Enterprise-128 basic file lister

#!/usr/bin/python
# -*- coding: UTF-8 -*-

from __future__ import division, absolute_import, print_function

"""
First try to display an Enteprise-128 BASIC program
(C)2012 LGB Gábor Lénárt
Nice information sources:
    http://ep128.hu/Ep_Konyv/Tippek.htm
    http://ep128.hu/Ep_Konyv/Gepi_kod_2.htm#342
"""

import sys


tokens = [
    "ALLOCATE", #  0
    "ASK",      #  1
    "AUTO",     #  2
    "CALL",     #  3
    "CAPTURE",  #  4
    "CASE",     #  5
    "CAUSE",    #  6
    "CLEAR",    #  7
    "CLOSE",    #  8
    "CODE",     #  9
    "CONTINUE", # 10
    "COPY",     # 11
    "DATA",     # 12
    "DEF",      # 13
    "DEF",      # 14
    "DELETE",   # 15
    "DIM",      # 16
    "DISPLAY",  # 17
    "DO",       # 18
    "CHAIN",    # 19
    "EDIT",     # 20
    "ELSE",     # 21
    "ELSE IF",  # 22
    "END",      # 23
    "END DEF",  # 24
    "END HANDLER",  # 25
    "END IF",   # 26
    "END SELECT",   # 27
    "END WHEN", # 28
    "ENVELOPE", # 29
    "EXIT",     # 30
    "FOR",      # 31
    "GOSUB",    # 32
    "GOTO",     # 33
    "GRAPHICS", # 34
    "HANDLER",  # 35
    "IMAGE",    # 36
    "IF",       # 37
    "IF",       # 38
    "INPUT",    # 39
    "LET",      # 40
    "LINE",     # 41
    "LIST",     # 42
    "LOAD",     # 43
    "LOOP",     # 44
    "MERGE",    # 45
    "NEW",      # 46
    "NEXT",     # 47
    "NUMERIC",  # 48
    "OPEN",     # 49
    "OPTION",   # 50
    "OK",       # 51
    "OUT",      # 52
    "PLOT",     # 53
    "POKE",     # 54
    "SPOKE",    # 55
    "PRINT",    # 56
    "PROGRAM",  # 57
    "RANDOMIZE",    # 58
    "READ",     # 59
    "REDIRECT", # 60
    "REM",      # 61
    "RENUMBER", # 62
    "RESTORE",  # 63
    "RETRY",    # 64
    "RETURN",   # 65
    "RUN",      # 66
    "SAVE",     # 67
    "SELECT",   # 68
    "SET",      # 69
    "SOUND",    # 70
    "START",    # 71
    "STOP",     # 72
    "INFO",     # 73
    "STRING",   # 74
    "TEXT",     # 75
    "TOGGLE",   # 76
    "TRACE",    # 77
    "TYPE",     # 78
    "VERIFY",   # 79
    "WHEN",     # 80
    "!",        # 81
    "LLIST",    # 82
    "LPRINT",   # 83
    "EXT",      # 84
    "GET",      # 85
    "FLUSH",    # 86
    "LOOK",     # 87
    "PING",     # 88
    "DATE",     # 89
    "TIME",     # 90
    "WAIT",     # 91
    "ON",       # 92
]

signs = [   # TODO: the first (0th ...) element and the ones marked with "?" ...
    " ", "!","\"", "#", "$", "%", "&", "'",
    "(", ")", "*", "+", ",", "-", ".", "/",
    ":", ";", "<", "=", ">","<>","<=",">=",
    "[","\\", "]", "^", "?", "?", "?", "?"
]


def die(msg = None):
    if msg is None: sys.exit(0)
    sys.stderr.write("EXIT ON FATAL: " + msg + "\n")
    sys.exit(1)

def hexbyte(h):
    return hex(h).replace("x", "0")[-2:].upper()

def hexdump(m):
    return " ".join([hexbyte(a) for a in m])


def list_line(line):
    #print(hexdump(line))
    o = str(line[0] | (line[1] << 8))
    o += " " + " " * line[2]    # one space + possibly other spaces based on the nesting level
    p = 3
    while p < len(line):
        b = line[p]
        #print("Pars[" + hex(p) + "] = " + hex(b))
        if b < 0x20: # special sign
            o += signs[b]
            p += 1
        elif b < 0x60:
            b &= 31
            o += str(line[p + 1:p + 1 + b])
            p += b + 1
        elif b == 0x80:
            b = line[p + 1]
            o += "\"" + str(line[p + 2:p + 2 + b]) + "\""
            p += b + 2
        elif b == 0x60:     # tokenized BASIC keyword
            b = line[p + 1]
            o += tokens[b]
            #print("Basic token=" + str(b) + " ~ " + hex(b))
            if b in [81, 61, 36, 12]: # REM or !, IMAGE, DATA marks start of "free" text till the end of the line (?)
                o += str(line[p + 2:])
                break
            else:
                o += " " # should I print a space after a basic token, always?
                p += 2
        elif b in [0xA2, 0xC2]:     # two byte (integer) constant
            o += str(line[p + 1] | (line[p + 2] << 8))
            p += 3
        elif b == 0xC6: # float constant, or so ...
            #print("FLOAT")
            #print(hexdump(line[p + 1:p + 7]))
            s = (hexbyte(line[p + 5]) + hexbyte(line[p + 4]) + hexbyte(line[p + 3]) + hexbyte(line[p + 2]) + hexbyte(line[p + 1]))
            b = line[p + 6] - 0x3F
            if b > 0: s = s[0:b] + "." + s[b:]
            else: s = "." + "0" * -b + s
            o += s.rstrip("0").rstrip(".")
            p += 7
        else:
            print("Parsing state = \"" + o + "\"")
            die("Unknown byte found in the stream: " + hex(b))
    return o


def list_program(prg):
    pos = 0
    while True:
        ll = prg[pos]
        #print("linelength=" + str(ll))
        if ll == 0: return
        if prg[pos + ll - 1] != 0: die("Invalid line, not terminated with zero")
        print(list_line(prg[pos + 1:pos + ll - 1]))
        pos += ll


if __name__ == "__main__":
    if len(sys.argv) != 2:
        die("Bad usage, give the file name")
    try:
        with open(sys.argv[1], "r") as f: prg = bytearray(f.read())
    except IOError as e:
        die("Reading input file: " + str(e))
    list_program(prg[0x10:])
    die()