From 1fa31332b607919d11f639543bdadf1c910c2c17 Mon Sep 17 00:00:00 2001
From: Peter Johnson <peter@tortall.net>
Date: Wed, 25 Jul 2001 00:33:10 +0000
Subject: [PATCH] Add lexing, parsing, and internal format for strings and
 character constants. Completely handle data declarations (db/dw/etc) and
 reserve space declarations (resb/resw/etc).

svn path=/trunk/yasm/; revision=129
---
 libyasm/bytecode.c                | 183 +++++++++++++++++++++++++++---
 libyasm/bytecode.h                |  41 ++++++-
 libyasm/errwarn.c                 |  12 +-
 libyasm/errwarn.h                 |  12 +-
 modules/parsers/nasm/bison.y.in   |  71 ++++++++++--
 modules/parsers/nasm/nasm-bison.y |  71 ++++++++++--
 modules/parsers/nasm/token.l.in   |  46 +++++++-
 src/bison.y.in                    |  71 ++++++++++--
 src/bytecode.c                    | 183 +++++++++++++++++++++++++++---
 src/bytecode.h                    |  41 ++++++-
 src/errwarn.c                     |  12 +-
 src/errwarn.h                     |  12 +-
 src/parsers/nasm/bison.y.in       |  71 ++++++++++--
 src/parsers/nasm/nasm-bison.y     |  71 ++++++++++--
 src/parsers/nasm/token.l.in       |  46 +++++++-
 src/token.l.in                    |  46 +++++++-
 16 files changed, 893 insertions(+), 96 deletions(-)

diff --git a/libyasm/bytecode.c b/libyasm/bytecode.c
index 4f3e631e..9af7ad0d 100644
--- a/libyasm/bytecode.c
+++ b/libyasm/bytecode.c
@@ -1,4 +1,4 @@
-/* $Id: bytecode.c,v 1.13 2001/07/11 23:16:50 peter Exp $
+/* $Id: bytecode.c,v 1.14 2001/07/25 00:33:10 peter Exp $
  * Bytecode utility functions
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -20,6 +20,7 @@
  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 #include <stdio.h>
+#include <stdlib.h>
 #include "globals.h"
 #include "bytecode.h"
 #include "errwarn.h"
@@ -336,21 +337,77 @@ BuildBC_JmpRel (bytecode      *bc,
     BuildBC_Common(bc);
 }
 
-/* TODO: implement.  Shouldn't be difficult. */
-unsigned char *
-ConvertBCInsnToBytes (unsigned char *ptr, bytecode *bc, int *len)
+void
+BuildBC_Data (bytecode      *bc,
+	      dataval       *data,
+	      unsigned long  size)
 {
-    if(bc->type != BC_INSN)
-	return (unsigned char *)NULL;
-    return (unsigned char *)NULL;
+    dataval *cur = data;
+
+    /* First check to see if all the data elements are valid for the size
+     * being set.
+     * Validity table:
+     *  db (1)  -> expr, string
+     *  dw (2)  -> expr, string
+     *  dd (4)  -> expr, float, string
+     *  dq (8)  -> expr, float, string
+     *  dt (10) -> float, string
+     *
+     * Once we calculate expr we'll have to validate it against the size
+     * and warn/error appropriately (symbol constants versus labels:
+     * constants (equ's) should always be legal, but labels should raise
+     * warnings when used in db or dq context at the minimum).
+     */
+    while(cur) {
+	switch(cur->type) {
+	    case DV_EMPTY:
+	    case DV_STRING:
+		/* string is valid in every size */
+		break;
+	    case DV_FLOAT:
+		if(size == 1)
+		    Error(ERR_DECLDATA_FLOAT, (char *)NULL, "DB");
+		else if(size == 2)
+		    Error(ERR_DECLDATA_FLOAT, (char *)NULL, "DW");
+		break;
+	    case DV_EXPR:
+		if(size == 10)
+		    Error(ERR_DECLDATA_EXPR, (char *)NULL, "DT");
+		break;
+	}
+	cur = cur->next;
+    }
+
+    bc->next = (bytecode *)NULL;
+    bc->type = BC_DATA;
+
+    bc->data.data.data = data;
+    bc->data.data.size = size;
+
+    BuildBC_Common(bc);
 }
 
 void
-DebugPrintBC (bytecode *bc)
+BuildBC_Reserve (bytecode      *bc,
+		 expr          *numitems,
+		 unsigned long  itemsize)
 {
-    unsigned long i;
+    bc->next = (bytecode *)NULL;
+    bc->type = BC_RESERVE;
 
+    bc->data.reserve.numitems = numitems;
+    bc->data.reserve.itemsize = itemsize;
+
+    BuildBC_Common(bc);
+}
+
+void
+DebugPrintBC (bytecode *bc)
+{
     switch(bc->type) {
+	case BC_EMPTY:
+	    printf("_Empty_\n");
+	    break;
 	case BC_INSN:
 	    printf("_Instruction_\n");
 	    printf("Effective Address:\n");
@@ -431,14 +488,17 @@ DebugPrintBC (bytecode *bc)
 	    break;
 	case BC_DATA:
 	    printf("_Data_\n");
-	    for(i=0; i<bc->len; i++) {
-		printf("%2x ", (unsigned int)bc->data.data.data[i]);
-		if((i & ~0xFFFF) == i)
-		    printf("\n");
-	    }
+	    printf("Final Element Size=%u\n",
+		(unsigned int)bc->data.data.size);
+	    printf("Elements:\n");
+	    dataval_print(bc->data.data.data);
 	    break;
 	case BC_RESERVE:
 	    printf("_Reserve_\n");
+	    printf("Num Items=");
+	    expr_print(bc->data.reserve.numitems);
+	    printf("\nItem Size=%u\n",
+		(unsigned int)bc->data.reserve.itemsize);
 	    break;
 	default:
 	    printf("_Unknown_\n");
@@ -448,3 +508,98 @@ DebugPrintBC (bytecode *bc)
 	bc->filename ? bc->filename : "<UNKNOWN>", bc->lineno);
     printf("Offset=%lx BITS=%u\n", bc->offset, bc->mode_bits);
 }
+
+dataval *
+dataval_new_expr (expr *exp)
+{
+    dataval *retval = malloc(sizeof(dataval));
+
+    if(!retval)
+	Fatal(FATAL_NOMEM);
+
+    retval->next = (dataval *)NULL;
+    retval->last = retval;
+    retval->type = DV_EXPR;
+    retval->data.exp = exp;
+
+    return retval;
+}
+
+dataval *
+dataval_new_float (double float_val)
+{
+    dataval *retval = malloc(sizeof(dataval));
+
+    if(!retval)
+	Fatal(FATAL_NOMEM);
+
+    retval->next = (dataval *)NULL;
+    retval->last = retval;
+    retval->type = DV_FLOAT;
+    retval->data.float_val = float_val;
+
+    return retval;
+}
+
+dataval *
+dataval_new_string (char *str_val)
+{
+    dataval *retval = malloc(sizeof(dataval));
+
+    if(!retval)
+	Fatal(FATAL_NOMEM);
+
+    retval->next = (dataval *)NULL;
+    retval->last = retval;
+    retval->type = DV_STRING;
+    retval->data.str_val = str_val;
+
+    return retval;
+}
+
+dataval *
+dataval_append (dataval *list, dataval *item)
+{
+    if(item)
+	item->next = (dataval *)NULL;
+
+    if(!list) {
+	item->last = item;
+	return item;
+    } else {
+	if(item) {
+	    list->last->next = item;
+	    list->last = item;
+	    item->last = (dataval *)NULL;
+	}
+	return list;
+    }
+}
+
+void
+dataval_print (dataval *start)
+{
+    dataval *cur = start;
+
+    while(cur) {
+	switch(cur->type) {
+	    case DV_EMPTY:
+		printf(" Empty\n");
+		break;
+	    case DV_EXPR:
+		printf(" Expr=");
+		expr_print(cur->data.exp);
+		printf("\n");
+		break;
+	    case DV_FLOAT:
+		printf(" Float=%e\n", cur->data.float_val);
+		break;
+	    case DV_STRING:
+		printf(" String=%s\n", cur->data.str_val);
+		break;
+	}
+
+	cur = cur->next;
+    }
+}
+
diff --git a/libyasm/bytecode.h b/libyasm/bytecode.h
index 6b971c33..fafb3b44 100644
--- a/libyasm/bytecode.h
+++ b/libyasm/bytecode.h
@@ -1,4 +1,4 @@
-/* $Id: bytecode.h,v 1.14 2001/07/11 23:16:50 peter Exp $
+/* $Id: bytecode.h,v 1.15 2001/07/25 00:33:10 peter Exp $
  * Bytecode utility functions header file
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -47,6 +47,18 @@ typedef struct immval_s {
     unsigned char f_sign;	/* 1 if final imm should be signed */
 } immval;
 
+typedef struct dataval_s {
+    struct dataval_s *next, *last;
+
+    enum { DV_EMPTY, DV_EXPR, DV_FLOAT, DV_STRING } type;
+
+    union {
+	struct expr_s *exp;
+	double float_val;
+	char *str_val;
+    } data;
+} dataval;
+
 typedef enum jmprel_opcode_sel_e {
     JR_NONE,
     JR_SHORT,
@@ -64,7 +76,7 @@ typedef struct targetval_s {
 typedef struct bytecode_s {
     struct bytecode_s *next;
 
-    enum { BC_INSN, BC_JMPREL, BC_DATA, BC_RESERVE } type;
+    enum { BC_EMPTY, BC_INSN, BC_JMPREL, BC_DATA, BC_RESERVE } type;
 
     union {
 	struct {
@@ -97,8 +109,15 @@ typedef struct bytecode_s {
 	    unsigned char lockrep_pre;	/* 0 indicates no prefix */
 	} jmprel;
 	struct {
-	    unsigned char *data;
+	    dataval *data;		/* non-converted data (linked list) */
+
+	    /* final (converted) size of each element (in bytes) */
+	    unsigned char size;
 	} data;
+	struct {
+	    struct expr_s *numitems;	/* number of items to reserve */
+	    unsigned char itemsize;	/* size of each item (in bytes) */
+	} reserve;
     } data;
 
     unsigned long len;	/* total length of entire bytecode */
@@ -154,8 +173,22 @@ void BuildBC_JmpRel(bytecode      *bc,
 		    unsigned char  near_op2,
 		    unsigned char  addrsize);
 
-unsigned char *ConvertBCInsnToBytes(unsigned char *ptr, bytecode *bc, int *len);
+void BuildBC_Data(bytecode      *bc,
+		  dataval       *data,
+		  unsigned long  size);
+
+void BuildBC_Reserve(bytecode      *bc,
+		     struct expr_s *numitems,
+		     unsigned long  itemsize);
 
 void DebugPrintBC(bytecode *bc);
 
+dataval *dataval_new_expr(struct expr_s *exp);
+dataval *dataval_new_float(double float_val);
+dataval *dataval_new_string(char *str_val);
+
+dataval *dataval_append(dataval *list, dataval *item);
+
+void dataval_print(dataval *start);
+
 #endif
diff --git a/libyasm/errwarn.c b/libyasm/errwarn.c
index cd296f64..b6c4e263 100644
--- a/libyasm/errwarn.c
+++ b/libyasm/errwarn.c
@@ -1,4 +1,4 @@
-/* $Id: errwarn.c,v 1.16 2001/07/11 23:16:50 peter Exp $
+/* $Id: errwarn.c,v 1.17 2001/07/25 00:33:10 peter Exp $
  * Error and warning reporting and related functions.
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -67,7 +67,12 @@ static char *err_msgs[] = {
     "expression syntax error",
     "duplicate definition of `%1'; previously defined line %2",
     "mismatch in operand sizes",
-    "no %s form of that jump instruction exists"
+    "no %s form of that jump instruction exists",
+    "unterminated string",
+    "unexpected end of file in string",
+    "expression syntax error",
+    "floating-point constant encountered in `%s'",
+    "non-floating-point value encountered in `%s'"
 };
 
 /* Warning messages.  Match up with warn_num enum in errwarn.h. */
@@ -78,7 +83,8 @@ static char *warn_msgs[] = {
     "multiple segment overrides, using leftmost",
     "multiple LOCK or REP prefixes, using leftmost",
     "no non-local label before '%s'",
-    "multiple SHORT or NEAR specifiers, using leftmost"
+    "multiple SHORT or NEAR specifiers, using leftmost",
+    "character constant too large, ignoring trailing characters"
 };
 
 /* I hate to define these strings as static buffers; a better solution would be
diff --git a/libyasm/errwarn.h b/libyasm/errwarn.h
index a1e2044b..51bf7175 100644
--- a/libyasm/errwarn.h
+++ b/libyasm/errwarn.h
@@ -1,4 +1,4 @@
-/* $Id: errwarn.h,v 1.10 2001/07/11 23:16:50 peter Exp $
+/* $Id: errwarn.h,v 1.11 2001/07/25 00:33:10 peter Exp $
  * Error and warning reporting and related functions header file.
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -45,7 +45,12 @@ typedef enum {
     ERR_EXP_SYNTAX,
     ERR_DUPLICATE_DEF,
     ERR_OP_SIZE_MISMATCH,
-    ERR_NO_JMPREL_FORM
+    ERR_NO_JMPREL_FORM,
+    ERR_STRING_UNTERM,
+    ERR_STRING_EOF,
+    ERR_EXPR_SYNTAX,
+    ERR_DECLDATA_FLOAT,
+    ERR_DECLDATA_EXPR
 } err_num;
 
 /* Warning constants.  Match up with warn_msgs in errwarn.c. */
@@ -56,7 +61,8 @@ typedef enum {
     WARN_MULT_SEG_OVERRIDE,
     WARN_MULT_LOCKREP_PREFIX,
     WARN_NO_BASE_LABEL,
-    WARN_MULT_SHORTNEAR
+    WARN_MULT_SHORTNEAR,
+    WARN_CHAR_CONST_TOO_BIG
 } warn_num;
 
 char *conv_unprint(char ch);
diff --git a/modules/parsers/nasm/bison.y.in b/modules/parsers/nasm/bison.y.in
index 5a4accf5..d0e1af44 100644
--- a/modules/parsers/nasm/bison.y.in
+++ b/modules/parsers/nasm/bison.y.in
@@ -1,4 +1,4 @@
-/* $Id: bison.y.in,v 1.21 2001/07/11 21:04:58 peter Exp $
+/* $Id: bison.y.in,v 1.22 2001/07/25 00:33:10 peter Exp $
  * Main bison parser
  *
  *  Copyright (C) 2001  Peter Johnson, Michael Urman
@@ -33,6 +33,7 @@
 void init_table(void);
 extern int yylex(void);
 extern void yyerror(char *);
+static unsigned long ConvertCharConstToInt(char *);
 
 %}
 
@@ -50,12 +51,13 @@ extern void yyerror(char *);
     expr *exp;
     immval im_val;
     targetval tgt_val;
+    dataval *data;
     bytecode bc;
-} 
+}
 
 %token <int_val> INTNUM
 %token <double_val> FLTNUM
-%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL
+%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL STRING
 %token <int_val> BYTE WORD DWORD QWORD TWORD DQWORD
 %token <int_val> DECLARE_DATA
 %token <int_val> RESERVE_SPACE
@@ -86,10 +88,11 @@ extern void yyerror(char *);
 %type <ea_val> rm8x rm16x rm32x /*rm64x rm128x*/
 %type <ea_val> rm8 rm16 rm32 rm64 rm128
 %type <im_val> imm imm8x imm16x imm32x imm8 imm16 imm32
-%type <exp> expr
+%type <exp> expr expr_no_string
 %type <syminfo> explabel
 %type <sym> label_id
 %type <tgt_val> target
+%type <data> dataval datavals
 
 %left '|'
 %left '^'
@@ -104,15 +107,34 @@ input: /* empty */
     | input line { OutputError(); OutputWarning(); line_number++; }
 ;
 
-line: '\n'	{ $$.len = 0; }
+line: '\n'	{ $$.type = BC_EMPTY; }
     | exp '\n' { DebugPrintBC(&$1); $$ = $1; }
     | label exp '\n' { DebugPrintBC(&$2); $$ = $2; }
     | label '\n'
-    | directive '\n' { }
-    | error '\n' { Error(ERR_INVALID_LINE, (char *)NULL); yyerrok; }
+    | directive '\n' { $$.type = BC_EMPTY; }
+    | error '\n' {
+	Error(ERR_INVALID_LINE, (char *)NULL);
+	$$.type = BC_EMPTY;
+	yyerrok;
+    }
 ;
 
 exp: instr
+    | DECLARE_DATA datavals	{ BuildBC_Data(&$$, $2, $1); }
+    | RESERVE_SPACE expr	{ BuildBC_Reserve(&$$, $2, $1); }
+;
+
+datavals: dataval
+    | datavals ',' dataval	{ $$ = dataval_append($1, $3); }
+;
+
+dataval: expr_no_string		{ $$ = dataval_new_expr($1); }
+    | FLTNUM			{ $$ = dataval_new_float($1); }
+    | STRING			{ $$ = dataval_new_string($1); }
+    | error			{
+	Error(ERR_EXPR_SYNTAX, (char *)NULL);
+	$$ = (dataval *)NULL;
+    }
 ;
 
 label: label_id { $1->value = 0; } /* TODO: calculate offset */
@@ -312,8 +334,7 @@ target: expr		{ $$.val = $1; $$.op_sel = JR_NONE; }
 ;
 
 /* expression trees */
-expr:
-	INTNUM		{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
+expr_no_string: INTNUM	{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
     | explabel		{ $$ = expr_new_ident (EXPR_SYM, ExprSym(sym_use_get ($1.name, SYM_LABEL))); }
     /*| expr '||' expr	{ $$ = expr_new_tree ($1, EXPR_LOR, $3); }*/
     | expr '|' expr	{ $$ = expr_new_tree ($1, EXPR_OR, $3); }
@@ -340,6 +361,10 @@ expr:
     | '(' expr ')'	{ $$ = $2; }
 ;
 
+expr: expr_no_string
+    | STRING		{ $$ = expr_new_ident (EXPR_NUM, ExprNum(ConvertCharConstToInt($1))); }
+;
+
 explabel: ID | SPECIAL_ID | LOCAL_ID ;
 
 instr: instrbase
@@ -360,3 +385,31 @@ instr: instrbase
 /* instruction grammars (dynamically generated) */
 /* @INSTRUCTIONS@ */
 
+%%
+
+static unsigned long
+ConvertCharConstToInt(char *cc)
+{
+    unsigned long retval = 0;
+    size_t len = strlen(cc);
+
+    if(len > 4)
+	Warning(WARN_CHAR_CONST_TOO_BIG, (char *)NULL);
+
+    switch(len) {
+	case 4:
+	    retval |= (unsigned long)cc[3];
+	    retval <<= 8;
+	case 3:
+	    retval |= (unsigned long)cc[2];
+	    retval <<= 8;
+	case 2:
+	    retval |= (unsigned long)cc[1];
+	    retval <<= 8;
+	case 1:
+	    retval |= (unsigned long)cc[0];
+    }
+
+    return retval;
+}
+
diff --git a/modules/parsers/nasm/nasm-bison.y b/modules/parsers/nasm/nasm-bison.y
index 364f84cf..94f11f12 100644
--- a/modules/parsers/nasm/nasm-bison.y
+++ b/modules/parsers/nasm/nasm-bison.y
@@ -1,4 +1,4 @@
-/* $Id: nasm-bison.y,v 1.21 2001/07/11 21:04:58 peter Exp $
+/* $Id: nasm-bison.y,v 1.22 2001/07/25 00:33:10 peter Exp $
  * Main bison parser
  *
  *  Copyright (C) 2001  Peter Johnson, Michael Urman
@@ -33,6 +33,7 @@
 void init_table(void);
 extern int yylex(void);
 extern void yyerror(char *);
+static unsigned long ConvertCharConstToInt(char *);
 
 %}
 
@@ -50,12 +51,13 @@ extern void yyerror(char *);
     expr *exp;
     immval im_val;
     targetval tgt_val;
+    dataval *data;
     bytecode bc;
-} 
+}
 
 %token <int_val> INTNUM
 %token <double_val> FLTNUM
-%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL
+%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL STRING
 %token <int_val> BYTE WORD DWORD QWORD TWORD DQWORD
 %token <int_val> DECLARE_DATA
 %token <int_val> RESERVE_SPACE
@@ -86,10 +88,11 @@ extern void yyerror(char *);
 %type <ea_val> rm8x rm16x rm32x /*rm64x rm128x*/
 %type <ea_val> rm8 rm16 rm32 rm64 rm128
 %type <im_val> imm imm8x imm16x imm32x imm8 imm16 imm32
-%type <exp> expr
+%type <exp> expr expr_no_string
 %type <syminfo> explabel
 %type <sym> label_id
 %type <tgt_val> target
+%type <data> dataval datavals
 
 %left '|'
 %left '^'
@@ -104,15 +107,34 @@ input: /* empty */
     | input line { OutputError(); OutputWarning(); line_number++; }
 ;
 
-line: '\n'	{ $$.len = 0; }
+line: '\n'	{ $$.type = BC_EMPTY; }
     | exp '\n' { DebugPrintBC(&$1); $$ = $1; }
     | label exp '\n' { DebugPrintBC(&$2); $$ = $2; }
     | label '\n'
-    | directive '\n' { }
-    | error '\n' { Error(ERR_INVALID_LINE, (char *)NULL); yyerrok; }
+    | directive '\n' { $$.type = BC_EMPTY; }
+    | error '\n' {
+	Error(ERR_INVALID_LINE, (char *)NULL);
+	$$.type = BC_EMPTY;
+	yyerrok;
+    }
 ;
 
 exp: instr
+    | DECLARE_DATA datavals	{ BuildBC_Data(&$$, $2, $1); }
+    | RESERVE_SPACE expr	{ BuildBC_Reserve(&$$, $2, $1); }
+;
+
+datavals: dataval
+    | datavals ',' dataval	{ $$ = dataval_append($1, $3); }
+;
+
+dataval: expr_no_string		{ $$ = dataval_new_expr($1); }
+    | FLTNUM			{ $$ = dataval_new_float($1); }
+    | STRING			{ $$ = dataval_new_string($1); }
+    | error			{
+	Error(ERR_EXPR_SYNTAX, (char *)NULL);
+	$$ = (dataval *)NULL;
+    }
 ;
 
 label: label_id { $1->value = 0; } /* TODO: calculate offset */
@@ -312,8 +334,7 @@ target: expr		{ $$.val = $1; $$.op_sel = JR_NONE; }
 ;
 
 /* expression trees */
-expr:
-	INTNUM		{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
+expr_no_string: INTNUM	{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
     | explabel		{ $$ = expr_new_ident (EXPR_SYM, ExprSym(sym_use_get ($1.name, SYM_LABEL))); }
     /*| expr '||' expr	{ $$ = expr_new_tree ($1, EXPR_LOR, $3); }*/
     | expr '|' expr	{ $$ = expr_new_tree ($1, EXPR_OR, $3); }
@@ -340,6 +361,10 @@ expr:
     | '(' expr ')'	{ $$ = $2; }
 ;
 
+expr: expr_no_string
+    | STRING		{ $$ = expr_new_ident (EXPR_NUM, ExprNum(ConvertCharConstToInt($1))); }
+;
+
 explabel: ID | SPECIAL_ID | LOCAL_ID ;
 
 instr: instrbase
@@ -360,3 +385,31 @@ instr: instrbase
 /* instruction grammars (dynamically generated) */
 /* @INSTRUCTIONS@ */
 
+%%
+
+static unsigned long
+ConvertCharConstToInt(char *cc)
+{
+    unsigned long retval = 0;
+    size_t len = strlen(cc);
+
+    if(len > 4)
+	Warning(WARN_CHAR_CONST_TOO_BIG, (char *)NULL);
+
+    switch(len) {
+	case 4:
+	    retval |= (unsigned long)cc[3];
+	    retval <<= 8;
+	case 3:
+	    retval |= (unsigned long)cc[2];
+	    retval <<= 8;
+	case 2:
+	    retval |= (unsigned long)cc[1];
+	    retval <<= 8;
+	case 1:
+	    retval |= (unsigned long)cc[0];
+    }
+
+    return retval;
+}
+
diff --git a/modules/parsers/nasm/token.l.in b/modules/parsers/nasm/token.l.in
index b9f29ab8..0cde489f 100644
--- a/modules/parsers/nasm/token.l.in
+++ b/modules/parsers/nasm/token.l.in
@@ -1,4 +1,4 @@
-/* $Id: token.l.in,v 1.11 2001/07/11 21:01:54 peter Exp $
+/* $Id: token.l.in,v 1.12 2001/07/25 00:33:10 peter Exp $
  * Main lexer
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -37,6 +37,15 @@
 #include "expr.h"
 #include "bison.h"
 
+/* starting size of string buffer */
+#define STRBUF_ALLOC_SIZE	128
+
+/* string buffer used when parsing strings/character constants */
+static char *strbuf = (char *)NULL;
+
+/* length of strbuf (including terminating NULL character) */
+static size_t strbuf_size = 0;
+
 %}
 %option noyywrap
 %option nounput
@@ -89,6 +98,41 @@ WS       [ \t\r]
     yylval.double_val = strtod(yytext, (char **)NULL);
     return FLTNUM;
 }
+
+    /* string/character constant values */
+["']	{
+    int inch, count;
+    char endch = yytext[0];
+
+    strbuf = malloc(STRBUF_ALLOC_SIZE);
+    if(!strbuf)
+	Fatal(FATAL_NOMEM);
+
+    strbuf_size = STRBUF_ALLOC_SIZE;
+    inch = input();
+    count = 0;
+    while(inch != EOF && inch != endch && inch != '\n') {
+	strbuf[count++] = inch;
+	if(count >= strbuf_size) {
+	    strbuf = realloc(strbuf, strbuf_size + STRBUF_ALLOC_SIZE);
+	    if(!strbuf)
+		Fatal(FATAL_NOMEM);
+	    strbuf_size += STRBUF_ALLOC_SIZE;
+	}
+	inch = input();
+    }
+
+    if(inch == '\n')
+	Error(ERR_STRING_UNTERM, (char *)NULL);
+    else if(inch == EOF)
+	Error(ERR_STRING_EOF, (char *)NULL);
+
+    strbuf[count] = '\0';
+
+    yylval.str_val = strbuf;
+    return STRING;
+}
+
     /* directive: [name value] */
 ^"["	{ BEGIN DIRECTIVE; return '['; }
 <DIRECTIVE>"]"	    { BEGIN INITIAL; return ']'; }
diff --git a/src/bison.y.in b/src/bison.y.in
index 5a4accf5..d0e1af44 100644
--- a/src/bison.y.in
+++ b/src/bison.y.in
@@ -1,4 +1,4 @@
-/* $Id: bison.y.in,v 1.21 2001/07/11 21:04:58 peter Exp $
+/* $Id: bison.y.in,v 1.22 2001/07/25 00:33:10 peter Exp $
  * Main bison parser
  *
  *  Copyright (C) 2001  Peter Johnson, Michael Urman
@@ -33,6 +33,7 @@
 void init_table(void);
 extern int yylex(void);
 extern void yyerror(char *);
+static unsigned long ConvertCharConstToInt(char *);
 
 %}
 
@@ -50,12 +51,13 @@ extern void yyerror(char *);
     expr *exp;
     immval im_val;
     targetval tgt_val;
+    dataval *data;
     bytecode bc;
-} 
+}
 
 %token <int_val> INTNUM
 %token <double_val> FLTNUM
-%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL
+%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL STRING
 %token <int_val> BYTE WORD DWORD QWORD TWORD DQWORD
 %token <int_val> DECLARE_DATA
 %token <int_val> RESERVE_SPACE
@@ -86,10 +88,11 @@ extern void yyerror(char *);
 %type <ea_val> rm8x rm16x rm32x /*rm64x rm128x*/
 %type <ea_val> rm8 rm16 rm32 rm64 rm128
 %type <im_val> imm imm8x imm16x imm32x imm8 imm16 imm32
-%type <exp> expr
+%type <exp> expr expr_no_string
 %type <syminfo> explabel
 %type <sym> label_id
 %type <tgt_val> target
+%type <data> dataval datavals
 
 %left '|'
 %left '^'
@@ -104,15 +107,34 @@ input: /* empty */
     | input line { OutputError(); OutputWarning(); line_number++; }
 ;
 
-line: '\n'	{ $$.len = 0; }
+line: '\n'	{ $$.type = BC_EMPTY; }
     | exp '\n' { DebugPrintBC(&$1); $$ = $1; }
     | label exp '\n' { DebugPrintBC(&$2); $$ = $2; }
     | label '\n'
-    | directive '\n' { }
-    | error '\n' { Error(ERR_INVALID_LINE, (char *)NULL); yyerrok; }
+    | directive '\n' { $$.type = BC_EMPTY; }
+    | error '\n' {
+	Error(ERR_INVALID_LINE, (char *)NULL);
+	$$.type = BC_EMPTY;
+	yyerrok;
+    }
 ;
 
 exp: instr
+    | DECLARE_DATA datavals	{ BuildBC_Data(&$$, $2, $1); }
+    | RESERVE_SPACE expr	{ BuildBC_Reserve(&$$, $2, $1); }
+;
+
+datavals: dataval
+    | datavals ',' dataval	{ $$ = dataval_append($1, $3); }
+;
+
+dataval: expr_no_string		{ $$ = dataval_new_expr($1); }
+    | FLTNUM			{ $$ = dataval_new_float($1); }
+    | STRING			{ $$ = dataval_new_string($1); }
+    | error			{
+	Error(ERR_EXPR_SYNTAX, (char *)NULL);
+	$$ = (dataval *)NULL;
+    }
 ;
 
 label: label_id { $1->value = 0; } /* TODO: calculate offset */
@@ -312,8 +334,7 @@ target: expr		{ $$.val = $1; $$.op_sel = JR_NONE; }
 ;
 
 /* expression trees */
-expr:
-	INTNUM		{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
+expr_no_string: INTNUM	{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
     | explabel		{ $$ = expr_new_ident (EXPR_SYM, ExprSym(sym_use_get ($1.name, SYM_LABEL))); }
     /*| expr '||' expr	{ $$ = expr_new_tree ($1, EXPR_LOR, $3); }*/
     | expr '|' expr	{ $$ = expr_new_tree ($1, EXPR_OR, $3); }
@@ -340,6 +361,10 @@ expr:
     | '(' expr ')'	{ $$ = $2; }
 ;
 
+expr: expr_no_string
+    | STRING		{ $$ = expr_new_ident (EXPR_NUM, ExprNum(ConvertCharConstToInt($1))); }
+;
+
 explabel: ID | SPECIAL_ID | LOCAL_ID ;
 
 instr: instrbase
@@ -360,3 +385,31 @@ instr: instrbase
 /* instruction grammars (dynamically generated) */
 /* @INSTRUCTIONS@ */
 
+%%
+
+static unsigned long
+ConvertCharConstToInt(char *cc)
+{
+    unsigned long retval = 0;
+    size_t len = strlen(cc);
+
+    if(len > 4)
+	Warning(WARN_CHAR_CONST_TOO_BIG, (char *)NULL);
+
+    switch(len) {
+	case 4:
+	    retval |= (unsigned long)cc[3];
+	    retval <<= 8;
+	case 3:
+	    retval |= (unsigned long)cc[2];
+	    retval <<= 8;
+	case 2:
+	    retval |= (unsigned long)cc[1];
+	    retval <<= 8;
+	case 1:
+	    retval |= (unsigned long)cc[0];
+    }
+
+    return retval;
+}
+
diff --git a/src/bytecode.c b/src/bytecode.c
index 4f3e631e..9af7ad0d 100644
--- a/src/bytecode.c
+++ b/src/bytecode.c
@@ -1,4 +1,4 @@
-/* $Id: bytecode.c,v 1.13 2001/07/11 23:16:50 peter Exp $
+/* $Id: bytecode.c,v 1.14 2001/07/25 00:33:10 peter Exp $
  * Bytecode utility functions
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -20,6 +20,7 @@
  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 #include <stdio.h>
+#include <stdlib.h>
 #include "globals.h"
 #include "bytecode.h"
 #include "errwarn.h"
@@ -336,21 +337,77 @@ BuildBC_JmpRel (bytecode      *bc,
     BuildBC_Common(bc);
 }
 
-/* TODO: implement.  Shouldn't be difficult. */
-unsigned char *
-ConvertBCInsnToBytes (unsigned char *ptr, bytecode *bc, int *len)
+void
+BuildBC_Data (bytecode      *bc,
+	      dataval       *data,
+	      unsigned long  size)
 {
-    if(bc->type != BC_INSN)
-	return (unsigned char *)NULL;
-    return (unsigned char *)NULL;
+    dataval *cur = data;
+
+    /* First check to see if all the data elements are valid for the size
+     * being set.
+     * Validity table:
+     *  db (1)  -> expr, string
+     *  dw (2)  -> expr, string
+     *  dd (4)  -> expr, float, string
+     *  dq (8)  -> expr, float, string
+     *  dt (10) -> float, string
+     *
+     * Once we calculate expr we'll have to validate it against the size
+     * and warn/error appropriately (symbol constants versus labels:
+     * constants (equ's) should always be legal, but labels should raise
+     * warnings when used in db or dq context at the minimum).
+     */
+    while(cur) {
+	switch(cur->type) {
+	    case DV_EMPTY:
+	    case DV_STRING:
+		/* string is valid in every size */
+		break;
+	    case DV_FLOAT:
+		if(size == 1)
+		    Error(ERR_DECLDATA_FLOAT, (char *)NULL, "DB");
+		else if(size == 2)
+		    Error(ERR_DECLDATA_FLOAT, (char *)NULL, "DW");
+		break;
+	    case DV_EXPR:
+		if(size == 10)
+		    Error(ERR_DECLDATA_EXPR, (char *)NULL, "DT");
+		break;
+	}
+	cur = cur->next;
+    }
+
+    bc->next = (bytecode *)NULL;
+    bc->type = BC_DATA;
+
+    bc->data.data.data = data;
+    bc->data.data.size = size;
+
+    BuildBC_Common(bc);
 }
 
 void
-DebugPrintBC (bytecode *bc)
+BuildBC_Reserve (bytecode      *bc,
+		 expr          *numitems,
+		 unsigned long  itemsize)
 {
-    unsigned long i;
+    bc->next = (bytecode *)NULL;
+    bc->type = BC_RESERVE;
 
+    bc->data.reserve.numitems = numitems;
+    bc->data.reserve.itemsize = itemsize;
+
+    BuildBC_Common(bc);
+}
+
+void
+DebugPrintBC (bytecode *bc)
+{
     switch(bc->type) {
+	case BC_EMPTY:
+	    printf("_Empty_\n");
+	    break;
 	case BC_INSN:
 	    printf("_Instruction_\n");
 	    printf("Effective Address:\n");
@@ -431,14 +488,17 @@ DebugPrintBC (bytecode *bc)
 	    break;
 	case BC_DATA:
 	    printf("_Data_\n");
-	    for(i=0; i<bc->len; i++) {
-		printf("%2x ", (unsigned int)bc->data.data.data[i]);
-		if((i & ~0xFFFF) == i)
-		    printf("\n");
-	    }
+	    printf("Final Element Size=%u\n",
+		(unsigned int)bc->data.data.size);
+	    printf("Elements:\n");
+	    dataval_print(bc->data.data.data);
 	    break;
 	case BC_RESERVE:
 	    printf("_Reserve_\n");
+	    printf("Num Items=");
+	    expr_print(bc->data.reserve.numitems);
+	    printf("\nItem Size=%u\n",
+		(unsigned int)bc->data.reserve.itemsize);
 	    break;
 	default:
 	    printf("_Unknown_\n");
@@ -448,3 +508,98 @@ DebugPrintBC (bytecode *bc)
 	bc->filename ? bc->filename : "<UNKNOWN>", bc->lineno);
     printf("Offset=%lx BITS=%u\n", bc->offset, bc->mode_bits);
 }
+
+dataval *
+dataval_new_expr (expr *exp)
+{
+    dataval *retval = malloc(sizeof(dataval));
+
+    if(!retval)
+	Fatal(FATAL_NOMEM);
+
+    retval->next = (dataval *)NULL;
+    retval->last = retval;
+    retval->type = DV_EXPR;
+    retval->data.exp = exp;
+
+    return retval;
+}
+
+dataval *
+dataval_new_float (double float_val)
+{
+    dataval *retval = malloc(sizeof(dataval));
+
+    if(!retval)
+	Fatal(FATAL_NOMEM);
+
+    retval->next = (dataval *)NULL;
+    retval->last = retval;
+    retval->type = DV_FLOAT;
+    retval->data.float_val = float_val;
+
+    return retval;
+}
+
+dataval *
+dataval_new_string (char *str_val)
+{
+    dataval *retval = malloc(sizeof(dataval));
+
+    if(!retval)
+	Fatal(FATAL_NOMEM);
+
+    retval->next = (dataval *)NULL;
+    retval->last = retval;
+    retval->type = DV_STRING;
+    retval->data.str_val = str_val;
+
+    return retval;
+}
+
+dataval *
+dataval_append (dataval *list, dataval *item)
+{
+    if(item)
+	item->next = (dataval *)NULL;
+
+    if(!list) {
+	item->last = item;
+	return item;
+    } else {
+	if(item) {
+	    list->last->next = item;
+	    list->last = item;
+	    item->last = (dataval *)NULL;
+	}
+	return list;
+    }
+}
+
+void
+dataval_print (dataval *start)
+{
+    dataval *cur = start;
+
+    while(cur) {
+	switch(cur->type) {
+	    case DV_EMPTY:
+		printf(" Empty\n");
+		break;
+	    case DV_EXPR:
+		printf(" Expr=");
+		expr_print(cur->data.exp);
+		printf("\n");
+		break;
+	    case DV_FLOAT:
+		printf(" Float=%e\n", cur->data.float_val);
+		break;
+	    case DV_STRING:
+		printf(" String=%s\n", cur->data.str_val);
+		break;
+	}
+
+	cur = cur->next;
+    }
+}
+
diff --git a/src/bytecode.h b/src/bytecode.h
index 6b971c33..fafb3b44 100644
--- a/src/bytecode.h
+++ b/src/bytecode.h
@@ -1,4 +1,4 @@
-/* $Id: bytecode.h,v 1.14 2001/07/11 23:16:50 peter Exp $
+/* $Id: bytecode.h,v 1.15 2001/07/25 00:33:10 peter Exp $
  * Bytecode utility functions header file
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -47,6 +47,18 @@ typedef struct immval_s {
     unsigned char f_sign;	/* 1 if final imm should be signed */
 } immval;
 
+typedef struct dataval_s {
+    struct dataval_s *next, *last;
+
+    enum { DV_EMPTY, DV_EXPR, DV_FLOAT, DV_STRING } type;
+
+    union {
+	struct expr_s *exp;
+	double float_val;
+	char *str_val;
+    } data;
+} dataval;
+
 typedef enum jmprel_opcode_sel_e {
     JR_NONE,
     JR_SHORT,
@@ -64,7 +76,7 @@ typedef struct targetval_s {
 typedef struct bytecode_s {
     struct bytecode_s *next;
 
-    enum { BC_INSN, BC_JMPREL, BC_DATA, BC_RESERVE } type;
+    enum { BC_EMPTY, BC_INSN, BC_JMPREL, BC_DATA, BC_RESERVE } type;
 
     union {
 	struct {
@@ -97,8 +109,15 @@ typedef struct bytecode_s {
 	    unsigned char lockrep_pre;	/* 0 indicates no prefix */
 	} jmprel;
 	struct {
-	    unsigned char *data;
+	    dataval *data;		/* non-converted data (linked list) */
+
+	    /* final (converted) size of each element (in bytes) */
+	    unsigned char size;
 	} data;
+	struct {
+	    struct expr_s *numitems;	/* number of items to reserve */
+	    unsigned char itemsize;	/* size of each item (in bytes) */
+	} reserve;
     } data;
 
     unsigned long len;	/* total length of entire bytecode */
@@ -154,8 +173,22 @@ void BuildBC_JmpRel(bytecode      *bc,
 		    unsigned char  near_op2,
 		    unsigned char  addrsize);
 
-unsigned char *ConvertBCInsnToBytes(unsigned char *ptr, bytecode *bc, int *len);
+void BuildBC_Data(bytecode      *bc,
+		  dataval       *data,
+		  unsigned long  size);
+
+void BuildBC_Reserve(bytecode      *bc,
+		     struct expr_s *numitems,
+		     unsigned long  itemsize);
 
 void DebugPrintBC(bytecode *bc);
 
+dataval *dataval_new_expr(struct expr_s *exp);
+dataval *dataval_new_float(double float_val);
+dataval *dataval_new_string(char *str_val);
+
+dataval *dataval_append(dataval *list, dataval *item);
+
+void dataval_print(dataval *start);
+
 #endif
diff --git a/src/errwarn.c b/src/errwarn.c
index cd296f64..b6c4e263 100644
--- a/src/errwarn.c
+++ b/src/errwarn.c
@@ -1,4 +1,4 @@
-/* $Id: errwarn.c,v 1.16 2001/07/11 23:16:50 peter Exp $
+/* $Id: errwarn.c,v 1.17 2001/07/25 00:33:10 peter Exp $
  * Error and warning reporting and related functions.
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -67,7 +67,12 @@ static char *err_msgs[] = {
     "expression syntax error",
     "duplicate definition of `%1'; previously defined line %2",
     "mismatch in operand sizes",
-    "no %s form of that jump instruction exists"
+    "no %s form of that jump instruction exists",
+    "unterminated string",
+    "unexpected end of file in string",
+    "expression syntax error",
+    "floating-point constant encountered in `%s'",
+    "non-floating-point value encountered in `%s'"
 };
 
 /* Warning messages.  Match up with warn_num enum in errwarn.h. */
@@ -78,7 +83,8 @@ static char *warn_msgs[] = {
     "multiple segment overrides, using leftmost",
     "multiple LOCK or REP prefixes, using leftmost",
     "no non-local label before '%s'",
-    "multiple SHORT or NEAR specifiers, using leftmost"
+    "multiple SHORT or NEAR specifiers, using leftmost",
+    "character constant too large, ignoring trailing characters"
 };
 
 /* I hate to define these strings as static buffers; a better solution would be
diff --git a/src/errwarn.h b/src/errwarn.h
index a1e2044b..51bf7175 100644
--- a/src/errwarn.h
+++ b/src/errwarn.h
@@ -1,4 +1,4 @@
-/* $Id: errwarn.h,v 1.10 2001/07/11 23:16:50 peter Exp $
+/* $Id: errwarn.h,v 1.11 2001/07/25 00:33:10 peter Exp $
  * Error and warning reporting and related functions header file.
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -45,7 +45,12 @@ typedef enum {
     ERR_EXP_SYNTAX,
     ERR_DUPLICATE_DEF,
     ERR_OP_SIZE_MISMATCH,
-    ERR_NO_JMPREL_FORM
+    ERR_NO_JMPREL_FORM,
+    ERR_STRING_UNTERM,
+    ERR_STRING_EOF,
+    ERR_EXPR_SYNTAX,
+    ERR_DECLDATA_FLOAT,
+    ERR_DECLDATA_EXPR
 } err_num;
 
 /* Warning constants.  Match up with warn_msgs in errwarn.c. */
@@ -56,7 +61,8 @@ typedef enum {
     WARN_MULT_SEG_OVERRIDE,
     WARN_MULT_LOCKREP_PREFIX,
     WARN_NO_BASE_LABEL,
-    WARN_MULT_SHORTNEAR
+    WARN_MULT_SHORTNEAR,
+    WARN_CHAR_CONST_TOO_BIG
 } warn_num;
 
 char *conv_unprint(char ch);
diff --git a/src/parsers/nasm/bison.y.in b/src/parsers/nasm/bison.y.in
index 5a4accf5..d0e1af44 100644
--- a/src/parsers/nasm/bison.y.in
+++ b/src/parsers/nasm/bison.y.in
@@ -1,4 +1,4 @@
-/* $Id: bison.y.in,v 1.21 2001/07/11 21:04:58 peter Exp $
+/* $Id: bison.y.in,v 1.22 2001/07/25 00:33:10 peter Exp $
  * Main bison parser
  *
  *  Copyright (C) 2001  Peter Johnson, Michael Urman
@@ -33,6 +33,7 @@
 void init_table(void);
 extern int yylex(void);
 extern void yyerror(char *);
+static unsigned long ConvertCharConstToInt(char *);
 
 %}
 
@@ -50,12 +51,13 @@ extern void yyerror(char *);
     expr *exp;
     immval im_val;
     targetval tgt_val;
+    dataval *data;
     bytecode bc;
-} 
+}
 
 %token <int_val> INTNUM
 %token <double_val> FLTNUM
-%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL
+%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL STRING
 %token <int_val> BYTE WORD DWORD QWORD TWORD DQWORD
 %token <int_val> DECLARE_DATA
 %token <int_val> RESERVE_SPACE
@@ -86,10 +88,11 @@ extern void yyerror(char *);
 %type <ea_val> rm8x rm16x rm32x /*rm64x rm128x*/
 %type <ea_val> rm8 rm16 rm32 rm64 rm128
 %type <im_val> imm imm8x imm16x imm32x imm8 imm16 imm32
-%type <exp> expr
+%type <exp> expr expr_no_string
 %type <syminfo> explabel
 %type <sym> label_id
 %type <tgt_val> target
+%type <data> dataval datavals
 
 %left '|'
 %left '^'
@@ -104,15 +107,34 @@ input: /* empty */
     | input line { OutputError(); OutputWarning(); line_number++; }
 ;
 
-line: '\n'	{ $$.len = 0; }
+line: '\n'	{ $$.type = BC_EMPTY; }
     | exp '\n' { DebugPrintBC(&$1); $$ = $1; }
     | label exp '\n' { DebugPrintBC(&$2); $$ = $2; }
     | label '\n'
-    | directive '\n' { }
-    | error '\n' { Error(ERR_INVALID_LINE, (char *)NULL); yyerrok; }
+    | directive '\n' { $$.type = BC_EMPTY; }
+    | error '\n' {
+	Error(ERR_INVALID_LINE, (char *)NULL);
+	$$.type = BC_EMPTY;
+	yyerrok;
+    }
 ;
 
 exp: instr
+    | DECLARE_DATA datavals	{ BuildBC_Data(&$$, $2, $1); }
+    | RESERVE_SPACE expr	{ BuildBC_Reserve(&$$, $2, $1); }
+;
+
+datavals: dataval
+    | datavals ',' dataval	{ $$ = dataval_append($1, $3); }
+;
+
+dataval: expr_no_string		{ $$ = dataval_new_expr($1); }
+    | FLTNUM			{ $$ = dataval_new_float($1); }
+    | STRING			{ $$ = dataval_new_string($1); }
+    | error			{
+	Error(ERR_EXPR_SYNTAX, (char *)NULL);
+	$$ = (dataval *)NULL;
+    }
 ;
 
 label: label_id { $1->value = 0; } /* TODO: calculate offset */
@@ -312,8 +334,7 @@ target: expr		{ $$.val = $1; $$.op_sel = JR_NONE; }
 ;
 
 /* expression trees */
-expr:
-	INTNUM		{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
+expr_no_string: INTNUM	{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
     | explabel		{ $$ = expr_new_ident (EXPR_SYM, ExprSym(sym_use_get ($1.name, SYM_LABEL))); }
     /*| expr '||' expr	{ $$ = expr_new_tree ($1, EXPR_LOR, $3); }*/
     | expr '|' expr	{ $$ = expr_new_tree ($1, EXPR_OR, $3); }
@@ -340,6 +361,10 @@ expr:
     | '(' expr ')'	{ $$ = $2; }
 ;
 
+expr: expr_no_string
+    | STRING		{ $$ = expr_new_ident (EXPR_NUM, ExprNum(ConvertCharConstToInt($1))); }
+;
+
 explabel: ID | SPECIAL_ID | LOCAL_ID ;
 
 instr: instrbase
@@ -360,3 +385,31 @@ instr: instrbase
 /* instruction grammars (dynamically generated) */
 /* @INSTRUCTIONS@ */
 
+%%
+
+static unsigned long
+ConvertCharConstToInt(char *cc)
+{
+    unsigned long retval = 0;
+    size_t len = strlen(cc);
+
+    if(len > 4)
+	Warning(WARN_CHAR_CONST_TOO_BIG, (char *)NULL);
+
+    switch(len) {
+	case 4:
+	    retval |= (unsigned long)cc[3];
+	    retval <<= 8;
+	case 3:
+	    retval |= (unsigned long)cc[2];
+	    retval <<= 8;
+	case 2:
+	    retval |= (unsigned long)cc[1];
+	    retval <<= 8;
+	case 1:
+	    retval |= (unsigned long)cc[0];
+    }
+
+    return retval;
+}
+
diff --git a/src/parsers/nasm/nasm-bison.y b/src/parsers/nasm/nasm-bison.y
index 364f84cf..94f11f12 100644
--- a/src/parsers/nasm/nasm-bison.y
+++ b/src/parsers/nasm/nasm-bison.y
@@ -1,4 +1,4 @@
-/* $Id: nasm-bison.y,v 1.21 2001/07/11 21:04:58 peter Exp $
+/* $Id: nasm-bison.y,v 1.22 2001/07/25 00:33:10 peter Exp $
  * Main bison parser
  *
  *  Copyright (C) 2001  Peter Johnson, Michael Urman
@@ -33,6 +33,7 @@
 void init_table(void);
 extern int yylex(void);
 extern void yyerror(char *);
+static unsigned long ConvertCharConstToInt(char *);
 
 %}
 
@@ -50,12 +51,13 @@ extern void yyerror(char *);
     expr *exp;
     immval im_val;
     targetval tgt_val;
+    dataval *data;
     bytecode bc;
-} 
+}
 
 %token <int_val> INTNUM
 %token <double_val> FLTNUM
-%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL
+%token <str_val> DIRECTIVE_NAME DIRECTIVE_VAL STRING
 %token <int_val> BYTE WORD DWORD QWORD TWORD DQWORD
 %token <int_val> DECLARE_DATA
 %token <int_val> RESERVE_SPACE
@@ -86,10 +88,11 @@ extern void yyerror(char *);
 %type <ea_val> rm8x rm16x rm32x /*rm64x rm128x*/
 %type <ea_val> rm8 rm16 rm32 rm64 rm128
 %type <im_val> imm imm8x imm16x imm32x imm8 imm16 imm32
-%type <exp> expr
+%type <exp> expr expr_no_string
 %type <syminfo> explabel
 %type <sym> label_id
 %type <tgt_val> target
+%type <data> dataval datavals
 
 %left '|'
 %left '^'
@@ -104,15 +107,34 @@ input: /* empty */
     | input line { OutputError(); OutputWarning(); line_number++; }
 ;
 
-line: '\n'	{ $$.len = 0; }
+line: '\n'	{ $$.type = BC_EMPTY; }
     | exp '\n' { DebugPrintBC(&$1); $$ = $1; }
     | label exp '\n' { DebugPrintBC(&$2); $$ = $2; }
     | label '\n'
-    | directive '\n' { }
-    | error '\n' { Error(ERR_INVALID_LINE, (char *)NULL); yyerrok; }
+    | directive '\n' { $$.type = BC_EMPTY; }
+    | error '\n' {
+	Error(ERR_INVALID_LINE, (char *)NULL);
+	$$.type = BC_EMPTY;
+	yyerrok;
+    }
 ;
 
 exp: instr
+    | DECLARE_DATA datavals	{ BuildBC_Data(&$$, $2, $1); }
+    | RESERVE_SPACE expr	{ BuildBC_Reserve(&$$, $2, $1); }
+;
+
+datavals: dataval
+    | datavals ',' dataval	{ $$ = dataval_append($1, $3); }
+;
+
+dataval: expr_no_string		{ $$ = dataval_new_expr($1); }
+    | FLTNUM			{ $$ = dataval_new_float($1); }
+    | STRING			{ $$ = dataval_new_string($1); }
+    | error			{
+	Error(ERR_EXPR_SYNTAX, (char *)NULL);
+	$$ = (dataval *)NULL;
+    }
 ;
 
 label: label_id { $1->value = 0; } /* TODO: calculate offset */
@@ -312,8 +334,7 @@ target: expr		{ $$.val = $1; $$.op_sel = JR_NONE; }
 ;
 
 /* expression trees */
-expr:
-	INTNUM		{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
+expr_no_string: INTNUM	{ $$ = expr_new_ident (EXPR_NUM, ExprNum($1)); }
     | explabel		{ $$ = expr_new_ident (EXPR_SYM, ExprSym(sym_use_get ($1.name, SYM_LABEL))); }
     /*| expr '||' expr	{ $$ = expr_new_tree ($1, EXPR_LOR, $3); }*/
     | expr '|' expr	{ $$ = expr_new_tree ($1, EXPR_OR, $3); }
@@ -340,6 +361,10 @@ expr:
     | '(' expr ')'	{ $$ = $2; }
 ;
 
+expr: expr_no_string
+    | STRING		{ $$ = expr_new_ident (EXPR_NUM, ExprNum(ConvertCharConstToInt($1))); }
+;
+
 explabel: ID | SPECIAL_ID | LOCAL_ID ;
 
 instr: instrbase
@@ -360,3 +385,31 @@ instr: instrbase
 /* instruction grammars (dynamically generated) */
 /* @INSTRUCTIONS@ */
 
+%%
+
+static unsigned long
+ConvertCharConstToInt(char *cc)
+{
+    unsigned long retval = 0;
+    size_t len = strlen(cc);
+
+    if(len > 4)
+	Warning(WARN_CHAR_CONST_TOO_BIG, (char *)NULL);
+
+    switch(len) {
+	case 4:
+	    retval |= (unsigned long)cc[3];
+	    retval <<= 8;
+	case 3:
+	    retval |= (unsigned long)cc[2];
+	    retval <<= 8;
+	case 2:
+	    retval |= (unsigned long)cc[1];
+	    retval <<= 8;
+	case 1:
+	    retval |= (unsigned long)cc[0];
+    }
+
+    return retval;
+}
+
diff --git a/src/parsers/nasm/token.l.in b/src/parsers/nasm/token.l.in
index b9f29ab8..0cde489f 100644
--- a/src/parsers/nasm/token.l.in
+++ b/src/parsers/nasm/token.l.in
@@ -1,4 +1,4 @@
-/* $Id: token.l.in,v 1.11 2001/07/11 21:01:54 peter Exp $
+/* $Id: token.l.in,v 1.12 2001/07/25 00:33:10 peter Exp $
  * Main lexer
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -37,6 +37,15 @@
 #include "expr.h"
 #include "bison.h"
 
+/* starting size of string buffer */
+#define STRBUF_ALLOC_SIZE	128
+
+/* string buffer used when parsing strings/character constants */
+static char *strbuf = (char *)NULL;
+
+/* length of strbuf (including terminating NULL character) */
+static size_t strbuf_size = 0;
+
 %}
 %option noyywrap
 %option nounput
@@ -89,6 +98,41 @@ WS       [ \t\r]
     yylval.double_val = strtod(yytext, (char **)NULL);
     return FLTNUM;
 }
+
+    /* string/character constant values */
+["']	{
+    int inch, count;
+    char endch = yytext[0];
+
+    strbuf = malloc(STRBUF_ALLOC_SIZE);
+    if(!strbuf)
+	Fatal(FATAL_NOMEM);
+
+    strbuf_size = STRBUF_ALLOC_SIZE;
+    inch = input();
+    count = 0;
+    while(inch != EOF && inch != endch && inch != '\n') {
+	strbuf[count++] = inch;
+	if(count >= strbuf_size) {
+	    strbuf = realloc(strbuf, strbuf_size + STRBUF_ALLOC_SIZE);
+	    if(!strbuf)
+		Fatal(FATAL_NOMEM);
+	    strbuf_size += STRBUF_ALLOC_SIZE;
+	}
+	inch = input();
+    }
+
+    if(inch == '\n')
+	Error(ERR_STRING_UNTERM, (char *)NULL);
+    else if(inch == EOF)
+	Error(ERR_STRING_EOF, (char *)NULL);
+
+    strbuf[count] = '\0';
+
+    yylval.str_val = strbuf;
+    return STRING;
+}
+
     /* directive: [name value] */
 ^"["	{ BEGIN DIRECTIVE; return '['; }
 <DIRECTIVE>"]"	    { BEGIN INITIAL; return ']'; }
diff --git a/src/token.l.in b/src/token.l.in
index b9f29ab8..0cde489f 100644
--- a/src/token.l.in
+++ b/src/token.l.in
@@ -1,4 +1,4 @@
-/* $Id: token.l.in,v 1.11 2001/07/11 21:01:54 peter Exp $
+/* $Id: token.l.in,v 1.12 2001/07/25 00:33:10 peter Exp $
  * Main lexer
  *
  *  Copyright (C) 2001  Peter Johnson
@@ -37,6 +37,15 @@
 #include "expr.h"
 #include "bison.h"
 
+/* starting size of string buffer */
+#define STRBUF_ALLOC_SIZE	128
+
+/* string buffer used when parsing strings/character constants */
+static char *strbuf = (char *)NULL;
+
+/* length of strbuf (including terminating NULL character) */
+static size_t strbuf_size = 0;
+
 %}
 %option noyywrap
 %option nounput
@@ -89,6 +98,41 @@ WS       [ \t\r]
     yylval.double_val = strtod(yytext, (char **)NULL);
     return FLTNUM;
 }
+
+    /* string/character constant values */
+["']	{
+    int inch, count;
+    char endch = yytext[0];
+
+    strbuf = malloc(STRBUF_ALLOC_SIZE);
+    if(!strbuf)
+	Fatal(FATAL_NOMEM);
+
+    strbuf_size = STRBUF_ALLOC_SIZE;
+    inch = input();
+    count = 0;
+    while(inch != EOF && inch != endch && inch != '\n') {
+	strbuf[count++] = inch;
+	if(count >= strbuf_size) {
+	    strbuf = realloc(strbuf, strbuf_size + STRBUF_ALLOC_SIZE);
+	    if(!strbuf)
+		Fatal(FATAL_NOMEM);
+	    strbuf_size += STRBUF_ALLOC_SIZE;
+	}
+	inch = input();
+    }
+
+    if(inch == '\n')
+	Error(ERR_STRING_UNTERM, (char *)NULL);
+    else if(inch == EOF)
+	Error(ERR_STRING_EOF, (char *)NULL);
+
+    strbuf[count] = '\0';
+
+    yylval.str_val = strbuf;
+    return STRING;
+}
+
     /* directive: [name value] */
 ^"["	{ BEGIN DIRECTIVE; return '['; }
 <DIRECTIVE>"]"	    { BEGIN INITIAL; return ']'; }
-- 
2.40.0