urlencode.asm

org 100h

start:
    lea esi, [sStr]
    lea edi, [sOut]

urlEncode:
  .processChar:
    mov al, [esi]

    ; Null terminated string
    cmp al, 00h
    je .done

    ; Allowed special chars
    cmp al, '-'
    je .nextChar
    cmp al, '.'
    je .nextChar
    cmp al, '_'
    je .nextChar
    cmp al, '~'
    je .nextChar

    ; AL >= '0' && AL <= '9' is valid otherwise encode
    cmp al, '0'
    jb .encode
    cmp al, '9'
    jbe .nextChar

    ; AL >= 'A' && AL <= 'Z' is valid otherwise encode
    cmp al, 'A'
    jb .encode
    cmp al, 'Z'
    jbe .nextChar

    ; AL >= 'a' && AL <= 'z' is valid otherwise encode
    cmp al, 'a'
    jb .encode
    cmp AL, 'z'
    jbe .nextChar

  .encode:
    mov [edi], byte '%' ; write % (%20, %1F, etc.)
    inc edi

    push esi
    and eax, 0xFF       ; only use AL
    call int2Hex        ; Get hex string to represent byte

    mov ax, [esi]       ; copy 2 char hex string
    mov [edi], ax
    add edi, 2

    pop esi

    inc esi
    jmp .processChar

  .nextChar:
    mov [edi], al
    inc esi
    inc edi
    jmp .processChar

  .done:
    mov [edi], byte '$' ; Add null terminator to output string

    mov bx, cs ;print output
    mov ds, bx
    lea dx, [sOut]
    mov ah, 09h
    int 21h

    ret

; ----------------------------------------------
; Converts a 32bit integer value to a hex string
;
; input:
;    eax = input value
;
; returns:
;    eax = length of string
;    esi = address of null-terminated string
;
; trashes: none
int2Hex:
    push ebx
    push ecx

    xor ecx, ecx      ; zero out length counter
    lea esi, [cgi86.tmpStr+9] ; Get address of output

  .nextDigit:
    mov ebx, eax
    and bl, 0x0F  ; Go one hex digit at a time

    cmp bl, 0x0A
    jb .isNum

    add bl, 'A'-0x0A ; convert to ASCII letter
    jmp .addDigit

  .isNum:
    add bl, '0'   ; convert to ASCII number

  .addDigit:
    inc ecx       ; increment char counter
    mov [esi], bl ; Write char to string

    shr eax, 4    ; Check if we're done
    jz .done

    dec esi       ; move to next char
    jmp .nextDigit

.done:
    mov eax, ecx
    pop ecx
    pop ebx
    ret


.data:
sStr: db "-._~abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890", 00h;"!@#$%^&*()", 00h;"Hello World #123", 00h
sOut: resb (sOut-sStr)*3

cgi86.tmpStr: db "4294967295", '$';00h, ; largest possible number we'll return