Shamusworld >> Repos - rmac/blob - token.c

   1 //
   2 // RMAC - Reboot's Macro Assembler for the Atari Jaguar Console System
   3 // TOKEN.C - Token Handling
   4 // Copyright (C) 199x Landon Dyer, 2011-2012 Reboot and Friends
   5 // RMAC derived from MADMAC v1.07 Written by Landon Dyer, 1986
   6 // Source utilised with the kind permission of Landon Dyer
   7 //
   8
   9 #include "token.h"
  10 #include "error.h"
  11 #include "macro.h"
  12 #include "procln.h"
  13 #include "symbol.h"
  14
  15 #define DECL_KW                         // Declare keyword arrays
  16 #define DEF_KW                          // Declare keyword values
  17 #include "kwtab.h"                      // Incl generated keyword tables & defs
  18
  19
  20 int lnsave;                                     // 1; strcpy() text of current line
  21 int curlineno;                          // Current line number
  22 int totlines;                           // Total # of lines
  23 int mjump_align = 0;            // mjump alignment flag
  24 char lntag;                                     // Line tag
  25 char * curfname;                        // Current filename
  26 char tolowertab[128];           // Uppercase ==> lowercase
  27 char hextab[128];                       // Table of hex values
  28 char dotxtab[128];                      // Table for ".b", ".s", etc.
  29 char irbuf[LNSIZ];                      // Text for .rept block line
  30 char lnbuf[LNSIZ];                      // Text of current line
  31 WORD filecount;                         // Unique file number counter
  32 WORD cfileno;                           // Current file number
  33 TOKEN * tok;                            // Ptr to current token
  34 TOKEN * etok;                           // Ptr past last token in tokbuf[]
  35 TOKEN tokeol[1] = {EOL};        // Bailout end-of-line token
  36 char * string[TOKBUFSIZE*2];    // Token buffer string pointer storage
  37
  38 // File record, used to maintain a list of every include file ever visited
  39 #define FILEREC struct _filerec
  40 FILEREC
  41 {
  42    FILEREC * frec_next;
  43    char * frec_name;
  44 };
  45
  46 FILEREC * filerec;
  47 FILEREC * last_fr;
  48
  49 INOBJ * cur_inobj;                                              // Ptr current input obj (IFILE/IMACRO)
  50 static INOBJ * f_inobj;                                 // Ptr list of free INOBJs
  51 static IFILE * f_ifile;                                 // Ptr list of free IFILEs
  52 static IMACRO * f_imacro;                               // Ptr list of free IMACROs
  53
  54 static TOKEN tokbuf[TOKBUFSIZE];                // Token buffer (stack-like, all files)
  55
  56 char chrtab[] = {
  57         ILLEG, ILLEG, ILLEG, ILLEG,                     // NUL SOH STX ETX
  58         ILLEG, ILLEG, ILLEG, ILLEG,                     // EOT ENQ ACK BEL
  59         ILLEG, WHITE, ILLEG, ILLEG,                     // BS HT LF VT
  60         WHITE, ILLEG, ILLEG, ILLEG,                     // FF CR SO SI
  61
  62         ILLEG, ILLEG, ILLEG, ILLEG,                     // DLE DC1 DC2 DC3
  63         ILLEG, ILLEG, ILLEG, ILLEG,                     // DC4 NAK SYN ETB
  64         ILLEG, ILLEG, ILLEG, ILLEG,                     // CAN EM SUB ESC
  65         ILLEG, ILLEG, ILLEG, ILLEG,                     // FS GS RS US
  66
  67         WHITE, MULTX, MULTX, SELF,                      // SP ! " #
  68         MULTX+CTSYM, MULTX, SELF, MULTX,        // $ % & '
  69         SELF, SELF, SELF, SELF,                         // ( ) * +
  70         SELF, SELF, STSYM, SELF,                        // , - . /
  71
  72         DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM,         // 0 1
  73         DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM,         // 2 3
  74         DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM,         // 4 5
  75         DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM,         // 6 7
  76         DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM,         // 8 9
  77         MULTX, MULTX,                                                           // : ;
  78         MULTX, MULTX, MULTX, STSYM+CTSYM,                       // < = > ?
  79
  80         MULTX, STSYM+CTSYM+HDIGIT,                                                                      // @ A
  81         (char)((BYTE)DOT)+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT,       // B C
  82         STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT,                                         // D E
  83         STSYM+CTSYM+HDIGIT, STSYM+CTSYM,                                                        // F G
  84         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM,                     // H I J K
  85         (char)((BYTE)DOT)+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM,   // L M N O
  86
  87         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM,   // P Q R S
  88         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM,   // T U V W
  89         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF,                            // X Y Z [
  90         SELF, SELF, MULTX, STSYM+CTSYM,                                                         // \ ] ^ _
  91
  92         ILLEG, STSYM+CTSYM+HDIGIT,                                                                      // ` a
  93         (char)((BYTE)DOT)+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT,       // b c
  94         STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT,                                         // d e
  95         STSYM+CTSYM+HDIGIT, STSYM+CTSYM,                                                        // f g
  96         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM,                     // h i j k
  97         (char)((BYTE)DOT)+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM,   // l m n o
  98
  99         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM,   // p q r s
 100         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM,   // t u v w
 101         STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF,                            // x y z {
 102         SELF, SELF, SELF, ILLEG                                                                         // | } ~ DEL
 103 };
 104
 105 // Names of registers
 106 static char * regname[] = {
 107         "d0", "d1",  "d2",  "d3", "d4", "d5", "d6", "d7",
 108         "a0", "a1",  "a2",  "a3", "a4", "a5", "a6", "a7",
 109         "pc", "ssp", "usp", "sr", "ccr"
 110 };
 111
 112 static char * riscregname[] = {
 113          "r0",  "r1",  "r2",  "r3",  "r4", "r5",   "r6",  "r7",
 114          "r8",  "r9", "r10", "r11", "r12", "r13", "r14", "r15",
 115         "r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
 116         "r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31"
 117 };
 118
 119
 120 //
 121 // Initialize tokenizer
 122 //
 123 void InitTokenizer(void)
 124 {
 125         int i;                                                                  // Iterator
 126         char * htab = "0123456789abcdefABCDEF"; // Hex character table
 127
 128         lnsave = 0;                                                             // Don't save lines
 129         curfname = "";                                                  // No file, empty filename
 130         filecount = (WORD)-1;
 131         cfileno = (WORD)-1;                                             // cfileno gets bumped to 0
 132         curlineno = 0;
 133         totlines = 0;
 134         etok = tokbuf;
 135         f_inobj = NULL;
 136         f_ifile = NULL;
 137         f_imacro = NULL;
 138         cur_inobj = NULL;
 139         filerec = NULL;
 140         last_fr = NULL;
 141         lntag = SPACE;
 142
 143         // Initialize hex, "dot" and tolower tables
 144         for(i=0; i<128; i++)
 145         {
 146                 hextab[i] = -1;
 147                 dotxtab[i] = 0;
 148                 tolowertab[i] = (char)i;
 149         }
 150
 151         for(i=0; htab[i]!=EOS; i++)
 152                 hextab[htab[i]] = (char)((i < 16) ? i : i - 6);
 153
 154         for(i='A'; i<='Z'; i++)
 155                 tolowertab[i] |= 0x20;
 156
 157         // These characters are legal immediately after a period
 158         dotxtab['b'] = DOTB;                                    // .b .B .s .S
 159         dotxtab['B'] = DOTB;
 160         dotxtab['s'] = DOTB;
 161         dotxtab['S'] = DOTB;
 162         dotxtab['w'] = DOTW;                                    // .w .W
 163         dotxtab['W'] = DOTW;
 164         dotxtab['l'] = DOTL;                                    // .l .L
 165         dotxtab['L'] = DOTL;
 166         dotxtab['i'] = DOTI;                                    // .i .I (???)
 167         dotxtab['I'] = DOTI;
 168 }
 169
 170
 171 void SetFilenameForErrorReporting(void)
 172 {
 173         WORD fnum = cfileno;
 174
 175         // Check for absolute top filename (this should never happen)
 176         if (fnum == -1)
 177         {
 178                 curfname = "(*top*)";
 179                 return;
 180         }
 181
 182         FILEREC * fr = filerec;
 183
 184         // Advance to the correct record...
 185         while (fr != NULL && fnum != 0)
 186         {
 187                 fr = fr->frec_next;
 188                 fnum--;
 189         }
 190
 191         // Check for file # record not found (this should never happen either)
 192         if (fr == NULL)
 193         {
 194                 curfname = "(*NOT FOUND*)";
 195                 return;
 196         }
 197
 198         curfname = fr->frec_name;
 199 }
 200
 201
 202 //
 203 // Allocate an IFILE or IMACRO
 204 //
 205 INOBJ * a_inobj(int typ)
 206 {
 207         INOBJ * inobj;
 208         IFILE * ifile;
 209         IMACRO * imacro;
 210
 211         // Allocate and initialize INOBJ first
 212         if (f_inobj == NULL)
 213                 inobj = malloc(sizeof(INOBJ));
 214         else
 215         {
 216                 inobj = f_inobj;
 217                 f_inobj = f_inobj->in_link;
 218         }
 219
 220         switch (typ)
 221         {
 222         case SRC_IFILE:                                                 // Alloc and init an IFILE
 223                 if (f_ifile == NULL)
 224                         ifile = malloc(sizeof(IFILE));
 225                 else
 226                 {
 227                         ifile = f_ifile;
 228                         f_ifile = f_ifile->if_link;
 229                 }
 230
 231                 inobj->inobj.ifile = ifile;
 232                 break;
 233         case SRC_IMACRO:                                                // Alloc and init an IMACRO
 234                 if (f_imacro == NULL)
 235                         imacro = malloc(sizeof(IMACRO));
 236                 else
 237                 {
 238                         imacro = f_imacro;
 239                         f_imacro = f_imacro->im_link;
 240                 }
 241
 242                 inobj->inobj.imacro = imacro;
 243                 break;
 244         case SRC_IREPT:                                                 // Alloc and init an IREPT
 245                 inobj->inobj.irept = malloc(sizeof(IREPT));
 246                 DEBUG printf("alloc IREPT\n");
 247                 break;
 248         }
 249
 250         // Install INOBJ on top of input stack
 251         inobj->in_ifent = ifent;                                // Record .if context on entry
 252         inobj->in_type = (WORD)typ;
 253         inobj->in_otok = tok;
 254         inobj->in_etok = etok;
 255         inobj->in_link = cur_inobj;
 256         cur_inobj = inobj;
 257
 258         return inobj;
 259 }
 260
 261
 262 //
 263 // Perform macro substitution from 'orig' to 'dest'. Return OK or some error.
 264 // A macro reference is in one of two forms:
 265 // \name <non-name-character>
 266 // \{name}
 267 // A doubled backslash (\\) is compressed to a single backslash (\).
 268 // Argument definitions have been pre-tokenized, so we have to turn them back
 269 // into text. This means that numbers, in particular, become hex, regardless of
 270 // their representation when the macro was invoked. This is a hack.
 271 // A label may appear at the beginning of the line:
 272 // :<name><whitespace>
 273 // (the colon must be in the first column). These labels are stripped before
 274 // macro expansion takes place.
 275 //
 276 int ExpandMacro(char * src, char * dest, int destsiz)
 277 {
 278         int i;
 279         int questmark;                  // \? for testing argument existence
 280         char mname[128];                // Assume max size of a formal arg name
 281         char numbuf[20];                // Buffer for text of CONSTs
 282         TOKEN * tk;
 283         SYM * arg;
 284         char ** symbolString;
 285
 286         DEBUG { printf("ExM: src=\"%s\"\n", src); }
 287
 288         IMACRO * imacro = cur_inobj->inobj.imacro;
 289         int macnum = (int)(imacro->im_macro->sattr);
 290
 291 //      destsiz--;
 292         char * dst = dest;                                              // Next dest slot
 293         char * edst = dest + destsiz - 1;               // End + 1(?) of dest buffer
 294
 295         // Check for (and skip over) any "label" on the line
 296         char * s = src;
 297         char * d = NULL;
 298
 299         if (*s == ':')
 300         {
 301                 while (*s != EOS && !(chrtab[*s] & WHITE))
 302                         s++;
 303
 304                 if (*s != EOS)
 305                         s++;                                                    // Skip first whitespace
 306         }
 307
 308         // Expand the rest of the line
 309         while (*s != EOS)
 310         {
 311                 // Copy single character
 312                 if (*s != '\\')
 313                 {
 314                         if (dst >= edst)
 315                                 goto overflow;
 316
 317                         *dst++ = *s++;
 318                 }
 319                 // Do macro expansion
 320                 else
 321                 {
 322                         questmark = 0;
 323
 324                         // Do special cases
 325                         switch (*++s)
 326                         {
 327                         case '\\':                                              // \\, \ (collapse to single backslash)
 328                                 if (dst >= edst)
 329                                         goto overflow;
 330
 331                                 *dst++ = *s++;
 332                                 continue;
 333                         case '?':                                               // \? <macro>  set `questmark' flag
 334                                 ++s;
 335                                 questmark = 1;
 336                                 break;
 337                         case '#':                                               // \#, number of arguments
 338                                 sprintf(numbuf, "%d", (int)imacro->im_nargs);
 339                                 goto copystr;
 340                         case '!':                                               // \! size suffix supplied on invocation
 341                                 switch ((int)imacro->im_siz)
 342                                 {
 343                                 case SIZN: d = "";   break;
 344                                 case SIZB: d = ".b"; break;
 345                                 case SIZW: d = ".w"; break;
 346                                 case SIZL: d = ".l"; break;
 347                                 }
 348
 349                                 goto copy_d;
 350                         case '~':                                               // ==> unique label string Mnnnn...
 351                                 sprintf(numbuf, "M%u", curuniq);
 352 copystr:
 353                                 d = numbuf;
 354 copy_d:
 355                                 s++;
 356
 357                                 while (*d != EOS)
 358                                 {
 359                                         if (dst >= edst)
 360                                                 goto overflow;
 361                                         else
 362                                                 *dst++ = *d++;
 363                                 }
 364
 365                                 continue;
 366                         case EOS:
 367                                 return error("missing argument name");
 368                         }
 369
 370                         // \n ==> argument number 'n', 0..9
 371                         if (chrtab[*s] & DIGIT)
 372                         {
 373                                 i = *s++ - '1';
 374
 375                                 if (i < 0)
 376                                         i = 9;
 377
 378                                 goto arg_num;
 379                         }
 380
 381                         // Get argument name: \name, \{name}
 382                         d = mname;
 383
 384                         // \label
 385                         if (*s != '{')
 386                         {
 387                                 do
 388                                 {
 389                                         *d++ = *s++;
 390                                 }
 391                                 while (chrtab[*s] & CTSYM);
 392                         }
 393                         // \\{label}
 394                         else
 395                         {
 396                                 for(++s; *s != EOS && *s != '}';)
 397                                         *d++ = *s++;
 398
 399                                 if (*s != '}')
 400                                         return error("missing '}'");
 401                                 else
 402                                         s++;
 403                         }
 404
 405                         *d = EOS;
 406
 407                         // Lookup the argument and copy its (string) value into the
 408                         // destination string
 409                         DEBUG printf("argument='%s'\n", mname);
 410
 411                         if ((arg = lookup(mname, MACARG, macnum)) == NULL)
 412                                 return errors("undefined argument: '%s'", mname);
 413                         else
 414                         {
 415                                 // Convert a string of tokens (terminated with EOL) back into
 416                                 // text. If an argument is out of range (not specified in the
 417                                 // macro invocation) then it is ignored.
 418                                 i = (int)arg->svalue;
 419 arg_num:
 420                                 DEBUG printf("~argnumber=%d (argBase=%u)\n", i, imacro->argBase);
 421                                 tk = NULL;
 422
 423                                 if (i < imacro->im_nargs)
 424                                 {
 425 #if 0
 426 //                                      tk = argp[i];
 427 //                                      tk = argPtrs[i];
 428                                         tk = argPtrs[imacro->argBase + i];
 429 #else
 430                                         tk = imacro->argument[i].token;
 431                                         symbolString = imacro->argument[i].string;
 432 //DEBUG
 433 //{
 434 //      printf("ExM: Preparing to parse argument #%u...\n", i);
 435 //      dumptok(tk);
 436 //}
 437 #endif
 438                                 }
 439
 440                                 // \?arg yields:
 441                                 //    0  if the argument is empty or non-existant,
 442                                 //    1  if the argument is not empty
 443                                 if (questmark)
 444                                 {
 445                                         if (tk == NULL || *tk == EOL)
 446                                                 questmark = 0;
 447
 448                                         if (dst >= edst)
 449                                                 goto overflow;
 450
 451                                         *dst++ = (char)(questmark + '0');
 452                                         continue;
 453                                 }
 454
 455                                 // Argument # is in range, so expand it
 456                                 if (tk != NULL)
 457                                 {
 458                                         while (*tk != EOL)
 459                                         {
 460                                                 // Reverse-translation from a token number to a string.
 461                                                 // This is a hack. It might be better table-driven.
 462                                                 d = NULL;
 463
 464                                                 if ((*tk >= KW_D0) && !rdsp && !rgpu)
 465                                                 {
 466                                                         d = regname[(int)*tk++ - KW_D0];
 467                                                         goto strcopy;
 468                                                 }
 469                                                 else if ((*tk >= KW_R0) && (*tk <= KW_R31))
 470                                                 {
 471                                                         d = riscregname[(int)*tk++ - KW_R0];
 472                                                         goto strcopy;
 473                                                 }
 474                                                 else
 475                                                 {
 476                                                         switch ((int)*tk++)
 477                                                         {
 478                                                         case SYMBOL:
 479 #if 0
 480 //                                                              d = (char *)*tk++;
 481                                                                 d = string[*tk++];
 482 #else
 483                                                                 // This fix should be done for strings too
 484                                                                 d = symbolString[*tk++];
 485 DEBUG printf("ExM: SYMBOL=\"%s\"", d);
 486 #endif
 487                                                                 break;
 488                                                         case STRING:
 489 #if 0
 490 //                                                              d = (char *)*tk++;
 491                                                                 d = string[*tk++];
 492 #else
 493                                                                 d = symbolString[*tk++];
 494 #endif
 495                                                                 if (dst >= edst)
 496                                                                         goto overflow;
 497
 498                                                                 *dst++ = '"';
 499
 500                                                                 while (*d != EOS)
 501                                                                 {
 502                                                                         if (dst >= edst)
 503                                                                                 goto overflow;
 504                                                                         else
 505                                                                                 *dst++ = *d++;
 506                                                                 }
 507
 508                                                                 if (dst >= edst)
 509                                                                         goto overflow;
 510
 511                                                                 *dst++ = '"';
 512                                                                 continue;
 513                                                                 break;
 514 // Shamus: Changing the format specifier from %lx to %ux caused
 515 //         the assembler to choke on legitimate code... Need to investigate
 516 //         this further before changing anything else here!
 517                                                         case CONST:
 518                                                                 sprintf(numbuf, "$%lx", (LONG)*tk++);
 519                                                                 d = numbuf;
 520                                                                 break;
 521                                                         case DEQUALS:
 522                                                                 d = "==";
 523                                                                 break;
 524                                                         case SET:
 525                                                                 d = "set";
 526                                                                 break;
 527                                                         case COLON:
 528                                                                 d = ":";
 529                                                                 break;
 530                                                         case DCOLON:
 531                                                                 d = "::";
 532                                                                 break;
 533                                                         case GE:
 534                                                                 d = ">=";
 535                                                                 break;
 536                                                         case LE:
 537                                                                 d = "<=";
 538                                                                 break;
 539                                                         case NE:
 540                                                                 d = "<>";
 541                                                                 break;
 542                                                         case SHR:
 543                                                                 d = ">>";
 544                                                                 break;
 545                                                         case SHL:
 546                                                                 d = "<<";
 547                                                                 break;
 548                                                         case DOTB:
 549                                                                 d = ".b";
 550                                                                 break;
 551                                                         case DOTW:
 552                                                                 d = ".w";
 553                                                                 break;
 554                                                         case DOTL:
 555                                                                 d = ".l";
 556                                                                 break;
 557                                                         case CR_ABSCOUNT:
 558                                                                 d = "^^abscount";
 559                                                                 break;
 560                                                         case CR_DATE:
 561                                                                 d = "^^date";
 562                                                                 break;
 563                                                         case CR_TIME:
 564                                                                 d = "^^time";
 565                                                                 break;
 566                                                         case CR_DEFINED:
 567                                                                 d = "^^defined ";
 568                                                                 break;
 569                                                         case CR_REFERENCED:
 570                                                                 d = "^^referenced ";
 571                                                                 break;
 572                                                         case CR_STREQ:
 573                                                                 d = "^^streq ";
 574                                                                 break;
 575                                                         case CR_MACDEF:
 576                                                                 d = "^^macdef ";
 577                                                                 break;
 578                                                         default:
 579                                                                 if (dst >= edst)
 580                                                                         goto overflow;
 581
 582                                                                 *dst++ = (char)*(tk - 1);
 583                                                                 break;
 584                                                         }
 585                                                 }
 586
 587                                                 // If 'd' != NULL, copy string to destination
 588                                                 if (d != NULL)
 589                                                 {
 590 strcopy:
 591                                                         DEBUG printf("d='%s'\n", d);
 592
 593                                                         while (*d != EOS)
 594                                                         {
 595                                                                 if (dst >= edst)
 596                                                                         goto overflow;
 597                                                                 else
 598                                                                         *dst++ = *d++;
 599                                                         }
 600                                                 }
 601                                         }
 602                                 }
 603                         }
 604                 }
 605         }
 606
 607         *dst = EOS;
 608         DEBUG { printf("ExM: dst=\"%s\"\n", dest); }
 609         return OK;
 610
 611 overflow:
 612         *dst = EOS;
 613         DEBUG printf("*** OVERFLOW LINE ***\n%s\n", dest);
 614         return fatal("line too long as a result of macro expansion");
 615 }
 616
 617
 618 //
 619 // Get next line of text from a macro
 620 //
 621 char * GetNextMacroLine(void)
 622 {
 623 //      unsigned source_addr;
 624
 625         IMACRO * imacro = cur_inobj->inobj.imacro;
 626 //      LONG * strp = imacro->im_nextln;
 627         struct LineList * strp = imacro->im_nextln;
 628
 629         if (strp == NULL)                                               // End-of-macro
 630                 return NULL;
 631
 632 //      imacro->im_nextln = (LONG *)*strp;
 633         imacro->im_nextln = strp->next;
 634 //      ExpandMacro((char *)(strp + 1), imacro->im_lnbuf, LNSIZ);
 635         ExpandMacro(strp->line, imacro->im_lnbuf, LNSIZ);
 636
 637         return imacro->im_lnbuf;
 638 }
 639
 640
 641 //
 642 // Get next line of text from a repeat block
 643 //
 644 char * GetNextRepeatLine(void)
 645 {
 646
 647         IREPT * irept = cur_inobj->inobj.irept;
 648         LONG * strp = irept->ir_nextln;                 // initial null
 649
 650         // Do repeat at end of .rept block's string list
 651         if (strp == NULL)
 652         {
 653                 DEBUG printf("back-to-top-of-repeat-block count=%d\n", (int)irept->ir_count);
 654                 irept->ir_nextln = irept->ir_firstln;   // copy first line
 655
 656                 if (irept->ir_count-- == 0)
 657                 {
 658                         DEBUG printf("end-repeat-block\n");
 659                         return NULL;
 660                 }
 661
 662                 strp = irept->ir_nextln;
 663         }
 664
 665         strcpy(irbuf, (char *)(irept->ir_nextln + 1));
 666         DEBUG printf("repeat line='%s'\n", irbuf);
 667         irept->ir_nextln = (LONG *)*strp;
 668
 669         return irbuf;
 670 }
 671
 672
 673 //
 674 // Include a source file used at the root, and for ".include" files
 675 //
 676 int include(int handle, char * fname)
 677 {
 678         IFILE * ifile;
 679         INOBJ * inobj;
 680         FILEREC * fr;
 681
 682         // Verbose mode
 683         if (verb_flag)
 684                 printf("[include: %s, cfileno=%u]\n", fname, cfileno);
 685
 686         // Alloc and initialize include-descriptors
 687         inobj = a_inobj(SRC_IFILE);
 688         ifile = inobj->inobj.ifile;
 689
 690         ifile->ifhandle = handle;                               // Setup file handle
 691         ifile->ifind = ifile->ifcnt = 0;                // Setup buffer indices
 692         ifile->ifoldlineno = curlineno;                 // Save old line number
 693         ifile->ifoldfname = curfname;                   // Save old filename
 694         ifile->ifno = cfileno;                                  // Save old file number
 695
 696 //      cfileno = filecount++;                                  // Compute new file number
 697         // NB: This *must* be preincrement, we're adding one to the filecount here!
 698         cfileno = ++filecount;                                  // Compute NEW file number
 699         curfname = strdup(fname);                               // Set current filename (alloc storage)
 700         curlineno = 0;                                                  // Start on line zero
 701
 702         // Add another file to the file-record
 703         fr = (FILEREC *)malloc(sizeof(FILEREC));
 704         fr->frec_next = NULL;
 705         fr->frec_name = curfname;
 706
 707         if (last_fr == NULL)
 708                 filerec = fr;                                           // Add first filerec
 709         else
 710                 last_fr->frec_next = fr;                        // Append to list of filerecs
 711
 712         last_fr = fr;
 713         DEBUG printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno);
 714
 715         return OK;
 716 }
 717
 718
 719 //
 720 // Pop the current input level
 721 //
 722 int fpop(void)
 723 {
 724         IFILE * ifile;
 725         IMACRO * imacro;
 726         LONG * p, * p1;
 727         INOBJ * inobj = cur_inobj;
 728
 729         if (inobj != NULL)
 730         {
 731                 // Pop IFENT levels until we reach the conditional assembly context we
 732                 // were at when the input object was entered.
 733                 while (ifent != inobj->in_ifent)
 734                 {
 735                         if (d_endif() != 0)                             // Something bad happened during endif parsing?
 736                                 return -1;                                      // If yes, bail instead of getting stuck in a loop
 737                 }
 738
 739                 tok = inobj->in_otok;                           // Restore tok and otok
 740                 etok = inobj->in_etok;
 741
 742                 switch (inobj->in_type)
 743                 {
 744                 case SRC_IFILE:                                         // Pop and release an IFILE
 745                         if (verb_flag)
 746                                 printf("[Leaving: %s]\n", curfname);
 747
 748                         ifile = inobj->inobj.ifile;
 749                         ifile->if_link = f_ifile;
 750                         f_ifile = ifile;
 751                         close(ifile->ifhandle);                 // Close source file
 752 if (verb_flag) printf("[fpop (pre):  curfname=%s]\n", curfname);
 753                         curfname = ifile->ifoldfname;   // Set current filename
 754 if (verb_flag) printf("[fpop (post): curfname=%s]\n", curfname);
 755 if (verb_flag) printf("[fpop: (pre)  cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
 756                         curlineno = ifile->ifoldlineno; // Set current line#
 757                         DEBUG printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno);
 758                         cfileno = ifile->ifno;                  // Restore current file number
 759 if (verb_flag) printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
 760                         break;
 761                 case SRC_IMACRO:                                        // Pop and release an IMACRO
 762                         imacro = inobj->inobj.imacro;
 763                         imacro->im_link = f_imacro;
 764                         f_imacro = imacro;
 765                         break;
 766                 case SRC_IREPT:                                         // Pop and release an IREPT
 767                         DEBUG printf("dealloc IREPT\n");
 768                         p = inobj->inobj.irept->ir_firstln;
 769
 770                         while (p != NULL)
 771                         {
 772                                 p1 = (LONG *)*p;
 773                                 p = p1;
 774                         }
 775
 776                         break;
 777                 }
 778
 779                 cur_inobj = inobj->in_link;
 780                 inobj->in_link = f_inobj;
 781                 f_inobj = inobj;
 782         }
 783
 784         return 0;
 785 }
 786
 787
 788 //
 789 // Get line from file into buf, return NULL on EOF or ptr to the start of a
 790 // null-term line
 791 //
 792 char * GetNextLine(void)
 793 {
 794         int i, j;
 795         char * p, * d;
 796         int readamt = -1;                                               // 0 if last read() yeilded 0 bytes
 797         IFILE * fl = cur_inobj->inobj.ifile;
 798
 799         for(;;)
 800         {
 801                 // Scan for next end-of-line; handle stupid text formats by treating
 802                 // \r\n the same as \n. (lone '\r' at end of buffer means we have to
 803                 // check for '\n').
 804                 d = &fl->ifbuf[fl->ifind];
 805
 806                 for(p=d, i=0, j=fl->ifcnt; i<j; i++, p++)
 807                 {
 808                         if (*p == '\r' || *p == '\n')
 809                         {
 810                                 i++;
 811
 812                                 if (*p == '\r')
 813                                 {
 814                                         if (i >= j)
 815                                                 break;  // Need to read more, then look for '\n' to eat
 816                                         else if (p[1] == '\n')
 817                                                 i++;
 818                                 }
 819
 820                                 // Cover up the newline with end-of-string sentinel
 821                                 *p = '\0';
 822
 823                                 fl->ifind += i;
 824                                 fl->ifcnt -= i;
 825                                 return d;
 826                         }
 827                 }
 828
 829                 // Handle hanging lines by ignoring them (Input file is exhausted, no
 830                 // \r or \n on last line)
 831                 // Shamus: This is retarded. Never ignore any input!
 832                 if (!readamt && fl->ifcnt)
 833                 {
 834 #if 0
 835                         fl->ifcnt = 0;
 836                         *p = '\0';
 837                         return NULL;
 838 #else
 839                         // Really should check to see if we're at the end of the buffer!
 840                         // :-P
 841                         fl->ifbuf[fl->ifind + fl->ifcnt] = '\0';
 842                         fl->ifcnt = 0;
 843                         return &fl->ifbuf[fl->ifind];
 844 #endif
 845                 }
 846
 847                 // Truncate and return absurdly long lines.
 848                 if (fl->ifcnt >= QUANTUM)
 849                 {
 850                         fl->ifbuf[fl->ifind + fl->ifcnt - 1] = '\0';
 851                         fl->ifcnt = 0;
 852                         return &fl->ifbuf[fl->ifind];
 853                 }
 854
 855                 // Relocate what's left of a line to the beginning of the buffer, and
 856                 // read some more of the file in; return NULL if the buffer's empty and
 857                 // on EOF.
 858                 if (fl->ifind != 0)
 859                 {
 860                         p = &fl->ifbuf[fl->ifind];
 861                         d = &fl->ifbuf[fl->ifcnt & 1];
 862
 863                         for(i=0; i<fl->ifcnt; i++)
 864                                 *d++ = *p++;
 865
 866                         fl->ifind = fl->ifcnt & 1;
 867                 }
 868
 869                 readamt = read(fl->ifhandle, &fl->ifbuf[fl->ifind + fl->ifcnt], QUANTUM);
 870
 871                 if (readamt < 0)
 872                         return NULL;
 873
 874                 if ((fl->ifcnt += readamt) == 0)
 875                         return NULL;
 876         }
 877 }
 878
 879
 880 //
 881 // Tokenize a line
 882 //
 883 int TokenizeLine(void)
 884 {
 885         char * ln = NULL;                       // Ptr to current position in line
 886         char * p;                                       // Random character ptr
 887         TOKEN * tk;                                     // Token-deposit ptr
 888         int state = 0;                          // State for keyword detector
 889         int j = 0;                                      // Var for keyword detector
 890         char c;                                         // Random char
 891         VALUE v;                                        // Random value
 892         char * nullspot = NULL;         // Spot to clobber for SYMBOL termination
 893         int stuffnull;                          // 1:terminate SYMBOL '\0' at *nullspot
 894         char c1;
 895         int stringNum = 0;                      // Pointer to string locations in tokenized line
 896
 897 retry:
 898
 899         if (cur_inobj == NULL)                                  // Return EOF if input stack is empty
 900                 return TKEOF;
 901
 902         // Get another line of input from the current input source: a file, a
 903         // macro, or a repeat-block
 904         switch (cur_inobj->in_type)
 905         {
 906         // Include-file:
 907         // o  handle EOF;
 908         // o  bump source line number;
 909         // o  tag the listing-line with a space;
 910         // o  kludge lines generated by Alcyon C.
 911         case SRC_IFILE:
 912                 if ((ln = GetNextLine()) == NULL)
 913                 {
 914 if (verb_flag) printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n");
 915                         fpop();                                                 // Pop input level
 916                         goto retry;                                             // Try for more lines
 917                 }
 918
 919                 curlineno++;                                            // Bump line number
 920                 lntag = SPACE;
 921
 922                 if (as68_flag)
 923                 {
 924                         // AS68 compatibility, throw away all lines starting with
 925                         // back-quotes, tildes, or '*'
 926                         // On other lines, turn the first '*' into a semi-colon.
 927                         if (*ln == '`' || *ln == '~' || *ln == '*')
 928                                 *ln = ';';
 929                         else
 930                         {
 931                                 for(p=ln; *p!=EOS; p++)
 932                                 {
 933                                         if (*p == '*')
 934                                         {
 935                                                 *p = ';';
 936                                                 break;
 937                                         }
 938                                 }
 939                         }
 940                 }
 941
 942                 break;
 943         // Macro-block:
 944         // o  Handle end-of-macro;
 945         // o  tag the listing-line with an at (@) sign.
 946         case SRC_IMACRO:
 947                 if ((ln = GetNextMacroLine()) == NULL)
 948                 {
 949                         if (ExitMacro() == 0)                   // Exit macro (pop args, do fpop(), etc)
 950                                 goto retry;                                     // Try for more lines...
 951                         else
 952                                 return TKEOF;                           // Oops, we got a non zero return code, signal EOF
 953                 }
 954
 955                 lntag = '@';
 956                 break;
 957         // Repeat-block:
 958         // o  Handle end-of-repeat-block;
 959         // o  tag the listing-line with a pound (#) sign.
 960         case SRC_IREPT:
 961                 if ((ln = GetNextRepeatLine()) == NULL)
 962                 {
 963 if (verb_flag) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n");
 964                         fpop();
 965                         goto retry;
 966                 }
 967
 968                 lntag = '#';
 969                 break;
 970         }
 971
 972         // Save text of the line.  We only do this during listings and within
 973         // macro-type blocks, since it is expensive to unconditionally copy every
 974         // line.
 975         if (lnsave)
 976                 strcpy(lnbuf, ln);
 977
 978         // General house-keeping
 979         tok = tokeol;                   // Set "tok" to EOL in case of error
 980         tk = etok;                              // Reset token ptr
 981         stuffnull = 0;                  // Don't stuff nulls
 982         totlines++;                             // Bump total #lines assembled
 983
 984         // See if the entire line is a comment. This is a win if the programmer
 985         // puts in lots of comments
 986         if (*ln == '*' || *ln == ';' || ((*ln == '/') && (*(ln + 1) == '/')))
 987                 goto goteol;
 988
 989         // Main tokenization loop;
 990         // o  skip whitespace;
 991         // o  handle end-of-line;
 992         // o  handle symbols;
 993         // o  handle single-character tokens (operators, etc.);
 994         // o  handle multiple-character tokens (constants, strings, etc.).
 995         for(; *ln!=EOS;)
 996         {
 997                 // Skip whitespace, handle EOL
 998                 while ((int)chrtab[*ln] & WHITE)
 999                         ln++;
1000
1001                 // Handle EOL, comment with ';'
1002                 if (*ln == EOS || *ln == ';'|| ((*ln == '/') && (*(ln + 1) == '/')))
1003                         break;
1004
1005                 // Handle start of symbol. Symbols are null-terminated in place. The
1006                 // termination is always one symbol behind, since there may be no place
1007                 // for a null in the case that an operator immediately follows the name.
1008                 c = chrtab[*ln];
1009
1010                 if (c & STSYM)
1011                 {
1012                         if (stuffnull)                  // Terminate old symbol from previous pass
1013                                 *nullspot = EOS;
1014
1015                         v = 0;                                  // Assume no DOT attrib follows symbol
1016                         stuffnull = 1;
1017                         p = nullspot = ln++;    // Nullspot -> start of this symbol
1018
1019                         // Find end of symbol (and compute its length)
1020                         for(j=1; (int)chrtab[*ln]&CTSYM; j++)
1021                                 ln++;
1022
1023                         // Handle "DOT" special forms (like ".b") that follow a normal
1024                         // symbol or keyword:
1025                         if (*ln == '.')
1026                         {
1027                                 *ln++ = EOS;            // Terminate symbol
1028                                 stuffnull = 0;          // And never try it again
1029
1030                                 // Character following the `.' must have a DOT attribute, and
1031                                 // the chararacter after THAT one must not have a start-symbol
1032                                 // attribute (to prevent symbols that look like, for example,
1033                                 // "zingo.barf", which might be a good idea anyway....)
1034                                 if ((((int)chrtab[*ln] & DOT) == 0) || ((int)dotxtab[*ln] <= 0))
1035                                         return error("[bwsl] must follow `.' in symbol");
1036
1037                                 v = (VALUE)dotxtab[*ln++];
1038
1039                                 if ((int)chrtab[*ln] & CTSYM)
1040                                         return error("misuse of `.', not allowed in symbols");
1041                         }
1042
1043                         // If the symbol is small, check to see if it's really the name of
1044                         // a register.
1045                         if (j <= KWSIZE)
1046                         {
1047                                 for(state=0; state>=0;)
1048                                 {
1049                                         j = (int)tolowertab[*p++];
1050                                         j += kwbase[state];
1051
1052                                         if (kwcheck[j] != state)
1053                                         {
1054                                                 j = -1;
1055                                                 break;
1056                                         }
1057
1058                                         if (*p == EOS || p == ln)
1059                                         {
1060                                                 j = kwaccept[j];
1061                                                 break;
1062                                         }
1063
1064                                         state = kwtab[j];
1065                                 }
1066                         }
1067                         else
1068                         {
1069                                 j = -1;
1070                         }
1071
1072                         // Make j = -1 if user tries to use a RISC register while in 68K mode
1073                         if (!(rgpu || rdsp) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31))
1074                         {
1075                                 j = -1;
1076                         }
1077
1078                         // Make j = -1 if time, date etc with no preceeding ^^
1079                         // defined, referenced, streq, macdef, date and time
1080                         switch ((TOKEN)j)
1081                         {
1082                         case 112:   // defined
1083                         case 113:   // referenced
1084                         case 118:   // streq
1085                         case 119:   // macdef
1086                         case 120:   // time
1087                         case 121:   // date
1088                                 j = -1;
1089                         }
1090
1091                         // If not tokenized keyword OR token was not found
1092                         if (j < 0 || state < 0)
1093                         {
1094                                 *tk++ = SYMBOL;
1095 //#warning
1096 //problem here: nullspot is a char * but TOKEN is a uint32_t. On a 64-bit
1097 //system, this will cause all kinds of mischief.
1098 #if 0
1099                                 *tk++ = (TOKEN)nullspot;
1100 #else
1101                                 string[stringNum] = nullspot;
1102                                 *tk++ = stringNum;
1103                                 stringNum++;
1104 #endif
1105                         }
1106                         else
1107                         {
1108                                 *tk++ = (TOKEN)j;
1109                                 stuffnull = 0;
1110                         }
1111
1112                         if (v)                                                  // Record attribute token (if any)
1113                                 *tk++ = (TOKEN)v;
1114
1115                         if (stuffnull)                                  // Arrange for string termination on next pass
1116                                 nullspot = ln;
1117
1118                         continue;
1119                 }
1120
1121                 // Handle identity tokens
1122                 if (c & SELF)
1123                 {
1124                         *tk++ = *ln++;
1125                         continue;
1126                 }
1127
1128                 // Handle multiple-character tokens
1129                 if (c & MULTX)
1130                 {
1131                         switch (*ln++)
1132                         {
1133                         case '!':               // ! or !=
1134                                 if (*ln == '=')
1135                                 {
1136                                         *tk++ = NE;
1137                                         ++ln;
1138                                 }
1139                                 else
1140                                         *tk++ = '!';
1141
1142                                 continue;
1143                         case '\'':              // 'string'
1144                         case '\"':              // "string"
1145                                 c1 = ln[-1];
1146                                 *tk++ = STRING;
1147 //#warning
1148 // More char * stuffing (8 bytes) into the space of 4 (TOKEN).
1149 // Need to figure out how to fix this crap.
1150 #if 0
1151                                 *tk++ = (TOKEN)ln;
1152 #else
1153                                 string[stringNum] = ln;
1154                                 *tk++ = stringNum;
1155                                 stringNum++;
1156 #endif
1157
1158                                 for(p=ln; *ln!=EOS && *ln!=c1;)
1159                                 {
1160                                         c = *ln++;
1161
1162                                         if (c == '\\')
1163                                         {
1164                                                 switch (*ln++)
1165                                                 {
1166                                                 case EOS:
1167                                                         return(error("unterminated string"));
1168                                                 case 'e':
1169                                                         c = '\033';
1170                                                         break;
1171                                                 case 'n':
1172                                                         c = '\n';
1173                                                         break;
1174                                                 case 'b':
1175                                                         c = '\b';
1176                                                         break;
1177                                                 case 't':
1178                                                         c = '\t';
1179                                                         break;
1180                                                 case 'r':
1181                                                         c = '\r';
1182                                                         break;
1183                                                 case 'f':
1184                                                         c = '\f';
1185                                                         break;
1186                                                 case '\"':
1187                                                         c = '\"';
1188                                                         break;
1189                                                 case '\'':
1190                                                         c = '\'';
1191                                                         break;
1192                                                 case '\\':
1193                                                         c = '\\';
1194                                                         break;
1195                                                 default:
1196                                                         warn("bad backslash code in string");
1197                                                         ln--;
1198                                                         break;
1199                                                 }
1200                                         }
1201
1202                                         *p++ = c;
1203                                 }
1204
1205                                 if (*ln++ != c1)
1206                                         return error("unterminated string");
1207
1208                                 *p++ = EOS;
1209                                 continue;
1210                         case '$':               // $, hex constant
1211                                 if ((int)chrtab[*ln] & HDIGIT)
1212                                 {
1213                                         v = 0;
1214
1215                                         // Parse the hex value
1216                                         while ((int)hextab[*ln] >= 0)
1217                                                 v = (v << 4) + (int)hextab[*ln++];
1218
1219                                         // ggn: Okay, some comments here are in order I think....
1220                                         // The original madmac sources didn't parse the size at
1221                                         // this point (i.e. .b/.w/.l). It was probably done at
1222                                         // another point, although it's unclear to me exactly
1223                                         // where. So why change this? My understanding (at least
1224                                         // from what SCPCD said on IRC) is that .w addressing
1225                                         // formats produce wrong code on jaguar (or doesn't execute
1226                                         // properly? something like that). So the code was changed
1227                                         // to mask off the upper bits depending on length (note: I
1228                                         // don't think .b is valid at all! I only know of .w/.l, so
1229                                         // this should probably be wiped). Then the code that
1230                                         // parses the constant and checks to see if it's between
1231                                         // $ffff0000 and $8000 never got triggered, so yay job
1232                                         // done! ...now say we want to assemble a st .prg. One of
1233                                         // the most widely spread optimisations is move.X expr.w,Y
1234                                         // (or vice versa, or both, anyway...) to access hardware
1235                                         // registers (which are mapped to $fxxxxx). This botchy
1236                                         // thing would create "hilarious" code while trying to
1237                                         // access hardware registers. So I made a condition to see
1238                                         // if st mode or jaguar is active and apply the both or
1239                                         // not. One last note: this is hardcoded to get optimised
1240                                         // for now on ST mode, i.e. it can't generate code like
1241                                         // move.w $00001234,d0 - it'll always get optimised to
1242                                         // move.w $1234.w,d0. It's probably ok, but maybe a warning
1243                                         // should be emitted? Or maybe finding a way to make it not
1244                                         // auto-optimise? I think it's ok for now...
1245                                         if (*ln == '.')
1246                                         {
1247                                                 if (obj_format == ALCYON)
1248                                                 {
1249                                                         if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B') || (*(ln + 1) == 'w') || (*(ln + 1) == 'W') || (*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1250                                                         {
1251                                                                 ln += 2;
1252                                                         }
1253                                                 }
1254                                                 else
1255                                                 {
1256                                                         if ((*(ln + 1) & 0xDF) == 'B')
1257                                                         {
1258                                                                 v &= 0x000000FF;
1259                                                                 ln += 2;
1260                                                         }
1261                                                         else if ((*(ln + 1) & 0xDF) == 'W')
1262                                                         {
1263                                                                 v &= 0x0000FFFF;
1264                                                                 ln += 2;
1265                                                         }
1266                                                         else if ((*(ln + 1) & 0xDF) == 'L')
1267                                                         {
1268                                                                 ln += 2;
1269                                                         }
1270                                                 }
1271                                         }
1272
1273                                         *tk++ = CONST;
1274                                         *tk++ = v;
1275                                 }
1276                                 else
1277                                         *tk++ = '$';
1278
1279                                 continue;
1280                         case '<':               // < or << or <> or <=
1281                                 switch (*ln)
1282                                 {
1283                                 case '<':
1284                                         *tk++ = SHL;
1285                                         ++ln;
1286                                         continue;
1287                                 case '>':
1288                                         *tk++ = NE;
1289                                         ++ln;
1290                                         continue;
1291                                 case '=':
1292                                         *tk++ = LE;
1293                                         ++ln;
1294                                         continue;
1295                                 default:
1296                                         *tk++ = '<';
1297                                         continue;
1298                                 }
1299                         case ':':               // : or ::
1300                                 if (*ln == ':')
1301                                 {
1302                                         *tk++ = DCOLON;
1303                                         ++ln;
1304                                 }
1305                                 else
1306                                         *tk++ = ':';
1307
1308                                 continue;
1309                         case '=':               // = or ==
1310                                 if (*ln == '=')
1311                                 {
1312                                         *tk++ = DEQUALS;
1313                                         ++ln;
1314                                 }
1315                                 else
1316                                         *tk++ = '=';
1317
1318                                 continue;
1319                         case '>':               // > or >> or >=
1320                                 switch (*ln)
1321                                 {
1322                                 case '>':
1323                                         *tk++ = SHR;
1324                                         ln++;
1325                                         continue;
1326                                 case '=':
1327                                         *tk++ = GE;
1328                                         ln++;
1329                                         continue;
1330                                 default:
1331                                         *tk++ = '>';
1332                                         continue;
1333                                 }
1334                         case '%':               // % or binary constant
1335                                 if (*ln < '0' || *ln > '1')
1336                                 {
1337                                         *tk++ = '%';
1338                                         continue;
1339                                 }
1340
1341                                 v = 0;
1342
1343                                 while (*ln >= '0' && *ln <= '1')
1344                                         v = (v << 1) + *ln++ - '0';
1345
1346                                 if (*ln == '.')
1347                                 {
1348                                         if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1349                                         {
1350                                                 v &= 0x000000FF;
1351                                                 ln += 2;
1352                                         }
1353
1354                                         if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1355                                         {
1356                                                 v &= 0x0000FFFF;
1357                                                 ln += 2;
1358                                         }
1359
1360                                         if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1361                                         {
1362                                                 ln += 2;
1363                                         }
1364                                 }
1365
1366                                 *tk++ = CONST;
1367                                 *tk++ = v;
1368                                 continue;
1369                         case '@':               // @ or octal constant
1370                                 if (*ln < '0' || *ln > '7')
1371                                 {
1372                                         *tk++ = '@';
1373                                         continue;
1374                                 }
1375
1376                                 v = 0;
1377
1378                                 while (*ln >= '0' && *ln <= '7')
1379                                         v = (v << 3) + *ln++ - '0';
1380
1381                                 if (*ln == '.')
1382                                 {
1383                                         if ((*(ln+1) == 'b') || (*(ln+1) == 'B'))
1384                                         {
1385                                                 v &= 0x000000FF;
1386                                                 ln += 2;
1387                                         }
1388
1389                                         if ((*(ln+1) == 'w') || (*(ln+1) == 'W'))
1390                                         {
1391                                                 v &= 0x0000FFFF;
1392                                                 ln += 2;
1393                                         }
1394
1395                                         if ((*(ln+1) == 'l') || (*(ln+1) == 'L'))
1396                                         {
1397                                                 ln += 2;
1398                                         }
1399                                 }
1400
1401                                 *tk++ = CONST;
1402                                 *tk++ = v;
1403                                 continue;
1404                         case '^':               // ^ or ^^ <operator-name>
1405                                 if (*ln != '^')
1406                                 {
1407                                         *tk++ = '^';
1408                                         continue;
1409                                 }
1410
1411                                 if (((int)chrtab[*++ln] & STSYM) == 0)
1412                                 {
1413                                         error("invalid symbol following ^^");
1414                                         continue;
1415                                 }
1416
1417                                 p = ln++;
1418
1419                                 while ((int)chrtab[*ln] & CTSYM)
1420                                         ++ln;
1421
1422                                 for(state=0; state>=0;)
1423                                 {
1424                                         // Get char, convert to lowercase
1425                                         j = *p++;
1426
1427                                         if (j >= 'A' && j <= 'Z')
1428                                                 j += 0x20;
1429
1430                                         j += kwbase[state];
1431
1432                                         if (kwcheck[j] != state)
1433                                         {
1434                                                 j = -1;
1435                                                 break;
1436                                         }
1437
1438                                         if (*p == EOS || p == ln)
1439                                         {
1440                                                 j = kwaccept[j];
1441                                                 break;
1442                                         }
1443
1444                                         state = kwtab[j];
1445                                 }
1446
1447                                 if (j < 0 || state < 0)
1448                                 {
1449                                         error("unknown symbol following ^^");
1450                                         continue;
1451                                 }
1452
1453                                 *tk++ = (TOKEN)j;
1454                                 continue;
1455                         default:
1456                                 interror(2);    // Bad MULTX entry in chrtab
1457                                 continue;
1458                         }
1459                 }
1460
1461                 // Handle decimal constant
1462                 if (c & DIGIT)
1463                 {
1464                         v = 0;
1465
1466                         while ((int)chrtab[*ln] & DIGIT)
1467                                 v = (v * 10) + *ln++ - '0';
1468
1469                         // See if there's a .[bwl] after the constant & deal with it if so
1470                         if (*ln == '.')
1471                         {
1472                                 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1473                                 {
1474                                         v &= 0x000000FF;
1475                                         ln += 2;
1476                                 }
1477                                 else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1478                                 {
1479                                         v &= 0x0000FFFF;
1480                                         ln += 2;
1481                                 }
1482                                 else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1483                                 {
1484                                         ln += 2;
1485                                 }
1486                         }
1487
1488                         *tk++ = CONST;
1489                         *tk++ = v;
1490 //printf("CONST: %i\n", v);
1491                         continue;
1492                 }
1493
1494                 // Handle illegal character
1495                 return error("illegal character");
1496         }
1497
1498         // Terminate line of tokens and return "success."
1499
1500 goteol:
1501         tok = etok;                                                             // Set tok to beginning of line
1502
1503         if (stuffnull)                                                  // Terminate last SYMBOL
1504                 *nullspot = EOS;
1505
1506         *tk++ = EOL;
1507
1508         return OK;
1509 }
1510
1511
1512 //
1513 // .GOTO <label>        goto directive
1514 //
1515 // The label is searched for starting from the first line of the current,
1516 // enclosing macro definition. If no enclosing macro exists, an error is
1517 // generated.
1518 //
1519 // A label is of the form:
1520 //
1521 // :<name><whitespace>
1522 //
1523 // The colon must appear in column 1.  The label is stripped prior to macro
1524 // expansion, and is NOT subject to macro expansion.  The whitespace may also
1525 // be EOL.
1526 //
1527 //int d_goto(WORD siz) {
1528 //int d_goto(void)
1529 int d_goto(WORD unused)
1530 {
1531         char * s1, * s2;
1532
1533         // Setup for the search
1534         if (*tok != SYMBOL)
1535                 return error("missing label");
1536
1537 //      sym = (char *)tok[1];
1538         char * sym = string[tok[1]];
1539         tok += 2;
1540
1541         if (cur_inobj->in_type != SRC_IMACRO)
1542                 return error("goto not in macro");
1543
1544         IMACRO * imacro = cur_inobj->inobj.imacro;
1545 //      defln = (LONG *)imacro->im_macro->svalue;
1546         struct LineList * defln = imacro->im_macro->lineList;
1547
1548         // Find the label, starting with the first line.
1549 //      for(; defln!=NULL; defln=(LONG *)*defln)
1550         for(; defln!=NULL; defln=defln->next)
1551         {
1552 //              if (*(char *)(defln + 1) == ':')
1553                 if (defln->line[0] == ':')
1554                 {
1555                         // Compare names (sleazo string compare)
1556                         // This string compare is not right. Doesn't check for lengths.
1557                         // (actually it does, but in a crappy, unclear way.)
1558 WARNING(!!!! Bad string comparison !!!)
1559                         s1 = sym;
1560 //                      s2 = (char *)(defln + 1) + 1;
1561                         s2 = defln->line;
1562
1563                         while (*s1 == *s2)
1564                         {
1565                                 if (*s1 == EOS)
1566                                         break;
1567                                 else
1568                                 {
1569                                         s1++;
1570                                         s2++;
1571                                 }
1572                         }
1573
1574                         // Found the label, set new macro next-line and return.
1575                         if ((*s2 == EOS) || ((int)chrtab[*s2] & WHITE))
1576                         {
1577                                 imacro->im_nextln = defln;
1578                                 return 0;
1579                         }
1580                 }
1581         }
1582
1583         return error("goto label not found");
1584 }
1585
1586
1587 void DumpTokenBuffer(void)
1588 {
1589         TOKEN * t;
1590         printf("Tokens [%X]: ", sloc);
1591
1592         for(t=tokbuf; *t!=EOL; t++)
1593         {
1594                 if (*t == COLON)
1595                         printf("[COLON]");
1596                 else if (*t == CONST)
1597                 {
1598                         t++;
1599                         printf("[CONST: $%X]", (uint32_t)*t);
1600                 }
1601                 else if (*t == ACONST)
1602                         printf("[ACONST]");
1603                 else if (*t == STRING)
1604                 {
1605                         t++;
1606                         printf("[STRING:\"%s\"]", string[*t]);
1607                 }
1608                 else if (*t == SYMBOL)
1609                 {
1610                         t++;
1611                         printf("[SYMBOL:\"%s\"]", string[*t]);
1612                 }
1613                 else if (*t == EOS)
1614                         printf("[EOS]");
1615                 else if (*t == TKEOF)
1616                         printf("[TKEOF]");
1617                 else if (*t == DEQUALS)
1618                         printf("[DEQUALS]");
1619                 else if (*t == SET)
1620                         printf("[SET]");
1621                 else if (*t == REG)
1622                         printf("[REG]");
1623                 else if (*t == DCOLON)
1624                         printf("[DCOLON]");
1625                 else if (*t == GE)
1626                         printf("[GE]");
1627                 else if (*t == LE)
1628                         printf("[LE]");
1629                 else if (*t == NE)
1630                         printf("[NE]");
1631                 else if (*t == SHR)
1632                         printf("[SHR]");
1633                 else if (*t == SHL)
1634                         printf("[SHL]");
1635                 else if (*t == UNMINUS)
1636                         printf("[UNMINUS]");
1637                 else if (*t == DOTB)
1638                         printf("[DOTB]");
1639                 else if (*t == DOTW)
1640                         printf("[DOTW]");
1641                 else if (*t == DOTL)
1642                         printf("[DOTL]");
1643                 else if (*t == DOTI)
1644                         printf("[DOTI]");
1645                 else if (*t == ENDEXPR)
1646                         printf("[ENDEXPR]");
1647                 else if (*t == CR_ABSCOUNT)
1648                         printf("[CR_ABSCOUNT]");
1649                 else if (*t == CR_DEFINED)
1650                         printf("[CR_DEFINED]");
1651                 else if (*t == CR_REFERENCED)
1652                         printf("[CR_REFERENCED]");
1653                 else if (*t == CR_STREQ)
1654                         printf("[CR_STREQ]");
1655                 else if (*t == CR_MACDEF)
1656                         printf("[CR_MACDEF]");
1657                 else if (*t == CR_TIME)
1658                         printf("[CR_TIME]");
1659                 else if (*t == CR_DATE)
1660                         printf("[CR_DATE]");
1661                 else if (*t >= 0x20 && *t <= 0x2F)
1662                         printf("[%c]", (char)*t);
1663                 else if (*t >= 0x3A && *t <= 0x3F)
1664                         printf("[%c]", (char)*t);
1665                 else if (*t >= 0x80 && *t <= 0x87)
1666                         printf("[D%u]", ((uint32_t)*t) - 0x80);
1667                 else if (*t >= 0x88 && *t <= 0x8F)
1668                         printf("[A%u]", ((uint32_t)*t) - 0x88);
1669                 else
1670                         printf("[%X:%c]", (uint32_t)*t, (char)*t);
1671         }
1672
1673         printf("[EOL]\n");
1674 }
1675