dt_lex.l

来自「Sun Solaris 10 中的 DTrace 组件的源代码。请参看: htt」· L 代码 · 共 794 行 · 第 1/2 页

%{
/*
 * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License, Version 1.0 only.
 * See the file usr/src/LICENSING.NOTICE in this distribution or
 * http://www.opensolaris.org/license/ for details.
 */

#pragma ident	"@(#)dt_lex.l	1.6	04/11/21 SMI"

#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include <assert.h>
#include <ctype.h>
#include <errno.h>

#include <dt_impl.h>
#include <dt_grammar.h>
#include <dt_parser.h>
#include <dt_string.h>

/*
 * We need to undefine lex's input and unput macros so that references to these
 * call the functions provided at the end of this source file.
 */
#undef input
#undef unput

static int id_or_type(const char *);
static int input(void);
static void unput(int);

/*
 * We first define a set of labeled states for use in the D lexer and then a
 * set of regular expressions to simplify things below.  The lexer states are:
 *
 * S0 - D program clause and expression lexing
 * S1 - D comments (i.e. skip everything until end of comment)
 * S2 - D program outer scope (probe specifiers and declarations)
 * S3 - D control line parsing (i.e. after ^# is seen but before \n)
 */
%}

%e 1400		/* maximum nodes */
%p 3700		/* maximum positions */

%s S0 S1 S2 S3

RGX_AGG		"@"[a-zA-Z_][0-9a-zA-Z_]*
RGX_PSPEC	[-$:a-zA-Z_.?*\\\[\]!][-$:0-9a-zA-Z_.`?*\\\[\]!]*
RGX_IDENT	[a-zA-Z_`][0-9a-zA-Z_`]*
RGX_INT		([0-9]+|0[xX][0-9A-Fa-f]+)[uU]?[lL]?[lL]?
RGX_FP		([0-9]+("."?)[0-9]*|"."[0-9]+)((e|E)("+"|-)?[0-9]+)?[fFlL]?
RGX_WS		[\f\n\r\t\v ]
RGX_STR		([^"\\\n]|\\[^"\n]|\\\")*
RGX_CHR		([^'\\\n]|\\[^'\n]|\\')*
RGX_INTERP	^[\f\t\v ]*#!.*
RGX_CTL		^[\f\t\v ]*#

%%

%{

/*
 * We insert a special prologue into yylex() itself: if the pcb contains a
 * context token, we return that prior to running the normal lexer.  This
 * allows libdtrace to force yacc into one of our two parsing contexts:
 * parsing a D expression (DT_CTX_DEXPR) or a D program (DT_CTX_DPROG).
 * Once the token is returned, we clear it so this only happens once.
 */
if (yypcb->pcb_token != 0) {
	int tok = yypcb->pcb_token;
	yypcb->pcb_token = 0;
	return (tok);
}

%}

<S0>auto	return (DT_KEY_AUTO);
<S0>break	return (DT_KEY_BREAK);
<S0>case	return (DT_KEY_CASE);
<S0>char	return (DT_KEY_CHAR);
<S0>const	return (DT_KEY_CONST);
<S0>continue	return (DT_KEY_CONTINUE);
<S0>counter	return (DT_KEY_COUNTER);
<S0>default	return (DT_KEY_DEFAULT);
<S0>do		return (DT_KEY_DO);
<S0>double	return (DT_KEY_DOUBLE);
<S0>else	return (DT_KEY_ELSE);
<S0>enum	return (DT_KEY_ENUM);
<S0>extern	return (DT_KEY_EXTERN);
<S0>float	return (DT_KEY_FLOAT);
<S0>for		return (DT_KEY_FOR);
<S0>goto	return (DT_KEY_GOTO);
<S0>if		return (DT_KEY_IF);
<S0>import	return (DT_KEY_IMPORT);
<S0>inline	return (DT_KEY_INLINE);
<S0>int		return (DT_KEY_INT);
<S0>long	return (DT_KEY_LONG);
<S0>offsetof	return (DT_TOK_OFFSETOF);
<S0>probe	return (DT_KEY_PROBE);
<S0>provider	return (DT_KEY_PROVIDER);
<S0>register	return (DT_KEY_REGISTER);
<S0>restrict	return (DT_KEY_RESTRICT);
<S0>return	return (DT_KEY_RETURN);
<S0>self	return (DT_KEY_SELF);
<S0>short	return (DT_KEY_SHORT);
<S0>signed	return (DT_KEY_SIGNED);
<S0>sizeof	return (DT_TOK_SIZEOF);
<S0>static	return (DT_KEY_STATIC);
<S0>string	return (DT_KEY_STRING);
<S0>stringof	return (DT_TOK_STRINGOF);
<S0>struct	return (DT_KEY_STRUCT);
<S0>switch	return (DT_KEY_SWITCH);
<S0>this	return (DT_KEY_THIS);
<S0>translator	return (DT_KEY_XLATOR);
<S0>typedef	return (DT_KEY_TYPEDEF);
<S0>union	return (DT_KEY_UNION);
<S0>unsigned	return (DT_KEY_UNSIGNED);
<S0>void	return (DT_KEY_VOID);
<S0>volatile	return (DT_KEY_VOLATILE);
<S0>while	return (DT_KEY_WHILE);
<S0>xlate	return (DT_TOK_XLATE);

<S2>auto	{ yybegin(YYS_EXPR);	return (DT_KEY_AUTO); }
<S2>char	{ yybegin(YYS_EXPR);	return (DT_KEY_CHAR); }
<S2>const	{ yybegin(YYS_EXPR);	return (DT_KEY_CONST); }
<S2>counter	{ yybegin(YYS_DEFINE);	return (DT_KEY_COUNTER); }
<S2>double	{ yybegin(YYS_EXPR);	return (DT_KEY_DOUBLE); }
<S2>enum	{ yybegin(YYS_EXPR);	return (DT_KEY_ENUM); }
<S2>extern	{ yybegin(YYS_EXPR);	return (DT_KEY_EXTERN); }
<S2>float	{ yybegin(YYS_EXPR);	return (DT_KEY_FLOAT); }
<S2>import	{ yybegin(YYS_EXPR);	return (DT_KEY_IMPORT); }
<S2>inline	{ yybegin(YYS_DEFINE);	return (DT_KEY_INLINE); }
<S2>int		{ yybegin(YYS_EXPR);	return (DT_KEY_INT); }
<S2>long	{ yybegin(YYS_EXPR);	return (DT_KEY_LONG); }
<S2>provider	{ yybegin(YYS_DEFINE);	return (DT_KEY_PROVIDER); }
<S2>register	{ yybegin(YYS_EXPR);	return (DT_KEY_REGISTER); }
<S2>restrict	{ yybegin(YYS_EXPR);	return (DT_KEY_RESTRICT); }
<S2>self	{ yybegin(YYS_EXPR);	return (DT_KEY_SELF); }
<S2>short	{ yybegin(YYS_EXPR);	return (DT_KEY_SHORT); }
<S2>signed	{ yybegin(YYS_EXPR);	return (DT_KEY_SIGNED); }
<S2>static	{ yybegin(YYS_EXPR);	return (DT_KEY_STATIC); }
<S2>string	{ yybegin(YYS_EXPR);	return (DT_KEY_STRING); }
<S2>struct	{ yybegin(YYS_EXPR);	return (DT_KEY_STRUCT); }
<S2>this	{ yybegin(YYS_EXPR);	return (DT_KEY_THIS); }
<S2>translator	{ yybegin(YYS_DEFINE);	return (DT_KEY_XLATOR); }
<S2>typedef	{ yybegin(YYS_EXPR);	return (DT_KEY_TYPEDEF); }
<S2>union	{ yybegin(YYS_EXPR);	return (DT_KEY_UNION); }
<S2>unsigned	{ yybegin(YYS_EXPR);	return (DT_KEY_UNSIGNED); }
<S2>void	{ yybegin(YYS_EXPR);	return (DT_KEY_VOID); }
<S2>volatile	{ yybegin(YYS_EXPR);	return (DT_KEY_VOLATILE); }

<S0>"$$"[0-9]+	{
			int i = atoi(yytext + 2);
			char *v = "";

			/*
			 * A macro argument reference substitutes the text of
			 * an argument in place of the current token.  When we
			 * see $$<d> we fetch the saved string from pcb_sargv
			 * (or use the default argument if the option has been
			 * set and the argument hasn't been specified) and
			 * return a token corresponding to this string.
			 */
			if (i < 0 || (i >= yypcb->pcb_sargc &&
			    !(yypcb->pcb_cflags & DTRACE_C_DEFARG))) {
				xyerror(D_MACRO_UNDEF, "macro argument %s is "
				    "not defined\n", yytext);
			}

			if (i < yypcb->pcb_sargc) {
				v = yypcb->pcb_sargv[i]; /* get val from pcb */
				yypcb->pcb_sflagv[i] |= DT_IDFLG_REF;
			}

			if ((yylval.l_str = strdup(v)) == NULL)
				longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM);

			(void) stresc2chr(yylval.l_str);
			return (DT_TOK_STRING);
		}

<S0>"$"[0-9]+	{
			int i = atoi(yytext + 1);
			char *p, *v = "0";

			/*
			 * A macro argument reference substitutes the text of
			 * one identifier or integer pattern for another.  When
			 * we see $<d> we fetch the saved string from pcb_sargv
			 * (or use the default argument if the option has been
			 * set and the argument hasn't been specified) and
			 * return a token corresponding to this string.
			 */
			if (i < 0 || (i >= yypcb->pcb_sargc &&
			    !(yypcb->pcb_cflags & DTRACE_C_DEFARG))) {
				xyerror(D_MACRO_UNDEF, "macro argument %s is "
				    "not defined\n", yytext);
			}

			if (i < yypcb->pcb_sargc) {
				v = yypcb->pcb_sargv[i]; /* get val from pcb */
				yypcb->pcb_sflagv[i] |= DT_IDFLG_REF;
			}

			/*
			 * If the macro text is not a valid integer or ident,
			 * then we treat it as a string.  The string may be
			 * optionally enclosed in quotes, which we strip.
			 */
			if (strbadidnum(v)) {
				size_t len = strlen(v);

				if (len != 1 && *v == '"' && v[len - 1] == '"')
					yylval.l_str = strndup(v + 1, len - 2);
				else
					yylval.l_str = strndup(v, len);

				if (yylval.l_str == NULL)
					longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM);

				(void) stresc2chr(yylval.l_str);
				return (DT_TOK_STRING);
			}

			/*
			 * If the macro text is not a string an begins with a
			 * digit or a +/- sign, process it as an integer token.
			 */
			if (isdigit(v[0]) || v[0] == '-' || v[0] == '+') {
				if (isdigit(v[0]))
					yyintprefix = 0;
				else
					yyintprefix = *v++;

				errno = 0;
				yylval.l_int = strtoull(v, &p, 0);
				(void) strncpy(yyintsuffix, p,
				    sizeof (yyintsuffix));
				yyintdecimal = *v != '0';

				if (errno == ERANGE) {
					xyerror(D_MACRO_OFLOW, "macro argument"
					    " %s constant %s results in integer"
					    " overflow\n", yytext, v);
				}

				return (DT_TOK_INT);
			}

			return (id_or_type(v));
		}

<S0>"$$"{RGX_IDENT} {
			dt_ident_t *idp = dt_idhash_lookup(
			    yypcb->pcb_hdl->dt_macros, yytext + 2);

			char s[16]; /* enough for UINT_MAX + \0 */

			if (idp == NULL) {
				xyerror(D_MACRO_UNDEF, "macro variable %s "
				    "is not defined\n", yytext);
			}

			/*
			 * For the moment, all current macro variables are of
			 * type id_t (refer to dtrace_update() for details).
			 */
			(void) snprintf(s, sizeof (s), "%u", idp->di_id);
			if ((yylval.l_str = strdup(s)) == NULL)
				longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM);

			return (DT_TOK_STRING);
		}

<S0>"$"{RGX_IDENT} {
			dt_ident_t *idp = dt_idhash_lookup(
			    yypcb->pcb_hdl->dt_macros, yytext + 1);

			if (idp == NULL) {
				xyerror(D_MACRO_UNDEF, "macro variable %s "
				    "is not defined\n", yytext);
			}

			/*
			 * For the moment, all current macro variables are of
			 * type id_t (refer to dtrace_update() for details).
			 */
			yylval.l_int = (intmax_t)(int)idp->di_id;
			yyintprefix = 0;
			yyintsuffix[0] = '\0';
			yyintdecimal = 1;

			return (DT_TOK_INT);
		}

<S0>{RGX_IDENT}	{
			return (id_or_type(yytext));
		}

<S0>{RGX_AGG}	{
			if ((yylval.l_str = strdup(yytext)) == NULL)
				longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM);
			return (DT_TOK_AGG);
		}

<S0>"@"		{
			if ((yylval.l_str = strdup("@_")) == NULL)
				longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM);
			return (DT_TOK_AGG);
		}

<S0>{RGX_INT}	|
<S2>{RGX_INT}	|
<S3>{RGX_INT}	{
			char *p;

			errno = 0;
			yylval.l_int = strtoull(yytext, &p, 0);
			yyintprefix = 0;
			(void) strncpy(yyintsuffix, p, sizeof (yyintsuffix));
			yyintdecimal = yytext[0] != '0';

			if (errno == ERANGE) {
				xyerror(D_INT_OFLOW, "constant %s results in "
				    "integer overflow\n", yytext);
			}

			if (*p != '\0' && strchr("uUlL", *p) == NULL) {
				xyerror(D_INT_DIGIT, "constant %s contains "
				    "invalid digit %c\n", yytext, *p);
			}

			if ((YYSTATE) != S3)
				return (DT_TOK_INT);

			yypragma = dt_node_link(yypragma,
			    dt_node_int(yylval.l_int));
		}

<S0>{RGX_FP}	yyerror("floating-point constants are not permitted\n");

<S0>\"{RGX_STR}$ |
<S3>\"{RGX_STR}$ xyerror(D_STR_NL, "newline encountered in string literal");

<S0>\"{RGX_STR}\" |
<S3>\"{RGX_STR}\" {
			/*
			 * Quoted string -- convert C escape sequences and
			 * return the string as a token.
			 */
			yylval.l_str = strndup(yytext + 1, yyleng - 2);

			if (yylval.l_str == NULL)
				longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM);

			(void) stresc2chr(yylval.l_str);
			if ((YYSTATE) != S3)
				return (DT_TOK_STRING);

			yypragma = dt_node_link(yypragma,
			    dt_node_string(yylval.l_str));
		}

<S0>'{RGX_CHR}$	xyerror(D_CHR_NL, "newline encountered in character constant");

<S0>'{RGX_CHR}'	{
			char *s, *p, *q;
			size_t nbytes;

			/*
			 * Character constant -- convert C escape sequences and
			 * return the character as an integer immediate value.
			 */
			if (yyleng == 2)
				xyerror(D_CHR_NULL, "empty character constant");

			s = yytext + 1;
			yytext[yyleng - 1] = '\0';
			nbytes = stresc2chr(s);
			yylval.l_int = 0;
			yyintprefix = 0;
			yyintsuffix[0] = '\0';
			yyintdecimal = 1;

			if (nbytes > sizeof (yylval.l_int)) {
				xyerror(D_CHR_OFLOW, "character constant is "
				    "too long");
			}
#ifdef _LITTLE_ENDIAN
			p = ((char *)&yylval.l_int) + nbytes - 1;
			for (q = s; nbytes != 0; nbytes--)
				*p-- = *q++;