polars-df 0.1.0 → 0.1.2

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.
@@ -3,8 +3,10 @@ use polars::chunked_array::ops::SortOptions;
3
3
  use polars::lazy::dsl;
4
4
  use polars::lazy::dsl::Operator;
5
5
  use polars::prelude::*;
6
+ use polars::series::ops::NullBehavior;
6
7
 
7
- use crate::conversion::parse_fill_null_strategy;
8
+ use crate::conversion::*;
9
+ use crate::lazy::apply::*;
8
10
  use crate::lazy::utils::rb_exprs_to_exprs;
9
11
  use crate::RbResult;
10
12
 
@@ -21,10 +23,30 @@ impl From<dsl::Expr> for RbExpr {
21
23
  }
22
24
 
23
25
  impl RbExpr {
26
+ pub fn add(&self, rhs: &RbExpr) -> RbResult<Self> {
27
+ Ok(dsl::binary_expr(self.inner.clone(), Operator::Plus, rhs.inner.clone()).into())
28
+ }
29
+
30
+ pub fn sub(&self, rhs: &RbExpr) -> RbResult<Self> {
31
+ Ok(dsl::binary_expr(self.inner.clone(), Operator::Minus, rhs.inner.clone()).into())
32
+ }
33
+
24
34
  pub fn mul(&self, rhs: &RbExpr) -> RbResult<Self> {
25
35
  Ok(dsl::binary_expr(self.inner.clone(), Operator::Multiply, rhs.inner.clone()).into())
26
36
  }
27
37
 
38
+ pub fn truediv(&self, rhs: &RbExpr) -> RbResult<Self> {
39
+ Ok(dsl::binary_expr(self.inner.clone(), Operator::TrueDivide, rhs.inner.clone()).into())
40
+ }
41
+
42
+ pub fn _mod(&self, rhs: &RbExpr) -> RbResult<Self> {
43
+ Ok(dsl::binary_expr(self.inner.clone(), Operator::Modulus, rhs.inner.clone()).into())
44
+ }
45
+
46
+ pub fn floordiv(&self, rhs: &RbExpr) -> RbResult<Self> {
47
+ Ok(dsl::binary_expr(self.inner.clone(), Operator::FloorDivide, rhs.inner.clone()).into())
48
+ }
49
+
28
50
  pub fn to_str(&self) -> String {
29
51
  format!("{:?}", self.inner)
30
52
  }
@@ -69,6 +91,22 @@ impl RbExpr {
69
91
  self.clone().inner.is_not_null().into()
70
92
  }
71
93
 
94
+ pub fn is_infinite(&self) -> Self {
95
+ self.clone().inner.is_infinite().into()
96
+ }
97
+
98
+ pub fn is_finite(&self) -> Self {
99
+ self.clone().inner.is_finite().into()
100
+ }
101
+
102
+ pub fn is_nan(&self) -> Self {
103
+ self.clone().inner.is_nan().into()
104
+ }
105
+
106
+ pub fn is_not_nan(&self) -> Self {
107
+ self.clone().inner.is_not_nan().into()
108
+ }
109
+
72
110
  pub fn min(&self) -> Self {
73
111
  self.clone().inner.min().into()
74
112
  }
@@ -77,6 +115,14 @@ impl RbExpr {
77
115
  self.clone().inner.max().into()
78
116
  }
79
117
 
118
+ pub fn nan_max(&self) -> Self {
119
+ self.clone().inner.nan_max().into()
120
+ }
121
+
122
+ pub fn nan_min(&self) -> Self {
123
+ self.clone().inner.nan_min().into()
124
+ }
125
+
80
126
  pub fn mean(&self) -> Self {
81
127
  self.clone().inner.mean().into()
82
128
  }
@@ -93,6 +139,10 @@ impl RbExpr {
93
139
  self.clone().inner.n_unique().into()
94
140
  }
95
141
 
142
+ pub fn arg_unique(&self) -> Self {
143
+ self.clone().inner.arg_unique().into()
144
+ }
145
+
96
146
  pub fn unique(&self) -> Self {
97
147
  self.clone().inner.unique().into()
98
148
  }
@@ -112,10 +162,46 @@ impl RbExpr {
112
162
  self.clone().inner.list().into()
113
163
  }
114
164
 
165
+ pub fn quantile(&self, quantile: f64, interpolation: Wrap<QuantileInterpolOptions>) -> Self {
166
+ self.clone()
167
+ .inner
168
+ .quantile(quantile, interpolation.0)
169
+ .into()
170
+ }
171
+
172
+ pub fn agg_groups(&self) -> Self {
173
+ self.clone().inner.agg_groups().into()
174
+ }
175
+
115
176
  pub fn count(&self) -> Self {
116
177
  self.clone().inner.count().into()
117
178
  }
118
179
 
180
+ pub fn value_counts(&self, multithreaded: bool, sorted: bool) -> Self {
181
+ self.inner
182
+ .clone()
183
+ .value_counts(multithreaded, sorted)
184
+ .into()
185
+ }
186
+
187
+ pub fn unique_counts(&self) -> Self {
188
+ self.inner.clone().unique_counts().into()
189
+ }
190
+
191
+ pub fn null_count(&self) -> Self {
192
+ self.inner.clone().null_count().into()
193
+ }
194
+
195
+ pub fn cast(&self, data_type: Wrap<DataType>, strict: bool) -> RbResult<Self> {
196
+ let dt = data_type.0;
197
+ let expr = if strict {
198
+ self.inner.clone().strict_cast(dt)
199
+ } else {
200
+ self.inner.clone().cast(dt)
201
+ };
202
+ Ok(expr.into())
203
+ }
204
+
119
205
  pub fn sort_with(&self, descending: bool, nulls_last: bool) -> Self {
120
206
  self.clone()
121
207
  .inner
@@ -126,11 +212,62 @@ impl RbExpr {
126
212
  .into()
127
213
  }
128
214
 
215
+ pub fn arg_sort(&self, reverse: bool, nulls_last: bool) -> Self {
216
+ self.clone()
217
+ .inner
218
+ .arg_sort(SortOptions {
219
+ descending: reverse,
220
+ nulls_last,
221
+ })
222
+ .into()
223
+ }
224
+
225
+ pub fn top_k(&self, k: usize, reverse: bool) -> Self {
226
+ self.inner.clone().top_k(k, reverse).into()
227
+ }
228
+
229
+ pub fn arg_max(&self) -> Self {
230
+ self.clone().inner.arg_max().into()
231
+ }
232
+
233
+ pub fn arg_min(&self) -> Self {
234
+ self.clone().inner.arg_min().into()
235
+ }
236
+
237
+ pub fn search_sorted(&self, element: &RbExpr) -> Self {
238
+ self.inner
239
+ .clone()
240
+ .search_sorted(element.inner.clone())
241
+ .into()
242
+ }
243
+
244
+ pub fn take(&self, idx: &RbExpr) -> Self {
245
+ self.clone().inner.take(idx.inner.clone()).into()
246
+ }
247
+
129
248
  pub fn sort_by(&self, by: RArray, reverse: Vec<bool>) -> RbResult<Self> {
130
249
  let by = rb_exprs_to_exprs(by)?;
131
250
  Ok(self.clone().inner.sort_by(by, reverse).into())
132
251
  }
133
252
 
253
+ pub fn backward_fill(&self, limit: FillNullLimit) -> Self {
254
+ self.clone().inner.backward_fill(limit).into()
255
+ }
256
+
257
+ pub fn forward_fill(&self, limit: FillNullLimit) -> Self {
258
+ self.clone().inner.forward_fill(limit).into()
259
+ }
260
+
261
+ pub fn shift(&self, periods: i64) -> Self {
262
+ self.clone().inner.shift(periods).into()
263
+ }
264
+ pub fn shift_and_fill(&self, periods: i64, fill_value: &RbExpr) -> Self {
265
+ self.clone()
266
+ .inner
267
+ .shift_and_fill(periods, fill_value.inner.clone())
268
+ .into()
269
+ }
270
+
134
271
  pub fn fill_null(&self, expr: &RbExpr) -> Self {
135
272
  self.clone().inner.fill_null(expr.inner.clone()).into()
136
273
  }
@@ -177,6 +314,26 @@ impl RbExpr {
177
314
  self.clone().inner.var(ddof).into()
178
315
  }
179
316
 
317
+ pub fn is_unique(&self) -> Self {
318
+ self.clone().inner.is_unique().into()
319
+ }
320
+
321
+ pub fn is_first(&self) -> Self {
322
+ self.clone().inner.is_first().into()
323
+ }
324
+
325
+ pub fn explode(&self) -> Self {
326
+ self.clone().inner.explode().into()
327
+ }
328
+
329
+ pub fn take_every(&self, n: usize) -> Self {
330
+ self.clone()
331
+ .inner
332
+ .map(move |s: Series| Ok(s.take_every(n)), GetOutput::same_type())
333
+ .with_fmt("take_every")
334
+ .into()
335
+ }
336
+
180
337
  pub fn tail(&self, n: Option<usize>) -> Self {
181
338
  self.clone().inner.tail(n).into()
182
339
  }
@@ -185,6 +342,115 @@ impl RbExpr {
185
342
  self.clone().inner.head(n).into()
186
343
  }
187
344
 
345
+ pub fn slice(&self, offset: &RbExpr, length: &RbExpr) -> Self {
346
+ self.inner
347
+ .clone()
348
+ .slice(offset.inner.clone(), length.inner.clone())
349
+ .into()
350
+ }
351
+
352
+ pub fn append(&self, other: &RbExpr, upcast: bool) -> Self {
353
+ self.inner
354
+ .clone()
355
+ .append(other.inner.clone(), upcast)
356
+ .into()
357
+ }
358
+
359
+ pub fn rechunk(&self) -> Self {
360
+ self.inner
361
+ .clone()
362
+ .map(|s| Ok(s.rechunk()), GetOutput::same_type())
363
+ .into()
364
+ }
365
+
366
+ pub fn round(&self, decimals: u32) -> Self {
367
+ self.clone().inner.round(decimals).into()
368
+ }
369
+
370
+ pub fn floor(&self) -> Self {
371
+ self.clone().inner.floor().into()
372
+ }
373
+
374
+ pub fn ceil(&self) -> Self {
375
+ self.clone().inner.ceil().into()
376
+ }
377
+
378
+ pub fn clip(&self, min: Value, max: Value) -> Self {
379
+ let min = min.try_convert::<Wrap<AnyValue>>().unwrap().0;
380
+ let max = max.try_convert::<Wrap<AnyValue>>().unwrap().0;
381
+ self.clone().inner.clip(min, max).into()
382
+ }
383
+
384
+ pub fn clip_min(&self, min: Value) -> Self {
385
+ let min = min.try_convert::<Wrap<AnyValue>>().unwrap().0;
386
+ self.clone().inner.clip_min(min).into()
387
+ }
388
+
389
+ pub fn clip_max(&self, max: Value) -> Self {
390
+ let max = max.try_convert::<Wrap<AnyValue>>().unwrap().0;
391
+ self.clone().inner.clip_max(max).into()
392
+ }
393
+
394
+ pub fn abs(&self) -> Self {
395
+ self.clone().inner.abs().into()
396
+ }
397
+
398
+ pub fn sin(&self) -> Self {
399
+ self.clone().inner.sin().into()
400
+ }
401
+
402
+ pub fn cos(&self) -> Self {
403
+ self.clone().inner.cos().into()
404
+ }
405
+
406
+ pub fn tan(&self) -> Self {
407
+ self.clone().inner.tan().into()
408
+ }
409
+
410
+ pub fn arcsin(&self) -> Self {
411
+ self.clone().inner.arcsin().into()
412
+ }
413
+
414
+ pub fn arccos(&self) -> Self {
415
+ self.clone().inner.arccos().into()
416
+ }
417
+
418
+ pub fn arctan(&self) -> Self {
419
+ self.clone().inner.arctan().into()
420
+ }
421
+
422
+ pub fn sinh(&self) -> Self {
423
+ self.clone().inner.sinh().into()
424
+ }
425
+
426
+ pub fn cosh(&self) -> Self {
427
+ self.clone().inner.cosh().into()
428
+ }
429
+
430
+ pub fn tanh(&self) -> Self {
431
+ self.clone().inner.tanh().into()
432
+ }
433
+
434
+ pub fn arcsinh(&self) -> Self {
435
+ self.clone().inner.arcsinh().into()
436
+ }
437
+
438
+ pub fn arccosh(&self) -> Self {
439
+ self.clone().inner.arccosh().into()
440
+ }
441
+
442
+ pub fn arctanh(&self) -> Self {
443
+ self.clone().inner.arctanh().into()
444
+ }
445
+
446
+ pub fn sign(&self) -> Self {
447
+ self.clone().inner.sign().into()
448
+ }
449
+
450
+ pub fn is_duplicated(&self) -> Self {
451
+ self.clone().inner.is_duplicated().into()
452
+ }
453
+
188
454
  pub fn over(&self, partition_by: RArray) -> RbResult<Self> {
189
455
  let partition_by = rb_exprs_to_exprs(partition_by)?;
190
456
  Ok(self.clone().inner.over(partition_by).into())
@@ -202,11 +468,130 @@ impl RbExpr {
202
468
  self.clone().inner.or(expr.inner.clone()).into()
203
469
  }
204
470
 
471
+ pub fn is_in(&self, expr: &RbExpr) -> Self {
472
+ self.clone().inner.is_in(expr.inner.clone()).into()
473
+ }
474
+
475
+ pub fn repeat_by(&self, by: &RbExpr) -> Self {
476
+ self.clone().inner.repeat_by(by.inner.clone()).into()
477
+ }
478
+
479
+ pub fn pow(&self, exponent: &RbExpr) -> Self {
480
+ self.clone().inner.pow(exponent.inner.clone()).into()
481
+ }
482
+
483
+ pub fn cumsum(&self, reverse: bool) -> Self {
484
+ self.clone().inner.cumsum(reverse).into()
485
+ }
486
+
487
+ pub fn cummax(&self, reverse: bool) -> Self {
488
+ self.clone().inner.cummax(reverse).into()
489
+ }
490
+
491
+ pub fn cummin(&self, reverse: bool) -> Self {
492
+ self.clone().inner.cummin(reverse).into()
493
+ }
494
+
495
+ pub fn cumprod(&self, reverse: bool) -> Self {
496
+ self.clone().inner.cumprod(reverse).into()
497
+ }
498
+
205
499
  pub fn product(&self) -> Self {
206
500
  self.clone().inner.product().into()
207
501
  }
208
502
 
209
- pub fn str_lengths(&self) -> RbExpr {
503
+ pub fn shrink_dtype(&self) -> Self {
504
+ self.inner.clone().shrink_dtype().into()
505
+ }
506
+
507
+ pub fn str_parse_date(&self, fmt: Option<String>, strict: bool, exact: bool) -> Self {
508
+ self.inner
509
+ .clone()
510
+ .str()
511
+ .strptime(StrpTimeOptions {
512
+ date_dtype: DataType::Date,
513
+ fmt,
514
+ strict,
515
+ exact,
516
+ })
517
+ .into()
518
+ }
519
+
520
+ pub fn str_parse_datetime(&self, fmt: Option<String>, strict: bool, exact: bool) -> Self {
521
+ let tu = match fmt {
522
+ Some(ref fmt) => {
523
+ if fmt.contains("%.9f")
524
+ || fmt.contains("%9f")
525
+ || fmt.contains("%f")
526
+ || fmt.contains("%.f")
527
+ {
528
+ TimeUnit::Nanoseconds
529
+ } else if fmt.contains("%.3f") || fmt.contains("%3f") {
530
+ TimeUnit::Milliseconds
531
+ } else {
532
+ TimeUnit::Microseconds
533
+ }
534
+ }
535
+ None => TimeUnit::Microseconds,
536
+ };
537
+ self.inner
538
+ .clone()
539
+ .str()
540
+ .strptime(StrpTimeOptions {
541
+ date_dtype: DataType::Datetime(tu, None),
542
+ fmt,
543
+ strict,
544
+ exact,
545
+ })
546
+ .into()
547
+ }
548
+
549
+ pub fn str_parse_time(&self, fmt: Option<String>, strict: bool, exact: bool) -> Self {
550
+ self.inner
551
+ .clone()
552
+ .str()
553
+ .strptime(StrpTimeOptions {
554
+ date_dtype: DataType::Time,
555
+ fmt,
556
+ strict,
557
+ exact,
558
+ })
559
+ .into()
560
+ }
561
+
562
+ pub fn str_strip(&self, matches: Option<char>) -> Self {
563
+ self.inner.clone().str().strip(matches).into()
564
+ }
565
+
566
+ pub fn str_rstrip(&self, matches: Option<char>) -> Self {
567
+ self.inner.clone().str().rstrip(matches).into()
568
+ }
569
+
570
+ pub fn str_lstrip(&self, matches: Option<char>) -> Self {
571
+ self.inner.clone().str().lstrip(matches).into()
572
+ }
573
+
574
+ pub fn str_slice(&self, start: i64, length: Option<u64>) -> Self {
575
+ let function = move |s: Series| {
576
+ let ca = s.utf8()?;
577
+ Ok(ca.str_slice(start, length)?.into_series())
578
+ };
579
+ self.clone()
580
+ .inner
581
+ .map(function, GetOutput::from_type(DataType::Utf8))
582
+ .with_fmt("str.slice")
583
+ .into()
584
+ }
585
+
586
+ pub fn str_to_uppercase(&self) -> Self {
587
+ self.inner.clone().str().to_uppercase().into()
588
+ }
589
+
590
+ pub fn str_to_lowercase(&self) -> Self {
591
+ self.inner.clone().str().to_lowercase().into()
592
+ }
593
+
594
+ pub fn str_lengths(&self) -> Self {
210
595
  let function = |s: Series| {
211
596
  let ca = s.utf8()?;
212
597
  Ok(ca.str_lengths().into_series())
@@ -218,6 +603,46 @@ impl RbExpr {
218
603
  .into()
219
604
  }
220
605
 
606
+ pub fn str_n_chars(&self) -> Self {
607
+ let function = |s: Series| {
608
+ let ca = s.utf8()?;
609
+ Ok(ca.str_n_chars().into_series())
610
+ };
611
+ self.clone()
612
+ .inner
613
+ .map(function, GetOutput::from_type(DataType::UInt32))
614
+ .with_fmt("str.n_chars")
615
+ .into()
616
+ }
617
+
618
+ pub fn str_replace(&self, pat: &RbExpr, val: &RbExpr, literal: bool) -> Self {
619
+ self.inner
620
+ .clone()
621
+ .str()
622
+ .replace(pat.inner.clone(), val.inner.clone(), literal)
623
+ .into()
624
+ }
625
+
626
+ pub fn str_replace_all(&self, pat: &RbExpr, val: &RbExpr, literal: bool) -> Self {
627
+ self.inner
628
+ .clone()
629
+ .str()
630
+ .replace_all(pat.inner.clone(), val.inner.clone(), literal)
631
+ .into()
632
+ }
633
+
634
+ pub fn str_zfill(&self, alignment: usize) -> Self {
635
+ self.clone().inner.str().zfill(alignment).into()
636
+ }
637
+
638
+ pub fn str_ljust(&self, width: usize, fillchar: char) -> Self {
639
+ self.clone().inner.str().ljust(width, fillchar).into()
640
+ }
641
+
642
+ pub fn str_rjust(&self, width: usize, fillchar: char) -> Self {
643
+ self.clone().inner.str().rjust(width, fillchar).into()
644
+ }
645
+
221
646
  pub fn str_contains(&self, pat: String, literal: Option<bool>) -> Self {
222
647
  match literal {
223
648
  Some(true) => self.inner.clone().str().contains_literal(pat).into(),
@@ -225,23 +650,717 @@ impl RbExpr {
225
650
  }
226
651
  }
227
652
 
228
- pub fn prefix(&self, prefix: String) -> RbExpr {
653
+ pub fn str_ends_with(&self, sub: String) -> Self {
654
+ self.inner.clone().str().ends_with(sub).into()
655
+ }
656
+
657
+ pub fn str_starts_with(&self, sub: String) -> Self {
658
+ self.inner.clone().str().starts_with(sub).into()
659
+ }
660
+
661
+ pub fn str_extract(&self, pat: String, group_index: usize) -> Self {
662
+ self.inner.clone().str().extract(&pat, group_index).into()
663
+ }
664
+
665
+ pub fn str_extract_all(&self, pat: String) -> Self {
666
+ self.inner.clone().str().extract_all(&pat).into()
667
+ }
668
+
669
+ pub fn count_match(&self, pat: String) -> Self {
670
+ self.inner.clone().str().count_match(&pat).into()
671
+ }
672
+
673
+ pub fn strftime(&self, fmt: String) -> Self {
674
+ self.inner.clone().dt().strftime(&fmt).into()
675
+ }
676
+
677
+ pub fn str_split(&self, by: String) -> Self {
678
+ self.inner.clone().str().split(&by).into()
679
+ }
680
+
681
+ pub fn str_split_inclusive(&self, by: String) -> Self {
682
+ self.inner.clone().str().split_inclusive(&by).into()
683
+ }
684
+
685
+ pub fn str_split_exact(&self, by: String, n: usize) -> Self {
686
+ self.inner.clone().str().split_exact(&by, n).into()
687
+ }
688
+
689
+ pub fn str_split_exact_inclusive(&self, by: String, n: usize) -> Self {
690
+ self.inner
691
+ .clone()
692
+ .str()
693
+ .split_exact_inclusive(&by, n)
694
+ .into()
695
+ }
696
+
697
+ pub fn str_splitn(&self, by: String, n: usize) -> Self {
698
+ self.inner.clone().str().splitn(&by, n).into()
699
+ }
700
+
701
+ pub fn arr_lengths(&self) -> Self {
702
+ self.inner.clone().arr().lengths().into()
703
+ }
704
+
705
+ pub fn arr_contains(&self, other: &RbExpr) -> Self {
706
+ self.inner
707
+ .clone()
708
+ .arr()
709
+ .contains(other.inner.clone())
710
+ .into()
711
+ }
712
+
713
+ pub fn year(&self) -> Self {
714
+ self.clone().inner.dt().year().into()
715
+ }
716
+
717
+ pub fn iso_year(&self) -> Self {
718
+ self.clone().inner.dt().iso_year().into()
719
+ }
720
+
721
+ pub fn quarter(&self) -> Self {
722
+ self.clone().inner.dt().quarter().into()
723
+ }
724
+
725
+ pub fn month(&self) -> Self {
726
+ self.clone().inner.dt().month().into()
727
+ }
728
+
729
+ pub fn week(&self) -> Self {
730
+ self.clone().inner.dt().week().into()
731
+ }
732
+
733
+ pub fn weekday(&self) -> Self {
734
+ self.clone().inner.dt().weekday().into()
735
+ }
736
+
737
+ pub fn day(&self) -> Self {
738
+ self.clone().inner.dt().day().into()
739
+ }
740
+
741
+ pub fn ordinal_day(&self) -> Self {
742
+ self.clone().inner.dt().ordinal_day().into()
743
+ }
744
+
745
+ pub fn hour(&self) -> Self {
746
+ self.clone().inner.dt().hour().into()
747
+ }
748
+
749
+ pub fn minute(&self) -> Self {
750
+ self.clone().inner.dt().minute().into()
751
+ }
752
+
753
+ pub fn second(&self) -> Self {
754
+ self.clone().inner.dt().second().into()
755
+ }
756
+
757
+ pub fn millisecond(&self) -> Self {
758
+ self.clone().inner.dt().millisecond().into()
759
+ }
760
+
761
+ pub fn microsecond(&self) -> Self {
762
+ self.clone().inner.dt().microsecond().into()
763
+ }
764
+
765
+ pub fn nanosecond(&self) -> Self {
766
+ self.clone().inner.dt().nanosecond().into()
767
+ }
768
+
769
+ pub fn duration_days(&self) -> Self {
770
+ self.inner
771
+ .clone()
772
+ .map(
773
+ |s| Ok(s.duration()?.days().into_series()),
774
+ GetOutput::from_type(DataType::Int64),
775
+ )
776
+ .into()
777
+ }
778
+
779
+ pub fn duration_hours(&self) -> Self {
780
+ self.inner
781
+ .clone()
782
+ .map(
783
+ |s| Ok(s.duration()?.hours().into_series()),
784
+ GetOutput::from_type(DataType::Int64),
785
+ )
786
+ .into()
787
+ }
788
+
789
+ pub fn duration_minutes(&self) -> Self {
790
+ self.inner
791
+ .clone()
792
+ .map(
793
+ |s| Ok(s.duration()?.minutes().into_series()),
794
+ GetOutput::from_type(DataType::Int64),
795
+ )
796
+ .into()
797
+ }
798
+
799
+ pub fn duration_seconds(&self) -> Self {
800
+ self.inner
801
+ .clone()
802
+ .map(
803
+ |s| Ok(s.duration()?.seconds().into_series()),
804
+ GetOutput::from_type(DataType::Int64),
805
+ )
806
+ .into()
807
+ }
808
+
809
+ pub fn duration_nanoseconds(&self) -> Self {
810
+ self.inner
811
+ .clone()
812
+ .map(
813
+ |s| Ok(s.duration()?.nanoseconds().into_series()),
814
+ GetOutput::from_type(DataType::Int64),
815
+ )
816
+ .into()
817
+ }
818
+
819
+ pub fn duration_microseconds(&self) -> Self {
820
+ self.inner
821
+ .clone()
822
+ .map(
823
+ |s| Ok(s.duration()?.microseconds().into_series()),
824
+ GetOutput::from_type(DataType::Int64),
825
+ )
826
+ .into()
827
+ }
828
+
829
+ pub fn duration_milliseconds(&self) -> Self {
830
+ self.inner
831
+ .clone()
832
+ .map(
833
+ |s| Ok(s.duration()?.milliseconds().into_series()),
834
+ GetOutput::from_type(DataType::Int64),
835
+ )
836
+ .into()
837
+ }
838
+
839
+ pub fn timestamp(&self, tu: Wrap<TimeUnit>) -> Self {
840
+ self.inner.clone().dt().timestamp(tu.0).into()
841
+ }
842
+
843
+ pub fn dt_offset_by(&self, by: String) -> Self {
844
+ let by = Duration::parse(&by);
845
+ self.inner.clone().dt().offset_by(by).into()
846
+ }
847
+
848
+ pub fn dt_epoch_seconds(&self) -> Self {
849
+ self.clone()
850
+ .inner
851
+ .map(
852
+ |s| {
853
+ s.timestamp(TimeUnit::Milliseconds)
854
+ .map(|ca| (ca / 1000).into_series())
855
+ },
856
+ GetOutput::from_type(DataType::Int64),
857
+ )
858
+ .into()
859
+ }
860
+
861
+ pub fn dt_with_time_unit(&self, tu: Wrap<TimeUnit>) -> Self {
862
+ self.inner.clone().dt().with_time_unit(tu.0).into()
863
+ }
864
+
865
+ pub fn dt_with_time_zone(&self, tz: Option<TimeZone>) -> Self {
866
+ self.inner.clone().dt().with_time_zone(tz).into()
867
+ }
868
+
869
+ pub fn dt_cast_time_unit(&self, tu: Wrap<TimeUnit>) -> Self {
870
+ self.inner.clone().dt().cast_time_unit(tu.0).into()
871
+ }
872
+
873
+ pub fn dt_cast_time_zone(&self, tz: String) -> Self {
874
+ self.inner.clone().dt().cast_time_zone(tz).into()
875
+ }
876
+
877
+ pub fn dt_tz_localize(&self, tz: String) -> Self {
878
+ self.inner.clone().dt().tz_localize(tz).into()
879
+ }
880
+
881
+ pub fn dt_truncate(&self, every: String, offset: String) -> Self {
882
+ self.inner.clone().dt().truncate(&every, &offset).into()
883
+ }
884
+
885
+ pub fn dt_round(&self, every: String, offset: String) -> Self {
886
+ self.inner.clone().dt().round(&every, &offset).into()
887
+ }
888
+
889
+ pub fn mode(&self) -> Self {
890
+ self.inner.clone().mode().into()
891
+ }
892
+
893
+ pub fn keep_name(&self) -> Self {
894
+ self.inner.clone().keep_name().into()
895
+ }
896
+
897
+ pub fn prefix(&self, prefix: String) -> Self {
229
898
  self.inner.clone().prefix(&prefix).into()
230
899
  }
231
900
 
232
- pub fn suffix(&self, suffix: String) -> RbExpr {
901
+ pub fn suffix(&self, suffix: String) -> Self {
233
902
  self.inner.clone().suffix(&suffix).into()
234
903
  }
235
904
 
236
- pub fn interpolate(&self) -> RbExpr {
905
+ pub fn exclude(&self, columns: Vec<String>) -> Self {
906
+ self.inner.clone().exclude(columns).into()
907
+ }
908
+
909
+ pub fn interpolate(&self) -> Self {
237
910
  self.inner.clone().interpolate().into()
238
911
  }
912
+
913
+ pub fn rolling_sum(
914
+ &self,
915
+ window_size: String,
916
+ weights: Option<Vec<f64>>,
917
+ min_periods: usize,
918
+ center: bool,
919
+ by: Option<String>,
920
+ closed: Option<Wrap<ClosedWindow>>,
921
+ ) -> Self {
922
+ let options = RollingOptions {
923
+ window_size: Duration::parse(&window_size),
924
+ weights,
925
+ min_periods,
926
+ center,
927
+ by,
928
+ closed_window: closed.map(|c| c.0),
929
+ };
930
+ self.inner.clone().rolling_sum(options).into()
931
+ }
932
+
933
+ pub fn rolling_min(
934
+ &self,
935
+ window_size: String,
936
+ weights: Option<Vec<f64>>,
937
+ min_periods: usize,
938
+ center: bool,
939
+ by: Option<String>,
940
+ closed: Option<Wrap<ClosedWindow>>,
941
+ ) -> Self {
942
+ let options = RollingOptions {
943
+ window_size: Duration::parse(&window_size),
944
+ weights,
945
+ min_periods,
946
+ center,
947
+ by,
948
+ closed_window: closed.map(|c| c.0),
949
+ };
950
+ self.inner.clone().rolling_min(options).into()
951
+ }
952
+
953
+ pub fn rolling_max(
954
+ &self,
955
+ window_size: String,
956
+ weights: Option<Vec<f64>>,
957
+ min_periods: usize,
958
+ center: bool,
959
+ by: Option<String>,
960
+ closed: Option<Wrap<ClosedWindow>>,
961
+ ) -> Self {
962
+ let options = RollingOptions {
963
+ window_size: Duration::parse(&window_size),
964
+ weights,
965
+ min_periods,
966
+ center,
967
+ by,
968
+ closed_window: closed.map(|c| c.0),
969
+ };
970
+ self.inner.clone().rolling_max(options).into()
971
+ }
972
+
973
+ pub fn rolling_mean(
974
+ &self,
975
+ window_size: String,
976
+ weights: Option<Vec<f64>>,
977
+ min_periods: usize,
978
+ center: bool,
979
+ by: Option<String>,
980
+ closed: Option<Wrap<ClosedWindow>>,
981
+ ) -> Self {
982
+ let options = RollingOptions {
983
+ window_size: Duration::parse(&window_size),
984
+ weights,
985
+ min_periods,
986
+ center,
987
+ by,
988
+ closed_window: closed.map(|c| c.0),
989
+ };
990
+
991
+ self.inner.clone().rolling_mean(options).into()
992
+ }
993
+
994
+ pub fn rolling_std(
995
+ &self,
996
+ window_size: String,
997
+ weights: Option<Vec<f64>>,
998
+ min_periods: usize,
999
+ center: bool,
1000
+ by: Option<String>,
1001
+ closed: Option<Wrap<ClosedWindow>>,
1002
+ ) -> Self {
1003
+ let options = RollingOptions {
1004
+ window_size: Duration::parse(&window_size),
1005
+ weights,
1006
+ min_periods,
1007
+ center,
1008
+ by,
1009
+ closed_window: closed.map(|c| c.0),
1010
+ };
1011
+
1012
+ self.inner.clone().rolling_std(options).into()
1013
+ }
1014
+
1015
+ pub fn rolling_var(
1016
+ &self,
1017
+ window_size: String,
1018
+ weights: Option<Vec<f64>>,
1019
+ min_periods: usize,
1020
+ center: bool,
1021
+ by: Option<String>,
1022
+ closed: Option<Wrap<ClosedWindow>>,
1023
+ ) -> Self {
1024
+ let options = RollingOptions {
1025
+ window_size: Duration::parse(&window_size),
1026
+ weights,
1027
+ min_periods,
1028
+ center,
1029
+ by,
1030
+ closed_window: closed.map(|c| c.0),
1031
+ };
1032
+
1033
+ self.inner.clone().rolling_var(options).into()
1034
+ }
1035
+
1036
+ pub fn rolling_median(
1037
+ &self,
1038
+ window_size: String,
1039
+ weights: Option<Vec<f64>>,
1040
+ min_periods: usize,
1041
+ center: bool,
1042
+ by: Option<String>,
1043
+ closed: Option<Wrap<ClosedWindow>>,
1044
+ ) -> Self {
1045
+ let options = RollingOptions {
1046
+ window_size: Duration::parse(&window_size),
1047
+ weights,
1048
+ min_periods,
1049
+ center,
1050
+ by,
1051
+ closed_window: closed.map(|c| c.0),
1052
+ };
1053
+ self.inner.clone().rolling_median(options).into()
1054
+ }
1055
+
1056
+ #[allow(clippy::too_many_arguments)]
1057
+ pub fn rolling_quantile(
1058
+ &self,
1059
+ quantile: f64,
1060
+ interpolation: Wrap<QuantileInterpolOptions>,
1061
+ window_size: String,
1062
+ weights: Option<Vec<f64>>,
1063
+ min_periods: usize,
1064
+ center: bool,
1065
+ by: Option<String>,
1066
+ closed: Option<Wrap<ClosedWindow>>,
1067
+ ) -> Self {
1068
+ let options = RollingOptions {
1069
+ window_size: Duration::parse(&window_size),
1070
+ weights,
1071
+ min_periods,
1072
+ center,
1073
+ by,
1074
+ closed_window: closed.map(|c| c.0),
1075
+ };
1076
+
1077
+ self.inner
1078
+ .clone()
1079
+ .rolling_quantile(quantile, interpolation.0, options)
1080
+ .into()
1081
+ }
1082
+
1083
+ pub fn rolling_skew(&self, window_size: usize, bias: bool) -> Self {
1084
+ self.inner
1085
+ .clone()
1086
+ .rolling_apply_float(window_size, move |ca| {
1087
+ ca.clone().into_series().skew(bias).unwrap()
1088
+ })
1089
+ .into()
1090
+ }
1091
+
1092
+ pub fn lower_bound(&self) -> Self {
1093
+ self.inner.clone().lower_bound().into()
1094
+ }
1095
+
1096
+ pub fn upper_bound(&self) -> Self {
1097
+ self.inner.clone().upper_bound().into()
1098
+ }
1099
+
1100
+ pub fn lst_max(&self) -> Self {
1101
+ self.inner.clone().arr().max().into()
1102
+ }
1103
+
1104
+ pub fn lst_min(&self) -> Self {
1105
+ self.inner.clone().arr().min().into()
1106
+ }
1107
+
1108
+ pub fn lst_sum(&self) -> Self {
1109
+ self.inner.clone().arr().sum().with_fmt("arr.sum").into()
1110
+ }
1111
+
1112
+ pub fn lst_mean(&self) -> Self {
1113
+ self.inner.clone().arr().mean().with_fmt("arr.mean").into()
1114
+ }
1115
+
1116
+ pub fn lst_sort(&self, reverse: bool) -> Self {
1117
+ self.inner
1118
+ .clone()
1119
+ .arr()
1120
+ .sort(SortOptions {
1121
+ descending: reverse,
1122
+ ..Default::default()
1123
+ })
1124
+ .with_fmt("arr.sort")
1125
+ .into()
1126
+ }
1127
+
1128
+ pub fn lst_reverse(&self) -> Self {
1129
+ self.inner
1130
+ .clone()
1131
+ .arr()
1132
+ .reverse()
1133
+ .with_fmt("arr.reverse")
1134
+ .into()
1135
+ }
1136
+
1137
+ pub fn lst_unique(&self) -> Self {
1138
+ self.inner
1139
+ .clone()
1140
+ .arr()
1141
+ .unique()
1142
+ .with_fmt("arr.unique")
1143
+ .into()
1144
+ }
1145
+
1146
+ pub fn lst_get(&self, index: &RbExpr) -> Self {
1147
+ self.inner.clone().arr().get(index.inner.clone()).into()
1148
+ }
1149
+
1150
+ pub fn lst_join(&self, separator: String) -> Self {
1151
+ self.inner.clone().arr().join(&separator).into()
1152
+ }
1153
+
1154
+ pub fn lst_arg_min(&self) -> Self {
1155
+ self.inner.clone().arr().arg_min().into()
1156
+ }
1157
+
1158
+ pub fn lst_arg_max(&self) -> Self {
1159
+ self.inner.clone().arr().arg_max().into()
1160
+ }
1161
+
1162
+ pub fn lst_diff(&self, n: usize, null_behavior: Wrap<NullBehavior>) -> RbResult<Self> {
1163
+ Ok(self.inner.clone().arr().diff(n, null_behavior.0).into())
1164
+ }
1165
+
1166
+ pub fn lst_shift(&self, periods: i64) -> Self {
1167
+ self.inner.clone().arr().shift(periods).into()
1168
+ }
1169
+
1170
+ pub fn lst_slice(&self, offset: &RbExpr, length: Option<&RbExpr>) -> Self {
1171
+ let length = match length {
1172
+ Some(i) => i.inner.clone(),
1173
+ None => dsl::lit(i64::MAX),
1174
+ };
1175
+ self.inner
1176
+ .clone()
1177
+ .arr()
1178
+ .slice(offset.inner.clone(), length)
1179
+ .into()
1180
+ }
1181
+
1182
+ pub fn lst_eval(&self, expr: &RbExpr, parallel: bool) -> Self {
1183
+ self.inner
1184
+ .clone()
1185
+ .arr()
1186
+ .eval(expr.inner.clone(), parallel)
1187
+ .into()
1188
+ }
1189
+
1190
+ pub fn cumulative_eval(&self, expr: &RbExpr, min_periods: usize, parallel: bool) -> Self {
1191
+ self.inner
1192
+ .clone()
1193
+ .cumulative_eval(expr.inner.clone(), min_periods, parallel)
1194
+ .into()
1195
+ }
1196
+
1197
+ pub fn rank(&self, method: Wrap<RankMethod>, reverse: bool) -> Self {
1198
+ let options = RankOptions {
1199
+ method: method.0,
1200
+ descending: reverse,
1201
+ };
1202
+ self.inner.clone().rank(options).into()
1203
+ }
1204
+
1205
+ pub fn diff(&self, n: usize, null_behavior: Wrap<NullBehavior>) -> Self {
1206
+ self.inner.clone().diff(n, null_behavior.0).into()
1207
+ }
1208
+
1209
+ pub fn pct_change(&self, n: usize) -> Self {
1210
+ self.inner.clone().pct_change(n).into()
1211
+ }
1212
+
1213
+ pub fn skew(&self, bias: bool) -> Self {
1214
+ self.inner.clone().skew(bias).into()
1215
+ }
1216
+
1217
+ pub fn kurtosis(&self, fisher: bool, bias: bool) -> Self {
1218
+ self.inner.clone().kurtosis(fisher, bias).into()
1219
+ }
1220
+
1221
+ pub fn str_concat(&self, delimiter: String) -> Self {
1222
+ self.inner.clone().str().concat(&delimiter).into()
1223
+ }
1224
+
1225
+ pub fn cat_set_ordering(&self, ordering: Wrap<CategoricalOrdering>) -> Self {
1226
+ self.inner.clone().cat().set_ordering(ordering.0).into()
1227
+ }
1228
+
1229
+ pub fn reshape(&self, dims: Vec<i64>) -> Self {
1230
+ self.inner.clone().reshape(&dims).into()
1231
+ }
1232
+
1233
+ pub fn cumcount(&self, reverse: bool) -> Self {
1234
+ self.inner.clone().cumcount(reverse).into()
1235
+ }
1236
+
1237
+ pub fn to_physical(&self) -> Self {
1238
+ self.inner
1239
+ .clone()
1240
+ .map(
1241
+ |s| Ok(s.to_physical_repr().into_owned()),
1242
+ GetOutput::map_dtype(|dt| dt.to_physical()),
1243
+ )
1244
+ .with_fmt("to_physical")
1245
+ .into()
1246
+ }
1247
+
1248
+ pub fn shuffle(&self, seed: Option<u64>) -> Self {
1249
+ self.inner.clone().shuffle(seed).into()
1250
+ }
1251
+
1252
+ pub fn sample_n(
1253
+ &self,
1254
+ n: usize,
1255
+ with_replacement: bool,
1256
+ shuffle: bool,
1257
+ seed: Option<u64>,
1258
+ ) -> Self {
1259
+ self.inner
1260
+ .clone()
1261
+ .sample_n(n, with_replacement, shuffle, seed)
1262
+ .into()
1263
+ }
1264
+
1265
+ pub fn sample_frac(
1266
+ &self,
1267
+ frac: f64,
1268
+ with_replacement: bool,
1269
+ shuffle: bool,
1270
+ seed: Option<u64>,
1271
+ ) -> Self {
1272
+ self.inner
1273
+ .clone()
1274
+ .sample_frac(frac, with_replacement, shuffle, seed)
1275
+ .into()
1276
+ }
1277
+
1278
+ pub fn ewm_mean(&self, alpha: f64, adjust: bool, min_periods: usize) -> Self {
1279
+ let options = EWMOptions {
1280
+ alpha,
1281
+ adjust,
1282
+ bias: false,
1283
+ min_periods,
1284
+ };
1285
+ self.inner.clone().ewm_mean(options).into()
1286
+ }
1287
+
1288
+ pub fn ewm_std(&self, alpha: f64, adjust: bool, bias: bool, min_periods: usize) -> Self {
1289
+ let options = EWMOptions {
1290
+ alpha,
1291
+ adjust,
1292
+ bias,
1293
+ min_periods,
1294
+ };
1295
+ self.inner.clone().ewm_std(options).into()
1296
+ }
1297
+
1298
+ pub fn ewm_var(&self, alpha: f64, adjust: bool, bias: bool, min_periods: usize) -> Self {
1299
+ let options = EWMOptions {
1300
+ alpha,
1301
+ adjust,
1302
+ bias,
1303
+ min_periods,
1304
+ };
1305
+ self.inner.clone().ewm_var(options).into()
1306
+ }
1307
+
1308
+ pub fn any(&self) -> Self {
1309
+ self.inner.clone().any().into()
1310
+ }
1311
+
1312
+ pub fn all(&self) -> Self {
1313
+ self.inner.clone().all().into()
1314
+ }
1315
+
1316
+ pub fn struct_field_by_name(&self, name: String) -> Self {
1317
+ self.inner.clone().struct_().field_by_name(&name).into()
1318
+ }
1319
+
1320
+ pub fn struct_field_by_index(&self, index: i64) -> Self {
1321
+ self.inner.clone().struct_().field_by_index(index).into()
1322
+ }
1323
+
1324
+ pub fn struct_rename_fields(&self, names: Vec<String>) -> Self {
1325
+ self.inner.clone().struct_().rename_fields(names).into()
1326
+ }
1327
+
1328
+ pub fn log(&self, base: f64) -> Self {
1329
+ self.inner.clone().log(base).into()
1330
+ }
1331
+
1332
+ pub fn exp(&self) -> Self {
1333
+ self.inner.clone().exp().into()
1334
+ }
239
1335
  }
240
1336
 
241
1337
  pub fn col(name: String) -> RbExpr {
242
1338
  dsl::col(&name).into()
243
1339
  }
244
1340
 
1341
+ pub fn count() -> RbExpr {
1342
+ dsl::count().into()
1343
+ }
1344
+
1345
+ pub fn first() -> RbExpr {
1346
+ dsl::first().into()
1347
+ }
1348
+
1349
+ pub fn last() -> RbExpr {
1350
+ dsl::last().into()
1351
+ }
1352
+
1353
+ pub fn cols(names: Vec<String>) -> RbExpr {
1354
+ dsl::cols(names).into()
1355
+ }
1356
+
1357
+ pub fn fold(acc: &RbExpr, lambda: Value, exprs: RArray) -> RbResult<RbExpr> {
1358
+ let exprs = rb_exprs_to_exprs(exprs)?;
1359
+
1360
+ let func = move |a: Series, b: Series| binary_lambda(lambda, a, b);
1361
+ Ok(polars::lazy::dsl::fold_exprs(acc.inner.clone(), func, exprs).into())
1362
+ }
1363
+
245
1364
  // TODO improve
246
1365
  pub fn lit(value: Value) -> RbResult<RbExpr> {
247
1366
  if value.is_nil() {
@@ -296,3 +1415,13 @@ impl RbWhenThen {
296
1415
  pub fn when(predicate: &RbExpr) -> RbWhen {
297
1416
  dsl::when(predicate.inner.clone()).into()
298
1417
  }
1418
+
1419
+ pub fn concat_str(s: RArray, sep: String) -> RbResult<RbExpr> {
1420
+ let s = rb_exprs_to_exprs(s)?;
1421
+ Ok(dsl::concat_str(s, &sep).into())
1422
+ }
1423
+
1424
+ pub fn concat_lst(s: RArray) -> RbResult<RbExpr> {
1425
+ let s = rb_exprs_to_exprs(s)?;
1426
+ Ok(dsl::concat_lst(s).into())
1427
+ }