eliminate copy for Unicode.normalize of substring (#38381)

eliminate copy for Unicode.normalize of substring

Author: stevengj

base/strings/unicode.jl

@@ -145,7 +145,7 @@ const UTF8PROC_STRIPMARK = (1<<13)
 
 utf8proc_error(result) = error(unsafe_string(ccall(:utf8proc_errmsg, Cstring, (Cssize_t,), result)))
 
-function utf8proc_map(str::String, options::Integer)
+function utf8proc_map(str::Union{String,SubString{String}}, options::Integer)
     nwords = ccall(:utf8proc_decompose, Int, (Ptr{UInt8}, Int, Ptr{UInt8}, Int, Cint),
                    str, sizeof(str), C_NULL, 0, options)
     nwords < 0 && utf8proc_error(nwords)

stdlib/Unicode/test/runtests.jl

@@ -6,7 +6,7 @@ using Unicode: normalize, isassigned
 
 @testset "string normalization" begin
     # normalize (Unicode normalization etc.):
-    @test normalize("\u006e\u0303", :NFC) == "\u00f1"
+    @test normalize("\u006e\u0303", :NFC) == "\u00f1" == normalize(SubString("ab\u006e\u0303cd",3,4), :NFC)
     @test "\u006e\u0303" == normalize("\u00f1", :NFD)
     @test normalize("\ufb00", :NFC) != "ff"
     @test normalize("\ufb00", :NFKC) == "ff"