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_DATE:
 558                                                                 d = "^^date";
 559                                                                 break;
 560                                                         case CR_TIME:
 561                                                                 d = "^^time";
 562                                                                 break;
 563                                                         case CR_DEFINED:
 564                                                                 d = "^^defined ";
 565                                                                 break;
 566                                                         case CR_REFERENCED:
 567                                                                 d = "^^referenced ";
 568                                                                 break;
 569                                                         case CR_STREQ:
 570                                                                 d = "^^streq ";
 571                                                                 break;
 572                                                         case CR_MACDEF:
 573                                                                 d = "^^macdef ";
 574                                                                 break;
 575                                                         default:
 576                                                                 if (dst >= edst)
 577                                                                         goto overflow;
 578
 579                                                                 *dst++ = (char)*(tk - 1);
 580                                                                 break;
 581                                                         }
 582                                                 }
 583
 584                                                 // If 'd' != NULL, copy string to destination
 585                                                 if (d != NULL)
 586                                                 {
 587 strcopy:
 588                                                         DEBUG printf("d='%s'\n", d);
 589
 590                                                         while (*d != EOS)
 591                                                         {
 592                                                                 if (dst >= edst)
 593                                                                         goto overflow;
 594                                                                 else
 595                                                                         *dst++ = *d++;
 596                                                         }
 597                                                 }
 598                                         }
 599                                 }
 600                         }
 601                 }
 602         }
 603
 604         *dst = EOS;
 605         DEBUG { printf("ExM: dst=\"%s\"\n", dest); }
 606         return OK;
 607
 608 overflow:
 609         *dst = EOS;
 610         DEBUG printf("*** OVERFLOW LINE ***\n%s\n", dest);
 611         return fatal("line too long as a result of macro expansion");
 612 }
 613
 614
 615 //
 616 // Get next line of text from a macro
 617 //
 618 char * GetNextMacroLine(void)
 619 {
 620 //      unsigned source_addr;
 621
 622         IMACRO * imacro = cur_inobj->inobj.imacro;
 623 //      LONG * strp = imacro->im_nextln;
 624         struct LineList * strp = imacro->im_nextln;
 625
 626         if (strp == NULL)                                               // End-of-macro
 627                 return NULL;
 628
 629 //      imacro->im_nextln = (LONG *)*strp;
 630         imacro->im_nextln = strp->next;
 631 //      ExpandMacro((char *)(strp + 1), imacro->im_lnbuf, LNSIZ);
 632         ExpandMacro(strp->line, imacro->im_lnbuf, LNSIZ);
 633
 634         return imacro->im_lnbuf;
 635 }
 636
 637
 638 //
 639 // Get next line of text from a repeat block
 640 //
 641 char * GetNextRepeatLine(void)
 642 {
 643
 644         IREPT * irept = cur_inobj->inobj.irept;
 645         LONG * strp = irept->ir_nextln;                 // initial null
 646
 647         // Do repeat at end of .rept block's string list
 648         if (strp == NULL)
 649         {
 650                 DEBUG printf("back-to-top-of-repeat-block count=%d\n", (int)irept->ir_count);
 651                 irept->ir_nextln = irept->ir_firstln;   // copy first line
 652
 653                 if (irept->ir_count-- == 0)
 654                 {
 655                         DEBUG printf("end-repeat-block\n");
 656                         return NULL;
 657                 }
 658
 659                 strp = irept->ir_nextln;
 660         }
 661
 662         strcpy(irbuf, (char *)(irept->ir_nextln + 1));
 663         DEBUG printf("repeat line='%s'\n", irbuf);
 664         irept->ir_nextln = (LONG *)*strp;
 665
 666         return irbuf;
 667 }
 668
 669
 670 //
 671 // Include a source file used at the root, and for ".include" files
 672 //
 673 int include(int handle, char * fname)
 674 {
 675         IFILE * ifile;
 676         INOBJ * inobj;
 677         FILEREC * fr;
 678
 679         // Verbose mode
 680         if (verb_flag)
 681                 printf("[include: %s, cfileno=%u]\n", fname, cfileno);
 682
 683         // Alloc and initialize include-descriptors
 684         inobj = a_inobj(SRC_IFILE);
 685         ifile = inobj->inobj.ifile;
 686
 687         ifile->ifhandle = handle;                               // Setup file handle
 688         ifile->ifind = ifile->ifcnt = 0;                // Setup buffer indices
 689         ifile->ifoldlineno = curlineno;                 // Save old line number
 690         ifile->ifoldfname = curfname;                   // Save old filename
 691         ifile->ifno = cfileno;                                  // Save old file number
 692
 693 //      cfileno = filecount++;                                  // Compute new file number
 694         // NB: This *must* be preincrement, we're adding one to the filecount here!
 695         cfileno = ++filecount;                                  // Compute NEW file number
 696         curfname = strdup(fname);                               // Set current filename (alloc storage)
 697         curlineno = 0;                                                  // Start on line zero
 698
 699         // Add another file to the file-record
 700         fr = (FILEREC *)malloc(sizeof(FILEREC));
 701         fr->frec_next = NULL;
 702         fr->frec_name = curfname;
 703
 704         if (last_fr == NULL)
 705                 filerec = fr;                                           // Add first filerec
 706         else
 707                 last_fr->frec_next = fr;                        // Append to list of filerecs
 708
 709         last_fr = fr;
 710         DEBUG printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno);
 711
 712         return OK;
 713 }
 714
 715
 716 //
 717 // Pop the current input level
 718 //
 719 int fpop(void)
 720 {
 721         IFILE * ifile;
 722         IMACRO * imacro;
 723         LONG * p, * p1;
 724         INOBJ * inobj = cur_inobj;
 725
 726         if (inobj != NULL)
 727         {
 728                 // Pop IFENT levels until we reach the conditional assembly context we
 729                 // were at when the input object was entered.
 730                 while (ifent != inobj->in_ifent)
 731                         d_endif();
 732
 733                 tok = inobj->in_otok;                           // Restore tok and otok
 734                 etok = inobj->in_etok;
 735
 736                 switch (inobj->in_type)
 737                 {
 738                 case SRC_IFILE:                                         // Pop and release an IFILE
 739                         if (verb_flag)
 740                                 printf("[Leaving: %s]\n", curfname);
 741
 742                         ifile = inobj->inobj.ifile;
 743                         ifile->if_link = f_ifile;
 744                         f_ifile = ifile;
 745                         close(ifile->ifhandle);                 // Close source file
 746 if (verb_flag) printf("[fpop (pre):  curfname=%s]\n", curfname);
 747                         curfname = ifile->ifoldfname;   // Set current filename
 748 if (verb_flag) printf("[fpop (post): curfname=%s]\n", curfname);
 749 if (verb_flag) printf("[fpop: (pre)  cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
 750                         curlineno = ifile->ifoldlineno; // Set current line#
 751                         DEBUG printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno);
 752                         cfileno = ifile->ifno;                  // Restore current file number
 753 if (verb_flag) printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
 754                         break;
 755                 case SRC_IMACRO:                                        // Pop and release an IMACRO
 756                         imacro = inobj->inobj.imacro;
 757                         imacro->im_link = f_imacro;
 758                         f_imacro = imacro;
 759                         break;
 760                 case SRC_IREPT:                                         // Pop and release an IREPT
 761                         DEBUG printf("dealloc IREPT\n");
 762                         p = inobj->inobj.irept->ir_firstln;
 763
 764                         while (p != NULL)
 765                         {
 766                                 p1 = (LONG *)*p;
 767                                 p = p1;
 768                         }
 769
 770                         break;
 771                 }
 772
 773                 cur_inobj = inobj->in_link;
 774                 inobj->in_link = f_inobj;
 775                 f_inobj = inobj;
 776         }
 777
 778         return 0;
 779 }
 780
 781
 782 //
 783 // Get line from file into buf, return NULL on EOF or ptr to the start of a
 784 // null-term line
 785 //
 786 char * GetNextLine(void)
 787 {
 788         int i, j;
 789         char * p, * d;
 790         int readamt = -1;                                               // 0 if last read() yeilded 0 bytes
 791         IFILE * fl = cur_inobj->inobj.ifile;
 792
 793         for(;;)
 794         {
 795                 // Scan for next end-of-line; handle stupid text formats by treating
 796                 // \r\n the same as \n. (lone '\r' at end of buffer means we have to
 797                 // check for '\n').
 798                 d = &fl->ifbuf[fl->ifind];
 799
 800                 for(p=d, i=0, j=fl->ifcnt; i<j; i++, p++)
 801                 {
 802                         if (*p == '\r' || *p == '\n')
 803                         {
 804                                 i++;
 805
 806                                 if (*p == '\r')
 807                                 {
 808                                         if (i >= j)
 809                                                 break;  // Need to read more, then look for '\n' to eat
 810                                         else if (p[1] == '\n')
 811                                                 i++;
 812                                 }
 813
 814                                 // Cover up the newline with end-of-string sentinel
 815                                 *p = '\0';
 816
 817                                 fl->ifind += i;
 818                                 fl->ifcnt -= i;
 819                                 return d;
 820                         }
 821                 }
 822
 823                 // Handle hanging lines by ignoring them (Input file is exhausted, no
 824                 // \r or \n on last line)
 825                 // Shamus: This is retarded. Never ignore any input!
 826                 if (!readamt && fl->ifcnt)
 827                 {
 828 #if 0
 829                         fl->ifcnt = 0;
 830                         *p = '\0';
 831                         return NULL;
 832 #else
 833                         // Really should check to see if we're at the end of the buffer!
 834                         // :-P
 835                         fl->ifbuf[fl->ifind + fl->ifcnt] = '\0';
 836                         fl->ifcnt = 0;
 837                         return &fl->ifbuf[fl->ifind];
 838 #endif
 839                 }
 840
 841                 // Truncate and return absurdly long lines.
 842                 if (fl->ifcnt >= QUANTUM)
 843                 {
 844                         fl->ifbuf[fl->ifind + fl->ifcnt - 1] = '\0';
 845                         fl->ifcnt = 0;
 846                         return &fl->ifbuf[fl->ifind];
 847                 }
 848
 849                 // Relocate what's left of a line to the beginning of the buffer, and
 850                 // read some more of the file in; return NULL if the buffer's empty and
 851                 // on EOF.
 852                 if (fl->ifind != 0)
 853                 {
 854                         p = &fl->ifbuf[fl->ifind];
 855                         d = &fl->ifbuf[fl->ifcnt & 1];
 856
 857                         for(i=0; i<fl->ifcnt; i++)
 858                                 *d++ = *p++;
 859
 860                         fl->ifind = fl->ifcnt & 1;
 861                 }
 862
 863                 readamt = read(fl->ifhandle, &fl->ifbuf[fl->ifind + fl->ifcnt], QUANTUM);
 864
 865                 if (readamt < 0)
 866                         return NULL;
 867
 868                 if ((fl->ifcnt += readamt) == 0)
 869                         return NULL;
 870         }
 871 }
 872
 873
 874 //
 875 // Tokenize a line
 876 //
 877 int TokenizeLine(void)
 878 {
 879         char * ln = NULL;                       // Ptr to current position in line
 880         char * p;                                       // Random character ptr
 881         TOKEN * tk;                                     // Token-deposit ptr
 882         int state = 0;                          // State for keyword detector
 883         int j = 0;                                      // Var for keyword detector
 884         char c;                                         // Random char
 885         VALUE v;                                        // Random value
 886         char * nullspot = NULL;         // Spot to clobber for SYMBOL termination
 887         int stuffnull;                          // 1:terminate SYMBOL '\0' at *nullspot
 888         char c1;
 889         int stringNum = 0;                      // Pointer to string locations in tokenized line
 890
 891 retry:
 892
 893         if (cur_inobj == NULL)                                  // Return EOF if input stack is empty
 894                 return TKEOF;
 895
 896         // Get another line of input from the current input source: a file, a
 897         // macro, or a repeat-block
 898         switch (cur_inobj->in_type)
 899         {
 900         // Include-file:
 901         // o  handle EOF;
 902         // o  bump source line number;
 903         // o  tag the listing-line with a space;
 904         // o  kludge lines generated by Alcyon C.
 905         case SRC_IFILE:
 906                 if ((ln = GetNextLine()) == NULL)
 907                 {
 908 if (verb_flag) printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n");
 909                         fpop();                                                 // Pop input level
 910                         goto retry;                                             // Try for more lines
 911                 }
 912
 913                 curlineno++;                                            // Bump line number
 914                 lntag = SPACE;
 915
 916                 if (as68_flag)
 917                 {
 918                         // AS68 compatibility, throw away all lines starting with
 919                         // back-quotes, tildes, or '*'
 920                         // On other lines, turn the first '*' into a semi-colon.
 921                         if (*ln == '`' || *ln == '~' || *ln == '*')
 922                                 *ln = ';';
 923                         else
 924                         {
 925                                 for(p=ln; *p!=EOS; p++)
 926                                 {
 927                                         if (*p == '*')
 928                                         {
 929                                                 *p = ';';
 930                                                 break;
 931                                         }
 932                                 }
 933                         }
 934                 }
 935
 936                 break;
 937         // Macro-block:
 938         // o  Handle end-of-macro;
 939         // o  tag the listing-line with an at (@) sign.
 940         case SRC_IMACRO:
 941                 if ((ln = GetNextMacroLine()) == NULL)
 942                 {
 943                         ExitMacro();                                    // Exit macro (pop args, do fpop(), etc)
 944                         goto retry;                                             // Try for more lines...
 945                 }
 946
 947                 lntag = '@';
 948                 break;
 949         // Repeat-block:
 950         // o  Handle end-of-repeat-block;
 951         // o  tag the listing-line with a pound (#) sign.
 952         case SRC_IREPT:
 953                 if ((ln = GetNextRepeatLine()) == NULL)
 954                 {
 955 if (verb_flag) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n");
 956                         fpop();
 957                         goto retry;
 958                 }
 959
 960                 lntag = '#';
 961                 break;
 962         }
 963
 964         // Save text of the line.  We only do this during listings and within
 965         // macro-type blocks, since it is expensive to unconditionally copy every
 966         // line.
 967         if (lnsave)
 968                 strcpy(lnbuf, ln);
 969
 970         // General house-keeping
 971         tok = tokeol;                   // Set "tok" to EOL in case of error
 972         tk = etok;                              // Reset token ptr
 973         stuffnull = 0;                  // Don't stuff nulls
 974         totlines++;                             // Bump total #lines assembled
 975
 976         // See if the entire line is a comment. This is a win if the programmer
 977         // puts in lots of comments
 978         if (*ln == '*' || *ln == ';' || ((*ln == '/') && (*(ln + 1) == '/')))
 979                 goto goteol;
 980
 981         // Main tokenization loop;
 982         // o  skip whitespace;
 983         // o  handle end-of-line;
 984         // o  handle symbols;
 985         // o  handle single-character tokens (operators, etc.);
 986         // o  handle multiple-character tokens (constants, strings, etc.).
 987         for(; *ln!=EOS;)
 988         {
 989                 // Skip whitespace, handle EOL
 990                 while ((int)chrtab[*ln] & WHITE)
 991                         ln++;
 992
 993                 // Handle EOL, comment with ';'
 994                 if (*ln == EOS || *ln == ';'|| ((*ln == '/') && (*(ln + 1) == '/')))
 995                         break;
 996
 997                 // Handle start of symbol. Symbols are null-terminated in place. The
 998                 // termination is always one symbol behind, since there may be no place
 999                 // for a null in the case that an operator immediately follows the name.
1000                 c = chrtab[*ln];
1001
1002                 if (c & STSYM)
1003                 {
1004                         if (stuffnull)                  // Terminate old symbol from previous pass
1005                                 *nullspot = EOS;
1006
1007                         v = 0;                                  // Assume no DOT attrib follows symbol
1008                         stuffnull = 1;
1009                         p = nullspot = ln++;    // Nullspot -> start of this symbol
1010
1011                         // Find end of symbol (and compute its length)
1012                         for(j=1; (int)chrtab[*ln]&CTSYM; j++)
1013                                 ln++;
1014
1015                         // Handle "DOT" special forms (like ".b") that follow a normal
1016                         // symbol or keyword:
1017                         if (*ln == '.')
1018                         {
1019                                 *ln++ = EOS;            // Terminate symbol
1020                                 stuffnull = 0;          // And never try it again
1021
1022                                 // Character following the `.' must have a DOT attribute, and
1023                                 // the chararacter after THAT one must not have a start-symbol
1024                                 // attribute (to prevent symbols that look like, for example,
1025                                 // "zingo.barf", which might be a good idea anyway....)
1026                                 if ((((int)chrtab[*ln] & DOT) == 0) || ((int)dotxtab[*ln] <= 0))
1027                                         return error("[bwsl] must follow `.' in symbol");
1028
1029                                 v = (VALUE)dotxtab[*ln++];
1030
1031                                 if ((int)chrtab[*ln] & CTSYM)
1032                                         return error("misuse of `.', not allowed in symbols");
1033                         }
1034
1035                         // If the symbol is small, check to see if it's really the name of
1036                         // a register.
1037                         if (j <= KWSIZE)
1038                         {
1039                                 for(state=0; state>=0;)
1040                                 {
1041                                         j = (int)tolowertab[*p++];
1042                                         j += kwbase[state];
1043
1044                                         if (kwcheck[j] != state)
1045                                         {
1046                                                 j = -1;
1047                                                 break;
1048                                         }
1049
1050                                         if (*p == EOS || p == ln)
1051                                         {
1052                                                 j = kwaccept[j];
1053                                                 break;
1054                                         }
1055
1056                                         state = kwtab[j];
1057                                 }
1058                         }
1059                         else
1060                         {
1061                                 j = -1;
1062                         }
1063
1064                         // Make j = -1 if user tries to use a RISC register while in 68K mode
1065                         if (!(rgpu || rdsp) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31))
1066                         {
1067                                 j = -1;
1068                         }
1069
1070                         // Make j = -1 if time, date etc with no preceeding ^^
1071                         // defined, referenced, streq, macdef, date and time
1072                         switch ((TOKEN)j)
1073                         {
1074                         case 112:   // defined
1075                         case 113:   // referenced
1076                         case 118:   // streq
1077                         case 119:   // macdef
1078                         case 120:   // time
1079                         case 121:   // date
1080                                 j = -1;
1081                         }
1082
1083                         // If not tokenized keyword OR token was not found
1084                         if (j < 0 || state < 0)
1085                         {
1086                                 *tk++ = SYMBOL;
1087 //#warning
1088 //problem here: nullspot is a char * but TOKEN is a uint32_t. On a 64-bit
1089 //system, this will cause all kinds of mischief.
1090 #if 0
1091                                 *tk++ = (TOKEN)nullspot;
1092 #else
1093                                 string[stringNum] = nullspot;
1094                                 *tk++ = stringNum;
1095                                 stringNum++;
1096 #endif
1097                         }
1098                         else
1099                         {
1100                                 *tk++ = (TOKEN)j;
1101                                 stuffnull = 0;
1102                         }
1103
1104                         if (v)                                                  // Record attribute token (if any)
1105                                 *tk++ = (TOKEN)v;
1106
1107                         if (stuffnull)                                  // Arrange for string termination on next pass
1108                                 nullspot = ln;
1109
1110                         continue;
1111                 }
1112
1113                 // Handle identity tokens
1114                 if (c & SELF)
1115                 {
1116                         *tk++ = *ln++;
1117                         continue;
1118                 }
1119
1120                 // Handle multiple-character tokens
1121                 if (c & MULTX)
1122                 {
1123                         switch (*ln++)
1124                         {
1125                         case '!':               // ! or !=
1126                                 if (*ln == '=')
1127                                 {
1128                                         *tk++ = NE;
1129                                         ++ln;
1130                                 }
1131                                 else
1132                                         *tk++ = '!';
1133
1134                                 continue;
1135                         case '\'':              // 'string'
1136                         case '\"':              // "string"
1137                                 c1 = ln[-1];
1138                                 *tk++ = STRING;
1139 //#warning
1140 // More char * stuffing (8 bytes) into the space of 4 (TOKEN).
1141 // Need to figure out how to fix this crap.
1142 #if 0
1143                                 *tk++ = (TOKEN)ln;
1144 #else
1145                                 string[stringNum] = ln;
1146                                 *tk++ = stringNum;
1147                                 stringNum++;
1148 #endif
1149
1150                                 for(p=ln; *ln!=EOS && *ln!=c1;)
1151                                 {
1152                                         c = *ln++;
1153
1154                                         if (c == '\\')
1155                                         {
1156                                                 switch (*ln++)
1157                                                 {
1158                                                 case EOS:
1159                                                         return(error("unterminated string"));
1160                                                 case 'e':
1161                                                         c = '\033';
1162                                                         break;
1163                                                 case 'n':
1164                                                         c = '\n';
1165                                                         break;
1166                                                 case 'b':
1167                                                         c = '\b';
1168                                                         break;
1169                                                 case 't':
1170                                                         c = '\t';
1171                                                         break;
1172                                                 case 'r':
1173                                                         c = '\r';
1174                                                         break;
1175                                                 case 'f':
1176                                                         c = '\f';
1177                                                         break;
1178                                                 case '\"':
1179                                                         c = '\"';
1180                                                         break;
1181                                                 case '\'':
1182                                                         c = '\'';
1183                                                         break;
1184                                                 case '\\':
1185                                                         c = '\\';
1186                                                         break;
1187                                                 default:
1188                                                         warn("bad backslash code in string");
1189                                                         ln--;
1190                                                         break;
1191                                                 }
1192                                         }
1193
1194                                         *p++ = c;
1195                                 }
1196
1197                                 if (*ln++ != c1)
1198                                         return error("unterminated string");
1199
1200                                 *p++ = EOS;
1201                                 continue;
1202                         case '$':               // $, hex constant
1203                                 if ((int)chrtab[*ln] & HDIGIT)
1204                                 {
1205                                         v = 0;
1206
1207                                         // Parse the hex value
1208                                         while ((int)hextab[*ln] >= 0)
1209                                                 v = (v << 4) + (int)hextab[*ln++];
1210
1211                                         // ggn: Okay, some comments here are in order I think....
1212                                         // The original madmac sources didn't parse the size at
1213                                         // this point (i.e. .b/.w/.l). It was probably done at
1214                                         // another point, although it's unclear to me exactly
1215                                         // where. So why change this? My understanding (at least
1216                                         // from what SCPCD said on IRC) is that .w addressing
1217                                         // formats produce wrong code on jaguar (or doesn't execute
1218                                         // properly? something like that). So the code was changed
1219                                         // to mask off the upper bits depending on length (note: I
1220                                         // don't think .b is valid at all! I only know of .w/.l, so
1221                                         // this should probably be wiped). Then the code that
1222                                         // parses the constant and checks to see if it's between
1223                                         // $ffff0000 and $8000 never got triggered, so yay job
1224                                         // done! ...now say we want to assemble a st .prg. One of
1225                                         // the most widely spread optimisations is move.X expr.w,Y
1226                                         // (or vice versa, or both, anyway...) to access hardware
1227                                         // registers (which are mapped to $fxxxxx). This botchy
1228                                         // thing would create "hilarious" code while trying to
1229                                         // access hardware registers. So I made a condition to see
1230                                         // if st mode or jaguar is active and apply the both or
1231                                         // not. One last note: this is hardcoded to get optimised
1232                                         // for now on ST mode, i.e. it can't generate code like
1233                                         // move.w $00001234,d0 - it'll always get optimised to
1234                                         // move.w $1234.w,d0. It's probably ok, but maybe a warning
1235                                         // should be emitted? Or maybe finding a way to make it not
1236                                         // auto-optimise? I think it's ok for now...
1237                                         if (*ln == '.')
1238                                         {
1239                                                 if (obj_format == ALCYON)
1240                                                 {
1241                                                         if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B') || (*(ln + 1) == 'w') || (*(ln + 1) == 'W') || (*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1242                                                         {
1243                                                                 ln += 2;
1244                                                         }
1245                                                 }
1246                                                 else
1247                                                 {
1248                                                         if ((*(ln + 1) & 0xDF) == 'B')
1249                                                         {
1250                                                                 v &= 0x000000FF;
1251                                                                 ln += 2;
1252                                                         }
1253                                                         else if ((*(ln + 1) & 0xDF) == 'W')
1254                                                         {
1255                                                                 v &= 0x0000FFFF;
1256                                                                 ln += 2;
1257                                                         }
1258                                                         else if ((*(ln + 1) & 0xDF) == 'L')
1259                                                         {
1260                                                                 ln += 2;
1261                                                         }
1262                                                 }
1263                                         }
1264
1265                                         *tk++ = CONST;
1266                                         *tk++ = v;
1267                                 }
1268                                 else
1269                                         *tk++ = '$';
1270
1271                                 continue;
1272                         case '<':               // < or << or <> or <=
1273                                 switch (*ln)
1274                                 {
1275                                 case '<':
1276                                         *tk++ = SHL;
1277                                         ++ln;
1278                                         continue;
1279                                 case '>':
1280                                         *tk++ = NE;
1281                                         ++ln;
1282                                         continue;
1283                                 case '=':
1284                                         *tk++ = LE;
1285                                         ++ln;
1286                                         continue;
1287                                 default:
1288                                         *tk++ = '<';
1289                                         continue;
1290                                 }
1291                         case ':':               // : or ::
1292                                 if (*ln == ':')
1293                                 {
1294                                         *tk++ = DCOLON;
1295                                         ++ln;
1296                                 }
1297                                 else
1298                                         *tk++ = ':';
1299
1300                                 continue;
1301                         case '=':               // = or ==
1302                                 if (*ln == '=')
1303                                 {
1304                                         *tk++ = DEQUALS;
1305                                         ++ln;
1306                                 }
1307                                 else
1308                                         *tk++ = '=';
1309
1310                                 continue;
1311                         case '>':               // > or >> or >=
1312                                 switch (*ln)
1313                                 {
1314                                 case '>':
1315                                         *tk++ = SHR;
1316                                         ln++;
1317                                         continue;
1318                                 case '=':
1319                                         *tk++ = GE;
1320                                         ln++;
1321                                         continue;
1322                                 default:
1323                                         *tk++ = '>';
1324                                         continue;
1325                                 }
1326                         case '%':               // % or binary constant
1327                                 if (*ln < '0' || *ln > '1')
1328                                 {
1329                                         *tk++ = '%';
1330                                         continue;
1331                                 }
1332
1333                                 v = 0;
1334
1335                                 while (*ln >= '0' && *ln <= '1')
1336                                         v = (v << 1) + *ln++ - '0';
1337
1338                                 if (*ln == '.')
1339                                 {
1340                                         if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1341                                         {
1342                                                 v &= 0x000000FF;
1343                                                 ln += 2;
1344                                         }
1345
1346                                         if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1347                                         {
1348                                                 v &= 0x0000FFFF;
1349                                                 ln += 2;
1350                                         }
1351
1352                                         if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1353                                         {
1354                                                 ln += 2;
1355                                         }
1356                                 }
1357
1358                                 *tk++ = CONST;
1359                                 *tk++ = v;
1360                                 continue;
1361                         case '@':               // @ or octal constant
1362                                 if (*ln < '0' || *ln > '7')
1363                                 {
1364                                         *tk++ = '@';
1365                                         continue;
1366                                 }
1367
1368                                 v = 0;
1369
1370                                 while (*ln >= '0' && *ln <= '7')
1371                                         v = (v << 3) + *ln++ - '0';
1372
1373                                 if (*ln == '.')
1374                                 {
1375                                         if ((*(ln+1) == 'b') || (*(ln+1) == 'B'))
1376                                         {
1377                                                 v &= 0x000000FF;
1378                                                 ln += 2;
1379                                         }
1380
1381                                         if ((*(ln+1) == 'w') || (*(ln+1) == 'W'))
1382                                         {
1383                                                 v &= 0x0000FFFF;
1384                                                 ln += 2;
1385                                         }
1386
1387                                         if ((*(ln+1) == 'l') || (*(ln+1) == 'L'))
1388                                         {
1389                                                 ln += 2;
1390                                         }
1391                                 }
1392
1393                                 *tk++ = CONST;
1394                                 *tk++ = v;
1395                                 continue;
1396                         case '^':               // ^ or ^^ <operator-name>
1397                                 if (*ln != '^')
1398                                 {
1399                                         *tk++ = '^';
1400                                         continue;
1401                                 }
1402
1403                                 if (((int)chrtab[*++ln] & STSYM) == 0)
1404                                 {
1405                                         error("invalid symbol following ^^");
1406                                         continue;
1407                                 }
1408
1409                                 p = ln++;
1410
1411                                 while ((int)chrtab[*ln] & CTSYM)
1412                                         ++ln;
1413
1414                                 for(state=0; state>=0;)
1415                                 {
1416                                         // Get char, convert to lowercase
1417                                         j = *p++;
1418
1419                                         if (j >= 'A' && j <= 'Z')
1420                                                 j += 0x20;
1421
1422                                         j += kwbase[state];
1423
1424                                         if (kwcheck[j] != state)
1425                                         {
1426                                                 j = -1;
1427                                                 break;
1428                                         }
1429
1430                                         if (*p == EOS || p == ln)
1431                                         {
1432                                                 j = kwaccept[j];
1433                                                 break;
1434                                         }
1435
1436                                         state = kwtab[j];
1437                                 }
1438
1439                                 if (j < 0 || state < 0)
1440                                 {
1441                                         error("unknown symbol following ^^");
1442                                         continue;
1443                                 }
1444
1445                                 *tk++ = (TOKEN)j;
1446                                 continue;
1447                         default:
1448                                 interror(2);    // Bad MULTX entry in chrtab
1449                                 continue;
1450                         }
1451                 }
1452
1453                 // Handle decimal constant
1454                 if (c & DIGIT)
1455                 {
1456                         v = 0;
1457
1458                         while ((int)chrtab[*ln] & DIGIT)
1459                                 v = (v * 10) + *ln++ - '0';
1460
1461                         // See if there's a .[bwl] after the constant & deal with it if so
1462                         if (*ln == '.')
1463                         {
1464                                 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1465                                 {
1466                                         v &= 0x000000FF;
1467                                         ln += 2;
1468                                 }
1469                                 else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1470                                 {
1471                                         v &= 0x0000FFFF;
1472                                         ln += 2;
1473                                 }
1474                                 else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1475                                 {
1476                                         ln += 2;
1477                                 }
1478                         }
1479
1480                         *tk++ = CONST;
1481                         *tk++ = v;
1482 //printf("CONST: %i\n", v);
1483                         continue;
1484                 }
1485
1486                 // Handle illegal character
1487                 return error("illegal character");
1488         }
1489
1490         // Terminate line of tokens and return "success."
1491
1492 goteol:
1493         tok = etok;                                                             // Set tok to beginning of line
1494
1495         if (stuffnull)                                                  // Terminate last SYMBOL
1496                 *nullspot = EOS;
1497
1498         *tk++ = EOL;
1499
1500         return OK;
1501 }
1502
1503
1504 //
1505 // .GOTO <label>        goto directive
1506 //
1507 // The label is searched for starting from the first line of the current,
1508 // enclosing macro definition. If no enclosing macro exists, an error is
1509 // generated.
1510 //
1511 // A label is of the form:
1512 //
1513 // :<name><whitespace>
1514 //
1515 // The colon must appear in column 1.  The label is stripped prior to macro
1516 // expansion, and is NOT subject to macro expansion.  The whitespace may also
1517 // be EOL.
1518 //
1519 //int d_goto(WORD siz) {
1520 //int d_goto(void)
1521 int d_goto(WORD unused)
1522 {
1523         char * s1, * s2;
1524
1525         // Setup for the search
1526         if (*tok != SYMBOL)
1527                 return error("missing label");
1528
1529 //      sym = (char *)tok[1];
1530         char * sym = string[tok[1]];
1531         tok += 2;
1532
1533         if (cur_inobj->in_type != SRC_IMACRO)
1534                 return error("goto not in macro");
1535
1536         IMACRO * imacro = cur_inobj->inobj.imacro;
1537 //      defln = (LONG *)imacro->im_macro->svalue;
1538         struct LineList * defln = imacro->im_macro->lineList;
1539
1540         // Find the label, starting with the first line.
1541 //      for(; defln!=NULL; defln=(LONG *)*defln)
1542         for(; defln!=NULL; defln=defln->next)
1543         {
1544 //              if (*(char *)(defln + 1) == ':')
1545                 if (defln->line[0] == ':')
1546                 {
1547                         // Compare names (sleazo string compare)
1548                         // This string compare is not right. Doesn't check for lengths.
1549                         // (actually it does, but in a crappy, unclear way.)
1550 WARNING(!!!! Bad string comparison !!!)
1551                         s1 = sym;
1552 //                      s2 = (char *)(defln + 1) + 1;
1553                         s2 = defln->line;
1554
1555                         while (*s1 == *s2)
1556                         {
1557                                 if (*s1 == EOS)
1558                                         break;
1559                                 else
1560                                 {
1561                                         s1++;
1562                                         s2++;
1563                                 }
1564                         }
1565
1566                         // Found the label, set new macro next-line and return.
1567                         if ((*s2 == EOS) || ((int)chrtab[*s2] & WHITE))
1568                         {
1569                                 imacro->im_nextln = defln;
1570                                 return 0;
1571                         }
1572                 }
1573         }
1574
1575         return error("goto label not found");
1576 }
1577
1578
1579 void DumpTokenBuffer(void)
1580 {
1581         TOKEN * t;
1582         printf("Tokens [%X]: ", sloc);
1583
1584         for(t=tokbuf; *t!=EOL; t++)
1585         {
1586                 if (*t == COLON)
1587                         printf("[COLON]");
1588                 else if (*t == CONST)
1589                 {
1590                         t++;
1591                         printf("[CONST: $%X]", (uint32_t)*t);
1592                 }
1593                 else if (*t == ACONST)
1594                         printf("[ACONST]");
1595                 else if (*t == STRING)
1596                 {
1597                         t++;
1598                         printf("[STRING:\"%s\"]", string[*t]);
1599                 }
1600                 else if (*t == SYMBOL)
1601                 {
1602                         t++;
1603                         printf("[SYMBOL:\"%s\"]", string[*t]);
1604                 }
1605                 else if (*t == EOS)
1606                         printf("[EOS]");
1607                 else if (*t == TKEOF)
1608                         printf("[TKEOF]");
1609                 else if (*t == DEQUALS)
1610                         printf("[DEQUALS]");
1611                 else if (*t == SET)
1612                         printf("[SET]");
1613                 else if (*t == REG)
1614                         printf("[REG]");
1615                 else if (*t == DCOLON)
1616                         printf("[DCOLON]");
1617                 else if (*t == GE)
1618                         printf("[GE]");
1619                 else if (*t == LE)
1620                         printf("[LE]");
1621                 else if (*t == NE)
1622                         printf("[NE]");
1623                 else if (*t == SHR)
1624                         printf("[SHR]");
1625                 else if (*t == SHL)
1626                         printf("[SHL]");
1627                 else if (*t == UNMINUS)
1628                         printf("[UNMINUS]");
1629                 else if (*t == DOTB)
1630                         printf("[DOTB]");
1631                 else if (*t == DOTW)
1632                         printf("[DOTW]");
1633                 else if (*t == DOTL)
1634                         printf("[DOTL]");
1635                 else if (*t == DOTI)
1636                         printf("[DOTI]");
1637                 else if (*t == ENDEXPR)
1638                         printf("[ENDEXPR]");
1639                 else if (*t == CR_DEFINED)
1640                         printf("[CR_DEFINED]");
1641                 else if (*t == CR_REFERENCED)
1642                         printf("[CR_REFERENCED]");
1643                 else if (*t == CR_STREQ)
1644                         printf("[CR_STREQ]");
1645                 else if (*t == CR_MACDEF)
1646                         printf("[CR_MACDEF]");
1647                 else if (*t == CR_TIME)
1648                         printf("[CR_TIME]");
1649                 else if (*t == CR_DATE)
1650                         printf("[CR_DATE]");
1651                 else if (*t >= 0x20 && *t <= 0x2F)
1652                         printf("[%c]", (char)*t);
1653                 else if (*t >= 0x3A && *t <= 0x3F)
1654                         printf("[%c]", (char)*t);
1655                 else if (*t >= 0x80 && *t <= 0x87)
1656                         printf("[D%u]", ((uint32_t)*t) - 0x80);
1657                 else if (*t >= 0x88 && *t <= 0x8F)
1658                         printf("[A%u]", ((uint32_t)*t) - 0x88);
1659                 else
1660                         printf("[%X:%c]", (uint32_t)*t, (char)*t);
1661         }
1662
1663         printf("[EOL]\n");
1664 }
1665