utf8.inc

一个用纯汇编 写的操作系统 源代码 是用 TASM 编译器写的

; UTF-8

macro du [args]
 { common local current,char
   virtual at 0
    db args
    count = $
   end virtual
   current = 0
   while current < count
    virtual at 0
    db args
    load char byte from current
    wide = char
    current = current + 1
    if char > 0C0h
     if char < 0E0h
      wide = char and 11111b
      load char byte from current
      wide = wide shl 6 + (char and 111111b)
      current = current + 1
     else if char < 0F0h
      wide = char and 1111b
      load char byte from current
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+1
      wide = wide shl 6 + (char and 111111b)
      current = current + 2
     else if char < 0F8h
      wide = char and 111b
      load char byte from current
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+1
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+2
      wide = wide shl 6 + (char and 111111b)
      current = current + 3
     else if char < 0FCh
      wide = char and 11b
      load char byte from current
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+1
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+2
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+3
      wide = wide shl 6 + (char and 111111b)
      current = current + 4
     else
      wide = char and 1
      load char byte from current
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+1
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+2
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+3
      wide = wide shl 6 + (char and 111111b)
      load char byte from current+4
      wide = wide shl 6 + (char and 111111b)
      current = current + 5
     end if
    end if
    end virtual
    if wide < 10000h
     dw wide
    else
     dw 0D7C0h + wide shr 10,0DC00h or (wide and 3FFh)
    end if
   end while }

struc du [args]
 { common label . word
   du args }