rbascomp.py

import argparse, string, sys

BCODES = {
    "line": (10,),
    "id": (11,),
    "const": (12,),
    "if": (13,0),
    "goto": (14,),
    "print": (15,0),
    "stop": (16,0),
    "+": (17,1),
    "-": (17,2),
    "<": (17,3),
    "=": (17,4)
}

def scan(intx):
    # Scans the Basic source and separate characters into a token list
    toks = []
    buf = ""
    st = 0 # start at whitespace state
    pos = 0
    ttx = intx.lower() # make this case insensitive
    while pos < len(ttx):
        i = ttx[pos]
        if st == 0: # whitespace
            if i in string.digits: st = 1
            elif i in string.ascii_letters: st = 2
            elif i in string.punctuation: st = 3
            else: pos += 1 # skip the current character
            continue # skip the below lines
        elif st == 1: # number digits
            if i in string.digits:
                buf += i # add the character to the buffer
                pos += 1 # advance to the next character
                continue
        elif st == 2: # letters
            if i in string.ascii_letters:
                buf += i
                pos += 1
                continue
        elif st == 3: # punctuations
            if i in string.punctuation:
                buf += i
                pos += 1
                continue
        # different type of character found
        toks.append((st,buf)) # add the current state and sequence to the list
        buf = ""
        st = 0 # set the state back to 0
    if len(buf) > 0: toks.append((st,buf)) # empty the buffer
    toks.append((-1,"EOF"))
    return toks

# non-terminals are in integer index of parse table
# terminals are in string
PARSE_TABLE = [
    [[["slnum"],[1,0]],[["EOF"],["EOF"]]], # pgm
    [[["slnum"],["slnum",2]]], # line
    [[["id"],[3]],[["if"],[7]],[["print"],[10]],[["goto"],[11]],[["stop"],["stop"]]], # stmt
    [[["id"],["id","=",4]]], # asgmt
    [[["id","+","-","const"],[6,5]]], # exp
    [[["+"],["+",6]],[["-"],["-",6]],[["EOF","lnum"],[]]], # exp'
    [[["id"],["id"]],[["const"],["const"]]], # term
    [[["if"],["if",8,"gtdlnum"]]], # if
    [[["id","const"],[6,9]]], # cond
    [[["<"],["<",6]],[["="],["=",6]]], # cond'
    [[["print"],["print","id"]]], # print
    [[["goto"],["goto","lnum"]]], # goto
]

def validate(tok,term):
    # Checks if the token type matches the terminal
    # returns the B-Code if match, None if not
    outb = None
    if tok[0] == -1 and term == "EOF": outb = ()
    if tok[0] == 1: # number
        num = int(tok[1])
        if term in ["lnum","slnum","gtdlnum"] and 1 <= num <= 1000:
            outb = (num,)
            if term == "slnum": outb = BCODES["line"]+outb
            if term == "gtdlnum": outb = BCODES["goto"]+outb
        if term == "const" and 0 <= num <= 100:
            outb = BCODES[term]+(num,)
    else: # letters and punctuations
        if term in ["if","goto","print","stop","+","-","<","="] and tok[1] == term:
            outb = BCODES[term]
        elif term == "id" and tok[0] == 2 and len(tok[1]) == 1: # id
            outb = BCODES["id"]+(ord(tok[1])-ord("a")+1,)
    return outb

def parse(toks):
    # Parses the loken list into a B-Code table
    pstk = ["EOF",0] # parse stack
    outb = []
    tpos = 0 # token position
    lpos = -1 # parsed line number
    while tpos < len(toks):
        ctok = toks[tpos]
        if pstk[-1] == "EOF" and ctok == (-1,"EOF"):
            break
        elif type(pstk[-1]) is int: # non-terminal
            found = False
            for i in PARSE_TABLE[pstk[-1]]:
                for j in i[0]: # valid terminals
                    found = validate(ctok,j) != None
                    if found:
                        pstk.pop()
                        break
                if found:
                    pstk += i[1][::-1] # push the sequence to the stack backwards
                    break
            if not found: raise Exception("Error in line {}".format(lpos))
        else: # terminal
            bcod = validate(ctok,pstk[-1])
            if bcod == None: raise Exception("Error in line {}".format(lpos))
            if pstk[-1] == "slnum": lpos = bcod[1]
            outb += bcod
            pstk.pop()
            tpos += 1
    return outb

if __name__ == "__main__":
    ap = argparse.ArgumentParser()
    ap.add_argument('fi', metavar='in', type=argparse.FileType('r'), help='Input file name')
    ap.add_argument('fo', metavar='out', type=argparse.FileType('w'), help='Output file name')
    args = ap.parse_args()
    toks = scan(args.fi.read())
    bcodes = []
    try:
        bcodes = parse(toks)
    except Exception as e:
        print(e)
        sys.exit()
    args.fo.write(" ".join([str(i) for i in bcodes]))