BatAs: a 8086 assembler written in MSDOS batch

BatAs

This is a 8086 assembler written in MSDOS batch. It depends on just two utilities: RPN.COM and APPFB.COM, the rest is completely in batch.

The source for the utilities is provided in DEBUG.EXE script form (RPN.DBG and APPFB.DBG) and they can be "built" by invoking MAKE.BAT. For your convenience the utilities have also been provided in binary form. You can verify their integrity using any 8086 disassembler. The same utilities are also provided in form accepted by AS.BAT (RPN.S and APPFB.S). They don't produce identical binaries, but they do produce identical assembly listings, as some instructions have more than one possible encoding.

Usage

AS [/P <Path\To\BatAs>] <Input.S> <Output.COM>

If running from a directory other than where AS.BAT is located, you need to supply the /P switch.

Due to the nature of the assembler (as described below), for larger programs you may wish to increase the environment block size of your interpreter (see the /E switch in the COMMAND.COM manual).

Language

Due to there being nearly no string processing functions in MSDOS batch, the input to the assembler is actually also batch script full of CALL commands that execute .BAT files from this directory.

For example, to move a value from one register to another, you can write:

CALL MOV BX AX

Register names are case sensitive. The CALL is necessary because otherwise our batch script terminates as soon as MOV.BAT does.

The syntax of various addressing modes is as follows:

CALL MOV [SI] AX
CALL MOV [BX+SI] AX
CALL MOV [BX+SI+BYTE] 123 AX
CALL MOV [BX+SI+WORD] 12345 AX
CALL MOV [WORD] 12345 AX

An immediate value in arithmetic opcodes has to be prefixed with the words BYTE or WORD, depending on the size of the operand. In 8086, arithmetic instructions allow an encoding where a 8-bit immediate is sign extended and then used in a 16-bit operation. Such immediates are to be prefixed with BTW:

CALL ADD AL BYTE 123
CALL ADD AX WORD 12345
CALL ADD AX BTW 123

Unconditional jump (JMP) defaults to a local jump with a 8-bit displacement. To override, use either JMP NEAR for a 16-bit displacement or JMP FAR for a far jump.

"CALL" is a batch builtin so the CALL 8086 mnemonic had to be spelled "CLL".

Labels

Unlike DEBUG.EXE, this assembler supports arbitrary backward- and forward-referencing labels. You can create a label with LABEL.BAT and further use it as a regular batch variable:

MOV AX [WORD] %Data%
CALL LABEL Test
CALL JMP %Test%
CALL LABEL Data
CALL DW 0

Example

Here's an example hello-world program:

REM Place the string in DS:DX
CALL PUSH CS
CALL POP DS
CALL MOV DX WORD %String%

REM AH=9 Output string
CALL MOV AH BYTE 9
CALL INT 33

REM Terminate program
CALL INT 32

CALL LABEL String
REM     H  e   l   l   o   ,     w
CALL DB 72 101 108 108 111 44 32 119
REM     o   r   l   d   !  $
CALL DB 111 114 108 100 33 36