#!/bin/bash
# co2do.sh
# by hackerb9, February 2026
#
# Given a .CO machine language file for a TRS-80 Model 100 (or
# similar), creates a .DO file containing a two-stage BASIC loader
# which will install the .CO to the correct address and start it.

# This program uses a small assembly routine to decode and write the
# bytes to memory, which is a hundred times faster than doing it in
# BASIC. (For details see decode.asm and embedasm which writes the
# decode() function below.)
#
# USAGE: co2do.sh FOO.CO FOO.DO
#        Transfer FOO.DO to M100.
#        On M100: run "FOO"
#
# Features:
#
# * BASIC loader writes .CO to memory very quickly (less than a second).
#
# * Inspired by Stephen Adolph's efficient encoding scheme which
#   increases storage size by at most 2x + k (where k is approx. 500),
#   and on average closer to 1.3x + k.
#
# * Works on all of the Kyotronic Sisters:
#     Kyocera Kyotronic 85, TRS-80 Model 100/102, Tandy 200, Olivetti M10,
#     NEC PC-8201/8201A/8300.
#
# * BASIC Automatically CLEARs the correct space and CALLs the program.
#
# * Uses .CO header to detect where to POKE, length mismatch, and CALL addr.
#
# * As a special bonus, if you use the -t option, it will display a
#   Unicode version of the program instead of writing to a .DO file.
#   (Requires the tandy-200.charmap file from hackerb9/tandy-locale.)

# Todo:
#       * Prevent writing to bad parts of RAM.
#         (E.g., POKE Q where Q<HIMEM or Q>=MAXRAM).
#

set +o errexit
set +o pipefail

function usage() {
    cat <<EOF
Usage: co2do.sh <INPUT.CO> [OUTPUT.DO]
EOF
}

function main() {
    declare -i TOP LEN EXE
    declare -i ROT=136		# Rotate character set for efficiency
    TOP=$1+$2*256
    LEN=$3+$4*256
    EXE=$5+$6*256
    shift 6
    if [[ ! $quiet ]]; then
	cat<<-EOF >&2
		TOP: $TOP
		END: $((TOP+LEN))
		EXE: $EXE
		ROT: $ROT
	EOF
    fi
    if [[ $# -ne $LEN ]]; then
	echo "Warning filesize mismatch! $# bytes found after header, not $LEN." >&2
	if [[ $# -gt $LEN ]]; then
	    echo "Truncating output to $LEN bytes." >&2
	    set -- "${@:1:LEN}"
	fi
    fi
    if (( TOP<62960 && TOP>32767 )); then clear=", $TOP"; else clear=""; fi
    # We'll be POKEing into and executing the fast M/L decoder
    # directly out of the space allocated for DC$, so we need to make
    # sure it doesn't move around in memory as BASIC is wont to do.
    cat <<-EOF
	10 MAXFILES=0: CLEAR 256$clear: NM$="${coname}"
	20 TP=$((TOP)): LN=$LEN: EX=$((EXE)): RT=$((ROT)): ID=PEEK(1)
	30 DC$=SPACE\$(255)
	40 GOSUB 12000: GOSUB 13000
	50 IF FRE(0) < LN THEN ?"Insufficient space to save "NM$: Q$=CHR\$(34): C$=",":?"Please try NEW then SAVEM "Q$;NM$;Q$;C$;TP;C$;TP+LN;C$;EX: END
	60 ?"Saving "NM$;:IF ID<>148 THEN SAVEM NM$,TP,TP+LN,EX ELSE BSAVE NM$,TP,LN,EX
	80 ?:?"Running "NM$:IF ID<>148 THEN CALL EX ELSE EXEC EX
	90 END
	EOF
    emitmldecode
    emitbasicdecode
    printf "14000 '${coname}"
    emitdata "$@"
    emitnecvarptr
    true			# XXX todo: call external warnmem util
    return			# Return result from warnmem
}

function emitdata() {
    local -i linenum=14000
    local -i v
    for v; do
	v=(v+$ROT)%256
	if (( count++ % 120 == 0 )); then
	    printf '\n%d DATA"' $((linenum+=10))
	fi
	# Escape Quote (34), Bang (33), DEL (127), and ctrl chars (except Tab)
	if (( (v<35 && v!=9 && v!=32) || v==127 )); then
	    v=v+128
	    printf "!"
	fi
	printf -v x "%x" $v
	printf "\x$x"
    done
    # End of DATA marked by '!!'
    printf '\n%d DATA !!\n' $((linenum+=10))
}

function emitmldecode() {
    # BASIC wrapper for the M/L routine to decode bang-code DATA strings.
    # Sets the BASIC variable DX to the address of the routine.
    # (See routine starting at lines 13000 for example usage.)
    #
    # NOTA BENE: DX is actually the address of DC$'s buffer. The
    # machine language is POKEd directly into and executed from the
    # string, so manipulation of other string variables in BASIC can
    # foul it up. (READ is okay as it merely sets variables to point
    # to the program line.)

    # USAGE: Initialize by calling DX with A set to the rotation
    # (charset offset) and HL set to the destination address for the
    # decoded data. In a loop, read encoded data into P$ and call DX+9
    # with HL set to the VARPTR(P$). Re-initialization is not
    # necessary for each P$; each call to DX+9 appends data to the
    # destination. While probably not useful, at the end of each call
    # the length of P$ is set to the number of decoded bytes written.

    cat <<-EOF
	12000 'Fast loader
	12010 IF PEEK(1)<>148 THEN DC=VARPTR(DC$): ELSE VY$="DC$": GOSUB 40000: DC=L+256*H
	12015 DX=PEEK(DC+1)+256*PEEK(DC+2): Q=DX
	12020 READ X
	12030 IF X=-1 THEN 12080
	12040 IF X=-7 THEN READ L: READ H: A=H*256+L+DX: X=INT(A/256): L=A-X*256: POKE Q, L: Q=Q+1
	12050 POKE Q, X
	12060 Q=Q+1
	12070 GOTO 12020
	12080 POKE DC, Q-DX: 'Set length of DC$
	12090 RETURN
	EOF
    decode
}

function emitbasicdecode() {
    # Calls M/L routine to decode and install each DATA string.
    # DX is the start address of decode.asm, set by emitmlcode().
    # RT is the amount of rotation for the character set.
    # TP is the destination address ("TOP"), set by main().
    cat <<-"EOF"
	13000 REM ?"Loading "NM$" ("LN" bytes) at "TP
	13020 IF PEEK(1)<>148 THEN CALL DX, RT, TP: ELSE H=INT(TP/256): L=TP-H*256: POKE 63911, RT: POKE 63912, L: POKE 63913, H: EXEC DX
	13030 READP$: IF P$="!!" GOTO 13080
	13040 IF PEEK(1)<>148 THEN P=VARPTR(P$): CALL DX+9, 0, P: ELSE VY$="P$": GOSUB 40030: POKE 63912, L: POKE 63913, H: EXEC DX+9
	13060 GOTO 13030
	13080 'All done
	13090 ?:RETURN
	EOF
}

emitnecvarptr() {
    # VARPTR for NEC 8201/8300 by Gary Weber of web8201.net
    # Entry: VY$ must contain the name of variable of interest
    # Exit: H & L contain variable's address, TY contains type
    # Example:
    #   100 A$="This is a sample string."
    #   110 VY$="A$":GOSUB 40000
    #   120 PRINT "VARPTR of ";VY$;" is ";L+(H*256)
    # NOTES From hackerb9: 
    # * Minor modifications to use less RAM and string space. --b9 2026
    # * M/L routine placed at 64448 for performing VARPTR.
    # * Address 64457 is an 8-byte, NULL terminated copy of VY$.
    # * This routine passes 64457 in HL when it EXECs 64448.
    # * The result is read out of HL (addresses 63912,3) and A (63911).
    cat <<-EOF
	39999 'VARPTR by Gary Weber for NEC 8201/8300
	40000 A=64448
	40010 POKEA,205:POKEA+1,175:POKEA+2,73:POKEA+3,235
	40020 POKEA+4,58:POKEA+5,139:POKEA+6,250:POKEA+8,201
	40030 IFVY$=""THENPRINT"VY$ not defined!":STOP
	40035 A=64457 ' HL register for EXEC = 201/251
	40040 FORH=1TOLEN(VY$):POKEA,ASC(MID\$(VY$,H,1)):A=A+1:NEXT:POKEA,0:POKE64464,0
	40050 POKE63912,201:POKE63913,251:EXEC64448
	40060 L=PEEK(63912):H=PEEK(63913):TY=PEEK(63911)
	40070 RETURN
	EOF
}

### BEGIN AUTOMATIC EMBED ### decode.asm.lst 2026-03-22 09:54:33
#### DO NOT EDIT! Added automatically by embedasm.
decode() {
    cat <<-'EOT'
	12500 DATA 50,-7,42,0,34,-7,7,0,201,0,0,229,70,14,0,35,94,35,86,42,-7,7,0,235,120,167,202,-7,52,0,126,254,33,194,-7,40,0,5,202,-7,52,0,35,126,238,128,235,214,0,119,235,35,19,12,5,194,-7,26,0,235,34,-7,7,0,235,225,113,201,-1
	EOT
}
### END AUTOMATIC EMBED ###

{    # CLI args
    tandycharset=cat
    quiet=""
    while getopts 'qt' name; do
	case $name in
	    q) quiet=: ;;
	    t) tandycharset="iconv -f $(dirname $0)/tandy-200.charmap"  ;;
	    *) exit 1 ;;
	esac
    done
    shift $(($OPTIND - 1))

    if [ -z "$1" ]; then usage; exit 1; fi
    if [[ "$1" == "-" ]]; then shift; set -- /dev/stdin "$@"; fi
    if [ ! -r "$1" ]; then
	echo "'$1' is not readable"
	exit 1
    else
	input="$1"
    fi

    output=/dev/stdout
    if [ "$2" ]; then
	output="$2"
	shift
    elif [[ $tandycharset == "cat" ]]; then
	if [[ $input == *.CO ]]; then
	    output=${input%.CO}.DO
	else
	    output=${input%.*}.do
	fi
    fi
    coname=$(basename "$output")
    coname=${coname%%.*}
    coname=${coname:0:6}
    coname=${coname@U}.CO
}

outputfilter() {
    # Add carriage returns since NEC PC-8201 uses Mac (or DOS?) line endings.
    # Append ^Z at end of file to signal M100 that file is finished.
    sed 's/$/\r/g'
    printf '\x1A'
}

main $(od -t u1 -v -An "$input") | $tandycharset | outputfilter > "$output"
err=${PIPESTATUS[0]}
if (( err != 0 )); then
    echo "Exiting with error $err" >&2
    exit $err
fi

if [[ $tandycharset == "cat" && -z $quiet ]]; then
    outnodo=${output##*/}
    outnodo=${outnodo%.[Dd][Oo]}
    cat <<-EOF
	Now transfer $output to your Model-T
	and in BASIC type
	    run "${outnodo:0:6}"
	EOF
fi

exit $err