/
file_reader.go
executable file
·775 lines (669 loc) · 20.8 KB
/
file_reader.go
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
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package pqarrow
import (
"context"
"fmt"
"io"
"sync"
"sync/atomic"
"github.com/apache/arrow/go/v9/arrow"
"github.com/apache/arrow/go/v9/arrow/array"
"github.com/apache/arrow/go/v9/arrow/arrio"
"github.com/apache/arrow/go/v9/arrow/memory"
"github.com/apache/arrow/go/v9/parquet"
"github.com/apache/arrow/go/v9/parquet/file"
"github.com/apache/arrow/go/v9/parquet/schema"
"golang.org/x/sync/errgroup"
"golang.org/x/xerrors"
)
type itrFactory func(int, *file.Reader) *columnIterator
type readerCtx struct {
rdr *file.Reader
mem memory.Allocator
colFactory itrFactory
filterLeaves bool
includedLeaves map[int]bool
}
func (r readerCtx) includesLeaf(idx int) bool {
_, ok := r.includedLeaves[idx]
return ok
}
// ReadTable is a convenience function to quickly and easily read a parquet file
// into an arrow table.
//
// The schema of the arrow table is generated based on the schema of the parquet file,
// including nested columns/lists/etc. in the same fashion as the FromParquetSchema
// function. This just encapsulates the logic of creating a separate file.Reader and
// pqarrow.FileReader to make a single easy function when you just want to construct
// a table from the entire parquet file rather than reading it piecemeal.
func ReadTable(ctx context.Context, r parquet.ReaderAtSeeker, props *parquet.ReaderProperties, arrProps ArrowReadProperties, mem memory.Allocator) (arrow.Table, error) {
pf, err := file.NewParquetReader(r, file.WithReadProps(props))
if err != nil {
return nil, err
}
reader, err := NewFileReader(pf, arrProps, mem)
if err != nil {
return nil, err
}
return reader.ReadTable(ctx)
}
// FileReader is the base object for reading a parquet file into arrow object
// types.
//
// It provides utility functions for reading record batches, a table, subsets of
// columns / rowgroups, and so on.
type FileReader struct {
mem memory.Allocator
rdr *file.Reader
Props ArrowReadProperties
Manifest *SchemaManifest
}
// NewFileReader constructs a reader for converting to Arrow objects from an existing
// parquet file reader object.
//
// Only returns an error if there is some error constructing the schema manifest from
// the parquet file metadata.
func NewFileReader(rdr *file.Reader, props ArrowReadProperties, mem memory.Allocator) (*FileReader, error) {
manifest, err := NewSchemaManifest(rdr.MetaData().Schema, rdr.MetaData().KeyValueMetadata(), &props)
if err != nil {
return nil, err
}
return &FileReader{
mem: mem,
rdr: rdr,
Props: props,
Manifest: manifest,
}, nil
}
// Schema returns the arrow schema representation of the underlying file's schema.
func (fr *FileReader) Schema() (*arrow.Schema, error) {
return FromParquet(fr.rdr.MetaData().Schema, &fr.Props, fr.rdr.MetaData().KeyValueMetadata())
}
type colReaderImpl interface {
LoadBatch(nrecs int64) error
BuildArray(boundedLen int64) (*arrow.Chunked, error)
GetDefLevels() ([]int16, error)
GetRepLevels() ([]int16, error)
Field() *arrow.Field
IsOrHasRepeatedChild() bool
Retain()
Release()
}
// ColumnReader is used for reading batches of data from a specific column
// across multiple row groups to return a chunked arrow array.
type ColumnReader struct {
colReaderImpl
}
// NextBatch returns a chunked array after reading `size` values, potentially
// across multiple row groups.
func (c *ColumnReader) NextBatch(size int64) (*arrow.Chunked, error) {
if err := c.LoadBatch(size); err != nil {
return nil, err
}
return c.BuildArray(size)
}
type rdrCtxKey struct{}
func readerCtxFromContext(ctx context.Context) readerCtx {
rdc := ctx.Value(rdrCtxKey{})
if rdc != nil {
return rdc.(readerCtx)
}
panic("no readerctx")
}
// ParquetReader returns the underlying parquet file reader that it was constructed with
func (fr *FileReader) ParquetReader() *file.Reader { return fr.rdr }
// GetColumn returns a reader for pulling the data of leaf column index i
// across all row groups in the file.
func (fr *FileReader) GetColumn(ctx context.Context, i int) (*ColumnReader, error) {
return fr.getColumnReader(ctx, i, fr.allRowGroupFactory())
}
func rowGroupFactory(rowGroups []int) itrFactory {
return func(i int, rdr *file.Reader) *columnIterator {
return &columnIterator{
index: i,
rdr: rdr,
schema: rdr.MetaData().Schema,
rowGroups: rowGroups,
}
}
}
func (fr *FileReader) allRowGroupFactory() itrFactory {
rowGroups := make([]int, fr.rdr.NumRowGroups())
for idx := range rowGroups {
rowGroups[idx] = idx
}
return rowGroupFactory(rowGroups)
}
// GetFieldReader returns a reader for the entire Field of index i which could potentially include reading
// multiple columns from the underlying parquet file if that field is a nested field.
//
// IncludedLeaves and RowGroups are used to specify precisely which leaf indexes and row groups to read a subset of.
func (fr *FileReader) GetFieldReader(ctx context.Context, i int, includedLeaves map[int]bool, rowGroups []int) (*ColumnReader, error) {
ctx = context.WithValue(ctx, rdrCtxKey{}, readerCtx{
rdr: fr.rdr,
mem: fr.mem,
colFactory: rowGroupFactory(rowGroups),
filterLeaves: true,
includedLeaves: includedLeaves,
})
return fr.getReader(ctx, &fr.Manifest.Fields[i], *fr.Manifest.Fields[i].Field)
}
// GetFieldReaders is for retrieving readers for multiple fields at one time for only the list
// of column indexes and rowgroups requested. It returns a slice of the readers and the corresponding
// arrow.Schema for those columns.
func (fr *FileReader) GetFieldReaders(ctx context.Context, colIndices, rowGroups []int) ([]*ColumnReader, *arrow.Schema, error) {
fieldIndices, err := fr.Manifest.GetFieldIndices(colIndices)
if err != nil {
return nil, nil, err
}
includedLeaves := make(map[int]bool)
for _, col := range colIndices {
includedLeaves[col] = true
}
out := make([]*ColumnReader, len(fieldIndices))
outFields := make([]arrow.Field, len(fieldIndices))
// Load batches in parallel
// When reading structs with large numbers of columns, the serial load is very slow.
// This is especially true when reading Cloud Storage. Loading concurrently
// greatly improves performance.
// GetFieldReader causes read operations, when issued serially on large numbers of columns,
// this is super time consuming. Get field readers concurrently.
g,gctx := errgroup.WithContext(ctx)
if !fr.Props.Parallel {
g.SetLimit(1)
}
for idx, fidx := range fieldIndices {
idx, fidx := idx, fidx // create concurrent copy
g.Go(func() error {
rdr, err := fr.GetFieldReader(gctx, fidx, includedLeaves, rowGroups)
if err != nil {
return err
}
outFields[idx] = *rdr.Field()
out[idx] = rdr
return nil
})
}
if err = g.Wait(); err != nil {
return nil, nil, err
}
return out, arrow.NewSchema(outFields, fr.Manifest.SchemaMeta), nil
}
// RowGroup creates a reader that will *only* read from the requested row group
func (fr *FileReader) RowGroup(idx int) RowGroupReader {
return RowGroupReader{fr, idx}
}
// ReadColumn reads data to create a chunked array only from the requested row groups.
func (fr *FileReader) ReadColumn(rowGroups []int, rdr *ColumnReader) (*arrow.Chunked, error) {
recs := int64(0)
for _, rg := range rowGroups {
recs += fr.rdr.MetaData().RowGroups[rg].GetNumRows()
}
return rdr.NextBatch(recs)
}
// ReadTable reads the entire file into an array.Table
func (fr *FileReader) ReadTable(ctx context.Context) (arrow.Table, error) {
var (
cols = []int{}
rgs = []int{}
)
for i := 0; i < fr.rdr.MetaData().Schema.NumColumns(); i++ {
cols = append(cols, i)
}
for i := 0; i < fr.rdr.NumRowGroups(); i++ {
rgs = append(rgs, i)
}
return fr.ReadRowGroups(ctx, cols, rgs)
}
func (fr *FileReader) checkCols(indices []int) (err error) {
for _, col := range indices {
if col < 0 || col >= fr.rdr.MetaData().Schema.NumColumns() {
err = fmt.Errorf("invalid column index specified %d out of %d", col, fr.rdr.MetaData().Schema.NumColumns())
break
}
}
return
}
func (fr *FileReader) checkRowGroups(indices []int) (err error) {
for _, rg := range indices {
if rg < 0 || rg >= fr.rdr.NumRowGroups() {
err = fmt.Errorf("invalid row group specified: %d, file only has %d row groups", rg, fr.rdr.NumRowGroups())
break
}
}
return
}
type readerInfo struct {
rdr *ColumnReader
idx int
}
type resultPair struct {
idx int
data *arrow.Chunked
err error
}
//! This is Super complicated. I would simpify the pattern, but it works and hesitant to change what works.
// ReadRowGroups is for generating an array.Table from the file but filtering to only read the requested
// columns and row groups rather than the entire file which ReadTable does.
func (fr *FileReader) ReadRowGroups(ctx context.Context, indices, rowGroups []int) (arrow.Table, error) {
if err := fr.checkRowGroups(rowGroups); err != nil {
return nil, err
}
if err := fr.checkCols(indices); err != nil {
return nil, err
}
// TODO(mtopol): add optimizations for pre-buffering data options
readers, sc, err := fr.GetFieldReaders(ctx, indices, rowGroups)
if err != nil {
return nil, err
}
// producer-consumer parallelization
var (
np = 1
wg sync.WaitGroup
ch = make(chan readerInfo, len(readers))
results = make(chan resultPair, 2)
)
if fr.Props.Parallel {
np = len(readers)
}
ctx, cancel := context.WithCancel(ctx)
defer cancel()
wg.Add(np) // fan-out to np readers
for i := 0; i < np; i++ {
go func() {
defer wg.Done()
for {
select {
case r, ok := <-ch:
if !ok {
return
}
chnked, err := fr.ReadColumn(rowGroups, r.rdr)
// pass the result column data to the result channel
// for the consumer goroutine to process
results <- resultPair{r.idx, chnked, err}
case <-ctx.Done(): // check if we cancelled
return
}
}
}()
}
go func() {
wg.Wait()
close(results) // close the result channel when there's no more
}()
// pass pairs of reader and column index to the channel for the
// goroutines to read the data
for idx, r := range readers {
defer func(r *ColumnReader) {
r.Release()
}(r)
ch <- readerInfo{r, idx}
}
close(ch)
// output slice of columns
columns := make([]arrow.Column, len(sc.Fields()))
for data := range results {
if data.err != nil {
err = data.err
cancel()
break
}
//lint:ignore SA9001 defer.
defer data.data.Release()
col := arrow.NewColumn(sc.Field(data.idx), data.data)
columns[data.idx] = *col
}
if err != nil {
// if we encountered an error, consume any waiting data on the channel
// so the goroutines don't leak and so memory can get cleaned up. we already
// cancelled the context so we're just consuming anything that was already queued up.
for data := range results {
//lint:ignore SA9001 defer.
defer data.data.Release()
}
return nil, err
}
var nrows int
if len(columns) > 0 {
nrows = columns[0].Len()
}
return array.NewTable(sc, columns, int64(nrows)), nil
}
func (fr *FileReader) getColumnReader(ctx context.Context, i int, colFactory itrFactory) (*ColumnReader, error) {
if i < 0 || i >= fr.rdr.MetaData().Schema.NumColumns() {
return nil, fmt.Errorf("invalid column index chosen %d, there are only %d columns", i, fr.rdr.MetaData().Schema.NumColumns())
}
ctx = context.WithValue(ctx, rdrCtxKey{}, readerCtx{
rdr: fr.rdr,
mem: fr.mem,
colFactory: colFactory,
filterLeaves: false,
})
return fr.getReader(ctx, &fr.Manifest.Fields[i], *fr.Manifest.Fields[i].Field)
}
// RecordReader is a Record Batch Reader that meets the interfaces for both
// array.RecordReader and arrio.Reader to allow easy progressive reading
// of record batches from the parquet file. Ideal for streaming.
type RecordReader interface {
array.RecordReader
arrio.Reader
}
// GetRecordReader returns a record reader that reads only the requested column indexes and row groups.
//
// For both cases, if you pass nil for column indexes or rowgroups it will default to reading all of them.
func (fr *FileReader) GetRecordReader(ctx context.Context, colIndices, rowGroups []int) (RecordReader, error) {
if err := fr.checkRowGroups(rowGroups); err != nil {
return nil, err
}
if rowGroups == nil {
rowGroups = make([]int, fr.rdr.NumRowGroups())
for idx := range rowGroups {
rowGroups[idx] = idx
}
}
if err := fr.checkCols(colIndices); err != nil {
return nil, err
}
if colIndices == nil {
colIndices = make([]int, fr.rdr.MetaData().Schema.NumColumns())
for idx := range colIndices {
colIndices[idx] = idx
}
}
// TODO(mtopol): add optimizations to pre-buffer data from the file
readers, sc, err := fr.GetFieldReaders(ctx, colIndices, rowGroups)
if err != nil {
return nil, err
}
if len(readers) == 0 {
return nil, xerrors.New("no leaf column readers matched col indices")
}
nrows := int64(0)
for _, rg := range rowGroups {
nrows += fr.rdr.MetaData().RowGroup(rg).NumRows()
}
return &recordReader{
numRows: nrows,
batchSize: fr.Props.BatchSize,
parallel: fr.Props.Parallel,
sc: sc,
fieldReaders: readers,
refCount: 1,
}, nil
}
func (fr *FileReader) getReader(ctx context.Context, field *SchemaField, arrowField arrow.Field) (out *ColumnReader, err error) {
rctx := readerCtxFromContext(ctx)
if len(field.Children) == 0 {
if !field.IsLeaf() {
return nil, xerrors.New("parquet non-leaf node has no children")
}
if rctx.filterLeaves && !rctx.includesLeaf(field.ColIndex) {
return nil, nil
}
out, err = newLeafReader(&rctx, field.Field, rctx.colFactory(field.ColIndex, rctx.rdr), field.LevelInfo, fr.Props)
return
}
switch arrowField.Type.ID() {
case arrow.EXTENSION:
return nil, xerrors.New("extension type not implemented")
case arrow.STRUCT:
childReaders := make([]*ColumnReader, len(field.Children))
childFields := make([]arrow.Field, len(field.Children))
// Get child field readers concurrently
// 'getReader' causes a read operation. Issue the 'reads' concurrently
// When reading structs with large numbers of columns, the serial load is very slow.
// This is especially true when reading Cloud Storage. Loading concurrently
// greatly improves performance.
g,gctx := errgroup.WithContext(ctx)
if !fr.Props.Parallel {
g.SetLimit(1)
}
for n, child := range field.Children {
n, child := n, child
g.Go(func() error {
reader, err := fr.getReader(gctx, &child, *child.Field)
if err != nil {
return err
}
if reader == nil {
return nil
}
childFields[n] = *child.Field
childReaders[n] = reader
return nil
})
}
if err = g.Wait(); err != nil {
return nil, err
}
// because we performed getReader concurrently, we need to prune out any empty readers
for n := len(childReaders) - 1; n >= 0; n-- {
if childReaders[n] == nil {
childReaders = append(childReaders[:n], childReaders[n+1:]...)
childFields = append(childFields[:n], childFields[n+1:]...)
}
}
if len(childFields) == 0 {
return nil, nil
}
filtered := arrow.Field{Name: arrowField.Name, Nullable: arrowField.Nullable,
Metadata: arrowField.Metadata, Type: arrow.StructOf(childFields...)}
out = newStructReader(&rctx, &filtered, field.LevelInfo, childReaders, fr.Props)
case arrow.LIST, arrow.FIXED_SIZE_LIST, arrow.MAP:
child := field.Children[0]
childReader, err := fr.getReader(ctx, &child, *child.Field)
if err != nil {
return nil, err
}
if childReader == nil {
return nil, nil
}
defer childReader.Release()
switch arrowField.Type.(type) {
case *arrow.MapType:
if len(child.Children) != 2 {
arrowField.Type = arrow.ListOf(childReader.Field().Type)
}
out = newListReader(&rctx, &arrowField, field.LevelInfo, childReader, fr.Props)
case *arrow.ListType:
out = newListReader(&rctx, &arrowField, field.LevelInfo, childReader, fr.Props)
case *arrow.FixedSizeListType:
out = newFixedSizeListReader(&rctx, &arrowField, field.LevelInfo, childReader, fr.Props)
default:
return nil, fmt.Errorf("unknown list type: %s", field.Field.String())
}
}
return
}
// RowGroupReader is a reader for getting data only from a single row group of the file
// rather than having to repeatedly pass the index to functions on the reader.
type RowGroupReader struct {
impl *FileReader
idx int
}
// ReadTable provides an array.Table consisting only of the columns requested for this rowgroup
func (rgr RowGroupReader) ReadTable(ctx context.Context, colIndices []int) (arrow.Table, error) {
return rgr.impl.ReadRowGroups(ctx, colIndices, []int{rgr.idx})
}
// Column creates a reader for just the requested column chunk in only this row group.
func (rgr RowGroupReader) Column(idx int) ColumnChunkReader {
return ColumnChunkReader{rgr.impl, idx, rgr.idx}
}
// ColumnChunkReader is a reader that reads only a single column chunk from a single
// column in a single row group
type ColumnChunkReader struct {
impl *FileReader
idx int
rowGroup int
}
func (ccr ColumnChunkReader) Read(ctx context.Context) (*arrow.Chunked, error) {
rdr, err := ccr.impl.getColumnReader(ctx, ccr.idx, rowGroupFactory([]int{ccr.rowGroup}))
if err != nil {
return nil, err
}
return ccr.impl.ReadColumn([]int{ccr.rowGroup}, rdr)
}
type columnIterator struct {
index int
rdr *file.Reader
schema *schema.Schema
rowGroups []int
}
func (c *columnIterator) NextChunk() (file.PageReader, error) {
if len(c.rowGroups) == 0 {
return nil, nil
}
rgr := c.rdr.RowGroup(c.rowGroups[0])
c.rowGroups = c.rowGroups[1:]
return rgr.GetColumnPageReader(c.index)
}
func (c *columnIterator) Descr() *schema.Column { return c.schema.Column(c.index) }
// implementation of arrio.Reader for streaming record batches
// from the parquet data.
type recordReader struct {
numRows int64
batchSize int64
parallel bool
sc *arrow.Schema
fieldReaders []*ColumnReader
cur arrow.Record
err error
refCount int64
}
func (r *recordReader) Retain() {
atomic.AddInt64(&r.refCount, 1)
}
func (r *recordReader) Release() {
if atomic.AddInt64(&r.refCount, -1) == 0 {
if r.cur != nil {
r.cur.Release()
r.cur = nil
}
if r.fieldReaders == nil {
return
}
for _, fr := range r.fieldReaders {
fr.Release()
}
r.fieldReaders = nil
}
}
func (r *recordReader) Schema() *arrow.Schema { return r.sc }
func (r *recordReader) next() bool {
cols := make([]arrow.Array, len(r.sc.Fields()))
defer func() {
for _, c := range cols {
if c != nil {
c.Release()
}
}
}()
readField := func(idx int, rdr *ColumnReader) error {
data, err := rdr.NextBatch(r.batchSize)
if err != nil {
return err
}
if data.Len() == 0 {
return io.EOF
}
arrdata, err := chunksToSingle(data)
if err != nil {
return err
}
cols[idx] = array.MakeFromData(arrdata)
return nil
}
if !r.parallel {
for idx, rdr := range r.fieldReaders {
if err := readField(idx, rdr); err != nil {
r.err = err
return false
}
}
r.cur = array.NewRecord(r.sc, cols, -1)
return true
}
var (
wg sync.WaitGroup
np = len(cols)
ch = make(chan int, np)
errch = make(chan error, np)
)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
wg.Add(np)
for i := 0; i < np; i++ {
go func() {
defer wg.Done()
for {
select {
case idx, ok := <-ch:
if !ok {
return
}
if err := readField(idx, r.fieldReaders[idx]); err != nil {
errch <- err
cancel()
return
}
case <-ctx.Done():
return
}
}
}()
}
for idx := range r.fieldReaders {
ch <- idx
}
close(ch)
wg.Wait()
close(errch)
var ok bool
// check for any errors
if r.err, ok = <-errch; ok {
// return the first error that was reported and drain
// any remaining errors from the channel before returning.
for range errch {
}
return false
}
r.cur = array.NewRecord(r.sc, cols, -1)
return true
}
func (r *recordReader) Next() bool {
if r.cur != nil {
r.cur.Release()
r.cur = nil
}
if r.err != nil {
return false
}
return r.next()
}
func (r *recordReader) Record() arrow.Record { return r.cur }
func (r *recordReader) Read() (arrow.Record, error) {
if r.cur != nil {
r.cur.Release()
r.cur = nil
}
if !r.next() {
return nil, r.err
}
return r.cur, nil
}