DeepConvert

convenient literal construction of values of large data types

Linux, OSX:   Windows:      

Most of the following is also in this Jupyter notebook.

This package provides convenient literal construction of values of large data types.

It defines macros that define functions to convert all numbers in an expression to a given numeric type and evaluate that expression. (Like deepcopy, it traverses the entire expression tree.) It is meant to allow a convenient way to input large numbers without overflow.

The macros @bigfloat, @bigint, and @int128 convert numeric literals in their argument to BigFloat, BigInt or Int128. See examples below.

Two examples of non-standard AbstractString literals are exported, bf and bi, which construct BigFloats and BigInts from strings. Note that these are not the same as the string literal big in Julia Base.

julia> BigInt[2^63, 2^64]
2-element Array{BigInt,1}:  # awesome, an array of BigInt's!
 -9223372036854775808
                    0

julia> using DeepConvert
julia> bi"[2^63, 2^64]"
2-element Array{BigInt,1}:
  9223372036854775808
 18446744073709551616

julia> @bigint [2^63, 2^64]
2-element Array{BigInt,1}:
  9223372036854775808
 18446744073709551616

julia> binomial(100, 50)
ERROR: OverflowError: binomial(100, 50) overflows

julia> @int128 binomial(100, 50)
100891344545564193334812497256

julia> round(bf"((1 + 2e-50) - (1 + 1e-50)) / (1e-50)")
1.0

deepbigfloat takes an expression or string as an argument and acts on literals

julia> a = parse("1 + 1")
:(1 + 1)

julia> deepbigfloat(a)
2e+00 with 256 bits of precision

julia> deepbigfloat("1 + 1")
2e+00 with 256 bits of precision

@mkdeepconvert(funcname, convfunc)

Example

Define a function that converts Reals in an expression to Int128

julia> using DeepConvert

julia> Int128(2 * (10^19 + 10^17))  # does not do what you want
1753255926290448384

julia> @mkdeepconvert(convint128, Int128)
convint128 (generic function with 3 methods)

julia> convint128(:( 2 * (10^19 + 10^17) )) # does what you want
20200000000000000000

julia> convint128("2 * (10^19 + 10^17)")  # can use double quotes, as well
20200000000000000000

julia> @mkdeepconvert(convuint64, UInt64)
convuint64 (generic function with 3 methods)

julia> convuint64("10^19 + 10^17")
0x8c2a687ce7720000

bi, bf non-standard AbstractString literals

bi is implemented by

@mkdeepconvert(deepbigint, BigInt)
macro bi_str(s) deepbigint(s) end

@bigint, @int128

Any Int and Int128 values in the expression following @bigint are converted to BigInt. A contrived example:

@bigint function g(x)
           return 2^64 * x
        end

Gives this,

julia> g(2.0)
3.6893488147419103232e+19

To override the macro, you have to ask for the smaller type,

@bigint function g(x)
           return Int(2)^Int(64) * x   # always returns zero
        end

In effect, this temporarily makes BigInt the default integer type.