UTF8 conversion routines test

utf8.swift

#!/usr/bin/env xcrun swift -O3
//
//  utf.swift
//  
//
//  Created by Ilya Nikokoshev on 8/22/14.
//
//

import Foundation

var value:UInt32 = 0x1FFFF

// The original.
func utf8_1(inout buffer: [UInt8]) {
    // value is of type UInt32
    
    if value <= 0x007F {
        buffer.append(UInt8(value))
    }
    else if 0x0080 <= value && value <= 0x07FF {
        buffer.append(UInt8(value / 64 + 192))
        buffer.append(UInt8(value % 64 + 128))
    }
    else if (0x0800 <= value && value <= 0xD7FF) || (0xE000 <= value && value <= 0xFFFF) {
        buffer.append(UInt8(value / 4096 + 224))
        buffer.append(UInt8((value % 4096) / 64 + 128))
        buffer.append(UInt8(value % 64 + 128))
    }
    else {
        buffer.append(UInt8(value / 262144 + 240))
        buffer.append(UInt8((value % 262144) / 4_096 + 128))
        buffer.append(UInt8((value % 4096) / 64 + 128))
        buffer.append(UInt8(value % 64 + 128))
    }
}

// Universal for any 32-bit number
func encode8(inout buffer:[UInt8]) {
    var result:[UInt8] = []
    var current = value
    
    func extract(bits:UInt32, plus:UInt32) {
        result += [UInt8(current % (1 << bits) | plus)]
        current >>= bits
    }
    
    // The ASCII set takes one byte.
    if current < 1 << 7 {
        extract(7, 0)
    } else {
        var factor:UInt32 = 1 << 6
        
        // Then we produce bytes of the form 10xxxxxx.
        while current >= factor {
            extract(6, 1 << 7)
            factor >>= 1 
        } 
        
        // And the first byte contains some metadata.
        extract(6, 1 << 8 - 2 * factor)
    }
    
    buffer.extend(reverse(result))
}

func utf8_$(inout buffer: [UInt8]) {
    // Adds a number that fits into UInt8 into buffer.
    let $:UInt32 -> () = { buffer.append(UInt8($0)) }
    
    if value <= 0x007F {
        $(value)
    }
    else if 0x0080 <= value && value <= 0x07FF {
        $(value / 64 + 192)
        $(value % 64 + 128)
    }
    else if (0x0800 <= value && value <= 0xD7FF) || (0xE000 <= value && value <= 0xFFFF) {
        $(value / 4096 + 224)
        $((value % 4096) / 64 + 128)
        $(value % 64 + 128)
    }
    else {
        $(value / 262144 + 240)
        $((value % 262144) / 4_096 + 128)
        $((value % 4096) / 64 + 128)
        $(value % 64 + 128)
    }
}


func utf8_2(inout buffer: [UInt8]) {
    
    if value <= 0x007F {
        buffer.append(UInt8(value))
    }
    else if 0x0080 <= value && value <= 0x07FF {
        buffer.append(UInt8(value &/ 64) &+ 192)
        buffer.append(UInt8(value &% 64) &+ 128)
    }
    else if (0x0800 <= value && value <= 0xD7FF) || (0xE000 <= value && value <= 0xFFFF) {
        buffer.append(UInt8(value &/ 4096) &+ 224)
        buffer.append(UInt8((value &% 4096) &/ 64) &+ 128)
        buffer.append(UInt8(value &% 64 &+ 128))
    }
    else {
        buffer.append(UInt8(value &/ 262144) &+ 240)
        buffer.append(UInt8((value &% 262144) &/ 4096) &+ 128)
        buffer.append(UInt8((value &% 4096) &/ 64) &+ 128)
        buffer.append(UInt8(value &% 64) &+ 128)
    }    
}

var scalar = UnicodeScalar(value)

func utf8std(inout buffer: [UInt8]) {
    buffer.extend(String(scalar).utf8)
}

var buffer_std:[UInt8] = []
utf8std(&buffer_std)

var buffer_test1:[UInt8] = []
var buffer_test2:[UInt8] = []

func time(f:()->()) -> NSTimeInterval {
    let start = NSDate()
    var count = 0
    while count < 1000 {
        count++
        f()
    }
    return -start.timeIntervalSinceNow
}

func test(prefix:String, f:(inout [UInt8]) -> ())  {    
    var buffer:[UInt8] = []
    println("\(prefix) \(time({f(&buffer)}))")
}

test("original  ",      utf8_1)
test("small cast",      utf8_2)
test("with $    ",      utf8_$)
test("universal ",      encode8)
test("standard  ",      utf8std)