micro-optimize UUID(::AbstractString) (#36049)

* microoptimize UUID(::AbstractString)

* Update base/uuid.jl

Co-authored-by: Jeff Bezanson <jeff.bezanson@gmail.com>

Co-authored-by: Jeff Bezanson <jeff.bezanson@gmail.com>

Author: KristofferC

base/parse.jl

@@ -89,6 +89,20 @@ function parseint_preamble(signed::Bool, base::Int, s::AbstractString, startpos:
     return sgn, base, j
 end
 
+@inline function __convert_digit(_c::UInt32, base)
+    _0 = UInt32('0')
+    _9 = UInt32('9')
+    _A = UInt32('A')
+    _a = UInt32('a')
+    _Z = UInt32('Z')
+    _z = UInt32('z')
+    a::UInt32 = base <= 36 ? 10 : 36
+    d = _0 <= _c <= _9 ? _c-_0             :
+        _A <= _c <= _Z ? _c-_A+ UInt32(10) :
+        _a <= _c <= _z ? _c-_a+a           : UInt32(base)
+end
+
+
 function tryparse_internal(::Type{T}, s::AbstractString, startpos::Int, endpos::Int, base_::Integer, raise::Bool) where T<:Integer
     sgn, base, i = parseint_preamble(T<:Signed, Int(base_), s, startpos, endpos)
     if sgn == 0 && base == 0 && i == 0
@@ -115,19 +129,10 @@ function tryparse_internal(::Type{T}, s::AbstractString, startpos::Int, endpos::
            base == 16 ? div(typemax(T) - T(15), T(16)) :
                         div(typemax(T) - base + 1, base)
     n::T = 0
-    a::Int = base <= 36 ? 10 : 36
-    _0 = UInt32('0')
-    _9 = UInt32('9')
-    _A = UInt32('A')
-    _a = UInt32('a')
-    _Z = UInt32('Z')
-    _z = UInt32('z')
     while n <= m
         # Fast path from `UInt32(::Char)`; non-ascii will be >= 0x80
         _c = reinterpret(UInt32, c) >> 24
-        d::T = _0 <= _c <= _9 ? _c-_0             :
-               _A <= _c <= _Z ? _c-_A+ UInt32(10) :
-               _a <= _c <= _z ? _c-_a+a           : base
+        d::T = __convert_digit(_c, base)
         if d >= base
             raise && throw(ArgumentError("invalid base $base digit $(repr(c)) in $(repr(SubString(s,startpos,endpos)))"))
             return nothing
@@ -145,9 +150,7 @@ function tryparse_internal(::Type{T}, s::AbstractString, startpos::Int, endpos::
     while !isspace(c)
         # Fast path from `UInt32(::Char)`; non-ascii will be >= 0x80
         _c = reinterpret(UInt32, c) >> 24
-        d::T = _0 <= _c <= _9 ? _c-_0             :
-               _A <= _c <= _Z ? _c-_A+ UInt32(10) :
-               _a <= _c <= _z ? _c-_a+a           : base
+        d::T = __convert_digit(_c, base)
         if d >= base
             raise && throw(ArgumentError("invalid base $base digit $(repr(c)) in $(repr(SubString(s,startpos,endpos)))"))
             return nothing

base/uuid.jl

@@ -30,35 +30,56 @@ end
 
 UInt128(u::UUID) = u.value
 
-let groupings = [1:8; 10:13; 15:18; 20:23; 25:36]
-    global UUID
-    function UUID(s::AbstractString)
-        s = lowercase(s)
-
-        if !occursin(r"^[0-9a-f]{8}(?:-[0-9a-f]{4}){3}-[0-9a-f]{12}$", s)
-            throw(ArgumentError("Malformed UUID string: $(repr(s))"))
-        end
+let
+@noinline throw_malformed_uuid(s) = throw(ArgumentError("Malformed UUID string: $(repr(s))"))
+@inline function uuid_kernel(s, i, u)
+    _c = UInt32(@inbounds codeunit(s, i))
+    d = __convert_digit(_c, UInt32(16))
+    d >= 16 && throw_malformed_uuid(s)
+    u <<= 4
+    u | d
+end
 
-        u = UInt128(0)
-        for i in groupings
-            u <<= 4
-            d = s[i] - '0'
-            u |= 0xf & (d - 39*(d > 9))
-        end
-        return UUID(u)
+global UUID
+function UUID(s::AbstractString)
+    u = UInt128(0)
+    ncodeunits(s) != 36 && throw_malformed_uuid(s)
+    for i in 1:8
+        u = uuid_kernel(s, i, u)
+    end
+    @inbounds codeunit(s, 9) == UInt8('-') || @goto error
+    for i in 10:13
+        u = uuid_kernel(s, i, u)
     end
+    @inbounds codeunit(s, 14) == UInt8('-') || @goto error
+    for i in 15:18
+        u = uuid_kernel(s, i, u)
+    end
+    @inbounds codeunit(s, 19) == UInt8('-') || @goto error
+    for i in 20:23
+        u = uuid_kernel(s, i, u)
+    end
+    @inbounds codeunit(s, 24) == UInt8('-') || @goto error
+    for i in 25:36
+        u = uuid_kernel(s, i, u)
+    end
+    return Base.UUID(u)
+    @label error
+    throw_malformed_uuid(s)
 end
+end
+
 
 let groupings = [36:-1:25; 23:-1:20; 18:-1:15; 13:-1:10; 8:-1:1]
     global string
     function string(u::UUID)
         u = u.value
         a = Base.StringVector(36)
         for i in groupings
-            a[i] = hex_chars[1 + u & 0xf]
+            @inbounds a[i] = hex_chars[1 + u & 0xf]
             u >>= 4
         end
-        a[24] = a[19] = a[14] = a[9] = '-'
+        @inbounds a[24] = a[19] = a[14] = a[9] = '-'
         return String(a)
     end
 end

stdlib/UUIDs/test/runtests.jl

@@ -62,3 +62,14 @@ u4 = uuid4()
 Random.seed!(Random.GLOBAL_RNG, 10)
 @test u1 != uuid1()
 @test u4 != uuid4()
+
+@test_throws ArgumentError UUID("22b4a8a1ae548-4eeb-9270-60426d66a48e")
+@test_throws ArgumentError UUID("22b4a8a1-e548a4eeb-9270-60426d66a48e")
+@test_throws ArgumentError UUID("22b4a8a1-e548-4eeba9270-60426d66a48e")
+@test_throws ArgumentError UUID("22b4a8a1-e548-4eeb-9270a60426d66a48e")
+str = "22b4a8a1-e548-4eeb-9270-60426d66a48e"
+@test UUID(uppercase(str)) == UUID(str)
+
+for r in rand(UInt128, 10^3)
+    @test UUID(r) == UUID(string(UUID(r)))
+end