Immutable arbitrary-precision integers and decimals; pure implementation avoiding delegation to platform BigInt or math classes. Exact decimal semantics, scale-aware equality, two's-complement bitwise operations.
Pure Kotlin arbitrary-precision integers and decimals with identical semantics across Kotlin Multiplatform targets.
BigNumber provides immutable BigInteger and BigDecimal values. All arithmetic is implemented once in commonMain;
the library does not delegate to Java numeric classes, JavaScript BigInt, decimal.js, or platform-specific
arbitrary-precision APIs.
Add Maven Central and BigNumber to the commonMain source set:
repositories {
mavenCentral()
}
kotlin {
sourceSets {
commonMain.dependencies {
implementation("dev.quwac:bignumber:<version>")
}
}
}Replace <version> with the release you want to use.
import dev.quwac.bignumber.*
val integer = "12345678901234567890".toBigInteger()
val radixValue = BigInteger("ff", radix = 16)
val amount = BigDecimal("10.50")
val divisor = 4.toBigDecimal()
val exact = amount / divisor // 2.625
val total = 4 + amount // 14.50
val rounded = amount.divide(
BigDecimal("3"),
MathContext(4, RoundingMode.HALF_EVEN),
)Deprecated Kotlin/Native target presets are not published.
BigDecimal follows the familiar Java decimal model without calling java.math in production code:
unscaledValue and scale define the representation.precision is the number of digits in the unscaled value; zero has precision one.equals and hashCode include scale, so BigDecimal("1.0") and BigDecimal("1.00") are distinct representations.compareTo compares numerical values, so those same values compare as equal./ and divide(other) require an exact terminating decimal result.setScale, and MathContext operations require an explicit RoundingMode when information may
be discarded.RoundingMode.UNNECESSARY throws ArithmeticException if rounding is required.Byte, Short, Int, Long, Float, Double, and the unsigned integer types provide concrete overloads for
+, -, *, /, %, and ordered comparison with BigDecimal in either operand order. Floating-point operands
use their canonical decimal string representation; NaN and infinities are rejected.toString uses canonical plain or scientific notation; toPlainString never uses an exponent and
toEngineeringString uses exponents divisible by three.Primitive conversions such as toInt() and toLong() return the low-order bits, matching the truncating behavior
familiar from java.math.BigInteger. Exact variants such as toIntExact() and toULongExact() throw
ArithmeticException on overflow.
Byte, Short, Int, Long, and the unsigned integer types provide concrete overloads for +, -, *, /, %,
and ordered comparison with BigInteger in either operand order. Float and Double provide the same operators with
BigInteger, returning BigDecimal so fractional results are preserved.
Bitwise operations use an infinite-width two's-complement representation, so -1 has 1 in every bit position rather
than a finite representation such as 11. and, or, xor, andNot, inv, shl, and shr therefore work
consistently for positive and negative values. The binary operations accept BigInteger or any concrete integral type
in either operand order. Individual bits can be read or changed with testBit, setBit, clearBit, and flipBit.
There is no unsigned right shift because an arbitrary-precision integer has no fixed most-significant bit.
val masked = BigInteger.of(-1) and BigInteger.of(5) // 5
val inverted = BigInteger.of(5).inv() // -6
val shifted = BigInteger.of(-5) shr 1 // -3String parsing accepts an optional sign. BigInteger supports radices from 2 through 36. Invalid text throws
IllegalArgumentException; nullable parsing functions return null. Division by zero and non-terminating exact decimal
division throw ArithmeticException.
The internal decimal-digit storage is not public API. It may be replaced without changing the numerical contract.
The repository uses mise to install JDK 17 and the project tools.
mise run test
mise run checkThe local publication uses the coordinates dev.quwac:bignumber:0.1.0-SNAPSHOT.
BigNumber is available under the MIT License.
Pure Kotlin arbitrary-precision integers and decimals with identical semantics across Kotlin Multiplatform targets.
BigNumber provides immutable BigInteger and BigDecimal values. All arithmetic is implemented once in commonMain;
the library does not delegate to Java numeric classes, JavaScript BigInt, decimal.js, or platform-specific
arbitrary-precision APIs.
Add Maven Central and BigNumber to the commonMain source set:
repositories {
mavenCentral()
}
kotlin {
sourceSets {
commonMain.dependencies {
implementation("dev.quwac:bignumber:<version>")
}
}
}Replace <version> with the release you want to use.
import dev.quwac.bignumber.*
val integer = "12345678901234567890".toBigInteger()
val radixValue = BigInteger("ff", radix = 16)
val amount = BigDecimal("10.50")
val divisor = 4.toBigDecimal()
val exact = amount / divisor // 2.625
val total = 4 + amount // 14.50
val rounded = amount.divide(
BigDecimal("3"),
MathContext(4, RoundingMode.HALF_EVEN),
)Deprecated Kotlin/Native target presets are not published.
BigDecimal follows the familiar Java decimal model without calling java.math in production code:
unscaledValue and scale define the representation.precision is the number of digits in the unscaled value; zero has precision one.equals and hashCode include scale, so BigDecimal("1.0") and BigDecimal("1.00") are distinct representations.compareTo compares numerical values, so those same values compare as equal./ and divide(other) require an exact terminating decimal result.setScale, and MathContext operations require an explicit RoundingMode when information may
be discarded.RoundingMode.UNNECESSARY throws ArithmeticException if rounding is required.Byte, Short, Int, Long, Float, Double, and the unsigned integer types provide concrete overloads for
+, -, *, /, %, and ordered comparison with BigDecimal in either operand order. Floating-point operands
use their canonical decimal string representation; NaN and infinities are rejected.toString uses canonical plain or scientific notation; toPlainString never uses an exponent and
toEngineeringString uses exponents divisible by three.Primitive conversions such as toInt() and toLong() return the low-order bits, matching the truncating behavior
familiar from java.math.BigInteger. Exact variants such as toIntExact() and toULongExact() throw
ArithmeticException on overflow.
Byte, Short, Int, Long, and the unsigned integer types provide concrete overloads for +, -, *, /, %,
and ordered comparison with BigInteger in either operand order. Float and Double provide the same operators with
BigInteger, returning BigDecimal so fractional results are preserved.
Bitwise operations use an infinite-width two's-complement representation, so -1 has 1 in every bit position rather
than a finite representation such as 11. and, or, xor, andNot, inv, shl, and shr therefore work
consistently for positive and negative values. The binary operations accept BigInteger or any concrete integral type
in either operand order. Individual bits can be read or changed with testBit, setBit, clearBit, and flipBit.
There is no unsigned right shift because an arbitrary-precision integer has no fixed most-significant bit.
val masked = BigInteger.of(-1) and BigInteger.of(5) // 5
val inverted = BigInteger.of(5).inv() // -6
val shifted = BigInteger.of(-5) shr 1 // -3String parsing accepts an optional sign. BigInteger supports radices from 2 through 36. Invalid text throws
IllegalArgumentException; nullable parsing functions return null. Division by zero and non-terminating exact decimal
division throw ArithmeticException.
The internal decimal-digit storage is not public API. It may be replaced without changing the numerical contract.
The repository uses mise to install JDK 17 and the project tools.
mise run test
mise run checkThe local publication uses the coordinates dev.quwac:bignumber:0.1.0-SNAPSHOT.
BigNumber is available under the MIT License.