awk.y

早期freebsd实现

	: variable ASSIGNOP 
		{ want_assign = 0; }
	  rexp
		{ $$ = node ($1, $2, $4); }
	| rexp LEX_AND rexp
		{ $$ = node ($1, Node_and, $3); }
	| rexp LEX_OR rexp
		{ $$ = node ($1, Node_or, $3); }
	| LEX_GETLINE opt_variable input_redir
		{
		  if (do_lint && ! io_allowed && $3 == NULL)
			warning("non-redirected getline undefined inside BEGIN or END action");
		  $$ = node ($2, Node_K_getline, $3);
		}
	| regexp
		{ $$ = $1; } 
	| '!' regexp %prec UNARY
		{ $$ = node((NODE *) NULL, Node_nomatch, $2); }
	| rexp MATCHOP rexp
		 { $$ = node ($1, $2, mk_rexp($3)); }
	| rexp LEX_IN NAME
		{ $$ = node (variable($3,1), Node_in_array, $1); }
	| rexp RELOP rexp
		{ $$ = node ($1, $2, $3); }
	| rexp '?' rexp ':' rexp
		{ $$ = node($1, Node_cond_exp, node($3, Node_if_branches, $5));}
	| simp_exp
		{ $$ = $1; }
	| rexp simp_exp %prec CONCAT_OP
		{ $$ = node ($1, Node_concat, $2); }
	;

simp_exp
	: non_post_simp_exp
	/* Binary operators in order of decreasing precedence.  */
	| simp_exp '^' simp_exp
		{ $$ = node ($1, Node_exp, $3); }
	| simp_exp '*' simp_exp
		{ $$ = node ($1, Node_times, $3); }
	| simp_exp '/' simp_exp
		{ $$ = node ($1, Node_quotient, $3); }
	| simp_exp '%' simp_exp
		{ $$ = node ($1, Node_mod, $3); }
	| simp_exp '+' simp_exp
		{ $$ = node ($1, Node_plus, $3); }
	| simp_exp '-' simp_exp
		{ $$ = node ($1, Node_minus, $3); }
	| variable INCREMENT
		{ $$ = node ($1, Node_postincrement, (NODE *)NULL); }
	| variable DECREMENT
		{ $$ = node ($1, Node_postdecrement, (NODE *)NULL); }
	;

non_post_simp_exp
	: '!' simp_exp %prec UNARY
		{ $$ = node ($2, Node_not,(NODE *) NULL); }
	| '(' exp r_paren
		{ $$ = $2; }
	| LEX_BUILTIN
	  '(' opt_expression_list r_paren
		{ $$ = snode ($3, Node_builtin, (int) $1); }
	| LEX_LENGTH '(' opt_expression_list r_paren
		{ $$ = snode ($3, Node_builtin, (int) $1); }
	| LEX_LENGTH
	  {
		if (do_lint)
			warning("call of `length' without parentheses is not portable");
		$$ = snode ((NODE *)NULL, Node_builtin, (int) $1);
		if (do_posix)
			warning( "call of `length' without parentheses is deprecated by POSIX");
	  }
	| FUNC_CALL '(' opt_expression_list r_paren
	  {
		$$ = node ($3, Node_func_call, make_string($1, strlen($1)));
	  }
	| variable
	| INCREMENT variable
		{ $$ = node ($2, Node_preincrement, (NODE *)NULL); }
	| DECREMENT variable
		{ $$ = node ($2, Node_predecrement, (NODE *)NULL); }
	| YNUMBER
		{ $$ = $1; }
	| YSTRING
		{ $$ = $1; }

	| '-' simp_exp    %prec UNARY
		{ if ($2->type == Node_val) {
			$2->numbr = -(force_number($2));
			$$ = $2;
		  } else
			$$ = node ($2, Node_unary_minus, (NODE *)NULL);
		}
	| '+' simp_exp    %prec UNARY
		{ $$ = $2; }
	;

opt_variable
	: /* empty */
		{ $$ = NULL; }
	| variable
		{ $$ = $1; }
	;

variable
	: NAME
		{ $$ = variable($1,1); }
	| NAME '[' expression_list ']'
		{
		if ($3->rnode == NULL) {
			$$ = node (variable($1,1), Node_subscript, $3->lnode);
			freenode($3);
		} else
			$$ = node (variable($1,1), Node_subscript, $3);
		}
	| '$' non_post_simp_exp
		{ $$ = node ($2, Node_field_spec, (NODE *)NULL); }
	;

l_brace
	: '{' opt_nls
	;

r_brace
	: '}' opt_nls	{ yyerrok; }
	;

r_paren
	: ')' { yyerrok; }
	;

opt_semi
	: /* empty */
	| semi
	;

semi
	: ';'	{ yyerrok; want_assign = 0; }
	;

comma	: ',' opt_nls	{ yyerrok; }
	;

%%

struct token {
	char *operator;		/* text to match */
	NODETYPE value;		/* node type */
	int class;		/* lexical class */
	unsigned flags;		/* # of args. allowed and compatability */
#	define	ARGS	0xFF	/* 0, 1, 2, 3 args allowed (any combination */
#	define	A(n)	(1<<(n))
#	define	VERSION	0xFF00	/* old awk is zero */
#	define	NOT_OLD		0x0100	/* feature not in old awk */
#	define	NOT_POSIX	0x0200	/* feature not in POSIX */
#	define	GAWKX		0x0400	/* gawk extension */
	NODE *(*ptr) ();	/* function that implements this keyword */
};

extern NODE
	*do_exp(),	*do_getline(),	*do_index(),	*do_length(),
	*do_sqrt(),	*do_log(),	*do_sprintf(),	*do_substr(),
	*do_split(),	*do_system(),	*do_int(),	*do_close(),
	*do_atan2(),	*do_sin(),	*do_cos(),	*do_rand(),
	*do_srand(),	*do_match(),	*do_tolower(),	*do_toupper(),
	*do_sub(),	*do_gsub(),	*do_strftime(),	*do_systime();

/* Tokentab is sorted ascii ascending order, so it can be binary searched. */

static struct token tokentab[] = {
{"BEGIN",	Node_illegal,	 LEX_BEGIN,	0,		0},
{"END",		Node_illegal,	 LEX_END,	0,		0},
{"atan2",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(2),	do_atan2},
{"break",	Node_K_break,	 LEX_BREAK,	0,		0},
{"close",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(1),	do_close},
{"continue",	Node_K_continue, LEX_CONTINUE,	0,		0},
{"cos",		Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(1),	do_cos},
{"delete",	Node_K_delete,	 LEX_DELETE,	NOT_OLD,	0},
{"do",		Node_K_do,	 LEX_DO,	NOT_OLD,	0},
{"else",	Node_illegal,	 LEX_ELSE,	0,		0},
{"exit",	Node_K_exit,	 LEX_EXIT,	0,		0},
{"exp",		Node_builtin,	 LEX_BUILTIN,	A(1),		do_exp},
{"for",		Node_K_for,	 LEX_FOR,	0,		0},
{"func",	Node_K_function, LEX_FUNCTION,	NOT_POSIX|NOT_OLD,	0},
{"function",	Node_K_function, LEX_FUNCTION,	NOT_OLD,	0},
{"getline",	Node_K_getline,	 LEX_GETLINE,	NOT_OLD,	0},
{"gsub",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(2)|A(3), do_gsub},
{"if",		Node_K_if,	 LEX_IF,	0,		0},
{"in",		Node_illegal,	 LEX_IN,	0,		0},
{"index",	Node_builtin,	 LEX_BUILTIN,	A(2),		do_index},
{"int",		Node_builtin,	 LEX_BUILTIN,	A(1),		do_int},
{"length",	Node_builtin,	 LEX_LENGTH,	A(0)|A(1),	do_length},
{"log",		Node_builtin,	 LEX_BUILTIN,	A(1),		do_log},
{"match",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(2),	do_match},
{"next",	Node_K_next,	 LEX_NEXT,	0,		0},
{"print",	Node_K_print,	 LEX_PRINT,	0,		0},
{"printf",	Node_K_printf,	 LEX_PRINTF,	0,		0},
{"rand",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(0),	do_rand},
{"return",	Node_K_return,	 LEX_RETURN,	NOT_OLD,	0},
{"sin",		Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(1),	do_sin},
{"split",	Node_builtin,	 LEX_BUILTIN,	A(2)|A(3),	do_split},
{"sprintf",	Node_builtin,	 LEX_BUILTIN,	0,		do_sprintf},
{"sqrt",	Node_builtin,	 LEX_BUILTIN,	A(1),		do_sqrt},
{"srand",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(0)|A(1), do_srand},
{"strftime",	Node_builtin,	 LEX_BUILTIN,	GAWKX|A(1)|A(2), do_strftime},
{"sub",		Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(2)|A(3), do_sub},
{"substr",	Node_builtin,	 LEX_BUILTIN,	A(2)|A(3),	do_substr},
{"system",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(1),	do_system},
{"systime",	Node_builtin,	 LEX_BUILTIN,	GAWKX|A(0),	do_systime},
{"tolower",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(1),	do_tolower},
{"toupper",	Node_builtin,	 LEX_BUILTIN,	NOT_OLD|A(1),	do_toupper},
{"while",	Node_K_while,	 LEX_WHILE,	0,		0},
};

/* VARARGS0 */
static void
yyerror(va_alist)
va_dcl
{
	va_list args;
	char *mesg = NULL;
	register char *bp, *cp;
	char *scan;
	char buf[120];

	errcount++;
	/* Find the current line in the input file */
	if (lexptr) {
		if (!thisline) {
			cp = lexeme;
			if (*cp == '\n') {
				cp--;
				mesg = "unexpected newline";
			}
			for ( ; cp != lexptr_begin && *cp != '\n'; --cp)
				;
			if (*cp == '\n')
				cp++;
			thisline = cp;
		}
		/* NL isn't guaranteed */
		bp = lexeme;
		while (bp < lexend && *bp && *bp != '\n')
			bp++;
	} else {
		thisline = "(END OF FILE)";
		bp = thisline + 13;
	}
	msg("%.*s", (int) (bp - thisline), thisline);
	bp = buf;
	cp = buf + sizeof(buf) - 24;	/* 24 more than longest msg. input */
	if (lexptr) {
		scan = thisline;
		while (bp < cp && scan < lexeme)
			if (*scan++ == '\t')
				*bp++ = '\t';
			else
				*bp++ = ' ';
		*bp++ = '^';
		*bp++ = ' ';
	}
	va_start(args);
	if (mesg == NULL)
		mesg = va_arg(args, char *);
	strcpy(bp, mesg);
	err("", buf, args);
	va_end(args);
	exit(2);
}

static char *
get_src_buf()
{
	static int samefile = 0;
	static int nextfile = 0;
	static char *buf = NULL;
	static int fd;
	int n;
	register char *scan;
	static int len = 0;
	static int did_newline = 0;
#	define	SLOP	128	/* enough space to hold most source lines */

	if (nextfile > numfiles)
		return NULL;

	if (srcfiles[nextfile].stype == CMDLINE) {
		if (len == 0) {
			len = strlen(srcfiles[nextfile].val);
			sourceline = 1;
			lexptr = lexptr_begin = srcfiles[nextfile].val;
			lexend = lexptr + len;
		} else if (!did_newline && *(lexptr-1) != '\n') {
			/*
			 * The following goop is to ensure that the source
			 * ends with a newline and that the entire current
			 * line is available for error messages.
			 */
			int offset;

			did_newline = 1;
			offset = lexptr - lexeme;
			for (scan = lexeme; scan > lexptr_begin; scan--)
				if (*scan == '\n') {
					scan++;
					break;
				}
			len = lexptr - scan;
			emalloc(buf, char *, len+1, "get_src_buf");
			memcpy(buf, scan, len);
			thisline = buf;
			lexptr = buf + len;
			*lexptr = '\n';
			lexeme = lexptr - offset;
			lexptr_begin = buf;
			lexend = lexptr + 1;
		} else {
			len = 0;
			lexeme = lexptr = lexptr_begin = NULL;
		}
		if (lexptr == NULL && ++nextfile <= numfiles)
			return get_src_buf();
		return lexptr;
	}
	if (!samefile) {
		source = srcfiles[nextfile].val;
		if (source == NULL) {
			if (buf) {
				free(buf);
				buf = NULL;
			}
			len = 0;
			return lexeme = lexptr = lexptr_begin = NULL;
		}
		fd = pathopen(source);
		if (fd == -1)
			fatal("can't open source file \"%s\" for reading (%s)",
				source, strerror(errno));
		len = optimal_bufsize(fd);
		if (buf)
			free(buf);
		emalloc(buf, char *, len + SLOP, "get_src_buf");
		lexptr_begin = buf + SLOP;
		samefile = 1;
		sourceline = 1;
	} else {
		/*
		 * Here, we retain the current source line (up to length SLOP)
		 * in the beginning of the buffer that was overallocated above
		 */
		int offset;
		int linelen;

		offset = lexptr - lexeme;
		for (scan = lexeme; scan > lexptr_begin; scan--)
			if (*scan == '\n') {
				scan++;
				break;
			}
		linelen = lexptr - scan;
		if (linelen > SLOP)
			linelen = SLOP;
		thisline = buf + SLOP - linelen;
		memcpy(thisline, scan, linelen);
		lexeme = buf + SLOP - offset;
		lexptr_begin = thisline;
	}
	n = read(fd, buf + SLOP, len);
	if (n == -1)
		fatal("can't read sourcefile \"%s\" (%s)",
			source, strerror(errno));
	if (n == 0) {
		samefile = 0;
		nextfile++;
		len = 0;
		return get_src_buf();
	}
	lexptr = buf + SLOP;
	lexend = lexptr + n;
	return buf;
}

#define	tokadd(x) (*token++ = (x), token == tokend ? tokexpand() : token)

char *
tokexpand()
{
	static int toksize = 60;
	int tokoffset;

	tokoffset = token - tokstart;
	toksize *= 2;
	if (tokstart)
		erealloc(tokstart, char *, toksize, "tokexpand");
	else
		emalloc(tokstart, char *, toksize, "tokexpand");
	tokend = tokstart + toksize;
	token = tokstart + tokoffset;
	return token;
}

#if DEBUG
char
nextc() {
	if (lexptr && lexptr < lexend)
		return *lexptr++;
	else if (get_src_buf())
		return *lexptr++;
	else
		return '\0';
}
#else
#define	nextc()	((lexptr && lexptr < lexend) ? \
			*lexptr++ : \
			(get_src_buf() ? *lexptr++ : '\0') \
		)
#endif
#define pushback() (lexptr && lexptr > lexptr_begin ? lexptr-- : lexptr)

/*
 * Read the input and turn it into tokens.
 */

static int
yylex()
{
	register int c;
	int seen_e = 0;		/* These are for numbers */
	int seen_point = 0;
	int esc_seen;		/* for literal strings */
	int low, mid, high;
	static int did_newline = 0;
	char *tokkey;

	if (!nextc())
		return 0;
	pushback();
	lexeme = lexptr;
	thisline = NULL;
	if (want_regexp) {
		int in_brack = 0;

		want_regexp = 0;
		token = tokstart;
		while ((c = nextc()) != 0) {
			switch (c) {
			case '[':
				in_brack = 1;
				break;
			case ']':
				in_brack = 0;
				break;
			case '\\':
				if ((c = nextc()) == '\0') {
					yyerror("unterminated regexp ends with \\ at end of file");
				} else if (c == '\n') {
					sourceline++;
					continue;
				} else
					tokadd('\\');
				break;
			case '/':	/* end of the regexp */
				if (in_brack)
					break;

				pushback();
				tokadd('\0');
				yylval.sval = tokstart;
				return REGEXP;
			case '\n':
				pushback();
				yyerror("unterminated regexp");
			case '\0':
				yyerror("unterminated regexp at end of file");
			}
			tokadd(c);
		}
	}
retry:
	while ((c = nextc()) == ' ' || c == '\t')
		;

	lexeme = lexptr ? lexptr - 1 : lexptr;
	thisline = NULL;
	token = tokstart;
	yylval.nodetypeval = Node_illegal;

	switch (c) {
	case 0:
		return 0;

	case '\n':
		sourceline++;
		return NEWLINE;

	case '#':		/* it's a comment */
		while ((c = nextc()) != '\n') {
			if (c == '\0')
				return 0;
		}
		sourceline++;
		return NEWLINE;

	case '\\':
#ifdef RELAXED_CONTINUATION
		if (!do_unix) {	/* strip trailing white-space and/or comment */
			while ((c = nextc()) == ' ' || c == '\t') continue;
			if (c == '#')
				while ((c = nextc()) != '\n') if (!c) break;
			pushback();
		}
#endif /*RELAXED_CONTINUATION*/
		if (nextc() == '\n') {
			sourceline++;
			goto retry;
		} else
			yyerror("inappropriate use of backslash");
		break;

	case '$':
		want_assign = 1;
		return '$';

	case ')':
	case ']':
	case '(':	
	case '[':
	case ';':
	case ':':
	case '?':
	case '{':
	case ',':
		return c;

	case '*':
		if ((c = nextc()) == '=') {
			yylval.nodetypeval = Node_assign_times;
			return ASSIGNOP;
		} else if (do_posix) {
			pushback();
			return '*';
		} else if (c == '*') {
			/* make ** and **= aliases for ^ and ^= */
			static int did_warn_op = 0, did_warn_assgn = 0;

			if (nextc() == '=') {
				if (do_lint && ! did_warn_assgn) {
					did_warn_assgn = 1;
					warning("**= is not allowed by POSIX");
				}
				yylval.nodetypeval = Node_assign_exp;
				return ASSIGNOP;
			} else {
				pushback();
				if (do_lint && ! did_warn_op) {
					did_warn_op = 1;
					warning("** is not allowed by POSIX");
				}
				return '^';
			}
		}
		pushback();
		return '*';

	case '/':
		if (want_assign) {
			if (nextc() == '=') {
				yylval.nodetypeval = Node_assign_quotient;
				return ASSIGNOP;
			}
			pushback();
		}
		return '/';

	case '%':
		if (nextc() == '=') {
			yylval.nodetypeval = Node_assign_mod;
			return ASSIGNOP;
		}
		pushback();
		return '%';

	case '^':
	{
		static int did_warn_op = 0, did_warn_assgn = 0;

		if (nextc() == '=') {

			if (do_lint && ! did_warn_assgn) {
				did_warn_assgn = 1;
				warning("operator `^=' is not supported in old awk");
			}
			yylval.nodetypeval = Node_assign_exp;
			return ASSIGNOP;
		}
		pushback();
		if (do_lint && ! did_warn_op) {
			did_warn_op = 1;
			warning("operator `^' is not supported in old awk");
		}
		return '^';
	}