caldempsey commented on code in PR #152:
URL: https://github.com/apache/spark-connect-go/pull/152#discussion_r2891248410
##########
spark/sql/types/rowiterator.go:
##########
@@ -0,0 +1,67 @@
+package types
+
+import (
+ "context"
+ "errors"
+ "io"
+ "iter"
+
+ "github.com/apache/arrow-go/v18/arrow"
+)
+
+// rowIterFromRecord converts an Arrow record into a row iterator,
+// releasing the record when iteration completes or the consumer stops.
+func rowIterFromRecord(rec arrow.Record) iter.Seq2[Row, error] {
+ return func(yield func(Row, error) bool) {
+ defer rec.Release()
+ rows, err := ReadArrowRecordToRows(rec)
+ if err != nil {
+ _ = yield(nil, err)
+ return
+ }
+ for _, row := range rows {
+ if !yield(row, nil) {
+ return
+ }
+ }
+ }
+}
+
+// NewRowSequence flattens record batches to a sequence of rows stream.
+func NewRowSequence(ctx context.Context, recordSeq iter.Seq2[arrow.Record,
error]) iter.Seq2[Row, error] {
+ return func(yield func(Row, error) bool) {
+ for rec, recErr := range recordSeq {
+ select {
+ case <-ctx.Done():
+ _ = yield(nil, ctx.Err())
+ return
+ default:
+ }
+
+ // Treat io.EOF as clean stream termination. Some Spark
+ // implementations (notably Databricks clusters as of
05/2025)
+ // yield EOF as an error value instead of ending the
sequence.
+ if errors.Is(recErr, io.EOF) {
+ return
+ }
+ if recErr != nil {
+ _ = yield(nil, recErr)
+ return
+ }
+ if rec == nil {
+ _ = yield(nil, errors.New("expected
arrow.Record to contain non-nil Rows, got nil"))
+ return
+ }
+
+ for row, err := range rowIterFromRecord(rec) {
+ if err != nil {
+ _ = yield(nil, err)
+ return
+ }
+ if !yield(row, nil) {
+ return
+ }
Review Comment:
yeah! i wasn't sure if we wanted the explicit form of this where you make
the `err != nil` flow _painfully_ clear
```go
for row, err := range rowIterFromRecord(rec) {
if !yield(row, err) || err != nil {
return
}
}
```
is another way to hash this. but honestly? i like your version most because
it preserves the semantics from a human readability standpoint better.
```go
for row, err := range rowIterFromRecord(rec) {
cont := yield(row, err)
if err != nil || !cont {
return
}
}
```
will update
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
