asm-i386:

started to support strings in db
fixed embarrassingly wrong parser for hexadecimal numbers

2 files changed, 88 insertions, 7 deletions

diff --git a/ecomp-c/asm-i386.c b/ecomp-c/asm-i386.c
index 2e33189..fd4ab1a 100644
--- a/ecomp-c/asm-i386.c
+++ b/ecomp-c/asm-i386.c
@@ -49,6 +49,7 @@
  * org 0x0000
  * dd  0x0000
  * db  0x00
+ * db  "abcd"
  */
 
 /* constants */
@@ -57,6 +58,7 @@ enum {
 	MAX_IDENT_LEN = 64,
 	MAX_NUMBER_LEN = 10,
 	MAX_HEXNUMBER_LEN = 8,
+	MAX_STRING_LEN = 64,	
 	MAX_PASSES = 10
 };
 
@@ -74,6 +76,7 @@ typedef enum {
 	S_org,
 	S_ident,
 	S_number,
+	S_string,
 	S_colon,
 	S_comma,
 	S_lbrak,
@@ -89,6 +92,7 @@ static char *symname[S_eof+1] = {
 	"org",
 	"ident",
 	"number",
+	"string",
 	":",
 	",",
 	"[",
@@ -104,6 +108,7 @@ static S_Symbol sym;
 
 static char ident[MAX_IDENT_LEN+1];
 static int num;
+static char str[MAX_STRING_LEN+1];
 
 static S_Symbol peek_sym = S_eof;
 
@@ -144,6 +149,18 @@ static void *Allocate( unsigned int size )
 	return p;
 }
 
+static char *AllocateAndCopyStr( char *s )
+{
+	char *d;
+	int len = strlen( s );
+	if( len > MAX_STRING_LEN ) {
+		Abort( "Too long string literal, should not happen" );
+	}
+	d = (char *)Allocate( len + 1 );
+	strlcpy( d, s, MAX_STRING_LEN );
+	return d;
+}
+
 static int getChar( void )
 {
 	int c;
@@ -233,16 +250,22 @@ static void hexnumber( void )
 	
 	look = getChar( );
 	if( isHexDigit( look ) ) {
-		num = look - '0';
+		if( isDigit( look ) ) {
+			num = look - '0';
+		} else if( look >= 'a' && look <= 'f' ) {
+			num = 10 + look - 'a';
+		} else {
+			num = 10 + look - 'A';
+		}
 		look = getChar( );
 		while( isHexDigit( look ) && n < MAX_HEXNUMBER_LEN ) {
 			n++;
 			if( isDigit( look ) ) {
 				num = 16 * num + ( look - '0' );
 			} else if( look >= 'a' && look <= 'f' ) {
-				num = 16 * num + ( look - 'a' );
+				num = 16 * num + ( 10 + look - 'a' );
 			} else {
-				num = 16 * num + ( look - 'A' );
+				num = 16 * num + ( 10 + look - 'A' );
 			}
 			look = getChar( );
 		}
@@ -253,6 +276,28 @@ static void hexnumber( void )
 	}
 }
 
+static void string( void )
+{
+	int n = 0;
+	
+	if( look == '"' ) {
+		look = getChar( );
+		while( ( isDigit( look ) || isAlpha( look ) || isSpecial( look ) ) && n < MAX_STRING_LEN ) {
+			str[n] = look;
+			n++;
+			look = getChar( );
+		}
+		if( n == MAX_STRING_LEN ) {
+			Abort( "String exceeds maximal length" );
+		}
+		str[n] = '\0';
+		if( look != '\"' ) {
+			Abort( "Unterminated string literal" );
+		}
+		look = getChar( );
+	}
+}
+
 static void identifier( void )
 {
 	int n = 0;
@@ -310,6 +355,10 @@ static S_Symbol getSym( void )
 			hexnumber( );
 			s = S_number;
 			break;
+		case '\"':
+			string( );
+			s = S_string;
+			break;
 		case '0':
 		case '1':
 		case '2':
@@ -443,6 +492,10 @@ static S_Symbol getSym( void )
 			case S_number:
 				fprintf( stderr, "sym -> %s( '%d' )\n", symname[s], num );
 				break;
+				
+			case S_string:
+				fprintf( stderr, "sym -> %s( '%s' )\n", symname[s], str );
+				break;
 
 			default:
 				fprintf( stderr, "sym -> %s\n", symname[s] );
@@ -673,6 +726,7 @@ static char *registername[5] = {
 typedef struct OperandInfo {
 	OperandType type;
 	int num;
+	char *str;
 	Register reg;
 	int addr;
 	Symbol *symbol;
@@ -835,6 +889,9 @@ static void free_opcode_info( OpcodeInfo *opcode_info )
 	ptr = opcode_info->operand;
 	while( ptr != NULL ) {
 		tmp = ptr->next;
+		if( ptr->str != NULL ) {
+			free( ptr->str );
+		}
 		free( ptr );
 		ptr = tmp;
 	}
@@ -882,6 +939,7 @@ static OperandInfo *parseOperand( void )
 	OperandInfo *operand_info;
 	
 	operand_info = Allocate( sizeof( OperandInfo ) );
+	operand_info->str = NULL;
 	operand_info->next = NULL;
 	operand_info->addr = ADDRESS_UNDEFINED;
 	operand_info->symbol = NULL;
@@ -901,6 +959,10 @@ static OperandInfo *parseOperand( void )
 		operand_info->type = OPERAND_ABSOLUTE;
 		operand_info->num = num;
 		sym = getSym( );
+	} else if( sym == S_string ) {
+		operand_info->type = OPERAND_ABSOLUTE;
+		operand_info->str = AllocateAndCopyStr( str );
+		sym = getSym( );
 	} else if( sym == S_lbrak ) {
 		/* pointer indirection as [a] */
 		Expect( S_lbrak );
@@ -1223,7 +1285,17 @@ static void print_opcodes( OpcodeInfo *opcode_info )
 		while( operand != NULL ) {
 			switch( operand->type ) {
 				case OPERAND_ABSOLUTE:
-					fprintf( stderr, "%s$%X", indent, operand->num );
+					if( operand->str == NULL ) {
+						fprintf( stderr, "%s$%X", indent, operand->num );
+					} else {
+						int i;
+						int len = strlen( operand->str );
+						fprintf( stderr, "%s\"", indent );
+						for( i = 0; i < len; i++ ) {
+							fprintf( stderr, "%c", operand->str[i] );
+						}
+						fprintf( stderr, "\"" );
+					}
 					break;
 				case OPERAND_REGISTER:
 					fprintf( stderr, "%s%s", indent, registername[operand->reg] );
@@ -1385,7 +1457,15 @@ static void emit_opcode( OpcodeInfo *opcode_info )
 			Emit_double_little_endian( opcode_info->operand->num );
 			break;
 		case OPCODE_PSEUDO_DB:
-			Emit_byte( opcode_info->operand->num );
+			if( opcode_info->operand->str == NULL ) {
+				Emit_byte( opcode_info->operand->num );
+			} else {
+				int i;
+				int len = strlen( opcode_info->operand->str );
+				for( i = 0; i < len; i++ ) {
+					Emit( "%c", opcode_info->operand->str[i] );
+				}
+			}
 			break;
 		case OPCODE_HLT:
 			Emit( "%c", 0xF4 );
diff --git a/ecomp-c/tests/asm-i386/mov.asm b/ecomp-c/tests/asm-i386/mov.asm
index 3be7224..7495e63 100644
--- a/ecomp-c/tests/asm-i386/mov.asm
+++ b/ecomp-c/tests/asm-i386/mov.asm
@@ -5,8 +5,9 @@ mov al, [flag1]
 mov eax, $ffffffff
 mov [flag2], al
 hlt
-db 'XXXXXXX'
+db "The"
 flag1: db $1
+db "Hello"
 flag2: db $2
-db 'XXXXXXX'
+db "World"
 dd $01020304