a simple interpreter for BASIC written in AWK

basic.sh

#!/usr/bin/gawk -f
# a simple interpreter for GWBASIC(subset) written in AWK 
#     following 'The AWK Programming Language' Chapter 6
# @date June 20,2012
# @link basic.sh https://gist.github.com/2959816
# @link test.sh https://gist.github.com/2959877
# TODO:
#   gosub/return...ok
#   if/else/elseif/endif...ok
#   while/wend...ok
#   more unit test
#   error warning
#   builtin functon
#   document
#   clean code
#   more features in gwbasic
# @ref 'The AWK Programming Language'
# @ref 'man mawk'
# @ref http://en.wikipedia.org/wiki/GW-BASIC
BEGIN{
    DEBUG_MODE = 0;
    printf("BASIC\n");
    code = "\n"
    #top = "top"
    ifstack[0] = 1;
    #ifstack[top] = 0;
    ifstack["top"] = 0;
    callstack[0] = 0;
    callstack["top"] = 0;
    #whilestack[0] = 0;
    #whilestack["top"] = 0;
}
!/^( |\t)*\#/{
    if($0 ~ /^ *[A-Za-z_0-9]+ *\:$/){
        gsub(/ |\:/,"");
        $0 = "@" $0;
    }
    code = code $0 "\n";
}
END{
    debug("<CODE>");
    debug(code,length(code));
    debug("<RUN>\n\n\n");
    line = code;
    codelen = length(code);
    advance();
    stat();
}
function push(Stack,x){
    Stack[++Stack["top"]] = x;
}
function pop(Stack){
    return Stack[Stack["top"]--];
}
function advance() {
    #if (tok == "(eof)") return "(eof)"
    #while (length(line) == 0)if (getline line == 0)return tok = "(eof)";
    if (line == "")
        return tok = "(eof)";
    if (match(line, /^[A-Za-z_][A-Za-z_0-9]*/) ||
        match(line, /^-?([0-9]+\.?[0-9]*|\.[0-9]+)/) ||
        #match(line, /^\"(\\.|[^"])*\"/) ||
        match(line, /^\"[^\"]*\"/) ||
        match(line, /^(<|<=|=|<>|>=|>)/) ||        
        #match(line, /^\*\/\+\-]/) ||
        match(line, /^./)) {
            tok = substr(line, 1, RLENGTH);
            line = substr(line, RLENGTH+1);
            if(tok ~ /^ +/)
                return advance();
            return tok;
        }
    error("advance>line " substr(line,1,10) " incomprehensible at " line);
}
function eat(s) {     # read next token if s == tok
    if (tok != s) error("line " substr(line,1,20) ": saw " tok ", expected " s)
    advance();
}
function nl() {       # absorb newlines and semicolons
    while(tok ~ /^(\n|\:)/)
        advance();
    #while (tok == "\n" || tok == ":")
        #advance();
}
function error(s) { print("Error: " s)| "cat 1>&2"; exit 1 }
function jump(label){
    if(label ~ /^[0-9]+/){
        if(!match(code,"\n *" tok " *"))
            error("goto line");
        line = substr(code, RSTART+RLENGTH)
        advance();
    }else if(label ~ /^[A-Za-z_][A-Za-z_0-9]*/){
        if(!match(code,"\n *@" tok " *(\n|$)"))
            error("goto label");
        line = substr(code, RSTART+RLENGTH)
        debug(RSTART","RLENGTH "goto>" "@" label " " line)
        advance();
    }else{error("goto");}
    
}
function ifskip(){
    #debug("ifskipline> " ifstack["top"])
    #print_array(ifstack)
    return (ifstack[ifstack["top"]]!=1)
}
function print_array(A,   i,len,s){
    len=length(A);
    s = A[0] "|"
    for(i=1;i<=len;i++)
        s = s " " A[i];
        debug(s);
}
function stat(   lhs,op,e) {
    if(tok=="(eof)")
        return;
    nl();
    if(tok ~ /^[0-9]+/)
        advance();   #linenum
    if(tok=="(eof)")
        return;
    if(tok == "@"){
        eat("@")
        if(1)#use F_if_block
            if(tok=="if"){
                debug(">if");
                eat("if");
                if(!ifskip()){
                    e = (conj()!=0);
                    eat("\n");
                    debug("if> " e);
                    ifstack[++ifstack["top"]] = e;
                }else{
                    ifstack[++ifstack["top"]] = -1;
                }
            }else if(tok=="else"){
                debug(">else");
                eat("else");
                eat("\n");
                if(ifstack[ifstack["top"]]==1){
                    ifstack[ifstack["top"]] = 0;
                }else if(ifstack[ifstack["top"]]==0)
                    ifstack[ifstack["top"]] = 1;
                else
                    ;            
            }else if(tok=="elseif"){
                debug(">elseif");
                eat("elseif");
                e = (conj()!=0)
                eat("\n");
                if(ifstack[ifstack["top"]]==1){
                    ifstack[ifstack["top"]] = 0;
                }else if(ifstack[ifstack["top"]]==0){
                    ifstack[ifstack["top"]] = e;
                }else
                    ;  
            }else if(tok=="endif"){
                debug(">endif");
                eat("endif");
                eat("\n");
                debug(ifstack["top"],ifstack[ifstack["top"]]);
                delete ifstack[ifstack["top"]];
                #debug(ifstack["top"]--)
                ifstack["top"]--;
                debug(ifstack["top"],ifstack[ifstack["top"]]);
            }else{
                #while(tok !~ /^\n/)
                while(tok!="\n")
                    advance();
            }
        else
            while(tok !~ /^\n/)
                advance();
    }else if(ifskip()){
        debug("skiptok>" tok);
        while(tok !="\n" && tok!="(eof)"){
            advance();
            debug("skiptok->" tok);
        }
    }else{
        #debug("eval ing...")
        lhs = lvalue();
        if (lhs == "print"){
            #eat("print")
            #debug(conj())
            #return ""
            if(tok !~ /\n|\:/){
                printf("%s",exprlist());
                if(tok==";"){
                    eat(";");
                }else
                    printf("\n")
                
            }else
                printf("\n");
            #debug(e)
        }else if(lhs == "let"){
            #eat("let")
            lhs = lvalue();
            eat("=");
            env[lhs] = conj();
            #return ""
        }else if(lhs == "label"){
            advance();
        }else if(lhs == "input"){
            if(tok ~ /^\".*\"$/){
                gsub(/\"/,"",tok);
                printf("%s",tok);
                advance();
                eat(",");  
            }
            e = lvalue();
            getline env[e]<"-"
            #advance();
        }else if(lhs == "rem"){
            while(tok !~ /^\n/)
                advance();
        }else if(lhs == "cls"){
            system("clear || cls");
        }else if(lhs == "if"){
            e = conj();
            debug("if " e ":");
            if(tok=="goto"){
                eat("goto");
                #debug("label>" tok "!")
                if(e)
                    jump(tok);
                else{
                    advance();
                    }
            }else if(tok == "then"){
                eat("then")
                if(tok ~ /[0-9]+/){
                    if(e)
                        jump(tok);
                    else
                        advance();
                }else{
                    if(e)
                        stat()
                    else
                        while(tok !~ /^\n/)
                            advance();
                }
            }else{error("if")}
        }else if(lhs == "goto"){
            jump(tok);
        }else if(lhs == "end"){
            exit 0;
        }else if(lhs == "gosub"){
            callstack[++callstack["top"]] = length(code)-length(line)+1
            debug("here is |" substr(code,callstack[callstack["top"]]));
            jump(tok);
        }else if(lhs == "return"){
            #jump(tok);
            line = substr(code,callstack[callstack["top"]--]);
            advance();
        }else if(lhs == "while"){
            e = conj();
            debug("while " e ":");
            if(e)#e!=0
                ;
            else{
                debug("skip while");
                e = 0;
                #!!!!!!!!!!!!!!!!!!!!!!!!!!
                while(e>=0){
                    debug("while level> " e);
                    if(tok == "wend")
                        e--;
                    else if(tok=="while")
                        e++;
                    advance();
                }
            }
            #jump(tok)
            #callstack[++callstack["top"]] = 0
        }else if(lhs == "wend"){
            debug("wend");
            goto_while(length(code)-length(line)+1);
            debug(substr(code,1,length(code)-length(line)) "{end}");
            #jump(tok)
            #line = substr(code,callstack[callstack["top"]--]);
            #advance();
        }else{
            eat("=");
            env[lhs] = conj();
        }
    }
    if(tok=="(eof)")
        return;
    else
        stat() # use while later
}
function goto_while(len,    i,l,offset,A){
    A["top"] = 0;
    offset = 1;
    line = substr(code,1,len);
    advance();
    while(tok != "(eof)"){
        if(tok=="while"){
            debug("see>while{" substr(code,offset,20) "...}");
            A[++A["top"]] = offset;}
        else if(tok=="wend"){
            A["top"]--;
        }
        offset = len-length(line);
        advance();
    }
    if(A["top"]>=0){
        line = substr(code,A[A["top"]+1]);
        advance();
    }else
        error();
    #exit 0;
}
function expr(e) {
    e = rel();
    while (tok ~ /<|<=|=|<>|>=|>|&/) {
        op = tok;
        advance();
        if(op=="=")       e = 0+e == 0+rel();
        else if(op=="<>") e = 0+e != 0+rel();
        else if(op=="<")  e = 0+e < 0+rel();
        else if(op=="<=") e = 0+e <= 0+rel();
        else if(op==">")  e = 0+e > 0+rel();
        else if(op==">=") e = 0+e >= 0+rel();
        else if(op=="&")  e = e "" rel();
    }
    return e
}
function conj(    op,e){
    e = expr();
    while (tok == "and" || tok == "or") {
        op = tok;
        advance();
        if(op=="and")
            e = e && expr();
        else if(op=="or")
            e = e || expr();
    }
    return e    
}
function rel(    op, e) {
    e = term();
    while (tok == "+" || tok == "-") {
        op = tok;
        advance();
        if(op=="+")
            e = e + term();
        else if(op=="-")
            e = e - term();
    }
    return e
}

function term(    op, e) {        # fact | fact [*/%] fact
    e = fact();
    while (tok == "*" || tok == "/") {
        op = tok;
        advance();
        if(op=="*")
            e = e * fact();
        else if(op=="/")
            e = e / fact();
    }
    return e;
}

function fact(    e) {
    debug("fact>" tok);
    if (tok == "(") {
        eat("("); e = expr(); eat(")");
        return e
    } else if (tok ~ /^[A-Za-z][A-Za-z0-9]*/) {
        return ident();
    } else if (tok ~ /^-?([0-9]+\.?[0-9]*|\.[0-9]+)/) {
        e = tok
        advance();
        return e
    }else if(match(tok, /^\"[^\"]*\"/)){
        e = substr(tok,2,length(tok)-2);
        debug("string>" e);
        advance();
        return e
    }else
        error("fact>unexpected " tok " at line " substr(line,1,10));
}

function ident(    id, e) {
    if (!match(tok, /^[A-Za-z_][A-Za-z_0-9]*/))
        error("ident>unexpected " tok " at line " substr(line,1,10));
    id = tok;
    advance();
    if (tok == "[") {         # array
        eat("["); e = expr(); eat("]");
        return env[id,e]
    } else if (tok == "(") {  # function
        eat("(");
        e = "";
        if (tok != ")") {
            e = exprlist();
            eat(")");
        } else eat(")");
        return builtin(id,e);
    } else
        return env[id];
}
function lvalue(    id,e){
    if (!match(tok, /^[A-Za-z_][A-Za-z_0-9]*/))
        error("lvalue>unexpected " tok " at line " substr(line,1,10));
    id = tok;
    advance();
    if (tok == "[") {
        eat("["); e = expr(); eat("]");
        return id SUBSEP e;
    }else
        return id;
}
function exprlist(    n, e) { 
    e = conj()        # has to be at least one
    for (n = 1; tok == ","; n++) {
        advance();
        if(tok ~ /\n|:/)
            e = e SUBSEP " ";
        else
            e = e SUBSEP " " conj();
    }
    return e
}
function builtin(fun_name,arg_list,    n,A){
    n = split(arg_list,A,SUBSEP " ");
    if(fun_name=="len"){
        return length(A[1]);
    }else if(fun_name=="mod"){
        return A[1] % A[2];
    }else error();
    return;
}
function debug(x,y,z){
    #if(x)
    if(DEBUG_MODE)
    printf("%s\n",x y z);
}