|
-
let x = downcast_array::<NullArray>(column);
|
454
|
-
vec![ParquetValue::Null; x.len()]
|
455
|
-
}
|
456
|
-
_ => {
|
457
|
-
return Err(format!("Unsupported data type: {:?}", column.data_type()));
|
458
|
-
}
|
459
|
-
};
|
460
|
-
Ok(ParquetValueVec(tmp_vec))
|
461
|
-
}
|
462
|
-
}
|
@@ -1,204 +0,0 @@
|
|
1
|
-
use itertools::Itertools;
|
2
|
-
|
3
|
-
use super::*;
|
4
|
-
|
5
|
-
#[derive(Debug)]
|
6
|
-
pub enum RowRecord<S: BuildHasher + Default> {
|
7
|
-
Vec(Vec<ParquetField>),
|
8
|
-
Map(HashMap<StringCacheKey, ParquetField, S>),
|
9
|
-
}
|
10
|
-
|
11
|
-
#[derive(Debug)]
|
12
|
-
pub enum ColumnRecord<S: BuildHasher + Default> {
|
13
|
-
Vec(Vec<Vec<ParquetValue>>),
|
14
|
-
Map(HashMap<StringCacheKey, Vec<ParquetValue>, S>),
|
15
|
-
}
|
16
|
-
|
17
|
-
#[derive(Debug)]
|
18
|
-
pub struct ParquetField(pub Field);
|
19
|
-
|
20
|
-
impl<S: BuildHasher + Default> IntoValue for RowRecord<S> {
|
21
|
-
fn into_value_with(self, handle: &Ruby) -> Value {
|
22
|
-
match self {
|
23
|
-
RowRecord::Vec(vec) => {
|
24
|
-
let ary = handle.ary_new_capa(vec.len());
|
25
|
-
vec.into_iter().try_for_each(|v| ary.push(v)).unwrap();
|
26
|
-
handle.into_value(ary)
|
27
|
-
}
|
28
|
-
RowRecord::Map(map) => {
|
29
|
-
let hash = handle.hash_new_capa(map.len());
|
30
|
-
|
31
|
-
let mut values: [Value; 128] = [handle.qnil().as_value(); 128];
|
32
|
-
let mut i = 0;
|
33
|
-
|
34
|
-
for chunk in &map.into_iter().chunks(64) {
|
35
|
-
// Reduced to 64 to ensure space for pairs
|
36
|
-
for (k, v) in chunk {
|
37
|
-
if i + 1 >= values.len() {
|
38
|
-
// Bulk insert current batch if array is full
|
39
|
-
hash.bulk_insert(&values[..i]).unwrap();
|
40
|
-
values[..i].fill(handle.qnil().as_value());
|
41
|
-
i = 0;
|
42
|
-
}
|
43
|
-
values[i] = handle.into_value(k);
|
44
|
-
values[i + 1] = handle.into_value(v);
|
45
|
-
i += 2;
|
46
|
-
}
|
47
|
-
// Insert any remaining pairs
|
48
|
-
if i > 0 {
|
49
|
-
hash.bulk_insert(&values[..i]).unwrap();
|
50
|
-
values[..i].fill(handle.qnil().as_value());
|
51
|
-
i = 0;
|
52
|
-
}
|
53
|
-
}
|
54
|
-
|
55
|
-
hash.into_value_with(handle)
|
56
|
-
}
|
57
|
-
}
|
58
|
-
}
|
59
|
-
}
|
60
|
-
|
61
|
-
impl<S: BuildHasher + Default> IntoValue for ColumnRecord<S> {
|
62
|
-
fn into_value_with(self, handle: &Ruby) -> Value {
|
63
|
-
match self {
|
64
|
-
ColumnRecord::Vec(vec) => {
|
65
|
-
let ary = handle.ary_new_capa(vec.len());
|
66
|
-
vec.into_iter()
|
67
|
-
.try_for_each(|v| {
|
68
|
-
let nested_ary = handle.ary_new_capa(v.len());
|
69
|
-
v.into_iter().try_for_each(|v| nested_ary.push(v)).unwrap();
|
70
|
-
ary.push(nested_ary.into_value_with(handle))
|
71
|
-
})
|
72
|
-
.unwrap();
|
73
|
-
ary.into_value_with(handle)
|
74
|
-
}
|
75
|
-
ColumnRecord::Map(map) => {
|
76
|
-
let hash = handle.hash_new_capa(map.len());
|
77
|
-
|
78
|
-
let mut values: [Value; 128] = [handle.qnil().as_value(); 128];
|
79
|
-
let mut i = 0;
|
80
|
-
|
81
|
-
for chunk in &map.into_iter().chunks(64) {
|
82
|
-
// Reduced to 64 to ensure space for pairs
|
83
|
-
for (k, v) in chunk {
|
84
|
-
if i + 1 >= values.len() {
|
85
|
-
// Bulk insert current batch if array is full
|
86
|
-
hash.bulk_insert(&values[..i]).unwrap();
|
87
|
-
values[..i].fill(handle.qnil().as_value());
|
88
|
-
i = 0;
|
89
|
-
}
|
90
|
-
values[i] = handle.into_value(k);
|
91
|
-
let ary = handle.ary_new_capa(v.len());
|
92
|
-
v.into_iter().try_for_each(|v| ary.push(v)).unwrap();
|
93
|
-
values[i + 1] = handle.into_value(ary);
|
94
|
-
i += 2;
|
95
|
-
}
|
96
|
-
// Insert any remaining pairs
|
97
|
-
if i > 0 {
|
98
|
-
hash.bulk_insert(&values[..i]).unwrap();
|
99
|
-
values[..i].fill(handle.qnil().as_value());
|
100
|
-
i = 0;
|
101
|
-
}
|
102
|
-
}
|
103
|
-
|
104
|
-
hash.into_value_with(handle)
|
105
|
-
}
|
106
|
-
}
|
107
|
-
}
|
108
|
-
}
|
109
|
-
|
110
|
-
impl IntoValue for ParquetField {
|
111
|
-
fn into_value_with(self, handle: &Ruby) -> Value {
|
112
|
-
match self.0 {
|
113
|
-
Field::Null => handle.qnil().as_value(),
|
114
|
-
Field::Bool(b) => b.into_value_with(handle),
|
115
|
-
Field::Short(s) => s.into_value_with(handle),
|
116
|
-
Field::Int(i) => i.into_value_with(handle),
|
117
|
-
Field::Long(l) => l.into_value_with(handle),
|
118
|
-
Field::UByte(ub) => ub.into_value_with(handle),
|
119
|
-
Field::UShort(us) => us.into_value_with(handle),
|
120
|
-
Field::UInt(ui) => ui.into_value_with(handle),
|
121
|
-
Field::ULong(ul) => ul.into_value_with(handle),
|
122
|
-
Field::Float16(f) => f32::from(f).into_value_with(handle),
|
123
|
-
Field::Float(f) => f.into_value_with(handle),
|
124
|
-
Field::Double(d) => d.into_value_with(handle),
|
125
|
-
Field::Str(s) => s.into_value_with(handle),
|
126
|
-
Field::Byte(b) => b.into_value_with(handle),
|
127
|
-
Field::Bytes(b) => handle.str_from_slice(b.data()).as_value(),
|
128
|
-
Field::Date(d) => {
|
129
|
-
let ts = jiff::Timestamp::from_second((d as i64) * 86400).unwrap();
|
130
|
-
let formatted = ts.strftime("%Y-%m-%d").to_string();
|
131
|
-
formatted.into_value_with(handle)
|
132
|
-
}
|
133
|
-
Field::TimestampMillis(ts) => {
|
134
|
-
let ts = jiff::Timestamp::from_millisecond(ts).unwrap();
|
135
|
-
let time_class = handle.class_time();
|
136
|
-
time_class
|
137
|
-
.funcall::<_, _, Value>("parse", (ts.to_string(),))
|
138
|
-
.unwrap()
|
139
|
-
.into_value_with(handle)
|
140
|
-
}
|
141
|
-
Field::TimestampMicros(ts) => {
|
142
|
-
let ts = jiff::Timestamp::from_microsecond(ts).unwrap();
|
143
|
-
let time_class = handle.class_time();
|
144
|
-
time_class
|
145
|
-
.funcall::<_, _, Value>("parse", (ts.to_string(),))
|
146
|
-
.unwrap()
|
147
|
-
.into_value_with(handle)
|
148
|
-
}
|
149
|
-
Field::ListInternal(list) => {
|
150
|
-
let elements = list.elements();
|
151
|
-
let ary = handle.ary_new_capa(elements.len());
|
152
|
-
elements
|
153
|
-
.iter()
|
154
|
-
.try_for_each(|e| ary.push(ParquetField(e.clone()).into_value_with(handle)))
|
155
|
-
.unwrap();
|
156
|
-
ary.into_value_with(handle)
|
157
|
-
}
|
158
|
-
Field::MapInternal(map) => {
|
159
|
-
let entries = map.entries();
|
160
|
-
let hash = handle.hash_new_capa(entries.len());
|
161
|
-
entries
|
162
|
-
.iter()
|
163
|
-
.try_for_each(|(k, v)| {
|
164
|
-
hash.aset(
|
165
|
-
ParquetField(k.clone()).into_value_with(handle),
|
166
|
-
ParquetField(v.clone()).into_value_with(handle),
|
167
|
-
)
|
168
|
-
})
|
169
|
-
.unwrap();
|
170
|
-
hash.into_value_with(handle)
|
171
|
-
}
|
172
|
-
Field::Decimal(d) => {
|
173
|
-
let value = match d {
|
174
|
-
Decimal::Int32 { value, scale, .. } => {
|
175
|
-
let unscaled = i32::from_be_bytes(value);
|
176
|
-
format!("{}e-{}", unscaled, scale)
|
177
|
-
}
|
178
|
-
Decimal::Int64 { value, scale, .. } => {
|
179
|
-
let unscaled = i64::from_be_bytes(value);
|
180
|
-
format!("{}e-{}", unscaled, scale)
|
181
|
-
}
|
182
|
-
Decimal::Bytes { value, scale, .. } => {
|
183
|
-
// Convert bytes to string representation of unscaled value
|
184
|
-
let unscaled = String::from_utf8_lossy(value.data());
|
185
|
-
format!("{}e-{}", unscaled, scale)
|
186
|
-
}
|
187
|
-
};
|
188
|
-
handle.eval(&format!("BigDecimal(\"{value}\")")).unwrap()
|
189
|
-
}
|
190
|
-
Field::Group(row) => {
|
191
|
-
let hash = handle.hash_new();
|
192
|
-
row.get_column_iter()
|
193
|
-
.try_for_each(|(k, v)| {
|
194
|
-
hash.aset(
|
195
|
-
k.clone().into_value_with(handle),
|
196
|
-
ParquetField(v.clone()).into_value_with(handle),
|
197
|
-
)
|
198
|
-
})
|
199
|
-
.unwrap();
|
200
|
-
hash.into_value_with(handle)
|
201
|
-
}
|
202
|
-
}
|
203
|
-
}
|
204
|
-
}
|