/************************************************************************/
/*									*/
/*	File:	pp8.c							*/
/*									*/
/*	Expression evaluator.						*/
/*									*/
/*	Written by:							*/
/*			Gary Oliver					*/
/*			3420 NW Elmwood Dr.				*/
/*			PO Box 826					*/
/*			Corvallis, Oregon 97339				*/
/*			(503)758-5549					*/
/*	Maintained by:							*/
/*			Kirk Bailey					*/
/*			Logical Systems					*/
/*			P.O. Box 1702					*/
/*			Corvallis, OR 97339				*/
/*			(503)753-9051					*/
/*									*/
/*	This program is hereby placed in the public domain.  In		*/
/*	contrast to other claims of "public domain", this means no	*/
/*	copyright is claimed and you may do anything you like with PP,	*/
/*	including selling it!  As a gesture of courtesy, please retain	*/
/*	the authorship information in the source code and		*/
/*	documentation.							*/
/*									*/
/*	Functions contained within this module:				*/
/*									*/
/*		eval		Top level eval function.		*/
/*		evaltern	Evaluate ?: operator.			*/
/*		evallor		Evaluate || operator.			*/
/*		evalland	Evaluate && operator			*/
/*		evalbor		Evaluate | operator.			*/
/*		evalbxor	Evaluate ^ operator.			*/
/*		evalband	Evaluate & operator.			*/
/*		evaleq		Evaluate == and != operators.		*/
/*		evalrel		Evaluate < <= > and >= operators.	*/
/*		evalsh		Evaluate << and >> operators.		*/
/*		evalsum		Evaluate + and - operators.		*/
/*		evalmdr		Evaluate * / and % operators.		*/
/*		evalfuns	Evaluate functions ("defined", etc.)	*/
/*		evalucom	Evaluate unary ~ operator.		*/
/*		evalunot	Evaluate unary ! operator.		*/
/*		evalumin	Evaluate unary - operator.		*/
/*		evalval		Evaluate a value.			*/
/*		hexbin		Convert hex digit to binary.		*/
/*		ishex		Returns TRUE if char is hex digit.	*/
/*		isoct		Returns TRUE if char is octal digit.	*/
/*		item		Fetch next token.			*/
/*		look		Look ahead for matching string.		*/
/*		match		Read tokens while match and return str.	*/
/*		readexpline	Read a line and macro expand for expr.	*/
/*		test		Test (read and absorb) matching tokens.	*/
/*									*/
/************************************************************************/

#include	"pp.h"
#include	"ppext.h"

/************************************************************************/
/*									*/
/*	eval								*/
/*									*/
/*	Recursive descent expression parser.  Returns the value		*/
/*	of the expression.  Errors detected during parsing cause	*/
/*	calls to perr().  Tokens are fetched using getnstoken().	*/
/*									*/
/************************************************************************/

EVALINT
eval()
	{
	char			buf[TOKENSIZE];
	EVALINT			val;

	pbcstr(readexpline(buf,TOKENSIZE));	/* Macro expand line */

	for(;;)
		{
		val = evaltern();	/* Top level ternary form */

		if(Token[0] != ',')
			{
			if(Token[0] != '\n')
				{
				non_fatal("Expression: Invalid operator: ",
					Token);
				}
			else
				pushback('\n');	/* Put eol back out */
			break;
			}
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	evaltern							*/
/*									*/
/*	?: operator							*/
/*									*/
/************************************************************************/

EVALINT
evaltern()
	{
	EVALINT			val;
	EVALINT			val_f;
	EVALINT			val_t;

	val = evallor();		/* Look for expression */
	if(! test("?"))			/* ? part of ?: */
		return (val);

	val_t = eval();			/* Get true-side expression */
	if(! test(":"))
		{
		non_fatal("Expression: ':' expected","");
		return ((EVALINT) 0);
		}
	val_f = eval();			/* Get false-side expression */

	return (val ? val_t : val_f );	/* Use it or loose it */
	}

/************************************************************************/
/*									*/
/*	|| operator							*/
/*									*/
/************************************************************************/

EVALINT
evallor()
	{
	EVALINT			val;

	for(val = evalland(); test("||") ;)
		{
		if(evalland() != 0)
			val = TRUE;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	&& operator							*/
/*									*/
/************************************************************************/

EVALINT
evalland()
	{
	EVALINT			val;

	for(val = evalbor(); test("&&") ;)
		{
/*lint -e503 lint doesn't like boolean arguments to relationals */
		if(! evalbor() != 0)
/*lint +e503 lint doesn't like boolean arguments to relationals */
			val = FALSE;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	| operator							*/
/*									*/
/************************************************************************/

EVALINT
evalbor()
	{
	EVALINT			val;

	for(val = evalbxor(); !look("||") && test("|"); )
		val = val | evalbxor();

	return (val);
	}

/************************************************************************/
/*									*/
/*	^ operator							*/
/*									*/
/************************************************************************/

EVALINT
evalbxor()
	{
	EVALINT			val;

	for(val = evalband(); test("^"); )
		val = val ^ evalband();

	return (val);
	}

/************************************************************************/
/*									*/
/*	& operator							*/
/*									*/
/************************************************************************/

EVALINT
evalband()
	{
	EVALINT			val;

	for(val = evaleq(); !look("&&") && test("&"); )
		val = val & evaleq();

	return (val);
	}

/************************************************************************/
/*									*/
/*	== and != operators						*/
/*									*/
/************************************************************************/

EVALINT
evaleq()
	{
	EVALINT			val;

	for(val = evalrel() ; ;)
		{
		if(test("=="))
			val = val == evalrel();
		else if(test("!="))
			val = val != evalrel();
		else
			break;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	< > <= and >= operators						*/
/*									*/
/************************************************************************/

EVALINT
evalrel()
	{
	EVALINT			val;

	for(val = evalsh() ; ; )
		{
		if(test("<="))
			val = val <= evalsh();
		else if(test(">="))
			val = val >= evalsh();
		else if(test("<"))
			val = val < evalsh();
		else if(test(">"))
			val = val > evalsh();
		else
			break;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	<< and >> operators						*/
/*									*/
/************************************************************************/

EVALINT
evalsh()
	{
	EVALINT			val;

	for(val = evalsum() ; ; )
		{
		if(test("<<"))
			val = val << evalsum();
		else if(test(">>"))
			val = val >> evalsum();
		else
			break;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	+ and - operators						*/
/*									*/
/************************************************************************/

EVALINT
evalsum()
	{
	EVALINT			val;

	for(val = evalmdr() ; ; )
		{
		if(test("+"))
			val = val + evalmdr();
		else if(test("-"))
			val = val - evalmdr();
		else
			break;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	* / and % operators						*/
/*									*/
/************************************************************************/

EVALINT
evalmdr()
	{
	EVALINT			temp;
	EVALINT			val;

	for(val = evalfuns() ; ; )
		{
		if(test("*"))
			val = val * evalfuns();
		else if(test("/"))
			{
			temp = evalfuns();
			if(temp == 0)
				{
				non_fatal("Expression: Division by zero","");
				val = 0;
				}
			else
				val = val / temp;
			}
		else if(test("%"))
			{
			temp = evalfuns();
			if(temp == 0)
				{
				non_fatal("Expression: Division by zero","");
				val = 0;
				}
			else
				val = val % temp;
			}
		else
			break;
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	evalfuns							*/
/*									*/
/*	This level evaluates the pseudo-functions.			*/
/*									*/
/*	sizeof()	This is not allowed by the ANSI Draft C		*/
/*			Standard.  It is here to catch its use		*/
/*			in old programs where it was allowed.		*/
/*									*/
/*	defined		This is a pseudo-function that tests for	*/
/*			the defined state of the next symbol token.	*/
/*									*/
/*	_isstring	Predicate returns 1 if argument is a string	*/
/*			constant.					*/
/*									*/
/*	_strsize	Returns the amount of storage a string constant	*/
/*			will use.					*/
/*									*/
/************************************************************************/

EVALINT
evalfuns()
	{
	int			count;
	int			pflag;
	int			rv;
	char			*str;
	int			t;

	if(test("sizeof"))
		{
		non_fatal("Expression: sizeof() not allowed","");

		while(((t = getnstoken(GT_STR)) != ')') &&
			(t != '\n') && (t != EOF))
			;
		return ((EVALINT) 0);
		}
	else if(test("defined"))
		{
		if(! (pflag = (getnstoken(GT_STR) == '('))) /* Latch '(' */
			pbstr(Token);	/* Put it back if not */

		if((t = getnstoken(GT_STR)) == LETTER) /* Get an Id */
			{
			rv = (int) (lookup(Token,NULL) != NULL);

			if(pflag && (getnstoken(GT_STR) != ')'))
				{
				non_fatal("Expression: Missing ')'","");
				pbstr(Token);
				}

			return ((EVALINT) rv);
			}
		else if(t == '\n')
			pushback('\n');
		else if(t == EOF)
			end_of_file();

		non_fatal("Expression: Not an identifier: ",Token);
		return ((EVALINT) FALSE);
		}
	else if(test("_isstring"))
		{
		rv = FALSE;
		if(! test("("))
			non_fatal("Expression: Missing '('","");
		else
			{
			if(item(getnstoken,GT_STR) && (Token[0] == '"'))
				rv = TRUE;
			t = Token[0];
			count = 0;	/* Paren nesting level */
			while((! ((t == ')') && (count == 0))) &&
				(t != '\n') && (t != EOF))
				{
				if(t == '(')
					count++;
				else if(t == ')')
					count--;
				t = getnstoken(GT_STR);
				}
			if(t != ')')
				{
				non_fatal("Expression: Missing ')'","");
				if(t == EOF)
					end_of_file();
				pbstr(Token);
				}
			}
		return ((EVALINT) rv);
		}
	else if(test("_strsize"))
		{
		rv = 1;
		if(! test("("))
			non_fatal("Expression: Missing '('","");
		else
			{
			if(item(getnstoken,GT_STR) && (Token[0] != '"'))
				{
				non_fatal("_strsize: Missing string","");
				t = Token[0];
				count = 0;	/* Paren nesting level */
				while((! ((t == ')') && (count == 0))) &&
					(t != '\n') && (t != EOF))
					{
					if(t == '(')
						count++;
					else if(t == ')')
						count--;
					t = getnstoken(GT_STR);
					}
				if(t != ')')
					{
					non_fatal("Expression: Missing ')'",
						"");
					if(t == EOF)
						end_of_file();
					pbstr(Token);
					}
				}
			else
				{
				for(;;)
					{
					str = &Token[1];
					for(; (*str != '\0') &&
						(*str != '\"'); rv++)
						{
/************************************************************************/
	if(*str == '\\')
		{
		if(*++str == 'x')
			{		/* Hex seq. >= 2 hex digits */
			count = 0;
			while(isxdigit(*++str) && (++count < 3))
				;
			}
		else
			{
			if(isdigit(*str))
				{ 	 /* Octal seq. >=3 octal dig */
				count = 0;
				while (isdigit(*++str) && (++count < 3))
					;
				}
			else
				str++;
			}
		}
	else
		str++;
/************************************************************************/
						}
					if((t = getnstoken(GT_STR)) == ')')
						break;	/* Done */
					if(t == '"')
						/* Concatenated string */
						continue;
					non_fatal("_strsize: Not a string","");
					count = 0;	/* Paren nesting lvl */
					while(((t != ')') || (count != 0)) &&
						(t != '\n') && (t != EOF))
						{
						if(t == '(')
							count++;
						else if(t == ')')
							count--;
						t = getnstoken(GT_STR);
						}
					if(t != ')')
						{
				non_fatal("Expression: Missing ')'","");
						if(t == '\n')
							pushback('\n');
						else if(t == EOF)
							end_of_file();
						}
					break;
					}
				}
			}
		return ((EVALINT) rv);
		}
	else
		return (evalucom());
	}

/************************************************************************/
/*									*/
/*	unary ~ operator						*/
/*									*/
/************************************************************************/

EVALINT
evalucom()
	{
	if(test("~"))
/*lint -e502 lint doesn't like complementing signed integral types */
		return (~evalfuns());
/*lint +e502 lint doesn't like complementing signed integral types */
	else
		return (evalunot());
	}

/************************************************************************/
/*									*/
/*	unary ! operator						*/
/*									*/
/************************************************************************/

EVALINT
evalunot()
	{
	if(test("!"))
		return ((EVALINT) !evalfuns());
	else
		return (evalumin());
	}

/************************************************************************/
/*									*/
/*	unary + and - operators						*/
/*									*/
/************************************************************************/

EVALINT
evalumin()
	{
	if(test("+"))
		return ((EVALINT) evalfuns());
	else if(test("-"))
		return ((EVALINT) -evalfuns());
	else
		return (evalval());
	}

/************************************************************************/
/*									*/
/*	Must be a value (number or 'x') or a '(' for subexpression.	*/
/*									*/
/************************************************************************/

EVALINT
evalval()
	{
	int			c;
	char			*p;
	EVALINT			val;

	if(test("("))
		{
		val = evaltern();	/* Get deeper expression */
		if(! test(")"))
			non_fatal("Expression: Mismatched \"()\"","");
		}
	else if(test("'"))		/* Quoted character constant ? */
		{
		val = getchn();		/* Get next character char */
		if(val == '\\')		/* Special test for escaped chars */
			{
			switch((char) (val = getchn()))
				{
		case 'a':	val = '\a';	break;	/* Alert char	*/
		case 'b':	val = '\b';	break;	/* Backspace	*/
		case 'f':	val = '\f';	break;	/* Form feed	*/
		case 'n':	val = '\n';	break;	/* Newline	*/
		case 'r':	val = '\r';	break;	/* Return	*/
		case 't':	val = '\t';	break;	/* Horiz. tab	*/
		case 'v':	val = '\v';	break;	/* Vert. tab	*/
		case '\\':	val = '\\';	break;
		case '\'':	val = '\'';	break;
		case '\"':	val = '\"';	break;
		case '?':	val = '?';	break;

		case '0': case '1': case '2': case '3':
		case '4': case '5': case '6': case '7':
/*
 *	Lead in to octal numeric field: get octal number.
 */
			for(val -= '0'; (c = getchn()) != '\'';)
				{
				if(c == '\n')
					{
				non_fatal("Expression: EOL in '' constant","");
					break;
					}
				else if(c == EOF)
					end_of_file();
				else if(isoct(c))
					val = val * 8 + c - '0';  /* Add in */
				else
					{
			non_fatal("Expression: Illegal octal digit","");
					break;
					}
				}
			pushback(c);
			break;

		case 'x':
/*
 *	Beginning of hex numeric field, get number.
 */
			for(val = 0; (c = getchn()) != '\'';)
				{
				if(c == '\n')
					{
				non_fatal("Expression: EOL in '' constant","");
					break;
					}
				else if(c == EOF)
					end_of_file();
				else if(ishex(c))
					val = val * 16 + hexbin(c);
				else
					{
			non_fatal("Expression: Illegal hex digit","");
					break;
					}
				}
			pushback(c);
			break;
		default:
			non_fatal("Expression: Illegal character escape","");
			break;
				}
			}
		if(! test("'"))
			non_fatal("Expression: Mismatched apostrophes","");
		}
	else
		{
		/* Pack a number */
		if(item(getnstoken,GT_STR))
			{
			if(isdigit(Token[0]))
				{
				val = 0;
				p = Token;
				if(*p == '0')
					{
					/* Octal or Hex number */
					if(*++p == 'x' || *p == 'X')
						{
						/* Hex number */
						for(++p; ishex(*p); ++p)
							{
							val = val * 16 +
								hexbin(*p);
							}
						}
					else
						{
						/* Octal number */
						for(; isoct(*p); ++p)
							{
							val = val * 8 +
								*p - '0';
							}
						}
					}
				else
					{
					/* Decimal number */
					for(; isdigit(*p); ++p)
						val = val * 10 + *p - '0';
					}
				if(*p != '\0')
					{
					non_fatal("Expression: Bad operand: ",
						Token);
					}
				}
			else
				{
				/* #if of undef'd id OK */
				if(! istype(Token[0] & 0xFF,C_L))
					{
			non_fatal("Expression: Expected operand: ",Token);
					}
				return ((EVALINT) 0);
				}
			}
		else
			{
			non_fatal("Expression: Expected operand: ",Token);
			return ((EVALINT) 0);
			}
		}
	return (val);
	}

/************************************************************************/
/*									*/
/*	hexbin								*/
/*									*/
/*	Convert hexadecimal digit to binary.				*/
/*									*/
/************************************************************************/

EVALINT
hexbin(ch)
	char			ch;
	{
	if(isdigit(ch))
		return ((EVALINT) (ch - '0'));
	else
		return ((EVALINT) ((isupper(ch) ? tolower(ch) : ch) - 'a'+10));
	}

/************************************************************************/
/*									*/
/*	ishex								*/
/*									*/
/*	Returns TRUE if character is a hexadecimal digit.		*/
/*									*/
/************************************************************************/

int
ishex(ch)
	char			ch;
	{
	return (isdigit(ch)			||
		((ch >= 'a') && (ch <= 'f'))	||
			((ch >= 'A') && (ch <= 'F')));
	}

/************************************************************************/
/*									*/
/*	isoct								*/
/*									*/
/*	Returns TRUE if characters is an octal digit.			*/
/*									*/
/************************************************************************/

int
isoct(ch)
	char			ch;
	{
	return ((ch >= '0') && (ch <= '7'));
	}

/************************************************************************/
/*									*/
/*	item								*/
/*									*/
/*	Skip to first non-blank token and fetch the token into the	*/
/*	token buffer <Token>.  Return TRUE if an item was fetched or	*/
/*	FALSE if at eol.						*/
/*									*/
/************************************************************************/

int 
item(fun,f)
	register int		(*fun)();
	int			f;
	{
	register int		t;

	while((t = (*fun)(f)) != '\n')
		{
		if(t == EOF)
			end_of_file();	/* Abort on end of file */
		return(TRUE);
		}
	pushback('\n');			/* Unfetch the newline */

	return (FALSE);
	}

/************************************************************************/
/*									*/
/*	look								*/
/*									*/
/*	Look ahead in for the specified token.	Returns TRUE if		*/
/*	token is found, else FALSE.  Always leaves tokens unfetched.	*/
/*									*/
/************************************************************************/

int
look(str)
	register char		*str;
	{
	register int		rc;
	char			tokenbuf[TOKENSIZE];

	rc = match(tokenbuf,str);
	pbstr(tokenbuf);		/* Unfetch the item */
	return (rc);
	}

/************************************************************************/
/*									*/
/*	match								*/
/*									*/
/*	Skip leading blanks and fetch tokens into passed <tbuf> until	*/
/*	a complete match with <str> is found or a failure to match	*/
/*	occurs.  If a complete match occurs, return TRUE else return	*/
/*	FALSE.  On return, <tbuf> has the successfully matched part.	*/
/*									*/
/************************************************************************/

int 
match(tbuf,str)
	register char		*tbuf;
	register char		*str;
	{
	register int		(*fun)();
	register char		*s;
	register char		*t;

	fun = getnstoken;		/* Initially skip spaces */
	t = tbuf;			/* Token packing buffer */
	s = str;			/* String to match */

	while((*s != '\0') && item(fun,0))
		{
		if(strncmp(Token,s,strlen(Token)) == EQUAL)
			{
			/* Build token */
			t = addstr(t,&tbuf[TOKENSIZE],
				"Expression: Token too long",Token);

			s += strlen(Token);	/* Move past matched part */
			if(*s == '\0')
				break;	/* Exit if end of string */
			}
		else
			{
			pbstr(Token);	/* Unfetch unmatching string */
			break;
			}
		fun = gettoken;		/* Don't skip spaces anymore */
		}

	*t = '\0';			/* End of token here */
 
	return (*s == '\0');		/* Return TRUE if end of string	*/
	}

/************************************************************************/
/*									*/
/*	readexpline							*/
/*									*/
/*	Read and edit a line into buffer with macro expansion.  Like	*/
/*	"readline" except knows about "defined" pseudo-function.	*/
/*									*/
/************************************************************************/

char	*
readexpline(buf,bufsize)
	register char		*buf;
	register int		bufsize;
	{
	static	char		rbo[] = "Read buffer overflow";

	register char		*bufp;
	register int		is_func;
	struct	symtab		*sy;
	register int		t;

	is_func = FALSE;

	for(bufp = buf; (t = gettoken(GT_STR)) != '\n'; )
		{
		if(t == EOF)
			end_of_file();
		if(t == LETTER)
			{
			if((! is_func) &&
				((sy = lookup(Token,NULL)) != NULL) &&
					(sy->disable != TRUE))
				{
				bufp = docall(sy,bufp,&buf[bufsize - 1]);
				}
			else
				{
				bufp = addstr(bufp,&buf[bufsize - 1],rbo,
					Token);
				if(is_func)
					is_func = FALSE;
				else if(strcmp(Token,"defined") == EQUAL)
					is_func = TRUE;
				}
			}
		else
			bufp = addstr(bufp,&buf[bufsize - 1],rbo,Token);
		}

	pushback('\n');
	*bufp = '\0';

	for(bufp = buf; istype(*bufp & 0xFF,C_W); ++bufp)
		;			/* Skip leading blanks */

	return(bufp);
	}

/************************************************************************/
/*									*/
/*	test								*/
/*									*/
/*	Skip to next non-blank token and test for match	against		*/
/*	specified token.  If match, return TRUE and advance past	*/
/*	the token.  Otherwise return FALSE, leaving pointer pointing	*/
/*	to first non-blank character in token.				*/
/*									*/
/************************************************************************/

int 
test(str)
	register char		*str;
	{
	char			tokenbuf[TOKENSIZE];

	if(match(tokenbuf,str))
		return (TRUE);
	else
		{
		pbstr(tokenbuf);	/* Lets look at it again later */
		return (FALSE);
		}
	}