Skip to content

Printf: avoid converting to Float64 or BigFloat #187

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.

Sign up for GitHub

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

Merged
stevengj merged 4 commits into from
Feb 9, 2025
Merged

Printf: avoid converting to Float64 or BigFloat #187

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
f5a4a84
Printf: avoid converting to Float64 or BigFloat
jmkuhn Feb 1, 2025
f9140a8
Add tests to improve Codecov
jmkuhn Feb 2, 2025
8aa8471
ArgumentError instead of InvalidFormatStringError
jmkuhn Feb 2, 2025
19cf5e0
Move printf to a separate file
jmkuhn Feb 8, 2025
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
2 changes: 1 addition & 1 deletion .github/workflows/CI.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ jobs:
fail-fast: false
matrix:
version:
- "1.6" # oldest supported version
- "1.7" # oldest supported version
- "1" # Latest Release
os:
- ubuntu-latest
Expand Down
2 changes: 1 addition & 1 deletion Project.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
[compat]
DecFP_jll = "2.0.3"
SpecialFunctions = "0.8, 0.9, 0.10, 0.11, 1, 2"
julia = "1.6"
julia = "1.7"
Printf = "<0.0.1, 1"
Random = "<0.0.1, 1"

Expand Down
357 changes: 286 additions & 71 deletions src/DecFP.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -399,77 +399,6 @@
return
end

function Printf.Printf.fix_dec(x::$BID, n::Int, digits)
if n > length(digits) - 1
n = length(digits) - 1
end
rounded = round(ldexp10(x, n), RoundNearestTiesAway)
if rounded == 0
digits[1] = UInt8('0')
return Int32(1), Int32(1), signbit(x)
end
tostring(rounded)
buffer = _stringbuffer()
trailing_zeros = 0
i = 2
while buffer[i] != UInt8('E')
digits[i - 1] = buffer[i]
if buffer[i] == UInt8('0')
trailing_zeros += 1
else
trailing_zeros = 0
end
i += 1
end
ndigits = i - 2
len = ndigits - trailing_zeros
i += 1
if buffer[i] == UInt8('+')
expsign = +1
elseif buffer[i] == UInt8('-')
expsign = -1
end
exponent = 0
i += 1
while buffer[i] != 0x00
exponent = exponent * 10 + buffer[i] - UInt8('0')
i += 1
end
exponent *= expsign
pt = ndigits + exponent - n
neg = signbit(x)
return Int32(len), Int32(pt), neg
end

function Printf.Printf.ini_dec(x::$BID, n::Int, digits)
if n > length(digits) - 1
n = length(digits) - 1
end
if x == 0
for i = 1:n
digits[i] = UInt8('0')
end
return Int32(1), Int32(1), signbit(x)
end
normalized_exponent = exponent10(x)
rounded = round(ldexp10(x, n - 1 - normalized_exponent), RoundNearestTiesAway)
rounded_exponent = exponent10(rounded)
tostring(rounded)
buffer = _stringbuffer()
i = 2
while buffer[i] != UInt8('E')
digits[i - 1] = buffer[i]
i += 1
end
while i <= n + 1
digits[i - 1] = UInt8('0')
i += 1
end
pt = normalized_exponent + rounded_exponent - n + 2
neg = signbit(x)
return Int32(n), Int32(pt), neg
end

function sigexp(x::$BID)
isnan(x) && throw(DomainError(x, "sigexp(x) is not defined for NaN."))
isinf(x) && throw(DomainError(x, "sigexp(x) is only defined for finite x."))
Expand Down Expand Up @@ -791,6 +720,292 @@
Base.widen(::Type{Dec32}) = Dec64
Base.widen(::Type{Dec64}) = Dec128

# Printf

# max exponent: `ceil(Int, log10(floatmax(T)))`
max_integer_part_width(::Type{Dec32}) = 97
max_integer_part_width(::Type{Dec64}) = 385
max_integer_part_width(::Type{Dec128}) = 6145

Printf.plength(f::Printf.Spec{S}, x::T) where {S<:Printf.Floats, T<:DecimalFloatingPoint} = max(f.width, max_integer_part_width(T) + f.precision + 2)

Printf.tofloat(x::DecimalFloatingPoint) = x

@inline function Printf.fmt(buf, pos, x::DecimalFloatingPoint, spec::Printf.Spec{T}) where {T<:Printf.Floats}
leftalign, plus, space, zero, hash, width, prec =
spec.leftalign, spec.plus, spec.space, spec.zero, spec.hash, spec.width, spec.precision
if T == Val{'e'} || T == Val{'E'}
newpos = Base.Ryu.writeexp(buf, pos, x, prec, plus, space, hash, Printf.char(T), UInt8('.'))
elseif T == Val{'f'} || T == Val{'F'}
newpos = Base.Ryu.writefixed(buf, pos, x, prec, plus, space, hash, UInt8('.'))
elseif T == Val{'g'} || T == Val{'G'}
if isinf(x) || isnan(x)
newpos = Base.Ryu.writeshortest(buf, pos, x, plus, space)
else
# C11-compliant general format
prec = prec == 0 ? 1 : prec
_, sig, dexp = sigexp(x)
numdig = ndigits(sig)
exp = dexp + numdig - 1
if -4 ≤ exp < prec
newpos = Base.Ryu.writefixed(buf, pos, x, prec - (exp + 1), plus, space, hash, UInt8('.'), !hash)
else
newpos = Base.Ryu.writeexp(buf, pos, x, prec - 1, plus, space, hash, T == Val{'g'} ? UInt8('e') : UInt8('E'), UInt8('.'), !hash)
end
end
elseif T == Val{'a'} || T == Val{'A'}
throw(ArgumentError("%a format specifier is not implemented for DecimalFloatingPoint"))
end
if newpos - pos < width
# need to pad
if leftalign
# easy case, just pad spaces after number
for _ = 1:(width - (newpos - pos))
buf[newpos] = UInt8(' ')
newpos += 1
end
else
# right aligned
n = width - (newpos - pos)
if zero && isfinite(x)
ex = (x < 0 || (plus | space)) + (T <: Union{Val{'a'}, Val{'A'}} ? 2 : 0)
so = pos + ex
len = (newpos - pos) - ex
copyto!(buf, so + n, buf, so, len)
for i = so:(so + n - 1)
buf[i] = UInt8('0')
end
newpos += n
else
copyto!(buf, pos + n, buf, pos, newpos - pos)
for i = pos:(pos + n - 1)
buf[i] = UInt8(' ')
end
newpos += n
end
end
end
return newpos
end

function Base.Ryu.writefixed(buf, pos, x::T, precision=-1, plus=false, space=false, hash=false, decchar=UInt8('.'), trimtrailingzeros=false) where {T<:DecimalFloatingPoint}
pos = Base.Ryu.append_sign(x, plus, space, buf, pos)

# special cases
if iszero(x)
buf[pos] = UInt8('0')
pos += 1
if precision > 0 && !trimtrailingzeros
buf[pos] = decchar
pos += 1
for _ = 1:precision
buf[pos] = UInt8('0')
pos += 1
end
elseif hash
buf[pos] = decchar
pos += 1
end
return pos
elseif isnan(x)
buf[pos] = UInt8('N')
buf[pos + 1] = UInt8('a')
buf[pos + 2] = UInt8('N')
return pos + 3
elseif !isfinite(x)
buf[pos] = UInt8('I')
buf[pos + 1] = UInt8('n')
buf[pos + 2] = UInt8('f')
return pos + 3
end

_, sig, exp = sigexp(x)
numdig = ndigits(sig)
digits = codeunits(string(sig))
i = numdig
while digits[i] == UInt8('0')
sig ÷= 10
exp += 1
numdig -= 1
i -= 1
end
if trimtrailingzeros && exp <= 0 && -exp < precision
precision = -exp
end
reqdig = numdig + exp + precision
if reqdig < 0
buf[pos] = UInt8('0')
pos += 1
if precision > 0 && !trimtrailingzeros
buf[pos] = decchar
pos += 1
for _ = 1:precision
buf[pos] = UInt8('0')
pos += 1
end
elseif hash
buf[pos] = decchar
pos += 1
end
return pos
end
if reqdig < numdig
diff = numdig - reqdig
if diff == 1
denominator = oftype(sig, 10)
elseif diff == 2
denominator = oftype(sig, 100)
elseif diff == 3
denominator = oftype(sig, 1000)
else
denominator = oftype(sig, 10)^Int32(diff)
end
sig = div(sig, denominator, RoundNearest)
exp += diff
end
digits = codeunits(string(sig))
if exp <= -length(digits)
buf[pos] = UInt8('0')
buf[pos + 1] = decchar
pos += 2
for i in 1:-(exp + length(digits))
buf[pos] = UInt8('0')
pos += 1
end
end
decindex = length(digits) + exp
for i in 1:length(digits)
buf[pos] = digits[i]
pos += 1
if i == decindex && (precision > 0 || hash)
buf[pos] = decchar
pos += 1
end
end
for i in numdig+1:reqdig
i > decindex && trimtrailingzeros && break
buf[pos] = UInt8('0')
pos += 1
if i == decindex && (hash || i != reqdig)
trimtrailingzeros && break
buf[pos] = decchar
pos += 1
end
end
return pos
end

function Base.Ryu.writeexp(buf, pos, x::T, precision=-1, plus=false, space=false, hash=false, expchar=UInt8('e'), decchar=UInt8('.'), trimtrailingzeros=false) where {T<:DecimalFloatingPoint}
pos = Base.Ryu.append_sign(x, plus, space, buf, pos)

# special cases
if iszero(x)
@inbounds buf[pos] = UInt8('0')
pos += 1
if precision > 0 && !trimtrailingzeros
@inbounds buf[pos] = decchar
pos += 1
for _ = 1:precision
@inbounds buf[pos] = UInt8('0')
pos += 1
end

Check warning on line 911 in src/DecFP.jl

View check run for this annotation

Codecov / codecov/patch

src/DecFP.jl#L906-L911 

Added lines #L906 - L911 were not covered by tests
elseif hash
@inbounds buf[pos] = decchar
pos += 1
end
@inbounds buf[pos] = expchar
@inbounds buf[pos + 1] = UInt8('+')
@inbounds buf[pos + 2] = UInt8('0')
@inbounds buf[pos + 3] = UInt8('0')
return pos + 4
elseif isnan(x)
@inbounds buf[pos] = UInt8('N')
@inbounds buf[pos + 1] = UInt8('a')
@inbounds buf[pos + 2] = UInt8('N')
return pos + 3
elseif !isfinite(x)
@inbounds buf[pos] = UInt8('I')
@inbounds buf[pos + 1] = UInt8('n')
@inbounds buf[pos + 2] = UInt8('f')
return pos + 3
end

_, sig, exp = sigexp(x)
numdig = ndigits(sig)
if numdig > precision + 1
diff = numdig - precision - 1
denominator = oftype(sig, 10)^Int32(diff)
sig = div(sig, denominator, RoundNearest)
exp += diff
end
numdig = ndigits(sig)
if numdig > precision + 1
# leading digit 9 rounded up to 10
sig ÷= 10
exp += 1
end
digits = codeunits(string(sig))
numdig = length(digits)
if trimtrailingzeros
i = numdig
while digits[i] == UInt8('0')
exp += 1
numdig -= 1
i -= 1
end
end
pexp = exp + numdig - 1
buf[pos] = digits[1]
pos += 1
if numdig > 1 || (precision > 0 && !trimtrailingzeros) || hash
buf[pos] = decchar
pos += 1
end
for i in 2:numdig
buf[pos] = digits[i]
pos += 1
end
if !trimtrailingzeros
for _ in numdig:precision
buf[pos] = UInt8('0')
pos += 1
end
end
buf[pos] = expchar
pos += 1
if pexp < 0
@inbounds buf[pos] = UInt8('-')
pos += 1
pexp = -pexp
else
@inbounds buf[pos] = UInt8('+')
pos += 1
end
if pexp >= 1000
c = (pexp % 100) % UInt8
@inbounds d100 = Base._dec_d100[div(pexp, 100) + 1]
@inbounds buf[pos] = d100 % UInt8
@inbounds buf[pos + 1] = (d100 >> 0x8) % UInt8
@inbounds d100 = Base._dec_d100[c + 1]
@inbounds buf[pos + 2] = d100 % UInt8
@inbounds buf[pos + 3] = (d100 >> 0x8) % UInt8
pos += 4
elseif pexp >= 100
c = (pexp % 10) % UInt8
@inbounds d100 = Base._dec_d100[div(pexp, 10) + 1]
@inbounds buf[pos] = d100 % UInt8
@inbounds buf[pos + 1] = (d100 >> 0x8) % UInt8
@inbounds buf[pos + 2] = UInt8('0') + c
pos += 3
else
@inbounds d100 = Base._dec_d100[pexp + 1]
@inbounds buf[pos] = d100 % UInt8
@inbounds buf[pos + 1] = (d100 >> 0x8) % UInt8
pos += 2
end
return pos
end

macro d_str(s, flags...) parse(Dec64, s) end
macro d32_str(s, flags...) parse(Dec32, s) end
macro d64_str(s, flags...) parse(Dec64, s) end
Expand Down
Loading
Loading