# LVTL-W: port of LVTL-R to Busybox/GNU AWK in under 130 SLOC
# should also work in any other AWK version with bitwise operations
#
# Usage: [busybox] awk -f lvtl.awk [prog.vtl -]
# (don't forget the - after the program or LVTL will exit upon preloading)
# 
# Any program valid for LVTL-W will also run on LVTL-R and LVTL-O and,
# if it doesn't use bitwise ops, on the original VTL-2 and VTL02 too
#
# Differences from the C version of LVTL-R (besides being much slower):
# - lines are stored in an AWK array where line numbers are keys
# - maximum line length is not enforced
# - the variables and array areas are fully separated from line memory
# - because of all this, & is always 0 and * is always 65535
# - extra whitespace before and after line numbers is accepted
# - only LFs are printed instead of CRLFs
# - strings are allowed in immediate statements
#
# Differences from the original VTL-2 for Altair:
# - both LF and CRLF are accepted (but not saved into RAM)
# - only LFs are printed instead of CRLFs
# - maximum line length is not enforced
# - all whitespace after the line number is fully ignored
# - any whitespace within expressions is also insignificant
# - parentheses are NOT auto-closed at the end of the statement
# - input with ? is NOT evaluated and only numbers are accepted
# - only 26 characters (A-Z) are valid generic variable names
# - supports all standard VTL-2 binary operators: +, -, *, /, =, >, < 
# - also supports bitwise operators: & (and), ! (or), # (xor)
# - jumps to itself (like 10 #=10) are prohibited and counted as nops
# - no file I/O (only the option to preload VTL code from a file)
# - self-modifying programs won't work correctly
#
# Created by Luxferre in 2023, released into public domain

# Global arrays and vars: ORD, PROGLINES, VARS, SCRATCH, PROCCHARS

# fatal error reporter
function trapout(errmsg) {printf("Fatal error: %s\n", errmsg); exit(1)}

# read a generic/system variable
function getvar(varname, c) {
  res = 0
  if(varname == "$") { # get a single character from stdin
    (cmd="c='';read -r -n 1 c;echo \"$c\"") | getline c
    close(cmd)
    res = ORD[substr(c,1,1)]
  }
  else if(varname == "?") { # get a number from stdin
    getline c # read the line
    res = and(int(c), 65535) # cast to 16-bit integer
  }
  else if(varname == "'") res = int(rand()*65536) # get a pseudorandom number
  else if(varname in VARS) res = VARS[varname] # regular or system variable
  return res
}

# set a generic/system variable
function setvar(varname, value, qi) {
  if(varname == "$") printf("%c", and(int(value), 255)) # output a character
  else if(varname == "?") { # output a string or a number
    if(VARS["\""]) { # we print a string
      qi = index(value, "\"") # find the ending quote
      printf("%s", substr(value, 1, qi - 1))
      # also print LF unless stated otherwise with ; modifier
      if(substr(value, qi+1, 1) != ";") printf("\n") 
      VARS["\""] = 0 # reset the printing flag
    }
    else printf("%u", and(int(value), 65535)) # we print a number, no LF
  }
  else if(varname in VARS) { # regular or system variable
    value = and(int(value), 65535) # cast to 16-bit integer
    # if the var is #, cache the return address before writing
    if(varname == "#" && value > 0) VARS["!"] = VARS["#"] + 1
    VARS[varname] = value # update the value
  }
  else trapout("invalid variable name!")
}
  
# strict LTR evaluator
function evalexpr(expr, tkn, i, l, opex, acc, cop, oprnd, subs, res) {
  opex = 1
  res = cop = acc = oprnd = 0
  l = length(expr) # GAWK doesn't support multi-init in fors, busybox does
  for(i=1;i<=l;i++) { # token position is stored in i
    tkn = substr(expr, i, 1) # get a single character token
    if(tkn == ")") break # ditch the rest of expression
    if(tkn == " " || tkn == "\t") continue # skip all whitespace
    if(opex = 1 - opex) { # we expect an operator
      if(index("+-*/=<>&!#", tkn)) cop = tkn # save the valid operator
      else trapout(sprintf("unexpected binary operator at %u\n", VARS["#"]))
    }
    else { # we expect a value, a variable, a subscript or a subexpr
      if(tkn == "\"") { # quote immediately prints and returns
        VARS[tkn] = 1
        return substr(expr, i+1) # return everything after the quote
      }
      # extract and evaluate a parens or subscript expression
      else if(tkn == ":" || tkn == "(") {
        subs = evalexpr(substr(expr, i+1)) # process everything after tkn
        # set the operand to array elem or eval result
        oprnd = (tkn == ":") ? SCRATCH[subs] : subs
        i += PROCCHARS # update the outer index
      }
      else if(index("0123456789", tkn)) { # it's a digit
        oprnd = int(substr(expr, i)) # read the integer
        i += length(oprnd) - 1 # skip the rest
      }
      else oprnd = getvar(tkn) # we assume it's a valid variable name
      # now, perform the calculation - no default branch!
      if(oprnd < 0) oprnd += 65536 # keep the operand positive
      oprnd = and(oprnd, 65535) # keep the operand within 16 bits
      if(cop == 0) acc = oprnd
      else if(cop == "+") acc += oprnd
      else if(cop == "-") acc -= oprnd
      else if(cop == "*") acc *= oprnd
      else if(cop == "/" && oprnd) {
        VARS["%"] = acc % oprnd # keep the remainder in the % sysvar
        acc = int(acc/oprnd)
      }
      else if(cop == "=") acc = (acc == oprnd) ? 1 : 0
      else if(cop == ">") acc = (acc >= oprnd) ? 1 : 0
      else if(cop == "<") acc = (acc < oprnd) ? 1 : 0
      else if(cop == "&") acc = and(acc, oprnd)
      else if(cop == "!") acc = or(acc, oprnd)
      else if(cop == "#") acc = xor(acc, oprnd)
      if(acc < 0) acc += 65536 # keep the accumulator positive
      res = acc = and(acc, 65535) # keep the accumulator within 16 bits
    }
  }
  PROCCHARS = i # because we started from 1 
  return res
}

# run the statement
function exec_stmt(stmt, ln, lcache, lhs, rhs, ei, vn) {
  lcache = ln # cache the line number
  ei = index(stmt, "=") # get the first occurrence of =
  lhs = substr(stmt, 1, ei - 1) # extract LHS as everything before =
  rhs = substr(stmt, ei + 1) # extract RHS as everything after =
  if(!length(rhs) || !length(lhs)) return # nop on invalid statements
  ei = evalexpr(rhs) # evaluate RHS
  vn = substr(lhs, 1, 1) # extract the varname
  # evaluate the array subscript and update the array
  if(vn == ":") SCRATCH[evalexpr(substr(lhs, 2))] = ei
  else setvar(vn, ei) # update the (pseudo) variable
  # seek the next statement
  ln = VARS["#"]
  # we're starting from interactive mode OR updating the number
  if((!ln && lcache) || (ln > 0 && ln == lcache)) ln = lcache + 1
  if(ln > 0) {
    for(ei=ln;ei<65536;ei++) if(ei in PROGLINES) {ln = ei; break} # search
    if(ei >= 65536) ln = 0 # line not found
    VARS["#"] = ln # save the closest line number
    if(ln) exec_stmt(PROGLINES[ln], ln) # pass new statement for execution
  }
}

BEGIN { # interpreter init
  "date +%N"|getline rseed;srand(rseed) # init the PRNG
  PROCCHARS = 0 # global cache to track processed characters
  for(i=0;i<256;i++) ORD[sprintf("%c", i)] = i # init char-to-ASCII mapping
  for(i=0;i<26;i++) VARS[sprintf("%c", i+ORD["A"])] = 0 # init the var memory
  # initialize sysvars
  VARS["!"] = VARS["\""] = VARS["#"] = VARS["%"] = VARS["&"] = 0
  VARS["*"] = 65535
  for(i=0;i<65536;i++) SCRATCH[i] = 0 # initialize the scratch area
  print "LVTL-W by Luxferre\nPress Ctrl+C to exit\n\nOK"
}

{ # main line-by-line processing
  ln = and(int($1), 65535) # attempt to scan line number (default to 0)
  if(NF > 1) $1 = "" # prepare to scan the statement
  gsub(/^[ \t]+|[ \t]+$/, "") # trim the statement
  if(ln > 0) { # we have a line number
    if($0 == "" || $0 == ln) delete PROGLINES[ln] # delete the line
    else PROGLINES[ln] = $0 # cache the statement
  }
  else { # it's an immediate statement
    if($0 == "0") { # list the program
      for(i=1;i<65536;i++) # start from 1 because 0 is not a valid line number
        if(i in PROGLINES) print i, PROGLINES[i] # list the found line
    }
    else exec_stmt($0, 0) # pass to immediate execution
    print "\nOK"
  }
}