Benchmark test of length, validation, convert

perftest.jl

VERSION < v"0.4-" && (typealias AbstractString String)
function tst{T <: AbstractString}(str::T, max::Int)
    local l
    for i=1:max
	l = length(str)
    end
    l
end
function cvt8{T <: AbstractString}(str::T, max::Int)
    local out
    for i=1:max
	out = utf8(str)
    end
    out
end
function cvt16{T <: AbstractString}(str::T, max::Int)
    local out
    for i=1:max
	out = utf16(str)
    end
    out
end
function cvt32{T <: AbstractString}(str::T, max::Int)
    local out
    for i=1:max
	out = utf32(str)
    end
    out
end

function tstva{T <: AbstractString}(str::T, max::Int)
    local l
    for i=1:max
	l = is_valid_ascii(str)
    end
    l
end
function tstv8{T <: AbstractString}(str::T, max::Int)
    local l
    for i=1:max
	l = is_valid_utf8(str)
    end
    l
end
function tstv16{T <: AbstractString}(str::T, max::Int)
    local l
    for i=1:max
	l = is_valid_utf16(str)
    end
    l
end
function tstv32{T <: AbstractString}(str::T, max::Int)
    local l
    for i=1:max
	l = is_valid_utf32(str)
    end
    l
end

function tstall(flg::Bool, str::String, n::Int, strUTF8::UTF8String, strUTF16::UTF16String, strUTF32::UTF32String)
    println("\n\n$str: Looping $n times, length=$(length(strUTF32))")
    println("UTF-8: $(sizeof(strUTF8)), UTF-16: $(sizeof(strUTF16)), UTF-32: $(sizeof(strUTF32))\n")
    tstlen(n, strUTF8, strUTF16, strUTF32)
    tstval(n, strUTF8, strUTF16, strUTF32)
    flg && tstcvt(n, strUTF8, strUTF16, strUTF32)
    ""
end

function tstval(n::Int, strUTF8::UTF8String, strUTF16::UTF16String, strUTF32::UTF32String)
    print("UTF-8 valid:   ")
    @time tstv8(strUTF8, n)
    print("UTF-16 valid:  ")
    @time tstv16(strUTF16, n)
    try
        is_valid_utf32(strUTF32)
        print("UTF-32 valid:  ")
        @time tstv32(strUTF32, n)
    catch ;
    end
    println()
    ""
end

function tstlen(n::Int, strUTF8::UTF8String, strUTF16::UTF16String, strUTF32::UTF32String)
    print("UTF-8 length:  ")
    @time tst(strUTF8,n)
    print("UTF-16 length: ")
    @time tst(strUTF16,n)
    print("UTF-32 length: ")
    @time tst(strUTF32,n)
    println()
    ""
end

function tstascii(n::Int, str::ASCIIString)
    println("\nASCII: Looping $n times, length=$(length(str))")
    print("length:           ")
    @time tst(str,n)
    print("is_valid_ascii:   ")
    @time tstva(str, n)
    print("Convert to UTF-8: ")
    @time cvt8(str,n)
    print("Convert to UTF-16:")
    @time cvt16(str,n)
    print("Convert to UTF-32:")
    @time cvt32(str,n)
    ""
end

function tstcvt(n::Int, strUTF8::UTF8String, strUTF16::UTF16String, strUTF32::UTF32String)
    print("UTF-8 convert to UTF-16:  ")
    @time cvt16(strUTF8,n)
    print("UTF-8 convert to UTF-32:  ")
    @time cvt32(strUTF8,n)
    print("UTF-16 convert to UTF-8:  ")
    @time cvt8(strUTF16,n)
    print("UTF-16 convert to UTF-32: ")
    @time cvt32(strUTF16,n)
    print("UTF-32 convert to UTF-8:  ")
    @time cvt8(strUTF32,n)
    print("UTF-32 convert to UTF-16: ")
    @time cvt16(strUTF32,n)
    ""
end

function tstsiz(flg,n,strAscii,strA_UTF8,strL_UTF8,str2_UTF8,str3_UTF8,str4_UTF8,strS_UTF8)
    strA_UTF16 = utf16(strA_UTF8)
    strL_UTF16 = utf16(strL_UTF8)
    str2_UTF16 = utf16(str2_UTF8)
    str3_UTF16 = utf16(str3_UTF8)
    str4_UTF16 = utf16(str4_UTF8)
    strS_UTF16 = utf16(strS_UTF8)

    strA_UTF32 = utf32(strA_UTF8)
    strL_UTF32 = utf32(strL_UTF8)
    str2_UTF32 = utf32(str2_UTF8)
    str3_UTF32 = utf32(str3_UTF8)
    str4_UTF32 = utf32(str4_UTF8)
    strS_UTF32 = utf32(strS_UTF8)

    tstascii(n,strAscii)
    tstall(flg,"ASCII:",n,strA_UTF8,strA_UTF16,strA_UTF32)
    tstall(flg,"Latin1:",n,strL_UTF8,strL_UTF16,strL_UTF32)
    tstall(flg,"2-byte:",n,str2_UTF8,str2_UTF16,str2_UTF32)
    tstall(flg,"3-byte:",n,str3_UTF8,str3_UTF16,str3_UTF32)
    tstall(flg,"4-byte:",n,str4_UTF8,str4_UTF16,str4_UTF32)
    tstall(flg,"Surrogates:",n,strS_UTF8,strS_UTF16,strS_UTF32)
end

function dotest(n,flg::Bool = true)
    # Create some ASCII, UTF8, UTF16, and UTF32 strings
    baseascii = "abcdefghijklmnop\uff"
    binstr    = b"abcdefghijk\xc3\xbf\xdf\xbf\xe7\xbf\xbf\xed\xa0\x80\xed\xb0\x80\xed\xaf\xbf\xed\xbf\xbf"
    strAscii  = "abcd"
    strA_UTF8 = baseascii[1:4]
    strL_UTF8 = "ab\uff\uff"
    str2_UTF8 = "ab\uff\u7ff"
    str3_UTF8 = "a\uff\u7ff\u7fff"
    str4_UTF8 = "a\uff\u7fff\U7ffff"
    strS_UTF8 = UTF8String(b"\xc3\xbf\xdf\xbf\xe7\xbf\xbf\xed\xa0\x80\xed\xb0\x80")

    strAscii = "abcdefgh"
    strA_UTF8 = baseascii[1:8]
    strL_UTF8 = "abcdef\uff\uff"
    str2_UTF8 = "abcd\uff\uff\u7ff\u7ff"
    str3_UTF8 = "abcd\uff\uff\u7fff\u7fff"
    str4_UTF8 = "abcd\uff\u7ff\u7fff\U7ffff"
    strS_UTF8 = UTF8String(b"abcd\xc3\xbf\xdf\xbf\xe7\xbf\xbf\xed\xa0\x80\xed\xb0\x80")
    tstsiz(flg,n,strAscii,strA_UTF8,strL_UTF8,str2_UTF8,str3_UTF8,str4_UTF8,strS_UTF8)

    strAscii = "abcdefghijklmnop"
    strA_UTF8 = baseascii[1:16]
    strL_UTF8 = "abcdefghijk\uff\uff\uff\uff\uff"
    str2_UTF8 = "abcdefghijk\uff\uff\uff\u7ff\u7ff"
    str3_UTF8 = "abcdefghijk\uff\uff\uff\u7fff\u7fff"
    str4_UTF8 = "abcdefghijk\uff\u7ff\u7fff\U7ffff\U0fffff"

    for i=1:9
	strAscii  ^= 4
	strA_UTF8 ^= 4
	strL_UTF8 ^= 4
	str2_UTF8 ^= 4
	str3_UTF8 ^= 4
	str4_UTF8 ^= 4
	binstr    = vcat(binstr,binstr,binstr,binstr)
	tstsiz(flg,n,strAscii,strA_UTF8,strL_UTF8,str2_UTF8,str3_UTF8,str4_UTF8,UTF8String(binstr))
    end
end
""