Improve performance of convert to UTF16Str and UTF32Str from Vector of 32-bit values

Author: ScottPJones

src/utf16.jl

@@ -465,7 +465,7 @@ end
 #convert(::Type{T},  str::MaybeSub{T}) where {T<:Str{<:Union{UCS2_CSEs, UTF32_CSEs}}} = str
 convert(::Type{<:Str{UTF16CSE}}, str::MaybeSub{<:Str{<:UCS2_CSEs}}) = Str(UTF16CSE, str.data)
 
-function convert(::Type{<:Str{UTF16CSE}}, dat::AbstractArray{UInt16})
+function convert(::Type{<:Str{UTF16CSE}}, dat::AbstractArray{C}) where {C<:Union{UInt16,UInt32}}
     is_empty(dat) && return empty_utf16
     len, flags, num4byte = unsafe_check_string(dat, 1, lastindex(dat))
     # Optimize case where no surrogate characters

src/utf32.jl

@@ -216,7 +216,20 @@ end
 
 function convert(::Type{<:Str{UTF32CSE}}, dat::Vector{<:Union{UInt32,Int32,Text4Chr}})
     is_valid(UTF32Str, dat) || strerror(StrErrors.INVALID)
-    @preserve dat Str(UTF32CSE, _copysub(pointer(dat), length(dat)))
+    Str(UTF32CSE, _copysub(dat))
+end
+
+function is_valid(::Type{<:Str{UTF32CSE}}, dat::Vector{<:Union{UInt32,Int32,Text4Chr}})
+    @preserve dat begin
+        pnt = pointer(dat)
+        fin = pnt + sizeof(dat)
+        while pnt < fin
+            ch = get_codeunit(pnt)
+            (!is_surrogate_codeunit(ch) && ch <= 0x10ffff) || return false
+            pnt += 4
+        end
+    end
+    true
 end
 
 # Not sure this is valid anymore, want to avoid type piracy