Method Next

pkg/parquetquery/iters.go:1018–1074

()

}

func (j *JoinIterator) Next() (*IteratorResult, error) {
outer:
	for {
		// This loop is doing two things:
		// On first-pass peek each iter and ensure it has at least one
		// result.  If any iter has no results we can exit early
		// without processing the remaining data in the others.
		// On subsequent passes the first iter is never nil except
		// when everything is fully exhausted. We check once more
		// and then exit.
		if j.peeks[0] == nil {
			for i := range j.iters {
				res, err := j.peek(i)
				if err != nil {
					return nil, err
				}
				if res == nil {
					return nil, nil
				}
			}
		}

		// The first iter is pointing at the next candidate row. Proceed through iters 2 to N looking
		// for matches.
		for iterNum := 1; iterNum < len(j.iters); iterNum++ {
			err := j.seek(iterNum, j.peeks[0].RowNumber, j.definitionLevel)
			if err != nil {
				return nil, err
			}

			if j.peeks[iterNum] == nil {
				return nil, nil
			}

			if CompareRowNumbers(j.definitionLevel, j.peeks[iterNum].RowNumber, j.peeks[0].RowNumber) == 1 {
				// This iterator has a higher row number than all previous iterators.  That means it might have
				// a higher filtering power, swap it to the top and restart the loop.
				j.iters[0], j.iters[iterNum] = j.iters[iterNum], j.iters[0]
				j.peeks[0], j.peeks[iterNum] = j.peeks[iterNum], j.peeks[0]
				continue outer
			}
		}

		// All iterators pointing at same row
		// Get the data
		result, err := j.collect(j.peeks[0].RowNumber)
		if err != nil {
			return nil, fmt.Errorf("join iterator collect failed: %w", err)
		}

		// Keep group?
		if j.pred == nil || j.pred.KeepGroup(result) {
			// Yes
			return result, nil
		}
	}
}