Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Return error if value scale is different from schema scale #489

Merged
merged 3 commits into from
Sep 10, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 5 additions & 39 deletions lib/debezium/converters/decimal.go
Original file line number Diff line number Diff line change
Expand Up @@ -2,58 +2,24 @@ package converters

import (
"fmt"
"log/slog"

"github.com/artie-labs/transfer/lib/debezium"
"github.com/artie-labs/transfer/lib/debezium/converters"
"github.com/artie-labs/transfer/lib/typing"
"github.com/cockroachdb/apd/v3"
)

// decimalWithNewExponent takes a [*apd.Decimal] and returns a new [*apd.Decimal] with a the given exponent.
// If the new exponent is less precise then the extra digits will be truncated.
func decimalWithNewExponent(decimal *apd.Decimal, newExponent int32) *apd.Decimal {
exponentDelta := newExponent - decimal.Exponent // Exponent is negative.

if exponentDelta == 0 {
return new(apd.Decimal).Set(decimal)
}

coefficient := new(apd.BigInt).Set(&decimal.Coeff)

if exponentDelta < 0 {
multiplier := new(apd.BigInt).Exp(apd.NewBigInt(10), apd.NewBigInt(int64(-exponentDelta)), nil)
coefficient.Mul(coefficient, multiplier)
} else if exponentDelta > 0 {
divisor := new(apd.BigInt).Exp(apd.NewBigInt(10), apd.NewBigInt(int64(exponentDelta)), nil)
coefficient.Div(coefficient, divisor)
}

return &apd.Decimal{
Form: decimal.Form,
Negative: decimal.Negative,
Exponent: newExponent,
Coeff: *coefficient,
}
}

// encodeDecimalWithScale is used to encode a [*apd.Decimal] to `org.apache.kafka.connect.data.Decimal`
// using a specific scale.
func encodeDecimalWithScale(decimal *apd.Decimal, scale int32) []byte {
func encodeDecimalWithScale(decimal *apd.Decimal, scale int32) ([]byte, error) {
targetExponent := -scale // Negate scale since [Decimal.Exponent] is negative.
if decimal.Exponent != targetExponent {
// TODO: We may be able to remove this conversion and just return an error to maintain parity with `org.apache.kafka.connect.data.Decimal`
// Return an error if the scales are different, this maintains parity with `org.apache.kafka.connect.data.Decimal`.
// https://github.com/a0x8o/kafka/blob/54eff6af115ee647f60129f2ce6a044cb17215d0/connect/api/src/main/java/org/apache/kafka/connect/data/Decimal.java#L69
slog.Warn("Value scale is different from schema scale",
slog.Any("value", decimal.Text('f')),
slog.Any("actual", -decimal.Exponent),
slog.Any("expected", scale),
)

decimal = decimalWithNewExponent(decimal, targetExponent)
return nil, fmt.Errorf("value scale (%d) is different from schema scale (%d)", -decimal.Exponent, scale)
}
bytes, _ := converters.EncodeDecimal(decimal)
return bytes
return bytes, nil
}

type decimalConverter struct {
Expand Down Expand Up @@ -93,7 +59,7 @@ func (d decimalConverter) Convert(value any) (any, error) {
return nil, fmt.Errorf(`unable to use %q as a decimal: %w`, stringValue, err)
}

return encodeDecimalWithScale(decimal, int32(d.scale)), nil
return encodeDecimalWithScale(decimal, int32(d.scale))
}

type VariableNumericConverter struct{}
Expand Down
56 changes: 12 additions & 44 deletions lib/debezium/converters/decimal_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -9,36 +9,22 @@ import (
"github.com/artie-labs/transfer/lib/numbers"
"github.com/artie-labs/transfer/lib/ptr"
"github.com/artie-labs/transfer/lib/typing/decimal"
"github.com/cockroachdb/apd/v3"
"github.com/stretchr/testify/assert"
)

func TestDecimalWithNewExponent(t *testing.T) {
assert.Equal(t, "0", decimalWithNewExponent(apd.New(0, 0), 0).Text('f'))
assert.Equal(t, "00", decimalWithNewExponent(apd.New(0, 1), 1).Text('f'))
assert.Equal(t, "0", decimalWithNewExponent(apd.New(0, 100), 0).Text('f'))
assert.Equal(t, "00", decimalWithNewExponent(apd.New(0, 0), 1).Text('f'))
assert.Equal(t, "0.0", decimalWithNewExponent(apd.New(0, 0), -1).Text('f'))

// Same exponent:
assert.Equal(t, "12.349", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), -3).Text('f'))
// More precise exponent:
assert.Equal(t, "12.3490", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), -4).Text('f'))
assert.Equal(t, "12.34900", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), -5).Text('f'))
// Lest precise exponent:
// Extra digits should be truncated rather than rounded.
assert.Equal(t, "12.34", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), -2).Text('f'))
assert.Equal(t, "12.3", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), -1).Text('f'))
assert.Equal(t, "12", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), 0).Text('f'))
assert.Equal(t, "10", decimalWithNewExponent(numbers.MustParseDecimal("12.349"), 1).Text('f'))
}

func TestEncodeDecimalWithScale(t *testing.T) {
mustEncodeAndDecodeDecimal := func(value string, scale int32) string {
bytes := encodeDecimalWithScale(numbers.MustParseDecimal(value), scale)
bytes, err := encodeDecimalWithScale(numbers.MustParseDecimal(value), scale)
assert.NoError(t, err)
return converters.DecodeDecimal(bytes, scale).String()
}

mustReturnError := func(value string, scale int32) error {
_, err := encodeDecimalWithScale(numbers.MustParseDecimal(value), scale)
assert.Error(t, err)
return err
}

// Whole numbers:
for i := range 100_000 {
strValue := fmt.Sprint(i)
Expand All @@ -52,28 +38,10 @@ func TestEncodeDecimalWithScale(t *testing.T) {
// Scale of 15 that is equal to the amount of decimal places in the value:
assert.Equal(t, "145.183000000000000", mustEncodeAndDecodeDecimal("145.183000000000000", 15))
assert.Equal(t, "-145.183000000000000", mustEncodeAndDecodeDecimal("-145.183000000000000", 15))
// If scale is smaller than the amount of decimal places then the extra places should be truncated without rounding:
assert.Equal(t, "145.18300000000000", mustEncodeAndDecodeDecimal("145.183000000000000", 14))
assert.Equal(t, "145.18300000000000", mustEncodeAndDecodeDecimal("145.183000000000005", 14))
assert.Equal(t, "-145.18300000000000", mustEncodeAndDecodeDecimal("-145.183000000000005", 14))
assert.Equal(t, "145.18300000000000", mustEncodeAndDecodeDecimal("145.183000000000009", 14))
assert.Equal(t, "-145.18300000000000", mustEncodeAndDecodeDecimal("-145.183000000000009", 14))
assert.Equal(t, "-145.18300000000000", mustEncodeAndDecodeDecimal("-145.183000000000000", 14))
assert.Equal(t, "145.18300000000000", mustEncodeAndDecodeDecimal("145.183000000000001", 14))
assert.Equal(t, "-145.18300000000000", mustEncodeAndDecodeDecimal("-145.183000000000001", 14))
assert.Equal(t, "145.18300000000000", mustEncodeAndDecodeDecimal("145.183000000000004", 14))
assert.Equal(t, "-145.18300000000000", mustEncodeAndDecodeDecimal("-145.183000000000004", 14))
// If scale is larger than the amount of decimal places then the extra places should be padded with zeros:
assert.Equal(t, "145.1830000000000000", mustEncodeAndDecodeDecimal("145.183000000000000", 16))
assert.Equal(t, "-145.1830000000000000", mustEncodeAndDecodeDecimal("-145.183000000000000", 16))
assert.Equal(t, "145.1830000000000010", mustEncodeAndDecodeDecimal("145.183000000000001", 16))
assert.Equal(t, "-145.1830000000000010", mustEncodeAndDecodeDecimal("-145.183000000000001", 16))
assert.Equal(t, "145.1830000000000040", mustEncodeAndDecodeDecimal("145.183000000000004", 16))
assert.Equal(t, "-145.1830000000000040", mustEncodeAndDecodeDecimal("-145.183000000000004", 16))
assert.Equal(t, "145.1830000000000050", mustEncodeAndDecodeDecimal("145.183000000000005", 16))
assert.Equal(t, "-145.1830000000000050", mustEncodeAndDecodeDecimal("-145.183000000000005", 16))
assert.Equal(t, "145.1830000000000090", mustEncodeAndDecodeDecimal("145.183000000000009", 16))
assert.Equal(t, "-145.1830000000000090", mustEncodeAndDecodeDecimal("-145.183000000000009", 16))
// If scale is smaller than the amount of decimal places then an error should be returned:
assert.ErrorContains(t, mustReturnError("145.183000000000000", 14), "value scale (15) is different from schema scale (14)")
// If scale is larger than the amount of decimal places then an error should be returned:
assert.ErrorContains(t, mustReturnError("-145.183000000000005", 16), "value scale (15) is different from schema scale (16)")

assert.Equal(t, "-9063701308.217222135", mustEncodeAndDecodeDecimal("-9063701308.217222135", 9))
assert.Equal(t, "-74961544796695.89960242", mustEncodeAndDecodeDecimal("-74961544796695.89960242", 8))
Expand Down
2 changes: 1 addition & 1 deletion lib/debezium/converters/money.go
Original file line number Diff line number Diff line change
Expand Up @@ -56,5 +56,5 @@ func (m MoneyConverter) Convert(value any) (any, error) {
return nil, fmt.Errorf(`unable to use %q as a money value: %w`, valString, err)
}

return encodeDecimalWithScale(decimal, int32(m.Scale())), nil
return encodeDecimalWithScale(decimal, int32(m.Scale()))
}
8 changes: 3 additions & 5 deletions lib/debezium/converters/money_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -70,11 +70,9 @@ func TestMoneyConverter_Convert(t *testing.T) {
assert.Equal(t, "1234.56", decodeValue(converted))
}
{
// string with $, comma, and no cents
converted, err := converter.Convert("$1000,234")
assert.NoError(t, err)
assert.Equal(t, []byte{0x5, 0xf6, 0x3c, 0x68}, converted)
assert.Equal(t, "1000234.00", decodeValue(converted))
// string with missing cents
_, err := converter.Convert("$1000,234")
assert.ErrorContains(t, err, "value scale (0) is different from schema scale (2)")
}
{
// Malformed string - empty string.
Expand Down