embedasm

#!/bin/bash

# Given a shell script in $1 and an assembly listing in $2,
# parse the assembly listing into bytes which will be copied unchanged
# and addresses which must be offset when the program is loaded.

# The assembly listing must be created by 'asmx -l -b0 foo.asm'.
# The shell script must contain two lines which read, 
# 	#### BEGIN AUTOMATIC EMBED  ####
# and
# 	#### END AUTOMATIC EMBED  ####

begin="### BEGIN AUTOMATIC EMBED ###"
end='### END AUTOMATIC EMBED ###'
main() {
    [[ $# -ge 2 ]] || die "Usage: $0 co2do decode.asm.lst"
    script="$1"
    list="$2"
    [[ -r "$script" ]] || die "'$script' to modify is not readable"
    [[ -r "$list" ]] || die "'$list' is not readable. Please create with asmx -l -b0 ${list%.lst}"
    grep -sq "$begin" "$script" \
	&& grep -sq "$end" "$script" \
	    || die "The shell script '$script' must contain the two lines:\n\n"\
		   "\t$begin\n" "\t$end\n"

    # Sed's C command requires every line of the replacement text to
    # end with a \ before the newline, EXCEPT the last line.
    automaticembed=$(cat <<EOF | sed 's/$/\\/; $ s/\\/\n/'
$begin $(TZ=GMT stat -c"%n %y" $list | sed -r 's/\.[0-9]{9} \+0000//')
#### DO NOT EDIT! Added automatically by $(basename "$0").
${list%%.*}() {
    cat <<-'EOT'
	$(coco "$list")
	EOT
}
$end
EOF
    )

    sed -i "/$begin/,/$end/ c $automaticembed"  "$script"
}

function coco() {
    # coco: 
    # Given the filename of an assembly listing (from `asmx -l -b0`), 
    # output the bytes as comma separated decimal in a DATA statement.
    # Addresses that will need to be relocated are prefixed with -7.
    # End of data is signaled by -1.
    local listfile="$1"
    echo -n "12500 DATA "
    cat "$listfile"  | parselst
    echo -1
    return
}

function parselst() {
    # Parse an assembly listing on stdin and output data values.
    # Whenever an address is seen, prefix it with -7.
    # This routine is fragile, but it only needs to work with one
    # assembler (asmx) and one program: decode.asm (which see). 
    sed -rn '/^[0-9A-Fa-f]{4} /p' | cut -c 6-19 |
	while read op arg; do
	    case ${#op} in
		0) continue ;;
		2) echo "ibase=16; $op" | bc ;;
		4) # Data for DW looks like a 16-bit opcode.
		    echo "ibase=16; ${op:0:2}" | bc
		    echo "ibase=16; ${op:2:2}" | bc
		    ;;
		*)
		    echo "parselst error: What is op in '$op $arg'?" >&2
	    esac
	    case ${#arg} in
		0) continue ;;
		2) echo "ibase=16; $arg" | bc ;;
		4) # Escape address for relocation with -7
		    echo -7
		    echo "ibase=16; ${arg:0:2}" | bc
		    echo "ibase=16; ${arg:2:2}" | bc
		    ;;
		*)
		    echo "parselst error: What is arg in '$op $arg'?" >&2
	    esac
	done | tr -s '\n' ','
    return
}

die() {
    printf "$(basename $0): ERROR: "
    echo -e "$@" >&2
    exit 1
}
    

main "$@"