1 /*****************************************************************************/
5 /* The scanner for the ca65 macroassembler */
9 /* (C) 1998-2004 Ullrich von Bassewitz */
11 /* D-70794 Filderstadt */
12 /* EMail: uz@cc65.org */
15 /* This software is provided 'as-is', without any expressed or implied */
16 /* warranty. In no event will the authors be held liable for any damages */
17 /* arising from the use of this software. */
19 /* Permission is granted to anyone to use this software for any purpose, */
20 /* including commercial applications, and to alter it and redistribute it */
21 /* freely, subject to the following restrictions: */
23 /* 1. The origin of this software must not be misrepresented; you must not */
24 /* claim that you wrote the original software. If you use this software */
25 /* in a product, an acknowledgment in the product documentation would be */
26 /* appreciated but is not required. */
27 /* 2. Altered source versions must be plainly marked as such, and must not */
28 /* be misrepresented as being the original software. */
29 /* 3. This notice may not be removed or altered from any source */
32 /*****************************************************************************/
41 #include <sys/types.h> /* EMX needs this */
66 /*****************************************************************************/
68 /*****************************************************************************/
72 enum Token Tok = TOK_NONE; /* Current token */
73 int WS; /* Flag: Whitespace before token */
74 long IVal; /* Integer token attribute */
75 char SVal [MAX_STR_LEN+1]; /* String token attribute */
77 FilePos CurPos = { 0, 0, 0 }; /* Name and position in current file */
81 /* Struct to handle include files. Note: The length of the input line may
82 * not exceed 255+1, since the column is stored in the file position struct
83 * as a character. Increasing this value means changing the FilePos struct,
84 * and the read and write routines in the assembler and linker.
86 typedef struct InputFile_ InputFile;
88 FILE* F; /* Input file descriptor */
89 FilePos Pos; /* Position in file */
90 enum Token Tok; /* Last token */
91 int C; /* Last character */
92 char Line[256]; /* The current input line */
93 InputFile* Next; /* Linked list of input files */
96 /* Struct to handle textual input data */
97 typedef struct InputData_ InputData;
99 char* Data; /* Pointer to the data */
100 const char* Pos; /* Pointer to current position */
101 int Malloced; /* Memory was malloced */
102 enum Token Tok; /* Last token */
103 int C; /* Last character */
104 InputData* Next; /* Linked list of input data */
107 /* Current input variables */
108 static InputFile* IFile = 0; /* Current input file */
109 static InputData* IData = 0; /* Current input memory data */
110 static unsigned ICount = 0; /* Count of input files */
111 static int C = 0; /* Current input character */
113 /* Force end of assembly */
116 /* List of dot keywords with the corresponding tokens */
118 const char* Key; /* MUST be first field */
123 { ".ADDR", TOK_ADDR },
124 { ".ALIGN", TOK_ALIGN },
125 { ".AND", TOK_BOOLAND },
126 { ".ASCIIZ", TOK_ASCIIZ },
127 { ".ASSERT", TOK_ASSERT },
128 { ".AUTOIMPORT", TOK_AUTOIMPORT },
129 { ".BANKBYTE", TOK_BANKBYTE },
130 { ".BITAND", TOK_AND },
131 { ".BITNOT", TOK_NOT },
132 { ".BITOR", TOK_OR },
133 { ".BITXOR", TOK_XOR },
134 { ".BLANK", TOK_BLANK },
136 { ".BYT", TOK_BYTE },
137 { ".BYTE", TOK_BYTE },
138 { ".CASE", TOK_CASE },
139 { ".CHARMAP", TOK_CHARMAP },
140 { ".CODE", TOK_CODE },
141 { ".CONCAT", TOK_CONCAT },
142 { ".CONDES", TOK_CONDES },
143 { ".CONST", TOK_CONST },
144 { ".CONSTRUCTOR", TOK_CONSTRUCTOR },
146 { ".DATA", TOK_DATA },
148 { ".DBYT", TOK_DBYT },
149 { ".DEBUGINFO", TOK_DEBUGINFO },
150 { ".DEF", TOK_DEFINED },
151 { ".DEFINE", TOK_DEFINE },
152 { ".DEFINED", TOK_DEFINED },
153 { ".DESTRUCTOR", TOK_DESTRUCTOR },
154 { ".DWORD", TOK_DWORD },
155 { ".ELSE", TOK_ELSE },
156 { ".ELSEIF", TOK_ELSEIF },
158 { ".ENDENUM", TOK_ENDENUM },
159 { ".ENDIF", TOK_ENDIF },
160 { ".ENDMAC", TOK_ENDMACRO },
161 { ".ENDMACRO", TOK_ENDMACRO },
162 { ".ENDPROC", TOK_ENDPROC },
163 { ".ENDREP", TOK_ENDREP },
164 { ".ENDREPEAT", TOK_ENDREP },
165 { ".ENDSCOPE", TOK_ENDSCOPE },
166 { ".ENDSTRUCT", TOK_ENDSTRUCT },
167 { ".ENDUNION", TOK_ENDUNION },
168 { ".ENUM", TOK_ENUM },
169 { ".ERROR", TOK_ERROR },
170 { ".EXITMAC", TOK_EXITMACRO },
171 { ".EXITMACRO", TOK_EXITMACRO },
172 { ".EXPORT", TOK_EXPORT },
173 { ".EXPORTZP", TOK_EXPORTZP },
174 { ".FARADDR", TOK_FARADDR },
175 { ".FEATURE", TOK_FEATURE },
176 { ".FILEOPT", TOK_FILEOPT },
177 { ".FOPT", TOK_FILEOPT },
178 { ".FORCEIMPORT", TOK_FORCEIMPORT },
179 { ".FORCEWORD", TOK_FORCEWORD },
180 { ".GLOBAL", TOK_GLOBAL },
181 { ".GLOBALZP", TOK_GLOBALZP },
182 { ".HIBYTE", TOK_HIBYTE },
183 { ".HIWORD", TOK_HIWORD },
187 { ".IFBLANK", TOK_IFBLANK },
188 { ".IFCONST", TOK_IFCONST },
189 { ".IFDEF", TOK_IFDEF },
190 { ".IFNBLANK", TOK_IFNBLANK },
191 { ".IFNCONST", TOK_IFNCONST },
192 { ".IFNDEF", TOK_IFNDEF },
193 { ".IFNREF", TOK_IFNREF },
194 { ".IFP02", TOK_IFP02 },
195 { ".IFP816", TOK_IFP816 },
196 { ".IFPC02", TOK_IFPC02 },
197 { ".IFPSC02", TOK_IFPSC02 },
198 { ".IFREF", TOK_IFREF },
199 { ".IMPORT", TOK_IMPORT },
200 { ".IMPORTZP", TOK_IMPORTZP },
201 { ".INCBIN", TOK_INCBIN },
202 { ".INCLUDE", TOK_INCLUDE },
203 { ".LEFT", TOK_LEFT },
204 { ".LINECONT", TOK_LINECONT },
205 { ".LIST", TOK_LIST },
206 { ".LISTBYTES", TOK_LISTBYTES },
207 { ".LOBYTE", TOK_LOBYTE },
208 { ".LOCAL", TOK_LOCAL },
209 { ".LOCALCHAR", TOK_LOCALCHAR },
210 { ".LOWORD", TOK_LOWORD },
211 { ".MAC", TOK_MACRO },
212 { ".MACPACK", TOK_MACPACK },
213 { ".MACRO", TOK_MACRO },
214 { ".MATCH", TOK_MATCH },
217 { ".NOT", TOK_BOOLNOT },
218 { ".NULL", TOK_NULL },
219 { ".OR", TOK_BOOLOR },
223 { ".P816", TOK_P816 },
224 { ".PAGELEN", TOK_PAGELENGTH },
225 { ".PAGELENGTH", TOK_PAGELENGTH },
226 { ".PARAMCOUNT", TOK_PARAMCOUNT },
227 { ".PC02", TOK_PC02 },
228 { ".POPSEG", TOK_POPSEG },
229 { ".PROC", TOK_PROC },
230 { ".PSC02", TOK_PSC02 },
231 { ".PUSHSEG", TOK_PUSHSEG },
232 { ".REF", TOK_REFERENCED },
233 { ".REFERENCED", TOK_REFERENCED },
234 { ".RELOC", TOK_RELOC },
235 { ".REPEAT", TOK_REPEAT },
237 { ".RIGHT", TOK_RIGHT },
238 { ".RODATA", TOK_RODATA },
239 { ".SCOPE", TOK_SCOPE },
240 { ".SEGMENT", TOK_SEGMENT },
241 { ".SETCPU", TOK_SETCPU },
244 { ".SIZEOF", TOK_SIZEOF },
245 { ".SMART", TOK_SMART },
246 { ".STRAT", TOK_STRAT },
247 { ".STRING", TOK_STRING },
248 { ".STRLEN", TOK_STRLEN },
249 { ".STRUCT", TOK_STRUCT },
250 { ".SUNPLUS", TOK_SUNPLUS },
252 { ".TCOUNT", TOK_TCOUNT },
253 { ".TIME", TOK_TIME },
254 { ".UNION", TOK_UNION },
255 { ".VERSION", TOK_VERSION },
256 { ".WARNING", TOK_WARNING },
257 { ".WORD", TOK_WORD },
258 { ".XMATCH", TOK_XMATCH },
259 { ".XOR", TOK_BOOLXOR },
260 { ".ZEROPAGE", TOK_ZEROPAGE },
265 /*****************************************************************************/
267 /*****************************************************************************/
271 static void NextChar (void);
272 /* Read the next character from the input file */
276 /*****************************************************************************/
277 /* Character classification functions */
278 /*****************************************************************************/
282 static int IsIdChar (int C)
283 /* Return true if the character is a valid character for an identifier */
285 return IsAlNum (C) ||
287 (C == '@' && AtInIdents) ||
288 (C == '$' && DollarInIdents);
293 static int IsIdStart (int C)
294 /* Return true if the character may start an identifier */
296 return IsAlpha (C) || C == '_';
301 /*****************************************************************************/
303 /*****************************************************************************/
307 void NewInputFile (const char* Name)
308 /* Open a new input file */
313 /* First try to open the file */
314 F = fopen (Name, "r");
319 /* Error (fatal error if this is the main file) */
321 Fatal ("Cannot open input file `%s': %s", Name, strerror (errno));
324 /* We are on include level. Search for the file in the include
327 PathName = FindInclude (Name);
328 if (PathName == 0 || (F = fopen (PathName, "r")) == 0) {
329 /* Not found or cannot open, print an error and bail out */
330 Error ("Cannot open include file `%s': %s", Name, strerror (errno));
333 /* Free the allocated memory */
338 /* check again if we do now have an open file */
343 /* Stat the file and remember the values */
345 if (fstat (fileno (F), &Buf) != 0) {
346 Fatal ("Cannot stat input file `%s': %s", Name, strerror (errno));
349 /* Add the file to the input file table and remember the index */
350 FileIdx = AddFile (Name, Buf.st_size, Buf.st_mtime);
352 /* Create a new state variable and initialize it */
353 I = xmalloc (sizeof (*I));
357 I->Pos.Name = FileIdx;
362 /* Use the new file */
374 void DoneInputFile (void)
375 /* Close the current input file */
379 /* Restore the old token */
383 /* Save a pointer to the current struct, then set it back */
387 /* Cleanup the current stuff */
395 void NewInputData (char* Data, int Malloced)
396 /* Add a chunk of input data to the input stream */
400 /* Create a new state variable and initialize it */
401 I = xmalloc (sizeof (*I));
404 I->Malloced = Malloced;
408 /* Use the new data */
418 static void DoneInputData (void)
419 /* End the current input data stream */
423 /* Restore the old token */
427 /* Save a pointer to the current struct, then set it back */
431 /* Cleanup the current stuff */
440 static unsigned DigitVal (unsigned char C)
441 /* Convert a digit into it's numerical representation */
446 return toupper (C) - 'A' + 10;
452 static void NextChar (void)
453 /* Read the next character from the input file */
455 /* If we have an input data structure, read from there */
460 /* End of input data, will set to last file char */
466 /* Check for end of line, read the next line if needed */
467 while (IFile->Line [IFile->Pos.Col] == '\0') {
469 /* End of current line reached, read next line */
470 if (fgets (IFile->Line, sizeof (IFile->Line), IFile->F) == 0) {
471 /* End of file. Add an empty line to the listing. This is a
472 * small hack needed to keep the PC output in sync.
474 NewListingLine ("", IFile->Pos.Name, ICount);
483 /* Remember the new line for the listing */
484 NewListingLine (IFile->Line, IFile->Pos.Name, ICount);
488 /* Return the next character from the file */
489 C = IFile->Line [IFile->Pos.Col++];
496 void LocaseSVal (void)
497 /* Make SVal lower case */
501 SVal [I] = tolower (SVal [I]);
508 void UpcaseSVal (void)
509 /* Make SVal upper case */
513 SVal [I] = toupper (SVal [I]);
520 static int CmpDotKeyword (const void* K1, const void* K2)
521 /* Compare function for the dot keyword search */
523 return strcmp (((struct DotKeyword*)K1)->Key, ((struct DotKeyword*)K2)->Key);
528 static unsigned char FindDotKeyword (void)
529 /* Find the dot keyword in SVal. Return the corresponding token if found,
530 * return TOK_NONE if not found.
533 static const struct DotKeyword K = { SVal, 0 };
534 struct DotKeyword* R;
536 /* If we aren't in ignore case mode, we have to uppercase the keyword */
541 /* Search for the keyword */
542 R = bsearch (&K, DotKeywords, sizeof (DotKeywords) / sizeof (DotKeywords [0]),
543 sizeof (DotKeywords [0]), CmpDotKeyword);
553 static void ReadIdent (unsigned Index)
554 /* Read an identifier from the current input position into Ident. Filling SVal
555 * starts at Index with the current character in C. It is assumed that any
556 * characters already filled in are ok, and the character in C is checked.
559 /* Read the identifier */
561 if (Index < MAX_STR_LEN) {
565 } while (IsIdChar (C));
568 /* If we should ignore case, convert the identifier to upper case */
576 static unsigned ReadStringConst (int StringTerm)
577 /* Read a string constant into SVal. Check for maximum string length and all
578 * other stuff. The length of the string is returned.
583 /* Skip the leading string terminator */
586 /* Read the string */
589 if (C == StringTerm) {
592 if (C == '\n' || C == EOF) {
593 Error ("Newline in string constant");
597 /* Check for string length, print an error message once */
598 if (I == MAX_STR_LEN) {
599 Error ("Maximum string size exceeded");
600 } else if (I < MAX_STR_LEN) {
605 /* Skip the character */
609 /* Skip the trailing terminator */
612 /* Terminate the string */
613 if (I >= MAX_STR_LEN) {
618 /* Return the length of the string */
624 void NextRawTok (void)
625 /* Read the next raw token from the input stream */
627 /* If we've a forced end of assembly, don't read further */
634 /* Check if we have tokens from another input source */
635 if (InputFromStack ()) {
640 /* Skip whitespace, remember if we had some */
641 if ((WS = IsBlank (C)) != 0) {
644 } while (IsBlank (C));
647 /* If we're reading from the file, update the location from where the
648 * next token will be read. If we're reading from input data, keep the
655 /* Hex number or PC symbol? */
659 /* Hex digit must follow or DollarIsPC must be enabled */
665 Error ("Hexadecimal digit expected");
669 /* Read the number */
671 while (IsXDigit (C)) {
672 if (IVal & 0xF0000000) {
673 Error ("Overflow in hexadecimal number");
676 IVal = (IVal << 4) + DigitVal (C);
680 /* This is an integer constant */
689 /* 0 or 1 must follow */
691 Error ("Binary digit expected");
694 /* Read the number */
696 while (IsBDigit (C)) {
697 if (IVal & 0x80000000) {
698 Error ("Overflow in binary number");
701 IVal = (IVal << 1) + DigitVal (C);
705 /* This is an integer constant */
710 /* Decimal number? */
713 /* Read the number */
715 while (IsDigit (C)) {
716 if (IVal > (long) (0xFFFFFFFFUL / 10)) {
717 Error ("Overflow in decimal number");
720 IVal = (IVal * 10) + DigitVal (C);
724 /* This is an integer constant */
729 /* Control command? */
732 /* Remember and skip the dot */
735 /* Check if it's just a dot */
736 if (!IsIdStart (C)) {
743 /* Read the remainder of the identifier */
747 /* Dot keyword, search for it */
748 Tok = FindDotKeyword ();
749 if (Tok == TOK_NONE) {
752 if (!LeadingDotInIdents) {
753 /* Invalid pseudo instruction */
754 Error ("`%s' is not a recognized control command", SVal);
758 /* An identifier with a dot. Check if it's a define style
761 if (IsDefine (SVal)) {
762 /* This is a define style macro - expand it */
767 /* Just an identifier with a dot */
776 if (C == LocalStart) {
778 /* Read the identifier */
781 /* Start character alone is not enough */
782 if (SVal [1] == '\0') {
783 Error ("Invalid cheap local symbol");
787 /* A local identifier */
788 Tok = TOK_LOCAL_IDENT;
793 /* Identifier or keyword? */
796 /* Read the identifier */
799 /* Check for special names. Bail out if we have identified the type of
800 * the token. Go on if the token is an identifier.
802 if (SVal[1] == '\0') {
803 switch (toupper (SVal [0])) {
808 Tok = TOK_OVERRIDE_ABS;
817 Tok = TOK_OVERRIDE_FAR;
837 Tok = TOK_OVERRIDE_ZP;
847 /* Check for define style macro */
848 if (IsDefine (SVal)) {
849 /* Macro - expand it */
859 /* Ok, let's do the switch */
953 while (C != '\n' && C != EOF) {
998 } else if (C == '<') {
1001 } else if (C == '>') {
1024 } else if (C == '>') {
1038 /* Hack: If we allow ' as terminating character for strings, read
1039 * the following stuff as a string, and check for a one character
1042 if (LooseStringTerm) {
1043 if (ReadStringConst ('\'') == 1) {
1050 /* Always a character constant */
1052 if (C == EOF || IsControl (C)) {
1053 Error ("Illegal character constant");
1060 if (!MissingCharTerm) {
1061 Error ("Illegal character constant");
1070 ReadStringConst ('\"');
1075 /* Line continuation? */
1079 /* Handle as white space */
1093 /* Check if we have any open .IFs in this file */
1095 /* Check if we have any open token lists in this file */
1098 /* If this was an include file, then close it and handle like a
1099 * separator. Do not close the main file, but return EOF.
1110 /* If we go here, we could not identify the current character. Skip it
1113 Error ("Invalid input character: 0x%02X", C & 0xFF);
1120 int TokHasSVal (enum Token Tok)
1121 /* Return true if the given token has an attached SVal */
1123 return (Tok == TOK_IDENT || TOK_LOCAL_IDENT || Tok == TOK_STRCON);
1128 int TokHasIVal (enum Token Tok)
1129 /* Return true if the given token has an attached IVal */
1131 return (Tok == TOK_INTCON || Tok == TOK_CHARCON);
1136 int GetSubKey (const char** Keys, unsigned Count)
1137 /* Search for a subkey in a table of keywords. The current token must be an
1138 * identifier and all keys must be in upper case. The identifier will be
1139 * uppercased in the process. The function returns the index of the keyword,
1140 * or -1 if the keyword was not found.
1145 /* Must have an identifier */
1146 PRECONDITION (Tok == TOK_IDENT);
1148 /* If we aren't in ignore case mode, we have to uppercase the identifier */
1153 /* Do a linear search (a binary search is not worth the effort) */
1154 for (I = 0; I < Count; ++I) {
1155 if (strcmp (SVal, Keys [I]) == 0) {
1167 unsigned char ParseAddrSize (void)
1168 /* Check if the next token is a keyword that denotes an address size specifier.
1169 * If so, return the corresponding address size constant, otherwise output an
1170 * error message and return ADDR_SIZE_DEFAULT.
1173 static const char* Keys[] = {
1174 "DIRECT", "ZEROPAGE", "ZP",
1175 "ABSOLUTE", "ABS", "NEAR",
1180 /* Check for an identifier */
1181 if (Tok != TOK_IDENT) {
1182 Error ("Address size specifier expected");
1183 return ADDR_SIZE_DEFAULT;
1186 /* Search for the attribute */
1187 switch (GetSubKey (Keys, sizeof (Keys) / sizeof (Keys [0]))) {
1190 case 2: return ADDR_SIZE_ZP;
1193 case 5: return ADDR_SIZE_ABS;
1194 case 6: return ADDR_SIZE_FAR;
1196 case 8: return ADDR_SIZE_LONG;
1198 Error ("Address size specifier expected");
1199 return ADDR_SIZE_DEFAULT;
1205 void InitScanner (const char* InFile)
1206 /* Initialize the scanner, open the given input file */
1208 /* Open the input file */
1209 NewInputFile (InFile);
1214 void DoneScanner (void)
1215 /* Release scanner resources */