2 // RMAC - Renamed Macro Assembler for all Atari computers
3 // TOKEN.C - Token Handling
4 // Copyright (C) 199x Landon Dyer, 2011-2021 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
19 #define DECL_KW // Declare keyword arrays
20 #define DEF_KW // Declare keyword values
21 #include "kwtab.h" // Incl generated keyword tables & defs
24 int lnsave; // 1; strcpy() text of current line
25 uint32_t curlineno; // Current line number (64K max currently)
26 int totlines; // Total # of lines
27 int mjump_align = 0; // mjump alignment flag
28 char lntag; // Line tag
29 char * curfname; // Current filename
30 char tolowertab[128]; // Uppercase ==> lowercase
31 int8_t hextab[128]; // Table of hex values
32 char dotxtab[128]; // Table for ".b", ".s", etc.
33 char irbuf[LNSIZ]; // Text for .rept block line
34 char lnbuf[LNSIZ]; // Text of current line
35 WORD filecount; // Unique file number counter
36 WORD cfileno; // Current file number
37 TOKEN * tok; // Ptr to current token
38 TOKEN * etok; // Ptr past last token in tokbuf[]
39 TOKEN tokeol[1] = {EOL}; // Bailout end-of-line token
40 char * string[TOKBUFSIZE*2];// Token buffer string pointer storage
41 int optimizeOff; // Optimization override flag
47 INOBJ * cur_inobj; // Ptr current input obj (IFILE/IMACRO)
48 static INOBJ * f_inobj; // Ptr list of free INOBJs
49 static IFILE * f_ifile; // Ptr list of free IFILEs
50 static IMACRO * f_imacro; // Ptr list of free IMACROs
52 static TOKEN tokbuf[TOKBUFSIZE]; // Token buffer (stack-like, all files)
54 uint8_t chrtab[0x100] = {
55 ILLEG, ILLEG, ILLEG, ILLEG, // NUL SOH STX ETX
56 ILLEG, ILLEG, ILLEG, ILLEG, // EOT ENQ ACK BEL
57 ILLEG, WHITE, ILLEG, ILLEG, // BS HT LF VT
58 WHITE, ILLEG, ILLEG, ILLEG, // FF CR SO SI
60 ILLEG, ILLEG, ILLEG, ILLEG, // DLE DC1 DC2 DC3
61 ILLEG, ILLEG, ILLEG, ILLEG, // DC4 NAK SYN ETB
62 ILLEG, ILLEG, ILLEG, ILLEG, // CAN EM SUB ESC
63 ILLEG, ILLEG, ILLEG, ILLEG, // FS GS RS US
65 WHITE, MULTX, MULTX, SELF, // SP ! " #
66 MULTX+CTSYM, MULTX, SELF, MULTX, // $ % & '
67 SELF, SELF, SELF, SELF, // ( ) * +
68 SELF, SELF, STSYM, SELF, // , - . /
70 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 0 1
71 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 2 3
72 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 4 5
73 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 6 7
74 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 8 9
76 MULTX, MULTX, MULTX, STSYM+CTSYM, // < = > ?
78 MULTX, STSYM+CTSYM+HDIGIT, // @ A
79 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // B C
80 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // D E
81 STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // F G
82 STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // H I J K
83 DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // L M N O
85 DOT+STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // P Q R S
86 STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // T U V W
87 STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF,// X Y Z [
88 SELF, SELF, MULTX, STSYM+CTSYM, // \ ] ^ _
90 ILLEG, STSYM+CTSYM+HDIGIT, // ` a
91 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // b c
92 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // d e
93 STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // f g
94 STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // h i j k
95 DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // l m n o
97 DOT+STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // p q r s
98 STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // t u v w
99 DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // x y z {
100 SELF, SELF, SELF, ILLEG, // | } ~ DEL
102 // Anything above $7F is illegal (and yes, we need to check for this,
103 // otherwise you get strange and spurious errors that will lead you astray)
104 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
105 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
106 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
107 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
108 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
109 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
110 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
111 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
112 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
113 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
114 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
115 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
116 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
117 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
118 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
119 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG
122 // Names of registers
123 static char * regname[] = {
124 "d0","d1","d2","d3","d4","d5","d6","d7", // 128,135
125 "a0","a1","a2","a3","a4","a5","a6","sp", // 136,143
126 "ssp","pc","sr","ccr","regequ","set","reg","r0", // 144,151
127 "r1","r2","r3","r4","r5","r6","r7","r8", // 152,159
128 "r9","r10","r11","r12","r13","r14","r15","r16", // 160,167
129 "r17","r18","r19","r20","r21","r22","r23","r24", // 168,175
130 "r25","r26","r27","r28","r29","r30","r31","ccdef", // 176,183
131 "usp","ic40","dc40","bc40","sfc","dfc","","vbr", // 184,191
132 "cacr","caar","msp","isp","tc","itt0","itt1","dtt0", // 192,199
133 "dtt1","mmusr","urp","srp","iacr0","iacr1","dacr0","dacr1", // 200,207
134 "tt0","tt1","crp","","","","","", // 208,215
135 "","","","","fpiar","fpsr","fpcr","", // 216,223
136 "fp0","fp1","fp2","fp3","fp4","fp5","fp6","fp7", // 224,231
137 "","","","","","","","", // 232,239
138 "","","","","","","","", // 240,247
139 "","","","","","","","", // 248,255
140 "","","","","x0","x1","y0","y1", // 256,263
141 "","b0","","b2","","b1","a","b", // 264,271
142 "mr","omr","la","lc","ssh","ssl","ss","", // 272,279
143 "n0","n1","n2","n3","n4","n5","n6","n7", // 280,287
144 "m0","m1","m2","m3","m4","m5","m6","m7", // 288,295
145 "","","","","","","l","p", // 296,303
146 "mr","omr","la","lc","ssh","ssl","ss","", // 304,311
147 "a10","b10","x","y","","","ab","ba" // 312,319
150 static char * riscregname[] = {
151 "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
152 "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
153 "r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
154 "r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31"
159 // Initialize tokenizer
161 void InitTokenizer(void)
164 char * htab = "0123456789abcdefABCDEF"; // Hex character table
166 lnsave = 0; // Don't save lines
167 curfname = ""; // No file, empty filename
168 filecount = (WORD)-1;
169 cfileno = (WORD)-1; // cfileno gets bumped to 0
181 // Initialize hex, "dot" and tolower tables
186 tolowertab[i] = (char)i;
189 for(i=0; htab[i]!=EOS; i++)
190 hextab[htab[i]] = (char)((i < 16) ? i : i - 6);
192 for(i='A'; i<='Z'; i++)
193 tolowertab[i] |= 0x20;
195 // These characters are legal immediately after a period
196 dotxtab['b'] = DOTB; // .b .B .s .S
198 //dotxtab['s'] = DOTB;
199 //dotxtab['S'] = DOTB;
200 dotxtab['w'] = DOTW; // .w .W
202 dotxtab['l'] = DOTL; // .l .L
204 dotxtab['i'] = DOTI; // .i .I (WTF is this???)
206 dotxtab['D'] = DOTD; // .d .D (double)
208 dotxtab['S'] = DOTS; // .s .S
210 dotxtab['Q'] = DOTQ; // .q .Q (quad word)
212 dotxtab['X'] = DOTX; // .x .x
214 dotxtab['P'] = DOTP; // .p .P
219 void SetFilenameForErrorReporting(void)
223 // Check for absolute top filename (this should never happen)
226 curfname = "(*top*)";
230 FILEREC * fr = filerec;
232 // Advance to the correct record...
233 while (fr != NULL && fnum != 0)
239 // Check for file # record not found (this should never happen either)
242 curfname = "(*NOT FOUND*)";
246 curfname = fr->frec_name;
251 // Allocate an IFILE or IMACRO
253 INOBJ * a_inobj(int typ)
259 // Allocate and initialize INOBJ first
261 inobj = malloc(sizeof(INOBJ));
265 f_inobj = f_inobj->in_link;
270 case SRC_IFILE: // Alloc and init an IFILE
272 ifile = malloc(sizeof(IFILE));
276 f_ifile = f_ifile->if_link;
279 inobj->inobj.ifile = ifile;
282 case SRC_IMACRO: // Alloc and init an IMACRO
283 if (f_imacro == NULL)
284 imacro = malloc(sizeof(IMACRO));
288 f_imacro = f_imacro->im_link;
291 inobj->inobj.imacro = imacro;
294 case SRC_IREPT: // Alloc and init an IREPT
295 inobj->inobj.irept = malloc(sizeof(IREPT));
296 DEBUG { printf("alloc IREPT\n"); }
300 // Install INOBJ on top of input stack
301 inobj->in_ifent = ifent; // Record .if context on entry
302 inobj->in_type = (WORD)typ;
303 inobj->in_otok = tok;
304 inobj->in_etok = etok;
305 inobj->in_link = cur_inobj;
313 // Perform macro substitution from 'orig' to 'dest'. Return OK or some error.
314 // A macro reference is in one of two forms:
315 // \name <non-name-character>
317 // A doubled backslash (\\) is compressed to a single backslash (\).
318 // Argument definitions have been pre-tokenized, so we have to turn them back
319 // into text. This means that numbers, in particular, become hex, regardless of
320 // their representation when the macro was invoked. This is a hack.
321 // A label may appear at the beginning of the line:
322 // :<name><whitespace>
323 // (the colon must be in the first column). These labels are stripped before
324 // macro expansion takes place.
326 int ExpandMacro(char * src, char * dest, int destsiz)
329 int questmark; // \? for testing argument existence
330 char mname[128]; // Assume max size of a formal arg name
331 char numbuf[20]; // Buffer for text of CONSTs
334 char ** symbolString;
336 DEBUG { printf("ExM: src=\"%s\"\n", src); }
338 IMACRO * imacro = cur_inobj->inobj.imacro;
339 int macnum = (int)(imacro->im_macro->sattr);
341 char * dst = dest; // Next dest slot
342 char * edst = dest + destsiz - 1; // End + 1(?) of dest buffer
344 // Check for (and skip over) any "label" on the line
350 while (*s != EOS && !(chrtab[*s] & WHITE))
354 s++; // Skip first whitespace
357 // Expand the rest of the line
360 // Copy single character
366 // Skip comments in case a loose @ or \ is in there
367 // In that case the tokeniser was trying to expand it.
368 if ((*s == ';') || ((*s == '/') && (*(s + 1) == '/')))
373 // Do macro expansion
381 case '\\': // \\, \ (collapse to single backslash)
387 case '?': // \? <macro> set `questmark' flag
391 case '#': // \#, number of arguments
392 sprintf(numbuf, "%d", (int)imacro->im_nargs);
394 case '!': // \! size suffix supplied on invocation
395 switch ((int)imacro->im_siz)
397 case SIZN: d = ""; break;
398 case SIZB: d = ".b"; break;
399 case SIZW: d = ".w"; break;
400 case SIZL: d = ".l"; break;
404 case '~': // ==> unique label string Mnnnn...
405 sprintf(numbuf, "M%u", curuniq);
421 return error("missing argument name");
424 // \n ==> argument number 'n', 0..9
425 if (chrtab[*s] & DIGIT)
435 // Get argument name: \name, \{name}
445 while (chrtab[*s] & CTSYM);
450 for(++s; *s != EOS && *s != '}';)
454 return error("missing closing brace ('}')");
461 // Lookup the argument and copy its (string) value into the
462 // destination string
463 DEBUG { printf("argument='%s'\n", mname); }
465 if ((arg = lookup(mname, MACARG, macnum)) == NULL)
466 return error("undefined argument: '%s'", mname);
469 // Convert a string of tokens (terminated with EOL) back into
470 // text. If an argument is out of range (not specified in the
471 // macro invocation) then it is ignored.
472 i = (int)arg->svalue;
474 DEBUG { printf("~argnumber=%d\n", i); }
477 if (i < imacro->im_nargs)
479 tk = imacro->argument[i].token;
480 symbolString = imacro->argument[i].string;
483 // printf("ExM: Preparing to parse argument #%u...\n", i);
489 // 0 if the argument is empty or non-existant,
490 // 1 if the argument is not empty
493 if (tk == NULL || *tk == EOL)
499 *dst++ = (char)(questmark + '0');
503 // Argument # is in range, so expand it
508 // Reverse-translation from a token number to a string.
509 // This is a hack. It might be better table-driven.
512 if ((*tk >= KW_D0) && !rdsp && !rgpu)
514 d = regname[(int)*tk++ - KW_D0];
517 else if ((*tk >= KW_R0) && (*tk <= KW_R31))
519 d = riscregname[(int)*tk++ - KW_R0];
527 d = symbolString[*tk++];
528 DEBUG { printf("ExM: SYMBOL=\"%s\"", d); }
531 d = symbolString[*tk++];
552 // Shamus: Changing the format specifier from %lx to %ux caused the assembler
553 // to choke on legitimate code... Need to investigate this further
554 // before changing anything else here!
556 // sprintf(numbuf, "$%lx", (uint64_t)*tk++);
557 sprintf(numbuf, "$%" PRIX64, (uint64_t)*tk++);
625 *dst++ = (char)*(tk - 1);
630 // If 'd' != NULL, copy string to destination
634 DEBUG printf("d='%s'\n", d);
653 DEBUG { printf("ExM: dst=\"%s\"\n", dest); }
658 DEBUG { printf("*** OVERFLOW LINE ***\n%s\n", dest); }
659 return fatal("line too long as a result of macro expansion");
664 // Get next line of text from a macro
666 char * GetNextMacroLine(void)
668 IMACRO * imacro = cur_inobj->inobj.imacro;
669 LLIST * strp = imacro->im_nextln;
671 if (strp == NULL) // End-of-macro
674 imacro->im_nextln = strp->next;
675 // ExpandMacro((char *)(strp + 1), imacro->im_lnbuf, LNSIZ);
676 ExpandMacro(strp->line, imacro->im_lnbuf, LNSIZ);
678 return imacro->im_lnbuf;
683 // Get next line of text from a repeat block
685 char * GetNextRepeatLine(void)
687 IREPT * irept = cur_inobj->inobj.irept;
688 // LONG * strp = irept->ir_nextln; // initial null
690 // Do repeat at end of .rept block's string list
692 if (irept->ir_nextln == NULL)
694 DEBUG { printf("back-to-top-of-repeat-block count=%d\n", (int)irept->ir_count); }
695 irept->ir_nextln = irept->ir_firstln; // copy first line
697 if (irept->ir_count-- == 0)
699 DEBUG { printf("end-repeat-block\n"); }
703 // strp = irept->ir_nextln;
705 // Mark the current macro line in the irept object
706 // This is probably overkill - a global variable
707 // would suffice here (it only gets used during
708 // error reporting anyway)
709 irept->lineno = irept->ir_nextln->lineno;
711 // Copy the rept lines verbatim, unless we're in nest level 0.
712 // Then, expand any \~ labels to unique numbers (Rn)
715 strcpy(irbuf, irept->ir_nextln->line);
719 uint32_t linelen = strlen(irept->ir_nextln->line);
720 uint8_t *p_line = irept->ir_nextln->line;
721 char *irbufwrite = irbuf;
722 for (int i = 0; i <= linelen; i++)
726 if (c == '\\' && *p_line == '~')
729 irbufwrite += sprintf(irbufwrite, "R%u", reptuniq);
738 DEBUG { printf("repeat line='%s'\n", irbuf); }
739 // irept->ir_nextln = (LONG *)*strp;
740 irept->ir_nextln = irept->ir_nextln->next;
747 // Include a source file used at the root, and for ".include" files
749 int include(int handle, char * fname)
752 DEBUG { printf("[include: %s, cfileno=%u]\n", fname, cfileno); }
754 // Alloc and initialize include-descriptors
755 INOBJ * inobj = a_inobj(SRC_IFILE);
756 IFILE * ifile = inobj->inobj.ifile;
758 ifile->ifhandle = handle; // Setup file handle
759 ifile->ifind = ifile->ifcnt = 0; // Setup buffer indices
760 ifile->ifoldlineno = curlineno; // Save old line number
761 ifile->ifoldfname = curfname; // Save old filename
762 ifile->ifno = cfileno; // Save old file number
764 // NB: This *must* be preincrement, we're adding one to the filecount here!
765 cfileno = ++filecount; // Compute NEW file number
766 curfname = strdup(fname); // Set current filename (alloc storage)
767 curlineno = 0; // Start on line zero
769 // Add another file to the file-record
770 FILEREC * fr = (FILEREC *)malloc(sizeof(FILEREC));
771 fr->frec_next = NULL;
772 fr->frec_name = curfname;
775 filerec = fr; // Add first filerec
777 last_fr->frec_next = fr; // Append to list of filerecs
780 DEBUG { printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno); }
787 // Pop the current input level
791 INOBJ * inobj = cur_inobj;
796 // Pop IFENT levels until we reach the conditional assembly context we
797 // were at when the input object was entered.
798 int numUnmatched = 0;
800 while (ifent != inobj->in_ifent)
802 if (d_endif() != 0) // Something bad happened during endif parsing?
803 return -1; // If yes, bail instead of getting stuck in a loop
808 // Give a warning to the user that we had to wipe their bum for them
809 if (numUnmatched > 0)
810 warn("missing %d .endif(s)", numUnmatched);
812 tok = inobj->in_otok; // Restore tok and etok
813 etok = inobj->in_etok;
815 switch (inobj->in_type)
817 case SRC_IFILE: // Pop and release an IFILE
819 DEBUG { printf("[Leaving: %s]\n", curfname); }
821 IFILE * ifile = inobj->inobj.ifile;
822 ifile->if_link = f_ifile;
824 close(ifile->ifhandle); // Close source file
825 DEBUG { printf("[fpop (pre): curfname=%s]\n", curfname); }
826 curfname = ifile->ifoldfname; // Set current filename
827 DEBUG { printf("[fpop (post): curfname=%s]\n", curfname); }
828 DEBUG { printf("[fpop: (pre) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); }
829 curlineno = ifile->ifoldlineno; // Set current line#
830 DEBUG { printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno); }
831 cfileno = ifile->ifno; // Restore current file number
832 DEBUG { printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); }
836 case SRC_IMACRO: // Pop and release an IMACRO
838 IMACRO * imacro = inobj->inobj.imacro;
839 imacro->im_link = f_imacro;
844 case SRC_IREPT: // Pop and release an IREPT
846 DEBUG { printf("dealloc IREPT\n"); }
847 LLIST * p = inobj->inobj.irept->ir_firstln;
849 // Deallocate repeat lines
860 cur_inobj = inobj->in_link;
861 inobj->in_link = f_inobj;
869 // Get line from file into buf, return NULL on EOF or ptr to the start of a
872 char * GetNextLine(void)
876 int readamt = -1; // 0 if last read() yeilded 0 bytes
877 IFILE * fl = cur_inobj->inobj.ifile;
881 // Scan for next end-of-line; handle stupid text formats by treating
882 // \r\n the same as \n. (lone '\r' at end of buffer means we have to
884 d = &fl->ifbuf[fl->ifind];
886 for(p=d, i=0, j=fl->ifcnt; i<j; i++, p++)
888 if (*p == '\r' || *p == '\n')
895 break; // Need to read more, then look for '\n' to eat
896 else if (p[1] == '\n')
900 // Cover up the newline with end-of-string sentinel
909 // Handle hanging lines by ignoring them (Input file is exhausted, no
910 // \r or \n on last line)
911 // Shamus: This is retarded. Never ignore any input!
912 if (!readamt && fl->ifcnt)
919 // Really should check to see if we're at the end of the buffer!
921 fl->ifbuf[fl->ifind + fl->ifcnt] = '\0';
923 return &fl->ifbuf[fl->ifind];
927 // Truncate and return absurdly long lines.
928 if (fl->ifcnt >= QUANTUM)
930 fl->ifbuf[fl->ifind + fl->ifcnt - 1] = '\0';
932 return &fl->ifbuf[fl->ifind];
935 // Relocate what's left of a line to the beginning of the buffer, and
936 // read some more of the file in; return NULL if the buffer's empty and
940 p = &fl->ifbuf[fl->ifind];
941 d = &fl->ifbuf[fl->ifcnt & 1];
943 for(i=0; i<fl->ifcnt; i++)
946 fl->ifind = fl->ifcnt & 1;
949 readamt = read(fl->ifhandle, &fl->ifbuf[fl->ifind + fl->ifcnt], QUANTUM);
954 if ((fl->ifcnt += readamt) == 0)
963 int TokenizeLine(void)
965 uint8_t * ln = NULL; // Ptr to current position in line
966 uint8_t * p; // Random character ptr
967 PTR tk; // Token-deposit ptr
968 int state = 0; // State for keyword detector
969 int j = 0; // Var for keyword detector
970 uint8_t c; // Random char
971 uint64_t v; // Random value
972 uint32_t cursize = 0; // Current line's size (.b, .w, .l, .s, .q, .d)
973 uint8_t * nullspot = NULL; // Spot to clobber for SYMBOL termination
974 int stuffnull; // 1:terminate SYMBOL '\0' at *nullspot
976 int stringNum = 0; // Pointer to string locations in tokenized line
977 SYM* sy; // For looking up symbols (.equr)
978 int equrundef = 0; // Flag for equrundef scanning
982 if (cur_inobj == NULL) // Return EOF if input stack is empty
985 // Get another line of input from the current input source: a file, a
986 // macro, or a repeat-block
987 switch (cur_inobj->in_type)
991 // o bump source line number;
992 // o tag the listing-line with a space;
993 // o kludge lines generated by Alcyon C.
995 if ((ln = GetNextLine()) == NULL)
997 DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); }
998 if (fpop() == 0) // Pop input level
999 goto retry; // Try for more lines
1002 ifent->if_prev = (IFENT *)-1; //Signal Assemble() that we have reached EOF with unbalanced if/endifs
1007 curlineno++; // Bump line number
1013 // o Handle end-of-macro;
1014 // o tag the listing-line with an at (@) sign.
1016 if ((ln = GetNextMacroLine()) == NULL)
1018 if (ExitMacro() == 0) // Exit macro (pop args, do fpop(), etc)
1019 goto retry; // Try for more lines...
1021 return TKEOF; // Oops, we got a non zero return code, signal EOF
1028 // o Handle end-of-repeat-block;
1029 // o tag the listing-line with a pound (#) sign.
1031 if ((ln = GetNextRepeatLine()) == NULL)
1033 DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); }
1042 // Save text of the line. We only do this during listings and within
1043 // macro-type blocks, since it is expensive to unconditionally copy every
1048 if (strlen(ln) > LNSIZ)
1049 return error("line too long (%d, max %d)", strlen(ln), LNSIZ);
1054 // General housekeeping
1055 tok = tokeol; // Set "tok" to EOL in case of error
1056 tk.u32 = etok; // Reset token ptr
1057 stuffnull = 0; // Don't stuff nulls
1058 totlines++; // Bump total #lines assembled
1060 // See if the entire line is a comment. This is a win if the programmer
1061 // puts in lots of comments
1062 if (*ln == '*' || *ln == ';' || ((*ln == '/') && (*(ln + 1) == '/')))
1065 // And here we have a very ugly hack for signalling a single line 'turn off
1066 // optimization'. There's really no nice way to do this, so hack it is!
1067 optimizeOff = 0; // Default is to take optimizations as they come
1071 optimizeOff = 1; // Signal that we don't want to optimize this line
1072 ln++; // & skip over the darned thing
1075 // Main tokenization loop;
1076 // o skip whitespace;
1077 // o handle end-of-line;
1078 // o handle symbols;
1079 // o handle single-character tokens (operators, etc.);
1080 // o handle multiple-character tokens (constants, strings, etc.).
1083 // Check to see if there's enough space in the token buffer
1084 if (tk.cp >= ((uint8_t *)(&tokbuf[TOKBUFSIZE])) - 20)
1086 return error("token buffer overrun");
1089 // Skip whitespace, handle EOL
1090 while (chrtab[*ln] & WHITE)
1093 // Handle EOL, comment with ';'
1094 if (*ln == EOS || *ln == ';'|| ((*ln == '/') && (*(ln + 1) == '/')))
1097 // Handle start of symbol. Symbols are null-terminated in place. The
1098 // termination is always one symbol behind, since there may be no place
1099 // for a null in the case that an operator immediately follows the name.
1104 if (stuffnull) // Terminate old symbol from previous pass
1107 v = 0; // Assume no DOT attrib follows symbol
1110 // In some cases, we need to check for a DOTx at the *beginning*
1111 // of a symbol, as the "start" of the line we're currently looking
1112 // at could be somewhere in the middle of that line!
1115 // Make sure that it's *only* a .[bwsl] following, and not the
1116 // start of a local symbol:
1117 if ((chrtab[*(ln + 1)] & DOT)
1118 && (dotxtab[*(ln + 1)] != 0)
1119 && !(chrtab[*(ln + 2)] & CTSYM))
1121 // We found a legitimate DOTx construct, so add it to the
1125 *tk.u32++ = (TOKEN)dotxtab[*ln++];
1130 p = nullspot = ln++; // Nullspot -> start of this symbol
1132 // Find end of symbol (and compute its length)
1133 for(j=1; (int)chrtab[*ln]&CTSYM; j++)
1136 // Handle "DOT" special forms (like ".b") that follow a normal
1137 // symbol or keyword:
1140 *ln++ = EOS; // Terminate symbol
1141 stuffnull = 0; // And never try it again
1143 // Character following the '.' must have a DOT attribute, and
1144 // the chararacter after THAT one must not have a start-symbol
1145 // attribute (to prevent symbols that look like, for example,
1146 // "zingo.barf", which might be a good idea anyway....)
1147 if (((chrtab[*ln] & DOT) == 0) || (dotxtab[*ln] == 0))
1148 return error("[bwsl] must follow '.' in symbol");
1150 v = (uint32_t)dotxtab[*ln++];
1151 cursize = (uint32_t)v;
1153 if (chrtab[*ln] & CTSYM)
1154 return error("misuse of '.'; not allowed in symbols");
1157 // If the symbol is small, check to see if it's really the name of
1161 for(state=0; state>=0;)
1163 j = (int)tolowertab[*p++];
1166 if (kwcheck[j] != state)
1172 if (*p == EOS || p == ln)
1186 // Make j = -1 if user tries to use a RISC register while in 68K mode
1187 if (!(rgpu || rdsp || dsp56001) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31))
1192 // Make j = -1 if time, date etc with no preceeding ^^
1193 // defined, referenced, streq, macdef, date and time
1196 case 112: // defined
1197 case 113: // referenced
1202 case KW_FILESIZE: // filesize
1206 // If we detected equrundef/regundef set relevant flag
1207 if (j == KW_EQURUNDEF)
1213 // If not tokenized keyword OR token was not found
1214 if ((j < 0) || (state < 0))
1216 // Only proceed if no equrundef has been detected. In that case we need to store the symbol
1217 // because the directive handler (d_equrundef) will run outside this loop, further into procln.c
1220 // Last attempt: let's see if this is an equated register.
1221 // If yes, then just store the register's keyword value instead of the symbol
1224 sy = lookup(nullspot, LABEL, 0);
1228 if (sy->sattre & EQUATEDREG)
1230 *tk.u32++ = sy->svalue;
1236 // Ok, that failed, let's store the symbol instead
1238 string[stringNum] = nullspot;
1239 *tk.u32++ = stringNum;
1244 *tk.u32++ = (TOKEN)j;
1248 if (v) // Record attribute token (if any)
1249 *tk.u32++ = (TOKEN)v;
1251 if (stuffnull) // Arrange for string termination on next pass
1257 // Handle identity tokens
1264 // Handle multiple-character tokens
1269 case '!': // ! or !=
1279 case '\'': // 'string'
1282 // Hardcoded for now, maybe this will change in the future
1283 *tk.u32++ = STRINGA8;
1287 case '\"': // "string"
1291 string[stringNum] = ln;
1292 *tk.u32++ = stringNum;
1295 for(p=ln; *ln!=EOS && *ln!=c1;)
1304 return(error("unterminated string"));
1333 // If we're evaluating a macro
1334 // this is valid because it's
1335 // a parameter expansion
1337 // If we're evaluating a macro
1338 // this is valid and expands to
1342 warn("bad backslash code in string");
1352 return error("unterminated string");
1356 case '$': // $, hex constant
1357 if (chrtab[*ln] & HDIGIT)
1361 // Parse the hex value
1362 while (hextab[*ln] >= 0)
1363 v = (v << 4) + (int)hextab[*ln++];
1370 if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1375 else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1386 case '<': // < or << or <> or <=
1405 case ':': // : or ::
1415 case '=': // = or ==
1418 *tk.u32++ = DEQUALS;
1425 case '>': // > or >> or >=
1440 case '%': // % or binary constant
1441 if (*ln < '0' || *ln > '1')
1449 while (*ln >= '0' && *ln <= '1')
1450 v = (v << 1) + *ln++ - '0';
1454 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1460 if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1466 if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1476 case '@': // @ or octal constant
1477 if (*ln < '0' || *ln > '7')
1485 while (*ln >= '0' && *ln <= '7')
1486 v = (v << 3) + *ln++ - '0';
1490 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1496 if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1502 if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1512 case '^': // ^ or ^^ <operator-name>
1519 if (((int)chrtab[*++ln] & STSYM) == 0)
1521 error("invalid symbol following ^^");
1527 while ((int)chrtab[*ln] & CTSYM)
1530 for(state=0; state>=0;)
1532 // Get char, convert to lowercase
1535 if (j >= 'A' && j <= 'Z')
1540 if (kwcheck[j] != state)
1546 if (*p == EOS || p == ln)
1555 if (j < 0 || state < 0)
1557 error("unknown symbol following ^^");
1561 *tk.u32++ = (TOKEN)j;
1564 interror(2); // Bad MULTX entry in chrtab
1569 // Handle decimal constant
1572 uint8_t * numStart = ln;
1575 while ((int)chrtab[*ln] & DIGIT)
1576 v = (v * 10) + *ln++ - '0';
1578 // See if there's a .[bwl] after the constant & deal with it if so
1581 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1589 else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1597 else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1605 else if ((int)chrtab[*(ln + 1)] & DIGIT)
1607 // Hey, more digits after the dot, so we assume it's a
1608 // floating point number of some kind... numEnd will point
1609 // to the first non-float character after it's done
1612 double f = strtod(numStart, &numEnd);
1613 ln = (uint8_t *)numEnd;
1616 return error("floating point parse error");
1618 // N.B.: We use the C compiler's internal double
1619 // representation for all internal float calcs and
1620 // are reasonably sure that the size of said double
1621 // is 8 bytes long (which we check for in fltpoint.c)
1634 //printf("CONST: %i\n", v);
1638 // Handle illegal character
1639 return error("illegal character $%02X found", *ln);
1642 // Terminate line of tokens and return "success."
1645 tok = etok; // Set tok to beginning of line
1647 if (stuffnull) // Terminate last SYMBOL
1657 // .GOTO <label> goto directive
1659 // The label is searched for starting from the first line of the current,
1660 // enclosing macro definition. If no enclosing macro exists, an error is
1663 // A label is of the form:
1665 // :<name><whitespace>
1667 // The colon must appear in column 1. The label is stripped prior to macro
1668 // expansion, and is NOT subject to macro expansion. The whitespace may also
1671 int d_goto(WORD unused)
1673 // Setup for the search
1675 return error("missing label");
1677 char * sym = string[tok[1]];
1680 if (cur_inobj->in_type != SRC_IMACRO)
1681 return error("goto not in macro");
1683 IMACRO * imacro = cur_inobj->inobj.imacro;
1684 LLIST * defln = imacro->im_macro->lineList;
1686 // Attempt to find the label, starting with the first line.
1687 for(; defln!=NULL; defln=defln->next)
1689 // Must start with a colon
1690 if (defln->line[0] == ':')
1692 // Compare names (sleazo string compare)
1694 char * s2 = defln->line + 1;
1696 // Either we will match the strings to EOS on both, or we will
1697 // match EOS on string 1 to whitespace on string 2. Otherwise, we
1699 while ((*s1 == *s2) || ((*s1 == EOS) && (chrtab[*s2] & WHITE)))
1701 // If we reached the end of string 1 (sym), we're done.
1702 // Note that we're also checking for the end of string 2 as
1703 // well, since we've established they're equal above.
1706 // Found the label, set new macro next-line and return.
1707 imacro->im_nextln = defln;
1717 return error("goto label not found");
1721 void DumpToken(TOKEN t)
1725 else if (t == CONST)
1727 else if (t == FCONST)
1729 else if (t == ACONST)
1731 else if (t == STRING)
1733 else if (t == SYMBOL)
1737 else if (t == TKEOF)
1739 else if (t == DEQUALS)
1740 printf("[DEQUALS]");
1745 else if (t == DCOLON)
1757 else if (t == UNMINUS)
1758 printf("[UNMINUS]");
1773 else if (t == ENDEXPR)
1774 printf("[ENDEXPR]");
1775 else if (t == CR_ABSCOUNT)
1776 printf("[CR_ABSCOUNT]");
1777 else if (t == CR_FILESIZE)
1778 printf("[CR_FILESIZE]");
1779 else if (t == CR_DEFINED)
1780 printf("[CR_DEFINED]");
1781 else if (t == CR_REFERENCED)
1782 printf("[CR_REFERENCED]");
1783 else if (t == CR_STREQ)
1784 printf("[CR_STREQ]");
1785 else if (t == CR_MACDEF)
1786 printf("[CR_MACDEF]");
1787 else if (t == CR_TIME)
1788 printf("[CR_TIME]");
1789 else if (t == CR_DATE)
1790 printf("[CR_DATE]");
1791 else if (t >= 0x20 && t <= 0x2F)
1792 printf("[%c]", (char)t);
1793 else if (t >= 0x3A && t <= 0x3F)
1794 printf("[%c]", (char)t);
1795 else if (t >= 0x80 && t <= 0x87)
1796 printf("[D%u]", ((uint32_t)t) - 0x80);
1797 else if (t >= 0x88 && t <= 0x8F)
1798 printf("[A%u]", ((uint32_t)t) - 0x88);
1800 printf("[%X:%c]", (uint32_t)t, (char)t);
1804 void DumpTokenBuffer(void)
1806 printf("Tokens [%X]: ", sloc);
1808 for(TOKEN * t=tokbuf; *t!=EOL; t++)
1812 else if (*t == CONST)
1816 printf("[CONST: $%lX]", *tp.u64);
1819 else if (*t == FCONST)
1823 printf("[FCONST: $%lX]", *tp.u64);
1826 else if (*t == ACONST)
1828 printf("[ACONST: $%X, $%X]", (uint32_t)t[1], (uint32_t)t[2]);
1831 else if (*t == STRING)
1834 printf("[STRING:\"%s\"]", string[*t]);
1836 else if (*t == SYMBOL)
1839 printf("[SYMBOL:\"%s\"]", string[*t]);
1843 else if (*t == TKEOF)
1845 else if (*t == DEQUALS)
1846 printf("[DEQUALS]");
1851 else if (*t == DCOLON)
1863 else if (*t == UNMINUS)
1864 printf("[UNMINUS]");
1865 else if (*t == DOTB)
1867 else if (*t == DOTW)
1869 else if (*t == DOTL)
1871 else if (*t == DOTQ)
1873 else if (*t == DOTS)
1875 else if (*t == DOTD)
1877 else if (*t == DOTI)
1879 else if (*t == ENDEXPR)
1880 printf("[ENDEXPR]");
1881 else if (*t == CR_ABSCOUNT)
1882 printf("[CR_ABSCOUNT]");
1883 else if (*t == CR_FILESIZE)
1884 printf("[CR_FILESIZE]");
1885 else if (*t == CR_DEFINED)
1886 printf("[CR_DEFINED]");
1887 else if (*t == CR_REFERENCED)
1888 printf("[CR_REFERENCED]");
1889 else if (*t == CR_STREQ)
1890 printf("[CR_STREQ]");
1891 else if (*t == CR_MACDEF)
1892 printf("[CR_MACDEF]");
1893 else if (*t == CR_TIME)
1894 printf("[CR_TIME]");
1895 else if (*t == CR_DATE)
1896 printf("[CR_DATE]");
1897 else if (*t >= 0x20 && *t <= 0x2F)
1898 printf("[%c]", (char)*t);
1899 else if (*t >= 0x3A && *t <= 0x3F)
1900 printf("[%c]", (char)*t);
1901 else if (*t >= 0x80 && *t <= 0x87)
1902 printf("[D%u]", ((uint32_t)*t) - 0x80);
1903 else if (*t >= 0x88 && *t <= 0x8F)
1904 printf("[A%u]", ((uint32_t)*t) - 0x88);
1906 printf("[%X:%c]", (uint32_t)*t, (char)*t);