Character maps.

Makefile

@@ -15,6 +15,7 @@ yacc_pre := \
 rgbasm_obj := \
 	src/asm/alloca.o \
 	src/asm/asmy.o \
+	src/asm/charmap.o \
 	src/asm/fstack.o \
 	src/asm/globlex.o \
 	src/asm/lexer.o \

include/asm/charmap.h

@@ -0,0 +1,18 @@
+#ifndef ASMOTOR_ASM_CHARMAP_H
+#define ASMOTOR_ASM_CHARMAP_H
+
+#define MAXCHARMAPS	512
+#define CHARMAPLENGTH	8
+
+struct Charmap {
+	int count;
+	char input[MAXCHARMAPS][CHARMAPLENGTH + 1];
+	char output[MAXCHARMAPS];
+};
+
+int readUTF8Char(char *destination, char *source);
+void charmap_Sort();
+int charmap_Add(char *input, UBYTE output);
+int charmap_Convert(char **input);
+
+#endif

include/asm/output.h

@@ -12,6 +12,7 @@ struct Section {
 	ULONG nBank;
 	struct Section *pNext;
 	struct Patch *pPatches;
+	struct Charmap *charmap;
 	UBYTE *tData;
 };
 
@@ -20,6 +21,7 @@ void out_SetFileName(char *s);
 void out_NewSection(char *pzName, ULONG secttype);
 void out_NewAbsSection(char *pzName, ULONG secttype, SLONG org, SLONG bank);
 void out_AbsByte(int b);
+void out_AbsByteGroup(char *s, int length);
 void out_RelByte(struct Expression * expr);
 void out_RelWord(struct Expression * expr);
 void out_PCRelByte(struct Expression * expr);

src/asm/charmap.c

@@ -0,0 +1,182 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+#include "asm/asm.h"
+#include "asm/charmap.h"
+#include "asm/main.h"
+#include "asm/output.h"
+
+struct Charmap globalCharmap;
+
+extern struct Section *pCurrentSection;
+
+int
+readUTF8Char(char *destination, char *source)
+{
+	int size;
+	UBYTE first;
+	first = source[0];
+
+	if(first >= 0xFC)
+	{
+		size = 6;
+	}
+	else if(first >= 0xF8)
+	{
+		size = 5;
+	}
+	else if(first >= 0xF0)
+	{
+		size = 4;
+	}
+	else if(first >= 0xE0)
+	{
+		size = 3;
+	}
+	else if(first >= 0xC0)
+	{
+		size = 2;
+	}
+	else
+	{
+		size = 1;
+	}
+	strncpy(destination, source, size);
+	destination[size] = 0;
+	return size;
+}
+
+int
+charmap_Add(char *input, UBYTE output)
+{
+	int i, input_length;
+	char temp1i[CHARMAPLENGTH + 1], temp2i[CHARMAPLENGTH + 1], temp1o, temp2o;
+
+	struct Charmap *charmap;
+
+	if(pCurrentSection)
+	{
+		if(pCurrentSection -> charmap)
+		{
+			charmap = pCurrentSection -> charmap;
+		}
+		else
+		{
+			if((charmap = (struct Charmap *) malloc(sizeof(struct Charmap))) == NULL)
+			{
+				fatalerror("Not enough memory for charmap");
+			}
+			pCurrentSection -> charmap = charmap;
+		}
+	}
+	else
+	{
+		charmap = &globalCharmap;
+	}
+
+	if(charmap -> count > MAXCHARMAPS || strlen(input) > CHARMAPLENGTH)
+	{
+		return -1;
+	}
+
+	input_length = strlen(input);
+	if(input_length > 1)
+	{
+		i = 0;
+		while(i < charmap -> count)
+		{
+			if(input_length > strlen(charmap -> input[i]))
+			{
+				memcpy(temp1i, charmap -> input[i], CHARMAPLENGTH + 1);
+				memcpy(charmap -> input[i], input, CHARMAPLENGTH + 1);
+				temp1o = charmap -> output[i];
+				charmap -> output[i] = output;
+				i++;
+				break;
+			}
+			i++;
+		}
+		while(i < charmap -> count)
+		{
+			memcpy(temp2i, charmap -> input[i], CHARMAPLENGTH + 1);
+			memcpy(charmap -> input[i], temp1i, CHARMAPLENGTH + 1);
+			memcpy(temp1i, temp2i, CHARMAPLENGTH + 1);
+			temp2o = charmap -> output[i];
+			charmap -> output[i] = temp1o;
+			temp1o = temp2o;
+			i++;
+		}
+		memcpy(charmap -> input[charmap -> count], temp1i, CHARMAPLENGTH + 1);
+		charmap -> output[charmap -> count] = temp1o;
+	}
+	else
+	{
+		memcpy(charmap -> input[charmap -> count - 1], input, CHARMAPLENGTH + 1);
+		charmap -> output[charmap -> count - 1] = output;
+	}
+	return ++charmap -> count;
+}
+
+int 
+charmap_Convert(char **input)
+{
+	struct Charmap *charmap;
+
+	char outchar[CHARMAPLENGTH + 1];
+	char *input_temp, *buffer;
+	int i, j, length;
+
+	if(pCurrentSection && pCurrentSection -> charmap)
+	{
+		charmap = pCurrentSection -> charmap;
+	}
+	else
+	{
+		charmap = &globalCharmap;
+	}
+
+	length = 0;
+	input_temp = *input;
+	if((buffer = (char *) malloc(strlen(*input))) == NULL)
+	{
+		fatalerror("Not enough memory for buffer");
+	}
+
+	length = 0;
+	while(**input)
+	{
+		j = 0;
+		for(i = 0; i < charmap -> count; i++)
+		{
+			j = strlen(charmap -> input[i]);
+			if(memcmp(*input, charmap -> input[i], j) == 0)
+			{
+				outchar[0] = charmap -> output[i];
+				outchar[1] = 0;
+				break;
+			}
+			j = 0;
+		}
+		if(!j)
+		{
+			j = readUTF8Char(outchar, *input);
+		}
+		if(!outchar[0])
+		{
+			buffer[length++] = 0;
+		}
+		else
+		{
+			for(i = 0; outchar[i]; i++)
+			{
+				buffer[length++] = outchar[i];
+			}
+		}
+		*input += j;
+	}
+	*input = input_temp;
+	*input = buffer;
+	return length;
+}
+

src/asm/globlex.c

@@ -298,6 +298,7 @@ struct sLexInitString staticstrings[] = {
 	{"rsset", T_POP_RSSET},
 
 	{"incbin", T_POP_INCBIN},
+	{"charmap", T_POP_CHARMAP},
 
 	{"fail", T_POP_FAIL},
 	{"warn", T_POP_WARN},

src/asm/output.c

@@ -11,6 +11,7 @@
 #include <string.h>
 
 #include "asm/asm.h"
+#include "asm/charmap.h"
 #include "asm/output.h"
 #include "asm/symbol.h"
 #include "asm/mylink.h"
@@ -643,6 +644,7 @@ out_FindSection(char *pzName, ULONG secttype, SLONG org,
 			pSect->nBank = bank;
 			pSect->pNext = NULL;
 			pSect->pPatches = NULL;
+			pSect->charmap = NULL;
 			pPatchSymbols = NULL;
 
 			if ((pSect->tData =
@@ -716,6 +718,14 @@ out_AbsByte(int b)
 	nPC += 1;
 	pPCSymbol->nValue += 1;
 }
+
+void 
+out_AbsByteGroup(char *s, int length)
+{
+	checkcodesection(length);
+	while (length--)
+		out_AbsByte(*s++);
+}
 /*
  * RGBAsm - OUTPUT.C - Outputs an objectfile
  *

src/asm/yaccprt1.y

@@ -8,6 +8,7 @@
 
 #include "asm/symbol.h"
 #include "asm/asm.h"
+#include "asm/charmap.h"
 #include "asm/output.h"
 #include "asm/mylink.h"
 #include "asm/fstack.h"
@@ -42,6 +43,21 @@ ULONG	str2int( char *s )
 	return( r );
 }
 
+ULONG	str2int2( char *s, int length )
+{
+	int i;
+	ULONG r=0;
+	i = (length - 4 < 0 ? 0 : length - 4);
+	while(i < length)
+	{
+		r<<=8;
+		r|=(UBYTE)(s[i]);
+		i++;
+		
+	}
+	return( r );
+}
+
 ULONG	isWhiteSpace( char s )
 {
 	return( s==' ' || s=='\t' || s=='\0' || s=='\n' );
@@ -401,6 +417,7 @@ void	if_skip_to_endc( void )
 %token	T_POP_ENDM
 %token	T_POP_RSRESET T_POP_RSSET
 %token	T_POP_INCBIN T_POP_REPT
+%token	T_POP_CHARMAP
 %token	T_POP_SHIFT
 %token	T_POP_ENDR
 %token	T_POP_FAIL

src/asm/yaccprt3.y

@@ -88,6 +88,7 @@ simple_pseudoop	:	include
 				|	rsreset
 				|	rsset
 				|	incbin
+				|	charmap
 				|	rept
 				|	shift
 				|	fail
@@ -280,6 +281,24 @@ incbin			:	T_POP_INCBIN string
 					}
 ;
 
+charmap			:	T_POP_CHARMAP string ',' string
+					{
+						if(charmap_Add($2, $4[0] & 0xFF) == -1)
+						{
+							fprintf(stderr, "Error parsing charmap. Either you've added too many (%i), or the input character length is too long (%i)' : %s\n", MAXCHARMAPS, CHARMAPLENGTH, strerror(errno));
+							yyerror("Error parsing charmap.");
+						}
+					}
+				|	T_POP_CHARMAP string ',' const
+					{
+						if(charmap_Add($2, $4 & 0xFF) == -1)
+						{
+							fprintf(stderr, "Error parsing charmap. Either you've added too many (%i), or the input character length is too long (%i)' : %s\n", MAXCHARMAPS, CHARMAPLENGTH, strerror(errno));
+							yyerror("Error parsing charmap.");
+						}
+					}
+;
+
 printt			:	T_POP_PRINTT string
 					{
 						if( nPass==1 )
@@ -339,7 +358,7 @@ constlist_8bit	:	constlist_8bit_entry
 
 constlist_8bit_entry	:               { out_Skip( 1 ); }
 						|	const_8bit	{ out_RelByte( &$1 ); }
-			   			|	string		{ out_String( $1 );  }
+			   			|	string		{ char *s; int length; s = $1; length = charmap_Convert(&s); out_AbsByteGroup(s, length); free(s); }
 ;
 
 constlist_16bit	:	constlist_16bit_entry
@@ -394,7 +413,7 @@ relocconst		:	T_ID
 				|	T_NUMBER
 						{ rpn_Number(&$$,$1);	$$.nVal = $1; }
 				|	string
-						{ ULONG r; r=str2int($1); rpn_Number(&$$,r); $$.nVal=r; }
+						{ char *s; int length; ULONG r; s = $1; length = charmap_Convert(&s); r = str2int2(s, length); free(s); rpn_Number(&$$,r); $$.nVal=r; }
 				|	T_OP_LOGICNOT relocconst %prec NEG
 						{ rpn_LOGNOT(&$$,&$2); }
 				|	relocconst T_OP_LOGICOR relocconst