1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
//! Lazy variant of a [DataFrame].
#[cfg(feature = "python")]
mod python;

mod cached_arenas;
mod err;
#[cfg(not(target_arch = "wasm32"))]
mod exitable;
#[cfg(feature = "pivot")]
pub mod pivot;

#[cfg(any(
    feature = "parquet",
    feature = "ipc",
    feature = "csv",
    feature = "json"
))]
use std::path::PathBuf;
use std::sync::{Arc, Mutex};

pub use anonymous_scan::*;
#[cfg(feature = "csv")]
pub use csv::*;
#[cfg(not(target_arch = "wasm32"))]
pub use exitable::*;
pub use file_list_reader::*;
#[cfg(feature = "ipc")]
pub use ipc::*;
#[cfg(feature = "json")]
pub use ndjson::*;
#[cfg(feature = "parquet")]
pub use parquet::*;
use polars_core::prelude::*;
use polars_io::RowIndex;
use polars_ops::frame::JoinCoalesce;
pub use polars_plan::frame::{AllowedOptimizations, OptState};
use polars_plan::global::FETCH_ROWS;
use smartstring::alias::String as SmartString;

use crate::frame::cached_arenas::CachedArena;
use crate::physical_plan::executors::Executor;
use crate::physical_plan::planner::{
    create_physical_expr, create_physical_plan, ExpressionConversionState,
};
#[cfg(feature = "streaming")]
use crate::physical_plan::streaming::insert_streaming_nodes;
use crate::prelude::*;

pub trait IntoLazy {
    fn lazy(self) -> LazyFrame;
}

impl IntoLazy for DataFrame {
    /// Convert the `DataFrame` into a `LazyFrame`
    fn lazy(self) -> LazyFrame {
        let lp = DslBuilder::from_existing_df(self).build();
        LazyFrame {
            logical_plan: lp,
            opt_state: Default::default(),
            cached_arena: Default::default(),
        }
    }
}

impl IntoLazy for LazyFrame {
    fn lazy(self) -> LazyFrame {
        self
    }
}

/// Lazy abstraction over an eager `DataFrame`.
/// It really is an abstraction over a logical plan. The methods of this struct will incrementally
/// modify a logical plan until output is requested (via [`collect`](crate::frame::LazyFrame::collect)).
#[derive(Clone, Default)]
#[must_use]
pub struct LazyFrame {
    pub logical_plan: DslPlan,
    pub(crate) opt_state: OptState,
    pub(crate) cached_arena: Arc<Mutex<Option<CachedArena>>>,
}

impl From<DslPlan> for LazyFrame {
    fn from(plan: DslPlan) -> Self {
        Self {
            logical_plan: plan,
            opt_state: OptState {
                file_caching: true,
                ..Default::default()
            },
            cached_arena: Default::default(),
        }
    }
}

impl LazyFrame {
    pub(crate) fn from_inner(
        logical_plan: DslPlan,
        opt_state: OptState,
        cached_arena: Arc<Mutex<Option<CachedArena>>>,
    ) -> Self {
        Self {
            logical_plan,
            opt_state,
            cached_arena,
        }
    }

    pub(crate) fn get_plan_builder(self) -> DslBuilder {
        DslBuilder::from(self.logical_plan)
    }

    fn get_opt_state(&self) -> OptState {
        self.opt_state
    }

    fn from_logical_plan(logical_plan: DslPlan, opt_state: OptState) -> Self {
        LazyFrame {
            logical_plan,
            opt_state,
            cached_arena: Default::default(),
        }
    }

    /// Get current optimizations.
    pub fn get_current_optimizations(&self) -> OptState {
        self.opt_state
    }

    /// Set allowed optimizations.
    pub fn with_optimizations(mut self, opt_state: OptState) -> Self {
        self.opt_state = opt_state;
        self
    }

    /// Turn off all optimizations.
    pub fn without_optimizations(self) -> Self {
        self.with_optimizations(OptState {
            projection_pushdown: false,
            predicate_pushdown: false,
            cluster_with_columns: false,
            type_coercion: true,
            simplify_expr: false,
            slice_pushdown: false,
            // will be toggled by a scan operation such as csv scan or parquet scan
            file_caching: false,
            #[cfg(feature = "cse")]
            comm_subplan_elim: false,
            #[cfg(feature = "cse")]
            comm_subexpr_elim: false,
            streaming: false,
            eager: false,
            fast_projection: false,
            row_estimate: false,
        })
    }

    /// Toggle projection pushdown optimization.
    pub fn with_projection_pushdown(mut self, toggle: bool) -> Self {
        self.opt_state.projection_pushdown = toggle;
        self
    }

    /// Toggle cluster with columns optimization.
    pub fn with_cluster_with_columns(mut self, toggle: bool) -> Self {
        self.opt_state.cluster_with_columns = toggle;
        self
    }

    /// Toggle predicate pushdown optimization.
    pub fn with_predicate_pushdown(mut self, toggle: bool) -> Self {
        self.opt_state.predicate_pushdown = toggle;
        self
    }

    /// Toggle type coercion optimization.
    pub fn with_type_coercion(mut self, toggle: bool) -> Self {
        self.opt_state.type_coercion = toggle;
        self
    }

    /// Toggle expression simplification optimization on or off.
    pub fn with_simplify_expr(mut self, toggle: bool) -> Self {
        self.opt_state.simplify_expr = toggle;
        self
    }

    /// Toggle common subplan elimination optimization on or off
    #[cfg(feature = "cse")]
    pub fn with_comm_subplan_elim(mut self, toggle: bool) -> Self {
        self.opt_state.comm_subplan_elim = toggle;
        self
    }

    /// Toggle common subexpression elimination optimization on or off
    #[cfg(feature = "cse")]
    pub fn with_comm_subexpr_elim(mut self, toggle: bool) -> Self {
        self.opt_state.comm_subexpr_elim = toggle;
        self
    }

    /// Toggle slice pushdown optimization.
    pub fn with_slice_pushdown(mut self, toggle: bool) -> Self {
        self.opt_state.slice_pushdown = toggle;
        self
    }

    /// Run nodes that are capably of doing so on the streaming engine.
    pub fn with_streaming(mut self, toggle: bool) -> Self {
        self.opt_state.streaming = toggle;
        self
    }

    /// Try to estimate the number of rows so that joins can determine which side to keep in memory.
    pub fn with_row_estimate(mut self, toggle: bool) -> Self {
        self.opt_state.row_estimate = toggle;
        self
    }

    /// Run every node eagerly. This turns off multi-node optimizations.
    pub fn _with_eager(mut self, toggle: bool) -> Self {
        self.opt_state.eager = toggle;
        self
    }

    /// Return a String describing the naive (un-optimized) logical plan.
    pub fn describe_plan(&self) -> PolarsResult<String> {
        Ok(self.clone().to_alp()?.describe())
    }

    /// Return a String describing the naive (un-optimized) logical plan in tree format.
    pub fn describe_plan_tree(&self) -> PolarsResult<String> {
        Ok(self.clone().to_alp()?.describe_tree_format())
    }

    // @NOTE: this is used because we want to set the `enable_fmt` flag of `optimize_with_scratch`
    // to `true` for describe.
    fn _describe_to_alp_optimized(mut self) -> PolarsResult<IRPlan> {
        let (mut lp_arena, mut expr_arena) = self.get_arenas();
        let node = self.optimize_with_scratch(&mut lp_arena, &mut expr_arena, &mut vec![], true)?;

        Ok(IRPlan::new(node, lp_arena, expr_arena))
    }

    /// Return a String describing the optimized logical plan.
    ///
    /// Returns `Err` if optimizing the logical plan fails.
    pub fn describe_optimized_plan(&self) -> PolarsResult<String> {
        Ok(self.clone()._describe_to_alp_optimized()?.describe())
    }

    /// Return a String describing the optimized logical plan in tree format.
    ///
    /// Returns `Err` if optimizing the logical plan fails.
    pub fn describe_optimized_plan_tree(&self) -> PolarsResult<String> {
        Ok(self
            .clone()
            ._describe_to_alp_optimized()?
            .describe_tree_format())
    }

    /// Return a String describing the logical plan.
    ///
    /// If `optimized` is `true`, explains the optimized plan. If `optimized` is `false,
    /// explains the naive, un-optimized plan.
    pub fn explain(&self, optimized: bool) -> PolarsResult<String> {
        if optimized {
            self.describe_optimized_plan()
        } else {
            self.describe_plan()
        }
    }

    /// Add a sort operation to the logical plan.
    ///
    /// Sorts the LazyFrame by the column name specified using the provided options.
    ///
    /// # Example
    ///
    /// Sort DataFrame by 'sepal_width' column:
    /// ```rust
    /// # use polars_core::prelude::*;
    /// # use polars_lazy::prelude::*;
    /// fn sort_by_a(df: DataFrame) -> LazyFrame {
    ///     df.lazy().sort(["sepal_width"], Default::default())
    /// }
    /// ```
    /// Sort by a single column with specific order:
    /// ```
    /// # use polars_core::prelude::*;
    /// # use polars_lazy::prelude::*;
    /// fn sort_with_specific_order(df: DataFrame, descending: bool) -> LazyFrame {
    ///     df.lazy().sort(
    ///         ["sepal_width"],
    ///         SortMultipleOptions::new()
    ///             .with_order_descending(descending)
    ///     )
    /// }
    /// ```
    /// Sort by multiple columns with specifying order for each column:
    /// ```
    /// # use polars_core::prelude::*;
    /// # use polars_lazy::prelude::*;
    /// fn sort_by_multiple_columns_with_specific_order(df: DataFrame) -> LazyFrame {
    ///     df.lazy().sort(
    ///         &["sepal_width", "sepal_length"],
    ///         SortMultipleOptions::new()
    ///             .with_order_descending_multi([false, true])
    ///     )
    /// }
    /// ```
    /// See [`SortMultipleOptions`] for more options.
    pub fn sort(self, by: impl IntoVec<SmartString>, sort_options: SortMultipleOptions) -> Self {
        let opt_state = self.get_opt_state();
        let lp = self
            .get_plan_builder()
            .sort(
                by.into_vec().into_iter().map(|x| col(&x)).collect(),
                sort_options,
            )
            .build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Add a sort operation to the logical plan.
    ///
    /// Sorts the LazyFrame by the provided list of expressions, which will be turned into
    /// concrete columns before sorting.
    ///
    /// See [`SortMultipleOptions`] for more options.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    ///
    /// /// Sort DataFrame by 'sepal_width' column
    /// fn example(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///         .sort_by_exprs(vec![col("sepal_width")], Default::default())
    /// }
    /// ```
    pub fn sort_by_exprs<E: AsRef<[Expr]>>(
        self,
        by_exprs: E,
        sort_options: SortMultipleOptions,
    ) -> Self {
        let by_exprs = by_exprs.as_ref().to_vec();
        if by_exprs.is_empty() {
            self
        } else {
            let opt_state = self.get_opt_state();
            let lp = self.get_plan_builder().sort(by_exprs, sort_options).build();
            Self::from_logical_plan(lp, opt_state)
        }
    }

    pub fn top_k<E: AsRef<[Expr]>>(
        self,
        k: IdxSize,
        by_exprs: E,
        sort_options: SortMultipleOptions,
    ) -> Self {
        // this will optimize to top-k
        self.sort_by_exprs(by_exprs, sort_options.with_order_reversed())
            .slice(0, k)
    }

    pub fn bottom_k<E: AsRef<[Expr]>>(
        self,
        k: IdxSize,
        by_exprs: E,
        sort_options: SortMultipleOptions,
    ) -> Self {
        // this will optimize to bottom-k
        self.sort_by_exprs(by_exprs, sort_options).slice(0, k)
    }

    /// Reverse the `DataFrame` from top to bottom.
    ///
    /// Row `i` becomes row `number_of_rows - i - 1`.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    ///
    /// fn example(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///         .reverse()
    /// }
    /// ```
    pub fn reverse(self) -> Self {
        self.select(vec![col("*").reverse()])
    }

    /// Rename columns in the DataFrame.
    ///
    /// `existing` and `new` are iterables of the same length containing the old and
    /// corresponding new column names. Renaming happens to all `existing` columns
    /// simultaneously, not iteratively. (In particular, all columns in `existing` must
    /// already exist in the `LazyFrame` when `rename` is called.)
    pub fn rename<I, J, T, S>(self, existing: I, new: J) -> Self
    where
        I: IntoIterator<Item = T>,
        J: IntoIterator<Item = S>,
        T: AsRef<str>,
        S: AsRef<str>,
    {
        let iter = existing.into_iter();
        let cap = iter.size_hint().0;
        let mut existing_vec: Vec<SmartString> = Vec::with_capacity(cap);
        let mut new_vec: Vec<SmartString> = Vec::with_capacity(cap);

        // TODO! should this error if `existing` and `new` have different lengths?
        // Currently, the longer of the two is truncated.
        for (existing, new) in iter.zip(new) {
            let existing = existing.as_ref();
            let new = new.as_ref();
            if new != existing {
                existing_vec.push(existing.into());
                new_vec.push(new.into());
            }
        }

        self.map_private(DslFunction::Rename {
            existing: existing_vec.into(),
            new: new_vec.into(),
        })
    }

    /// Removes columns from the DataFrame.
    /// Note that it's better to only select the columns you need
    /// and let the projection pushdown optimize away the unneeded columns.
    pub fn drop<I, T>(self, columns: I) -> Self
    where
        I: IntoIterator<Item = T>,
        T: AsRef<str>,
    {
        let to_drop = columns
            .into_iter()
            .map(|s| s.as_ref().to_string())
            .collect::<PlHashSet<_>>();

        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().drop(to_drop).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Shift the values by a given period and fill the parts that will be empty due to this operation
    /// with `Nones`.
    ///
    /// See the method on [Series](polars_core::series::SeriesTrait::shift) for more info on the `shift` operation.
    pub fn shift<E: Into<Expr>>(self, n: E) -> Self {
        self.select(vec![col("*").shift(n.into())])
    }

    /// Shift the values by a given period and fill the parts that will be empty due to this operation
    /// with the result of the `fill_value` expression.
    ///
    /// See the method on [Series](polars_core::series::SeriesTrait::shift) for more info on the `shift` operation.
    pub fn shift_and_fill<E: Into<Expr>, IE: Into<Expr>>(self, n: E, fill_value: IE) -> Self {
        self.select(vec![col("*").shift_and_fill(n.into(), fill_value.into())])
    }

    /// Fill None values in the DataFrame with an expression.
    pub fn fill_null<E: Into<Expr>>(self, fill_value: E) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().fill_null(fill_value.into()).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Fill NaN values in the DataFrame with an expression.
    pub fn fill_nan<E: Into<Expr>>(self, fill_value: E) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().fill_nan(fill_value.into()).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Caches the result into a new LazyFrame.
    ///
    /// This should be used to prevent computations running multiple times.
    pub fn cache(self) -> Self {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().cache().build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Cast named frame columns, resulting in a new LazyFrame with updated dtypes
    pub fn cast(self, dtypes: PlHashMap<&str, DataType>, strict: bool) -> Self {
        let cast_cols: Vec<Expr> = dtypes
            .into_iter()
            .map(|(name, dt)| {
                if strict {
                    col(name).strict_cast(dt)
                } else {
                    col(name).cast(dt)
                }
            })
            .collect();

        if cast_cols.is_empty() {
            self.clone()
        } else {
            self.with_columns(cast_cols)
        }
    }

    /// Cast all frame columns to the given dtype, resulting in a new LazyFrame
    pub fn cast_all(self, dtype: DataType, strict: bool) -> Self {
        self.with_columns(vec![if strict {
            col("*").strict_cast(dtype)
        } else {
            col("*").cast(dtype)
        }])
    }

    /// Fetch is like a collect operation, but it overwrites the number of rows read by every scan
    /// operation. This is a utility that helps debug a query on a smaller number of rows.
    ///
    /// Note that the fetch does not guarantee the final number of rows in the DataFrame.
    /// Filter, join operations and a lower number of rows available in the scanned file influence
    /// the final number of rows.
    pub fn fetch(self, n_rows: usize) -> PolarsResult<DataFrame> {
        FETCH_ROWS.with(|fetch_rows| fetch_rows.set(Some(n_rows)));
        let res = self.collect();
        FETCH_ROWS.with(|fetch_rows| fetch_rows.set(None));
        res
    }

    pub fn optimize(
        self,
        lp_arena: &mut Arena<IR>,
        expr_arena: &mut Arena<AExpr>,
    ) -> PolarsResult<Node> {
        self.optimize_with_scratch(lp_arena, expr_arena, &mut vec![], false)
    }

    pub fn to_alp_optimized(mut self) -> PolarsResult<IRPlan> {
        let (mut lp_arena, mut expr_arena) = self.get_arenas();
        let node =
            self.optimize_with_scratch(&mut lp_arena, &mut expr_arena, &mut vec![], false)?;

        Ok(IRPlan::new(node, lp_arena, expr_arena))
    }

    pub fn to_alp(mut self) -> PolarsResult<IRPlan> {
        let (mut lp_arena, mut expr_arena) = self.get_arenas();
        let node = to_alp(
            self.logical_plan,
            &mut expr_arena,
            &mut lp_arena,
            true,
            true,
        )?;
        let plan = IRPlan::new(node, lp_arena, expr_arena);
        Ok(plan)
    }

    pub(crate) fn optimize_with_scratch(
        self,
        lp_arena: &mut Arena<IR>,
        expr_arena: &mut Arena<AExpr>,
        scratch: &mut Vec<Node>,
        enable_fmt: bool,
    ) -> PolarsResult<Node> {
        #[allow(unused_mut)]
        let mut opt_state = self.opt_state;
        let streaming = self.opt_state.streaming;
        #[cfg(feature = "cse")]
        if streaming && self.opt_state.comm_subplan_elim {
            polars_warn!(
                "Cannot combine 'streaming' with 'comm_subplan_elim'. CSE will be turned off."
            );
            opt_state.comm_subplan_elim = false;
        }
        let lp_top = optimize(
            self.logical_plan,
            opt_state,
            lp_arena,
            expr_arena,
            scratch,
            Some(&|expr, expr_arena| {
                let phys_expr = create_physical_expr(
                    expr,
                    Context::Default,
                    expr_arena,
                    None,
                    &mut ExpressionConversionState::new(true, 0),
                )
                .ok()?;
                let io_expr = phys_expr_to_io_expr(phys_expr);
                Some(io_expr)
            }),
        )?;

        if streaming {
            #[cfg(feature = "streaming")]
            {
                insert_streaming_nodes(
                    lp_top,
                    lp_arena,
                    expr_arena,
                    scratch,
                    enable_fmt,
                    true,
                    opt_state.row_estimate,
                )?;
            }
            #[cfg(not(feature = "streaming"))]
            {
                _ = enable_fmt;
                panic!("activate feature 'streaming'")
            }
        }

        Ok(lp_top)
    }

    fn prepare_collect_post_opt<P>(
        mut self,
        check_sink: bool,
        post_opt: P,
    ) -> PolarsResult<(ExecutionState, Box<dyn Executor>, bool)>
    where
        P: Fn(Node, &mut Arena<IR>, &mut Arena<AExpr>) -> PolarsResult<()>,
    {
        let (mut lp_arena, mut expr_arena) = self.get_arenas();

        let mut scratch = vec![];
        let lp_top =
            self.optimize_with_scratch(&mut lp_arena, &mut expr_arena, &mut scratch, false)?;

        post_opt(lp_top, &mut lp_arena, &mut expr_arena)?;

        // sink should be replaced
        let no_file_sink = if check_sink {
            !matches!(lp_arena.get(lp_top), IR::Sink { .. })
        } else {
            true
        };
        let physical_plan = create_physical_plan(lp_top, &mut lp_arena, &mut expr_arena)?;

        let state = ExecutionState::new();
        Ok((state, physical_plan, no_file_sink))
    }

    // post_opt: A function that is called after optimization. This can be used to modify the IR jit.
    pub fn _collect_post_opt<P>(self, post_opt: P) -> PolarsResult<DataFrame>
    where
        P: Fn(Node, &mut Arena<IR>, &mut Arena<AExpr>) -> PolarsResult<()>,
    {
        let (mut state, mut physical_plan, _) = self.prepare_collect_post_opt(false, post_opt)?;
        physical_plan.execute(&mut state)
    }

    #[allow(unused_mut)]
    fn prepare_collect(
        self,
        check_sink: bool,
    ) -> PolarsResult<(ExecutionState, Box<dyn Executor>, bool)> {
        self.prepare_collect_post_opt(check_sink, |_, _, _| Ok(()))
    }

    /// Execute all the lazy operations and collect them into a [`DataFrame`].
    ///
    /// The query is optimized prior to execution.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    ///
    /// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
    ///     df.lazy()
    ///       .group_by([col("foo")])
    ///       .agg([col("bar").sum(), col("ham").mean().alias("avg_ham")])
    ///       .collect()
    /// }
    /// ```
    pub fn collect(self) -> PolarsResult<DataFrame> {
        self._collect_post_opt(|_, _, _| Ok(()))
    }

    /// Profile a LazyFrame.
    ///
    /// This will run the query and return a tuple
    /// containing the materialized DataFrame and a DataFrame that contains profiling information
    /// of each node that is executed.
    ///
    /// The units of the timings are microseconds.
    pub fn profile(self) -> PolarsResult<(DataFrame, DataFrame)> {
        let (mut state, mut physical_plan, _) = self.prepare_collect(false)?;
        state.time_nodes();
        let out = physical_plan.execute(&mut state)?;
        let timer_df = state.finish_timer()?;
        Ok((out, timer_df))
    }

    /// Stream a query result into a parquet file. This is useful if the final result doesn't fit
    /// into memory. This methods will return an error if the query cannot be completely done in a
    /// streaming fashion.
    #[cfg(feature = "parquet")]
    pub fn sink_parquet(self, path: PathBuf, options: ParquetWriteOptions) -> PolarsResult<()> {
        self.sink(
            SinkType::File {
                path: Arc::new(path),
                file_type: FileType::Parquet(options),
            },
            "collect().write_parquet()",
        )
    }

    /// Stream a query result into a parquet file on an ObjectStore-compatible cloud service. This is useful if the final result doesn't fit
    /// into memory, and where you do not want to write to a local file but to a location in the cloud.
    /// This method will return an error if the query cannot be completely done in a
    /// streaming fashion.
    #[cfg(all(feature = "cloud_write", feature = "parquet"))]
    pub fn sink_parquet_cloud(
        self,
        uri: String,
        cloud_options: Option<polars_io::cloud::CloudOptions>,
        parquet_options: ParquetWriteOptions,
    ) -> PolarsResult<()> {
        self.sink(
            SinkType::Cloud {
                uri: Arc::new(uri),
                cloud_options,
                file_type: FileType::Parquet(parquet_options),
            },
            "collect().write_parquet()",
        )
    }

    /// Stream a query result into an ipc/arrow file. This is useful if the final result doesn't fit
    /// into memory. This methods will return an error if the query cannot be completely done in a
    /// streaming fashion.
    #[cfg(feature = "ipc")]
    pub fn sink_ipc(self, path: PathBuf, options: IpcWriterOptions) -> PolarsResult<()> {
        self.sink(
            SinkType::File {
                path: Arc::new(path),
                file_type: FileType::Ipc(options),
            },
            "collect().write_ipc()",
        )
    }

    /// Stream a query result into an ipc/arrow file on an ObjectStore-compatible cloud service.
    /// This is useful if the final result doesn't fit
    /// into memory, and where you do not want to write to a local file but to a location in the cloud.
    /// This method will return an error if the query cannot be completely done in a
    /// streaming fashion.
    #[cfg(all(feature = "cloud_write", feature = "ipc"))]
    pub fn sink_ipc_cloud(
        mut self,
        uri: String,
        cloud_options: Option<polars_io::cloud::CloudOptions>,
        ipc_options: IpcWriterOptions,
    ) -> PolarsResult<()> {
        self.opt_state.streaming = true;
        self.logical_plan = DslPlan::Sink {
            input: Arc::new(self.logical_plan),
            payload: SinkType::Cloud {
                uri: Arc::new(uri),
                cloud_options,
                file_type: FileType::Ipc(ipc_options),
            },
        };
        let (mut state, mut physical_plan, is_streaming) = self.prepare_collect(true)?;
        polars_ensure!(
            is_streaming,
            ComputeError: "cannot run the whole query in a streaming order; \
                           use `collect().write_ipc()` instead"
        );
        let _ = physical_plan.execute(&mut state)?;
        Ok(())
    }

    /// Stream a query result into an csv file. This is useful if the final result doesn't fit
    /// into memory. This methods will return an error if the query cannot be completely done in a
    /// streaming fashion.
    #[cfg(feature = "csv")]
    pub fn sink_csv(self, path: PathBuf, options: CsvWriterOptions) -> PolarsResult<()> {
        self.sink(
            SinkType::File {
                path: Arc::new(path),
                file_type: FileType::Csv(options),
            },
            "collect().write_csv()",
        )
    }

    /// Stream a query result into a json file. This is useful if the final result doesn't fit
    /// into memory. This methods will return an error if the query cannot be completely done in a
    /// streaming fashion.
    #[cfg(feature = "json")]
    pub fn sink_json(self, path: PathBuf, options: JsonWriterOptions) -> PolarsResult<()> {
        self.sink(
            SinkType::File {
                path: Arc::new(path),
                file_type: FileType::Json(options),
            },
            "collect().write_ndjson()` or `collect().write_json()",
        )
    }

    #[cfg(any(
        feature = "ipc",
        feature = "parquet",
        feature = "cloud_write",
        feature = "csv",
        feature = "json",
    ))]
    fn sink(mut self, payload: SinkType, msg_alternative: &str) -> Result<(), PolarsError> {
        self.opt_state.streaming = true;
        self.logical_plan = DslPlan::Sink {
            input: Arc::new(self.logical_plan),
            payload,
        };
        let (mut state, mut physical_plan, is_streaming) = self.prepare_collect(true)?;
        polars_ensure!(
            is_streaming,
            ComputeError: format!("cannot run the whole query in a streaming order; \
            use `{msg_alternative}` instead", msg_alternative=msg_alternative)
        );
        let _ = physical_plan.execute(&mut state)?;
        Ok(())
    }

    /// Filter by some predicate expression.
    ///
    /// The expression must yield boolean values.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    ///
    /// fn example(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///         .filter(col("sepal_width").is_not_null())
    ///         .select(&[col("sepal_width"), col("sepal_length")])
    /// }
    /// ```
    pub fn filter(self, predicate: Expr) -> Self {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().filter(predicate).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Select (and optionally rename, with [`alias`](crate::dsl::Expr::alias)) columns from the query.
    ///
    /// Columns can be selected with [`col`];
    /// If you want to select all columns use `col("*")`.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    ///
    /// /// This function selects column "foo" and column "bar".
    /// /// Column "bar" is renamed to "ham".
    /// fn example(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///         .select(&[col("foo"),
    ///                   col("bar").alias("ham")])
    /// }
    ///
    /// /// This function selects all columns except "foo"
    /// fn exclude_a_column(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///         .select(&[col("*").exclude(["foo"])])
    /// }
    /// ```
    pub fn select<E: AsRef<[Expr]>>(self, exprs: E) -> Self {
        let exprs = exprs.as_ref().to_vec();
        self.select_impl(
            exprs,
            ProjectionOptions {
                run_parallel: true,
                duplicate_check: true,
                should_broadcast: true,
            },
        )
    }

    pub fn select_seq<E: AsRef<[Expr]>>(self, exprs: E) -> Self {
        let exprs = exprs.as_ref().to_vec();
        self.select_impl(
            exprs,
            ProjectionOptions {
                run_parallel: false,
                duplicate_check: true,
                should_broadcast: true,
            },
        )
    }

    fn select_impl(self, exprs: Vec<Expr>, options: ProjectionOptions) -> Self {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().project(exprs, options).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Performs a "group-by" on a `LazyFrame`, producing a [`LazyGroupBy`], which can subsequently be aggregated.
    ///
    /// Takes a list of expressions to group on.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// use arrow::legacy::prelude::QuantileInterpolOptions;
    ///
    /// fn example(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///        .group_by([col("date")])
    ///        .agg([
    ///            col("rain").min().alias("min_rain"),
    ///            col("rain").sum().alias("sum_rain"),
    ///            col("rain").quantile(lit(0.5), QuantileInterpolOptions::Nearest).alias("median_rain"),
    ///        ])
    /// }
    /// ```
    pub fn group_by<E: AsRef<[IE]>, IE: Into<Expr> + Clone>(self, by: E) -> LazyGroupBy {
        let keys = by
            .as_ref()
            .iter()
            .map(|e| e.clone().into())
            .collect::<Vec<_>>();
        let opt_state = self.get_opt_state();

        #[cfg(feature = "dynamic_group_by")]
        {
            LazyGroupBy {
                logical_plan: self.logical_plan,
                opt_state,
                keys,
                maintain_order: false,
                dynamic_options: None,
                rolling_options: None,
            }
        }

        #[cfg(not(feature = "dynamic_group_by"))]
        {
            LazyGroupBy {
                logical_plan: self.logical_plan,
                opt_state,
                keys,
                maintain_order: false,
            }
        }
    }

    /// Create rolling groups based on a time column.
    ///
    /// Also works for index values of type UInt32, UInt64, Int32, or Int64.
    ///
    /// Different from a [`group_by_dynamic`][`Self::group_by_dynamic`], the windows are now determined by the
    /// individual values and are not of constant intervals. For constant intervals use
    /// *group_by_dynamic*
    #[cfg(feature = "dynamic_group_by")]
    pub fn rolling<E: AsRef<[Expr]>>(
        mut self,
        index_column: Expr,
        group_by: E,
        mut options: RollingGroupOptions,
    ) -> LazyGroupBy {
        if let Expr::Column(name) = index_column {
            options.index_column = name.as_ref().into();
        } else {
            let output_field = index_column
                .to_field(&self.schema().unwrap(), Context::Default)
                .unwrap();
            return self.with_column(index_column).rolling(
                Expr::Column(Arc::from(output_field.name().as_str())),
                group_by,
                options,
            );
        }
        let opt_state = self.get_opt_state();
        LazyGroupBy {
            logical_plan: self.logical_plan,
            opt_state,
            keys: group_by.as_ref().to_vec(),
            maintain_order: true,
            dynamic_options: None,
            rolling_options: Some(options),
        }
    }

    /// Group based on a time value (or index value of type Int32, Int64).
    ///
    /// Time windows are calculated and rows are assigned to windows. Different from a
    /// normal group_by is that a row can be member of multiple groups. The time/index
    /// window could be seen as a rolling window, with a window size determined by
    /// dates/times/values instead of slots in the DataFrame.
    ///
    /// A window is defined by:
    ///
    /// - every: interval of the window
    /// - period: length of the window
    /// - offset: offset of the window
    ///
    /// The `group_by` argument should be empty `[]` if you don't want to combine this
    /// with a ordinary group_by on these keys.
    #[cfg(feature = "dynamic_group_by")]
    pub fn group_by_dynamic<E: AsRef<[Expr]>>(
        mut self,
        index_column: Expr,
        group_by: E,
        mut options: DynamicGroupOptions,
    ) -> LazyGroupBy {
        if let Expr::Column(name) = index_column {
            options.index_column = name.as_ref().into();
        } else {
            let output_field = index_column
                .to_field(&self.schema().unwrap(), Context::Default)
                .unwrap();
            return self.with_column(index_column).group_by_dynamic(
                Expr::Column(Arc::from(output_field.name().as_str())),
                group_by,
                options,
            );
        }
        let opt_state = self.get_opt_state();
        LazyGroupBy {
            logical_plan: self.logical_plan,
            opt_state,
            keys: group_by.as_ref().to_vec(),
            maintain_order: true,
            dynamic_options: Some(options),
            rolling_options: None,
        }
    }

    /// Similar to [`group_by`][`Self::group_by`], but order of the DataFrame is maintained.
    pub fn group_by_stable<E: AsRef<[IE]>, IE: Into<Expr> + Clone>(self, by: E) -> LazyGroupBy {
        let keys = by
            .as_ref()
            .iter()
            .map(|e| e.clone().into())
            .collect::<Vec<_>>();
        let opt_state = self.get_opt_state();

        #[cfg(feature = "dynamic_group_by")]
        {
            LazyGroupBy {
                logical_plan: self.logical_plan,
                opt_state,
                keys,
                maintain_order: true,
                dynamic_options: None,
                rolling_options: None,
            }
        }

        #[cfg(not(feature = "dynamic_group_by"))]
        {
            LazyGroupBy {
                logical_plan: self.logical_plan,
                opt_state,
                keys,
                maintain_order: true,
            }
        }
    }

    /// Left anti join this query with another lazy query.
    ///
    /// Matches on the values of the expressions `left_on` and `right_on`. For more
    /// flexible join logic, see [`join`](LazyFrame::join) or
    /// [`join_builder`](LazyFrame::join_builder).
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn anti_join_dataframes(ldf: LazyFrame, other: LazyFrame) -> LazyFrame {
    ///         ldf
    ///         .anti_join(other, col("foo"), col("bar").cast(DataType::String))
    /// }
    /// ```
    #[cfg(feature = "semi_anti_join")]
    pub fn anti_join<E: Into<Expr>>(self, other: LazyFrame, left_on: E, right_on: E) -> LazyFrame {
        self.join(
            other,
            [left_on.into()],
            [right_on.into()],
            JoinArgs::new(JoinType::Anti),
        )
    }

    /// Creates the Cartesian product from both frames, preserving the order of the left keys.
    #[cfg(feature = "cross_join")]
    pub fn cross_join(self, other: LazyFrame, suffix: Option<String>) -> LazyFrame {
        self.join(
            other,
            vec![],
            vec![],
            JoinArgs::new(JoinType::Cross).with_suffix(suffix),
        )
    }

    /// Left outer join this query with another lazy query.
    ///
    /// Matches on the values of the expressions `left_on` and `right_on`. For more
    /// flexible join logic, see [`join`](LazyFrame::join) or
    /// [`join_builder`](LazyFrame::join_builder).
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn left_join_dataframes(ldf: LazyFrame, other: LazyFrame) -> LazyFrame {
    ///         ldf
    ///         .left_join(other, col("foo"), col("bar"))
    /// }
    /// ```
    pub fn left_join<E: Into<Expr>>(self, other: LazyFrame, left_on: E, right_on: E) -> LazyFrame {
        self.join(
            other,
            [left_on.into()],
            [right_on.into()],
            JoinArgs::new(JoinType::Left),
        )
    }

    /// Inner join this query with another lazy query.
    ///
    /// Matches on the values of the expressions `left_on` and `right_on`. For more
    /// flexible join logic, see [`join`](LazyFrame::join) or
    /// [`join_builder`](LazyFrame::join_builder).
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn inner_join_dataframes(ldf: LazyFrame, other: LazyFrame) -> LazyFrame {
    ///         ldf
    ///         .inner_join(other, col("foo"), col("bar").cast(DataType::String))
    /// }
    /// ```
    pub fn inner_join<E: Into<Expr>>(self, other: LazyFrame, left_on: E, right_on: E) -> LazyFrame {
        self.join(
            other,
            [left_on.into()],
            [right_on.into()],
            JoinArgs::new(JoinType::Inner),
        )
    }

    /// Full outer join this query with another lazy query.
    ///
    /// Matches on the values of the expressions `left_on` and `right_on`. For more
    /// flexible join logic, see [`join`](LazyFrame::join) or
    /// [`join_builder`](LazyFrame::join_builder).
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn full_join_dataframes(ldf: LazyFrame, other: LazyFrame) -> LazyFrame {
    ///         ldf
    ///         .full_join(other, col("foo"), col("bar"))
    /// }
    /// ```
    pub fn full_join<E: Into<Expr>>(self, other: LazyFrame, left_on: E, right_on: E) -> LazyFrame {
        self.join(
            other,
            [left_on.into()],
            [right_on.into()],
            JoinArgs::new(JoinType::Full),
        )
    }

    /// Left semi join this query with another lazy query.
    ///
    /// Matches on the values of the expressions `left_on` and `right_on`. For more
    /// flexible join logic, see [`join`](LazyFrame::join) or
    /// [`join_builder`](LazyFrame::join_builder).
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn semi_join_dataframes(ldf: LazyFrame, other: LazyFrame) -> LazyFrame {
    ///         ldf
    ///         .semi_join(other, col("foo"), col("bar").cast(DataType::String))
    /// }
    /// ```
    #[cfg(feature = "semi_anti_join")]
    pub fn semi_join<E: Into<Expr>>(self, other: LazyFrame, left_on: E, right_on: E) -> LazyFrame {
        self.join(
            other,
            [left_on.into()],
            [right_on.into()],
            JoinArgs::new(JoinType::Semi),
        )
    }

    /// Generic function to join two LazyFrames.
    ///
    /// `join` can join on multiple columns, given as two list of expressions, and with a
    /// [`JoinType`] specified by `how`. Non-joined column names in the right DataFrame
    /// that already exist in this DataFrame are suffixed with `"_right"`. For control
    /// over how columns are renamed and parallelization options, use
    /// [`join_builder`](LazyFrame::join_builder).
    ///
    /// Any provided `args.slice` parameter is not considered, but set by the internal optimizer.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    ///
    /// fn example(ldf: LazyFrame, other: LazyFrame) -> LazyFrame {
    ///         ldf
    ///         .join(other, [col("foo"), col("bar")], [col("foo"), col("bar")], JoinArgs::new(JoinType::Inner))
    /// }
    /// ```
    pub fn join<E: AsRef<[Expr]>>(
        mut self,
        other: LazyFrame,
        left_on: E,
        right_on: E,
        args: JoinArgs,
    ) -> LazyFrame {
        // if any of the nodes reads from files we must activate this this plan as well.
        self.opt_state.file_caching |= other.opt_state.file_caching;

        let left_on = left_on.as_ref().to_vec();
        let right_on = right_on.as_ref().to_vec();

        let mut builder = self
            .join_builder()
            .with(other)
            .left_on(left_on)
            .right_on(right_on)
            .how(args.how)
            .validate(args.validation)
            .coalesce(args.coalesce)
            .join_nulls(args.join_nulls);

        if let Some(suffix) = args.suffix {
            builder = builder.suffix(suffix);
        }
        // Note: args.slice is set by the optimizer
        builder.finish()
    }

    /// Consume `self` and return a [`JoinBuilder`] to customize a join on this LazyFrame.
    ///
    /// After the `JoinBuilder` has been created and set up, calling
    /// [`finish()`](JoinBuilder::finish) on it will give back the `LazyFrame`
    /// representing the `join` operation.
    pub fn join_builder(self) -> JoinBuilder {
        JoinBuilder::new(self)
    }

    /// Add or replace a column, given as an expression, to a DataFrame.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn add_column(df: DataFrame) -> LazyFrame {
    ///     df.lazy()
    ///         .with_column(
    ///             when(col("sepal_length").lt(lit(5.0)))
    ///             .then(lit(10))
    ///             .otherwise(lit(1))
    ///             .alias("new_column_name"),
    ///         )
    /// }
    /// ```
    pub fn with_column(self, expr: Expr) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self
            .get_plan_builder()
            .with_columns(
                vec![expr],
                ProjectionOptions {
                    run_parallel: false,
                    duplicate_check: true,
                    should_broadcast: true,
                },
            )
            .build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Add or replace multiple columns, given as expressions, to a DataFrame.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// fn add_columns(df: DataFrame) -> LazyFrame {
    ///     df.lazy()
    ///         .with_columns(
    ///             vec![lit(10).alias("foo"), lit(100).alias("bar")]
    ///          )
    /// }
    /// ```
    pub fn with_columns<E: AsRef<[Expr]>>(self, exprs: E) -> LazyFrame {
        let exprs = exprs.as_ref().to_vec();
        self.with_columns_impl(
            exprs,
            ProjectionOptions {
                run_parallel: true,
                duplicate_check: true,
                should_broadcast: true,
            },
        )
    }

    /// Add or replace multiple columns to a DataFrame, but evaluate them sequentially.
    pub fn with_columns_seq<E: AsRef<[Expr]>>(self, exprs: E) -> LazyFrame {
        let exprs = exprs.as_ref().to_vec();
        self.with_columns_impl(
            exprs,
            ProjectionOptions {
                run_parallel: false,
                duplicate_check: true,
                should_broadcast: true,
            },
        )
    }

    fn with_columns_impl(self, exprs: Vec<Expr>, options: ProjectionOptions) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().with_columns(exprs, options).build();
        Self::from_logical_plan(lp, opt_state)
    }

    pub fn with_context<C: AsRef<[LazyFrame]>>(self, contexts: C) -> LazyFrame {
        let contexts = contexts
            .as_ref()
            .iter()
            .map(|lf| lf.logical_plan.clone())
            .collect();
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().with_context(contexts).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Aggregate all the columns as their maximum values.
    ///
    /// Aggregated columns will have the same names as the original columns.
    pub fn max(self) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Max))
    }

    /// Aggregate all the columns as their minimum values.
    ///
    /// Aggregated columns will have the same names as the original columns.
    pub fn min(self) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Min))
    }

    /// Aggregate all the columns as their sum values.
    ///
    /// Aggregated columns will have the same names as the original columns.
    ///
    /// - Boolean columns will sum to a `u32` containing the number of `true`s.
    /// - For integer columns, the ordinary checks for overflow are performed:
    ///   if running in `debug` mode, overflows will panic, whereas in `release` mode overflows will
    ///   silently wrap.
    /// - String columns will sum to None.
    pub fn sum(self) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Sum))
    }

    /// Aggregate all the columns as their mean values.
    ///
    /// - Boolean and integer columns are converted to `f64` before computing the mean.
    /// - String columns will have a mean of None.
    pub fn mean(self) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Mean))
    }

    /// Aggregate all the columns as their median values.
    ///
    /// - Boolean and integer results are converted to `f64`. However, they are still
    ///   susceptible to overflow before this conversion occurs.
    /// - String columns will sum to None.
    pub fn median(self) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Median))
    }

    /// Aggregate all the columns as their quantile values.
    pub fn quantile(self, quantile: Expr, interpol: QuantileInterpolOptions) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Quantile {
            quantile,
            interpol,
        }))
    }

    /// Aggregate all the columns as their standard deviation values.
    ///
    /// `ddof` is the "Delta Degrees of Freedom"; `N - ddof` will be the denominator when
    /// computing the variance, where `N` is the number of rows.
    /// > In standard statistical practice, `ddof=1` provides an unbiased estimator of the
    /// > variance of a hypothetical infinite population. `ddof=0` provides a maximum
    /// > likelihood estimate of the variance for normally distributed variables. The
    /// > standard deviation computed in this function is the square root of the estimated
    /// > variance, so even with `ddof=1`, it will not be an unbiased estimate of the
    /// > standard deviation per se.
    ///
    /// Source: [Numpy](https://numpy.org/doc/stable/reference/generated/numpy.std.html#)
    pub fn std(self, ddof: u8) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Std { ddof }))
    }

    /// Aggregate all the columns as their variance values.
    ///
    /// `ddof` is the "Delta Degrees of Freedom"; `N - ddof` will be the denominator when
    /// computing the variance, where `N` is the number of rows.
    /// > In standard statistical practice, `ddof=1` provides an unbiased estimator of the
    /// > variance of a hypothetical infinite population. `ddof=0` provides a maximum
    /// > likelihood estimate of the variance for normally distributed variables.
    ///
    /// Source: [Numpy](https://numpy.org/doc/stable/reference/generated/numpy.var.html#)
    pub fn var(self, ddof: u8) -> Self {
        self.map_private(DslFunction::Stats(StatsFunction::Var { ddof }))
    }

    /// Apply explode operation. [See eager explode](polars_core::frame::DataFrame::explode).
    pub fn explode<E: AsRef<[IE]>, IE: Into<Expr> + Clone>(self, columns: E) -> LazyFrame {
        let columns = columns
            .as_ref()
            .iter()
            .map(|e| e.clone().into())
            .collect::<Vec<_>>();
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().explode(columns).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Aggregate all the columns as the sum of their null value count.
    pub fn null_count(self) -> LazyFrame {
        self.select(vec![col("*").null_count()])
    }

    /// Drop non-unique rows and maintain the order of kept rows.
    ///
    /// `subset` is an optional `Vec` of column names to consider for uniqueness; if
    /// `None`, all columns are considered.
    pub fn unique_stable(
        self,
        subset: Option<Vec<String>>,
        keep_strategy: UniqueKeepStrategy,
    ) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let options = DistinctOptions {
            subset: subset.map(Arc::new),
            maintain_order: true,
            keep_strategy,
            ..Default::default()
        };
        let lp = self.get_plan_builder().distinct(options).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Drop non-unique rows without maintaining the order of kept rows.
    ///
    /// The order of the kept rows may change; to maintain the original row order, use
    /// [`unique_stable`](LazyFrame::unique_stable).
    ///
    /// `subset` is an optional `Vec` of column names to consider for uniqueness; if None,
    /// all columns are considered.
    pub fn unique(
        self,
        subset: Option<Vec<String>>,
        keep_strategy: UniqueKeepStrategy,
    ) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let options = DistinctOptions {
            subset: subset.map(Arc::new),
            maintain_order: false,
            keep_strategy,
            ..Default::default()
        };
        let lp = self.get_plan_builder().distinct(options).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Drop rows containing None.
    ///
    /// `subset` is an optional `Vec` of column names to consider for nulls; if None, all
    /// columns are considered.
    pub fn drop_nulls(self, subset: Option<Vec<Expr>>) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().drop_nulls(subset).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Slice the DataFrame using an offset (starting row) and a length.
    ///
    /// If `offset` is negative, it is counted from the end of the DataFrame. For
    /// instance, `lf.slice(-5, 3)` gets three rows, starting at the row fifth from the
    /// end.
    ///
    /// If `offset` and `len` are such that the slice extends beyond the end of the
    /// DataFrame, the portion between `offset` and the end will be returned. In this
    /// case, the number of rows in the returned DataFrame will be less than `len`.
    pub fn slice(self, offset: i64, len: IdxSize) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().slice(offset, len).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Get the first row.
    ///
    /// Equivalent to `self.slice(0, 1)`.
    pub fn first(self) -> LazyFrame {
        self.slice(0, 1)
    }

    /// Get the last row.
    ///
    /// Equivalent to `self.slice(-1, 1)`.
    pub fn last(self) -> LazyFrame {
        self.slice(-1, 1)
    }

    /// Get the last `n` rows.
    ///
    /// Equivalent to `self.slice(-(n as i64), n)`.
    pub fn tail(self, n: IdxSize) -> LazyFrame {
        let neg_tail = -(n as i64);
        self.slice(neg_tail, n)
    }

    /// Melt the DataFrame from wide to long format.
    ///
    /// See [`MeltArgs`] for information on how to melt a DataFrame.
    pub fn melt(self, args: MeltArgs) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().melt(args).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Limit the DataFrame to the first `n` rows.
    ///
    /// Note if you don't want the rows to be scanned, use [`fetch`](LazyFrame::fetch).
    pub fn limit(self, n: IdxSize) -> LazyFrame {
        self.slice(0, n)
    }

    /// Apply a function/closure once the logical plan get executed.
    ///
    /// The function has access to the whole materialized DataFrame at the time it is
    /// called.
    ///
    /// To apply specific functions to specific columns, use [`Expr::map`] in conjunction
    /// with `LazyFrame::with_column` or `with_columns`.
    ///
    /// ## Warning
    /// This can blow up in your face if the schema is changed due to the operation. The
    /// optimizer relies on a correct schema.
    ///
    /// You can toggle certain optimizations off.
    pub fn map<F>(
        self,
        function: F,
        optimizations: AllowedOptimizations,
        schema: Option<Arc<dyn UdfSchema>>,
        name: Option<&'static str>,
    ) -> LazyFrame
    where
        F: 'static + Fn(DataFrame) -> PolarsResult<DataFrame> + Send + Sync,
    {
        let opt_state = self.get_opt_state();
        let lp = self
            .get_plan_builder()
            .map(
                function,
                optimizations,
                schema,
                name.unwrap_or("ANONYMOUS UDF"),
            )
            .build();
        Self::from_logical_plan(lp, opt_state)
    }

    #[cfg(feature = "python")]
    pub fn map_python(
        self,
        function: polars_plan::prelude::python_udf::PythonFunction,
        optimizations: AllowedOptimizations,
        schema: Option<SchemaRef>,
        validate_output: bool,
    ) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self
            .get_plan_builder()
            .map_python(function, optimizations, schema, validate_output)
            .build();
        Self::from_logical_plan(lp, opt_state)
    }

    pub(crate) fn map_private(self, function: DslFunction) -> LazyFrame {
        let opt_state = self.get_opt_state();
        let lp = self.get_plan_builder().map_private(function).build();
        Self::from_logical_plan(lp, opt_state)
    }

    /// Add a new column at index 0 that counts the rows.
    ///
    /// `name` is the name of the new column. `offset` is where to start counting from; if
    /// `None`, it is set to `0`.
    ///
    /// # Warning
    /// This can have a negative effect on query performance. This may for instance block
    /// predicate pushdown optimization.
    pub fn with_row_index(mut self, name: &str, offset: Option<IdxSize>) -> LazyFrame {
        let add_row_index_in_map = match &mut self.logical_plan {
            DslPlan::Scan {
                file_options: options,
                scan_type,
                ..
            } if !matches!(scan_type, FileScan::Anonymous { .. }) => {
                options.row_index = Some(RowIndex {
                    name: Arc::from(name),
                    offset: offset.unwrap_or(0),
                });
                false
            },
            _ => true,
        };

        if add_row_index_in_map {
            self.map_private(DslFunction::RowIndex {
                name: Arc::from(name),
                offset,
            })
        } else {
            self
        }
    }

    /// Return the number of non-null elements for each column.
    pub fn count(self) -> LazyFrame {
        self.select(vec![col("*").count()])
    }

    /// Unnest the given `Struct` columns: the fields of the `Struct` type will be
    /// inserted as columns.
    #[cfg(feature = "dtype-struct")]
    pub fn unnest<I: IntoIterator<Item = S>, S: AsRef<str>>(self, cols: I) -> Self {
        self.map_private(DslFunction::FunctionNode(FunctionNode::Unnest {
            columns: cols.into_iter().map(|s| Arc::from(s.as_ref())).collect(),
        }))
    }

    #[cfg(feature = "merge_sorted")]
    pub fn merge_sorted(self, other: LazyFrame, key: &str) -> PolarsResult<LazyFrame> {
        // The two DataFrames are temporary concatenated
        // this indicates until which chunk the data is from the left df
        // this trick allows us to reuse the `Union` architecture to get map over
        // two DataFrames
        let left = self.map_private(DslFunction::FunctionNode(FunctionNode::Rechunk));
        let q = concat(
            &[left, other],
            UnionArgs {
                rechunk: false,
                parallel: true,
                ..Default::default()
            },
        )?;
        Ok(
            q.map_private(DslFunction::FunctionNode(FunctionNode::MergeSorted {
                column: Arc::from(key),
            })),
        )
    }
}

/// Utility struct for lazy group_by operation.
#[derive(Clone)]
pub struct LazyGroupBy {
    pub logical_plan: DslPlan,
    opt_state: OptState,
    keys: Vec<Expr>,
    maintain_order: bool,
    #[cfg(feature = "dynamic_group_by")]
    dynamic_options: Option<DynamicGroupOptions>,
    #[cfg(feature = "dynamic_group_by")]
    rolling_options: Option<RollingGroupOptions>,
}

impl From<LazyGroupBy> for LazyFrame {
    fn from(lgb: LazyGroupBy) -> Self {
        Self {
            logical_plan: lgb.logical_plan,
            opt_state: lgb.opt_state,
            cached_arena: Default::default(),
        }
    }
}

impl LazyGroupBy {
    /// Group by and aggregate.
    ///
    /// Select a column with [col] and choose an aggregation.
    /// If you want to aggregate all columns use `col("*")`.
    ///
    /// # Example
    ///
    /// ```rust
    /// use polars_core::prelude::*;
    /// use polars_lazy::prelude::*;
    /// use arrow::legacy::prelude::QuantileInterpolOptions;
    ///
    /// fn example(df: DataFrame) -> LazyFrame {
    ///       df.lazy()
    ///        .group_by_stable([col("date")])
    ///        .agg([
    ///            col("rain").min().alias("min_rain"),
    ///            col("rain").sum().alias("sum_rain"),
    ///            col("rain").quantile(lit(0.5), QuantileInterpolOptions::Nearest).alias("median_rain"),
    ///        ])
    /// }
    /// ```
    pub fn agg<E: AsRef<[Expr]>>(self, aggs: E) -> LazyFrame {
        #[cfg(feature = "dynamic_group_by")]
        let lp = DslBuilder::from(self.logical_plan)
            .group_by(
                self.keys,
                aggs,
                None,
                self.maintain_order,
                self.dynamic_options,
                self.rolling_options,
            )
            .build();

        #[cfg(not(feature = "dynamic_group_by"))]
        let lp = DslBuilder::from(self.logical_plan)
            .group_by(self.keys, aggs, None, self.maintain_order)
            .build();
        LazyFrame::from_logical_plan(lp, self.opt_state)
    }

    /// Return first n rows of each group
    pub fn head(self, n: Option<usize>) -> LazyFrame {
        let keys = self
            .keys
            .iter()
            .filter_map(|expr| expr_output_name(expr).ok())
            .collect::<Vec<_>>();

        self.agg([col("*").exclude(&keys).head(n)])
            .explode([col("*").exclude(&keys)])
    }

    /// Return last n rows of each group
    pub fn tail(self, n: Option<usize>) -> LazyFrame {
        let keys = self
            .keys
            .iter()
            .filter_map(|expr| expr_output_name(expr).ok())
            .collect::<Vec<_>>();

        self.agg([col("*").exclude(&keys).tail(n)])
            .explode([col("*").exclude(&keys)])
    }

    /// Apply a function over the groups as a new DataFrame.
    ///
    /// **It is not recommended that you use this as materializing the DataFrame is very
    /// expensive.**
    pub fn apply<F>(self, f: F, schema: SchemaRef) -> LazyFrame
    where
        F: 'static + Fn(DataFrame) -> PolarsResult<DataFrame> + Send + Sync,
    {
        #[cfg(feature = "dynamic_group_by")]
        let options = GroupbyOptions {
            dynamic: self.dynamic_options,
            rolling: self.rolling_options,
            slice: None,
        };

        #[cfg(not(feature = "dynamic_group_by"))]
        let options = GroupbyOptions { slice: None };

        let lp = DslPlan::GroupBy {
            input: Arc::new(self.logical_plan),
            keys: self.keys,
            aggs: vec![],
            apply: Some((Arc::new(f), schema)),
            maintain_order: self.maintain_order,
            options: Arc::new(options),
        };
        LazyFrame::from_logical_plan(lp, self.opt_state)
    }
}

#[must_use]
pub struct JoinBuilder {
    lf: LazyFrame,
    how: JoinType,
    other: Option<LazyFrame>,
    left_on: Vec<Expr>,
    right_on: Vec<Expr>,
    allow_parallel: bool,
    force_parallel: bool,
    suffix: Option<String>,
    validation: JoinValidation,
    coalesce: JoinCoalesce,
    join_nulls: bool,
}
impl JoinBuilder {
    /// Create the `JoinBuilder` with the provided `LazyFrame` as the left table.
    pub fn new(lf: LazyFrame) -> Self {
        Self {
            lf,
            other: None,
            how: JoinType::Inner,
            left_on: vec![],
            right_on: vec![],
            allow_parallel: true,
            force_parallel: false,
            join_nulls: false,
            suffix: None,
            validation: Default::default(),
            coalesce: Default::default(),
        }
    }

    /// The right table in the join.
    pub fn with(mut self, other: LazyFrame) -> Self {
        self.other = Some(other);
        self
    }

    /// Select the join type.
    pub fn how(mut self, how: JoinType) -> Self {
        self.how = how;
        self
    }

    pub fn validate(mut self, validation: JoinValidation) -> Self {
        self.validation = validation;
        self
    }

    /// The expressions you want to join both tables on.
    ///
    /// The passed expressions must be valid in both `LazyFrame`s in the join.
    pub fn on<E: AsRef<[Expr]>>(mut self, on: E) -> Self {
        let on = on.as_ref().to_vec();
        self.left_on.clone_from(&on);
        self.right_on = on;
        self
    }

    /// The expressions you want to join the left table on.
    ///
    /// The passed expressions must be valid in the left table.
    pub fn left_on<E: AsRef<[Expr]>>(mut self, on: E) -> Self {
        self.left_on = on.as_ref().to_vec();
        self
    }

    /// The expressions you want to join the right table on.
    ///
    /// The passed expressions must be valid in the right table.
    pub fn right_on<E: AsRef<[Expr]>>(mut self, on: E) -> Self {
        self.right_on = on.as_ref().to_vec();
        self
    }

    /// Allow parallel table evaluation.
    pub fn allow_parallel(mut self, allow: bool) -> Self {
        self.allow_parallel = allow;
        self
    }

    /// Force parallel table evaluation.
    pub fn force_parallel(mut self, force: bool) -> Self {
        self.force_parallel = force;
        self
    }

    /// Join on null values. By default null values will never produce matches.
    pub fn join_nulls(mut self, join_nulls: bool) -> Self {
        self.join_nulls = join_nulls;
        self
    }

    /// Suffix to add duplicate column names in join.
    /// Defaults to `"_right"` if this method is never called.
    pub fn suffix<S: AsRef<str>>(mut self, suffix: S) -> Self {
        self.suffix = Some(suffix.as_ref().to_string());
        self
    }

    /// Whether to coalesce join columns.
    pub fn coalesce(mut self, coalesce: JoinCoalesce) -> Self {
        self.coalesce = coalesce;
        self
    }

    /// Finish builder
    pub fn finish(self) -> LazyFrame {
        let mut opt_state = self.lf.opt_state;
        let other = self.other.expect("with not set");

        // if any of the nodes reads from files we must activate this this plan as well.
        opt_state.file_caching |= other.opt_state.file_caching;

        let args = JoinArgs {
            how: self.how,
            validation: self.validation,
            suffix: self.suffix,
            slice: None,
            join_nulls: self.join_nulls,
            coalesce: self.coalesce,
        };

        let lp = self
            .lf
            .get_plan_builder()
            .join(
                other.logical_plan,
                self.left_on,
                self.right_on,
                JoinOptions {
                    allow_parallel: self.allow_parallel,
                    force_parallel: self.force_parallel,
                    args,
                    ..Default::default()
                }
                .into(),
            )
            .build();
        LazyFrame::from_logical_plan(lp, opt_state)
    }
}