]>
Commit | Line | Data |
---|---|---|
1d09f67e TL |
1 | // Licensed to the Apache Software Foundation (ASF) under one |
2 | // or more contributor license agreements. See the NOTICE file | |
3 | // distributed with this work for additional information | |
4 | // regarding copyright ownership. The ASF licenses this file | |
5 | // to you under the Apache License, Version 2.0 (the | |
6 | // "License"); you may not use this file except in compliance | |
7 | // with the License. You may obtain a copy of the License at | |
8 | // | |
9 | // http://www.apache.org/licenses/LICENSE-2.0 | |
10 | // | |
11 | // Unless required by applicable law or agreed to in writing, software | |
12 | // distributed under the License is distributed on an "AS IS" BASIS, | |
13 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | |
14 | // See the License for the specific language governing permissions and | |
15 | // limitations under the License. | |
16 | ||
17 | package scalar | |
18 | ||
19 | import ( | |
20 | "fmt" | |
21 | "reflect" | |
22 | "time" | |
23 | "unsafe" | |
24 | ||
25 | "github.com/apache/arrow/go/v6/arrow" | |
26 | "golang.org/x/xerrors" | |
27 | ) | |
28 | ||
29 | type op int8 | |
30 | ||
31 | const ( | |
32 | convDIVIDE = iota | |
33 | convMULTIPLY | |
34 | ) | |
35 | ||
36 | var timestampConversion = [...][4]struct { | |
37 | op op | |
38 | factor int64 | |
39 | }{ | |
40 | arrow.Nanosecond: { | |
41 | arrow.Nanosecond: {convMULTIPLY, int64(time.Nanosecond)}, | |
42 | arrow.Microsecond: {convDIVIDE, int64(time.Microsecond)}, | |
43 | arrow.Millisecond: {convDIVIDE, int64(time.Millisecond)}, | |
44 | arrow.Second: {convDIVIDE, int64(time.Second)}, | |
45 | }, | |
46 | arrow.Microsecond: { | |
47 | arrow.Nanosecond: {convMULTIPLY, int64(time.Microsecond)}, | |
48 | arrow.Microsecond: {convMULTIPLY, 1}, | |
49 | arrow.Millisecond: {convDIVIDE, int64(time.Millisecond / time.Microsecond)}, | |
50 | arrow.Second: {convDIVIDE, int64(time.Second / time.Microsecond)}, | |
51 | }, | |
52 | arrow.Millisecond: { | |
53 | arrow.Nanosecond: {convMULTIPLY, int64(time.Millisecond)}, | |
54 | arrow.Microsecond: {convMULTIPLY, int64(time.Millisecond / time.Microsecond)}, | |
55 | arrow.Millisecond: {convMULTIPLY, 1}, | |
56 | arrow.Second: {convDIVIDE, int64(time.Second / time.Millisecond)}, | |
57 | }, | |
58 | arrow.Second: { | |
59 | arrow.Nanosecond: {convMULTIPLY, int64(time.Second)}, | |
60 | arrow.Microsecond: {convMULTIPLY, int64(time.Second / time.Microsecond)}, | |
61 | arrow.Millisecond: {convMULTIPLY, int64(time.Second / time.Millisecond)}, | |
62 | arrow.Second: {convMULTIPLY, 1}, | |
63 | }, | |
64 | } | |
65 | ||
66 | func ConvertTimestampValue(in, out arrow.TimeUnit, value int64) int64 { | |
67 | conv := timestampConversion[int(in)][int(out)] | |
68 | switch conv.op { | |
69 | case convMULTIPLY: | |
70 | return value * conv.factor | |
71 | case convDIVIDE: | |
72 | return value / conv.factor | |
73 | } | |
74 | ||
75 | return 0 | |
76 | } | |
77 | ||
78 | func temporalToString(s TemporalScalar) string { | |
79 | switch s := s.(type) { | |
80 | case *Date32: | |
81 | return time.Unix(0, 0).UTC().AddDate(0, 0, int(s.Value)).Format("2006-01-02") | |
82 | case *Date64: | |
83 | days := int(int64(s.Value) / (time.Hour * 24).Milliseconds()) | |
84 | return time.Unix(0, 0).UTC().AddDate(0, 0, days).Format("2006-01-02") | |
85 | case *Duration: | |
86 | return fmt.Sprint(time.Duration(s.Value) * s.Unit().Multiplier()) | |
87 | case *Time32: | |
88 | return time.Unix(0, int64(s.Value)*int64(s.Unit().Multiplier())).UTC().Format("15:04:05.999") | |
89 | case *Time64: | |
90 | return time.Unix(0, int64(s.Value)*int64(s.Unit().Multiplier())).UTC().Format("15:04:05.999999999") | |
91 | case *Timestamp: | |
92 | return time.Unix(0, int64(s.Value)*int64(s.Unit().Multiplier())).UTC().Format("2006-01-02 15:04:05.999999999") | |
93 | } | |
94 | return "..." | |
95 | } | |
96 | ||
97 | type TemporalScalar interface { | |
98 | Scalar | |
99 | temporal() | |
100 | } | |
101 | ||
102 | type Duration struct { | |
103 | scalar | |
104 | Value arrow.Duration | |
105 | } | |
106 | ||
107 | func (Duration) temporal() {} | |
108 | func (s *Duration) value() interface{} { return s.Value } | |
109 | func (s *Duration) CastTo(to arrow.DataType) (Scalar, error) { return castTemporal(s, to) } | |
110 | func (s *Duration) String() string { | |
111 | if !s.Valid { | |
112 | return "null" | |
113 | } | |
114 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
115 | if err != nil { | |
116 | return "..." | |
117 | } | |
118 | return string(val.(*String).Value.Bytes()) | |
119 | } | |
120 | ||
121 | func (s *Duration) equals(rhs Scalar) bool { | |
122 | return s.Value == rhs.(*Duration).Value | |
123 | } | |
124 | ||
125 | func (s *Duration) Unit() arrow.TimeUnit { | |
126 | return s.DataType().(*arrow.TimestampType).Unit | |
127 | } | |
128 | func (s *Duration) Data() []byte { | |
129 | return (*[arrow.DurationSizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
130 | } | |
131 | ||
132 | func NewDurationScalar(val arrow.Duration, typ arrow.DataType) *Duration { | |
133 | return &Duration{scalar{typ, true}, val} | |
134 | } | |
135 | ||
136 | type DateScalar interface { | |
137 | TemporalScalar | |
138 | date() | |
139 | } | |
140 | ||
141 | type TimeScalar interface { | |
142 | TemporalScalar | |
143 | Unit() arrow.TimeUnit | |
144 | time() | |
145 | } | |
146 | ||
147 | type IntervalScalar interface { | |
148 | TemporalScalar | |
149 | interval() | |
150 | } | |
151 | ||
152 | const millisecondsInDay = (time.Hour * 24) / time.Millisecond | |
153 | ||
154 | func castTemporal(from TemporalScalar, to arrow.DataType) (Scalar, error) { | |
155 | if arrow.TypeEqual(from.DataType(), to) { | |
156 | return from, nil | |
157 | } | |
158 | ||
159 | if !from.IsValid() { | |
160 | return MakeNullScalar(to), nil | |
161 | } | |
162 | ||
163 | if r, ok := numericMap[to.ID()]; ok { | |
164 | return convertToNumeric(reflect.ValueOf(from.value()), r.valueType, r.scalarFunc), nil | |
165 | } | |
166 | ||
167 | if to.ID() == arrow.STRING { | |
168 | return NewStringScalar(temporalToString(from)), nil | |
169 | } | |
170 | ||
171 | switch s := from.(type) { | |
172 | case DateScalar: | |
173 | if to.ID() == arrow.TIMESTAMP { | |
174 | var newValue int64 | |
175 | switch s := s.(type) { | |
176 | case *Date32: | |
177 | newValue = int64(s.Value) * int64(millisecondsInDay) | |
178 | case *Date64: | |
179 | newValue = int64(s.Value) | |
180 | } | |
181 | return NewTimestampScalar(arrow.Timestamp(ConvertTimestampValue(arrow.Millisecond, to.(*arrow.TimestampType).Unit, newValue)), to), nil | |
182 | } | |
183 | ||
184 | switch s := s.(type) { | |
185 | case *Date32: | |
186 | if to.ID() == arrow.DATE64 { | |
187 | return NewDate64Scalar(arrow.Date64(s.Value) * arrow.Date64(millisecondsInDay)), nil | |
188 | } | |
189 | case *Date64: | |
190 | if to.ID() == arrow.DATE32 { | |
191 | return NewDate32Scalar(arrow.Date32(s.Value / arrow.Date64(millisecondsInDay))), nil | |
192 | } | |
193 | } | |
194 | case *Timestamp: | |
195 | switch to := to.(type) { | |
196 | case *arrow.TimestampType: | |
197 | return NewTimestampScalar(arrow.Timestamp(ConvertTimestampValue(s.Unit(), to.Unit, int64(s.Value))), to), nil | |
198 | case *arrow.Date32Type: | |
199 | millis := ConvertTimestampValue(s.Unit(), arrow.Millisecond, int64(s.Value)) | |
200 | return NewDate32Scalar(arrow.Date32(millis / int64(millisecondsInDay))), nil | |
201 | case *arrow.Date64Type: | |
202 | millis := ConvertTimestampValue(s.Unit(), arrow.Millisecond, int64(s.Value)) | |
203 | return NewDate64Scalar(arrow.Date64(millis - millis%int64(millisecondsInDay))), nil | |
204 | } | |
205 | case TimeScalar: | |
206 | switch to := to.(type) { | |
207 | case *arrow.Time32Type: | |
208 | return NewTime32Scalar(arrow.Time32(ConvertTimestampValue(s.Unit(), to.Unit, int64(s.value().(arrow.Time64)))), to), nil | |
209 | case *arrow.Time64Type: | |
210 | return NewTime64Scalar(arrow.Time64(ConvertTimestampValue(s.Unit(), to.Unit, int64(s.value().(arrow.Time32)))), to), nil | |
211 | } | |
212 | ||
213 | case *Duration: | |
214 | switch to := to.(type) { | |
215 | case *arrow.StringType: | |
216 | ||
217 | case *arrow.DurationType: | |
218 | return NewDurationScalar(arrow.Duration(ConvertTimestampValue(s.Unit(), to.Unit, int64(s.Value))), to), nil | |
219 | } | |
220 | } | |
221 | ||
222 | return nil, xerrors.Errorf("") | |
223 | } | |
224 | ||
225 | type Date32 struct { | |
226 | scalar | |
227 | Value arrow.Date32 | |
228 | } | |
229 | ||
230 | func (Date32) temporal() {} | |
231 | func (Date32) date() {} | |
232 | func (s *Date32) value() interface{} { return s.Value } | |
233 | func (s *Date32) Data() []byte { | |
234 | return (*[arrow.Date32SizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
235 | } | |
236 | func (s *Date32) equals(rhs Scalar) bool { | |
237 | return s.Value == rhs.(*Date32).Value | |
238 | } | |
239 | func (s *Date32) CastTo(to arrow.DataType) (Scalar, error) { return castTemporal(s, to) } | |
240 | func (s *Date32) String() string { | |
241 | if !s.Valid { | |
242 | return "null" | |
243 | } | |
244 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
245 | if err != nil { | |
246 | return "..." | |
247 | } | |
248 | return string(val.(*String).Value.Bytes()) | |
249 | } | |
250 | ||
251 | func NewDate32Scalar(val arrow.Date32) *Date32 { | |
252 | return &Date32{scalar{arrow.FixedWidthTypes.Date32, true}, val} | |
253 | } | |
254 | ||
255 | type Date64 struct { | |
256 | scalar | |
257 | Value arrow.Date64 | |
258 | } | |
259 | ||
260 | func (Date64) temporal() {} | |
261 | func (Date64) date() {} | |
262 | func (s *Date64) value() interface{} { return s.Value } | |
263 | func (s *Date64) CastTo(to arrow.DataType) (Scalar, error) { return castTemporal(s, to) } | |
264 | func (s *Date64) Data() []byte { | |
265 | return (*[arrow.Date64SizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
266 | } | |
267 | func (s *Date64) equals(rhs Scalar) bool { | |
268 | return s.Value == rhs.(*Date64).Value | |
269 | } | |
270 | func (s *Date64) String() string { | |
271 | if !s.Valid { | |
272 | return "null" | |
273 | } | |
274 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
275 | if err != nil { | |
276 | return "..." | |
277 | } | |
278 | return string(val.(*String).Value.Bytes()) | |
279 | } | |
280 | ||
281 | func NewDate64Scalar(val arrow.Date64) *Date64 { | |
282 | return &Date64{scalar{arrow.FixedWidthTypes.Date64, true}, val} | |
283 | } | |
284 | ||
285 | type Time32 struct { | |
286 | scalar | |
287 | Value arrow.Time32 | |
288 | } | |
289 | ||
290 | func (Time32) temporal() {} | |
291 | func (Time32) time() {} | |
292 | func (s *Time32) value() interface{} { return s.Value } | |
293 | func (s *Time32) CastTo(to arrow.DataType) (Scalar, error) { return castTemporal(s, to) } | |
294 | func (s *Time32) Unit() arrow.TimeUnit { | |
295 | return s.DataType().(*arrow.Time32Type).Unit | |
296 | } | |
297 | func (s *Time32) equals(rhs Scalar) bool { | |
298 | return s.Value == rhs.(*Time32).Value | |
299 | } | |
300 | func (s *Time32) String() string { | |
301 | if !s.Valid { | |
302 | return "null" | |
303 | } | |
304 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
305 | if err != nil { | |
306 | return "..." | |
307 | } | |
308 | return string(val.(*String).Value.Bytes()) | |
309 | } | |
310 | ||
311 | func (s *Time32) Data() []byte { | |
312 | return (*[arrow.Time32SizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
313 | } | |
314 | ||
315 | func NewTime32Scalar(val arrow.Time32, typ arrow.DataType) *Time32 { | |
316 | return &Time32{scalar{typ, true}, val} | |
317 | } | |
318 | ||
319 | type Time64 struct { | |
320 | scalar | |
321 | Value arrow.Time64 | |
322 | } | |
323 | ||
324 | func (Time64) temporal() {} | |
325 | func (Time64) time() {} | |
326 | func (s *Time64) value() interface{} { return s.Value } | |
327 | func (s *Time64) CastTo(to arrow.DataType) (Scalar, error) { return castTemporal(s, to) } | |
328 | func (s *Time64) Unit() arrow.TimeUnit { | |
329 | return s.DataType().(*arrow.Time64Type).Unit | |
330 | } | |
331 | func (s *Time64) Data() []byte { | |
332 | return (*[arrow.Time64SizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
333 | } | |
334 | func (s *Time64) equals(rhs Scalar) bool { | |
335 | return s.Value == rhs.(*Time64).Value | |
336 | } | |
337 | func (s *Time64) String() string { | |
338 | if !s.Valid { | |
339 | return "null" | |
340 | } | |
341 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
342 | if err != nil { | |
343 | return "..." | |
344 | } | |
345 | return string(val.(*String).Value.Bytes()) | |
346 | } | |
347 | ||
348 | func NewTime64Scalar(val arrow.Time64, typ arrow.DataType) *Time64 { | |
349 | return &Time64{scalar{typ, true}, val} | |
350 | } | |
351 | ||
352 | type Timestamp struct { | |
353 | scalar | |
354 | Value arrow.Timestamp | |
355 | } | |
356 | ||
357 | func (Timestamp) temporal() {} | |
358 | func (Timestamp) time() {} | |
359 | func (s *Timestamp) value() interface{} { return s.Value } | |
360 | func (s *Timestamp) CastTo(to arrow.DataType) (Scalar, error) { return castTemporal(s, to) } | |
361 | func (s *Timestamp) Unit() arrow.TimeUnit { | |
362 | return s.DataType().(*arrow.TimestampType).Unit | |
363 | } | |
364 | func (s *Timestamp) Data() []byte { | |
365 | return (*[arrow.TimestampSizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
366 | } | |
367 | func (s *Timestamp) equals(rhs Scalar) bool { | |
368 | return s.Value == rhs.(*Timestamp).Value | |
369 | } | |
370 | func (s *Timestamp) String() string { | |
371 | if !s.Valid { | |
372 | return "null" | |
373 | } | |
374 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
375 | if err != nil { | |
376 | return "..." | |
377 | } | |
378 | return string(val.(*String).Value.Bytes()) | |
379 | } | |
380 | ||
381 | func NewTimestampScalar(val arrow.Timestamp, typ arrow.DataType) *Timestamp { | |
382 | return &Timestamp{scalar{typ, true}, val} | |
383 | } | |
384 | ||
385 | type MonthInterval struct { | |
386 | scalar | |
387 | Value arrow.MonthInterval | |
388 | } | |
389 | ||
390 | func (MonthInterval) temporal() {} | |
391 | func (MonthInterval) interval() {} | |
392 | func (s *MonthInterval) value() interface{} { return s.Value } | |
393 | func (s *MonthInterval) CastTo(to arrow.DataType) (Scalar, error) { | |
394 | if !s.Valid { | |
395 | return MakeNullScalar(to), nil | |
396 | } | |
397 | ||
398 | if !arrow.TypeEqual(s.DataType(), to) { | |
399 | return nil, xerrors.Errorf("non-null monthinterval scalar cannot be cast to anything other than monthinterval") | |
400 | } | |
401 | ||
402 | return s, nil | |
403 | } | |
404 | func (s *MonthInterval) String() string { | |
405 | if !s.Valid { | |
406 | return "null" | |
407 | } | |
408 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
409 | if err != nil { | |
410 | return "..." | |
411 | } | |
412 | return string(val.(*String).Value.Bytes()) | |
413 | } | |
414 | func (s *MonthInterval) equals(rhs Scalar) bool { | |
415 | return s.Value == rhs.(*MonthInterval).Value | |
416 | } | |
417 | func (s *MonthInterval) Data() []byte { | |
418 | return (*[arrow.MonthIntervalSizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
419 | } | |
420 | ||
421 | func NewMonthIntervalScalar(val arrow.MonthInterval) *MonthInterval { | |
422 | return &MonthInterval{scalar{arrow.FixedWidthTypes.MonthInterval, true}, val} | |
423 | } | |
424 | ||
425 | type DayTimeInterval struct { | |
426 | scalar | |
427 | Value arrow.DayTimeInterval | |
428 | } | |
429 | ||
430 | func (DayTimeInterval) temporal() {} | |
431 | func (DayTimeInterval) interval() {} | |
432 | func (s *DayTimeInterval) value() interface{} { return s.Value } | |
433 | func (s *DayTimeInterval) Data() []byte { | |
434 | return (*[arrow.DayTimeIntervalSizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
435 | } | |
436 | func (s *DayTimeInterval) String() string { | |
437 | if !s.Valid { | |
438 | return "null" | |
439 | } | |
440 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
441 | if err != nil { | |
442 | return "..." | |
443 | } | |
444 | return string(val.(*String).Value.Bytes()) | |
445 | } | |
446 | ||
447 | func (s *DayTimeInterval) CastTo(to arrow.DataType) (Scalar, error) { | |
448 | if !s.Valid { | |
449 | return MakeNullScalar(to), nil | |
450 | } | |
451 | ||
452 | if !arrow.TypeEqual(s.DataType(), to) { | |
453 | return nil, xerrors.Errorf("non-null daytimeinterval scalar cannot be cast to anything other than monthinterval") | |
454 | } | |
455 | ||
456 | return s, nil | |
457 | } | |
458 | ||
459 | func (s *DayTimeInterval) equals(rhs Scalar) bool { | |
460 | return s.Value == rhs.(*DayTimeInterval).Value | |
461 | } | |
462 | ||
463 | func NewDayTimeIntervalScalar(val arrow.DayTimeInterval) *DayTimeInterval { | |
464 | return &DayTimeInterval{scalar{arrow.FixedWidthTypes.DayTimeInterval, true}, val} | |
465 | } | |
466 | ||
467 | type MonthDayNanoInterval struct { | |
468 | scalar | |
469 | Value arrow.MonthDayNanoInterval | |
470 | } | |
471 | ||
472 | func (MonthDayNanoInterval) temporal() {} | |
473 | func (MonthDayNanoInterval) interval() {} | |
474 | func (s *MonthDayNanoInterval) value() interface{} { return s.Value } | |
475 | func (s *MonthDayNanoInterval) Data() []byte { | |
476 | return (*[arrow.MonthDayNanoIntervalSizeBytes]byte)(unsafe.Pointer(&s.Value))[:] | |
477 | } | |
478 | func (s *MonthDayNanoInterval) String() string { | |
479 | if !s.Valid { | |
480 | return "null" | |
481 | } | |
482 | val, err := s.CastTo(arrow.BinaryTypes.String) | |
483 | if err != nil { | |
484 | return "..." | |
485 | } | |
486 | return string(val.(*String).Value.Bytes()) | |
487 | } | |
488 | ||
489 | func (s *MonthDayNanoInterval) CastTo(to arrow.DataType) (Scalar, error) { | |
490 | if !s.Valid { | |
491 | return MakeNullScalar(to), nil | |
492 | } | |
493 | ||
494 | if !arrow.TypeEqual(s.DataType(), to) { | |
495 | return nil, xerrors.Errorf("non-null month_day_nano_interval scalar cannot be cast to anything other than monthinterval") | |
496 | } | |
497 | ||
498 | return s, nil | |
499 | } | |
500 | ||
501 | func (s *MonthDayNanoInterval) equals(rhs Scalar) bool { | |
502 | return s.Value == rhs.(*MonthDayNanoInterval).Value | |
503 | } | |
504 | ||
505 | func NewMonthDayNanoIntervalScalar(val arrow.MonthDayNanoInterval) *MonthDayNanoInterval { | |
506 | return &MonthDayNanoInterval{scalar{arrow.FixedWidthTypes.MonthDayNanoInterval, true}, val} | |
507 | } | |
508 | ||
509 | var ( | |
510 | _ Scalar = (*Date32)(nil) | |
511 | ) |