2 // RMAC - Reboot's Macro Assembler for all Atari computers
3 // TOKEN.C - Token Handling
4 // Copyright (C) 199x Landon Dyer, 2011-2018 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 uint16_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
43 // File record, used to maintain a list of every include file ever visited
44 #define FILEREC struct _filerec
54 INOBJ * cur_inobj; // Ptr current input obj (IFILE/IMACRO)
55 static INOBJ * f_inobj; // Ptr list of free INOBJs
56 static IFILE * f_ifile; // Ptr list of free IFILEs
57 static IMACRO * f_imacro; // Ptr list of free IMACROs
59 static TOKEN tokbuf[TOKBUFSIZE]; // Token buffer (stack-like, all files)
61 uint8_t chrtab[0x100] = {
62 ILLEG, ILLEG, ILLEG, ILLEG, // NUL SOH STX ETX
63 ILLEG, ILLEG, ILLEG, ILLEG, // EOT ENQ ACK BEL
64 ILLEG, WHITE, ILLEG, ILLEG, // BS HT LF VT
65 WHITE, ILLEG, ILLEG, ILLEG, // FF CR SO SI
67 ILLEG, ILLEG, ILLEG, ILLEG, // DLE DC1 DC2 DC3
68 ILLEG, ILLEG, ILLEG, ILLEG, // DC4 NAK SYN ETB
69 ILLEG, ILLEG, ILLEG, ILLEG, // CAN EM SUB ESC
70 ILLEG, ILLEG, ILLEG, ILLEG, // FS GS RS US
72 WHITE, MULTX, MULTX, SELF, // SP ! " #
73 MULTX+CTSYM, MULTX, SELF, MULTX, // $ % & '
74 SELF, SELF, SELF, SELF, // ( ) * +
75 SELF, SELF, STSYM, SELF, // , - . /
77 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 0 1
78 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 2 3
79 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 4 5
80 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 6 7
81 DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 8 9
83 MULTX, MULTX, MULTX, STSYM+CTSYM, // < = > ?
85 MULTX, STSYM+CTSYM+HDIGIT, // @ A
86 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // B C
87 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // D E
88 STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // F G
89 STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // H I J K
90 DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // L M N O
92 DOT+STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // P Q R S
93 STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // T U V W
94 STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF,// X Y Z [
95 SELF, SELF, MULTX, STSYM+CTSYM, // \ ] ^ _
97 ILLEG, STSYM+CTSYM+HDIGIT, // ` a
98 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // b c
99 DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // d e
100 STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // f g
101 STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // h i j k
102 DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // l m n o
104 DOT+STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // p q r s
105 STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // t u v w
106 DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // x y z {
107 SELF, SELF, SELF, ILLEG, // | } ~ DEL
109 // Anything above $7F is illegal (and yes, we need to check for this,
110 // otherwise you get strange and spurious errors that will lead you astray)
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,
120 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
121 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
122 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
123 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
124 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
125 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG,
126 ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG
129 // Names of registers
130 static char * regname[] = {
131 "d0","d1","d2","d3","d4","d5","d6","d7", // 128,135
132 "a0","a1","a2","a3","a4","a5","a6","sp", // 136,143
133 "ssp","pc","sr","ccr","regequ","set","reg","r0", // 144,151
134 "r1","r2","r3","r4","r5","r6","r7","r8", // 152,159
135 "r9","r10","r11","r12","r13","r14","r15","r16", // 160,167
136 "r17","r18","r19","r20","r21","r22","r23","r24", // 168,175
137 "r25","r26","r27","r28","r29","r30","r31","ccdef", // 176,183
138 "usp","ic40","dc40","bc40","sfc","dfc","","vbr", // 184,191
139 "cacr","caar","msp","isp","tc","itt0","itt1","dtt0", // 192,199
140 "dtt1","mmusr","urp","srp","iacr0","iacr1","dacr0","dacr1", // 200,207
141 "tt0","tt1","crp","","","","","", // 208,215
142 "","","","","fpiar","fpsr","fpcr","", // 216,223
143 "fp0","fp1","fp2","fp3","fp4","fp5","fp6","fp7", // 224,231
144 "","","","","","","","", // 232,239
145 "","","","","","","","", // 240,247
146 "","","","","","","","", // 248,255
147 "","","","","x0","x1","y0","y1", // 256,263
148 "","b0","","b2","","b1","a","b", // 264,271
149 "mr","omr","la","lc","ssh","ssl","ss","", // 272,279
150 "n0","n1","n2","n3","n4","n5","n6","n7", // 280,287
151 "m0","m1","m2","m3","m4","m5","m6","m7", // 288,295
152 "","","","","","","l","p", // 296,303
153 "mr","omr","la","lc","ssh","ssl","ss","", // 304,311
154 "a10","b10","x","y","","","ab","ba" // 312,319
157 static char * riscregname[] = {
158 "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
159 "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
160 "r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
161 "r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31"
166 // Initialize tokenizer
168 void InitTokenizer(void)
171 char * htab = "0123456789abcdefABCDEF"; // Hex character table
173 lnsave = 0; // Don't save lines
174 curfname = ""; // No file, empty filename
175 filecount = (WORD)-1;
176 cfileno = (WORD)-1; // cfileno gets bumped to 0
188 // Initialize hex, "dot" and tolower tables
193 tolowertab[i] = (char)i;
196 for(i=0; htab[i]!=EOS; i++)
197 hextab[htab[i]] = (char)((i < 16) ? i : i - 6);
199 for(i='A'; i<='Z'; i++)
200 tolowertab[i] |= 0x20;
202 // These characters are legal immediately after a period
203 dotxtab['b'] = DOTB; // .b .B .s .S
205 //dotxtab['s'] = DOTB;
206 //dotxtab['S'] = DOTB;
207 dotxtab['w'] = DOTW; // .w .W
209 dotxtab['l'] = DOTL; // .l .L
211 dotxtab['i'] = DOTI; // .i .I (WTF is this???)
213 dotxtab['D'] = DOTD; // .d .D (double)
215 dotxtab['S'] = DOTS; // .s .S
217 dotxtab['Q'] = DOTQ; // .q .Q (quad word)
219 dotxtab['X'] = DOTX; // .x .x
221 dotxtab['P'] = DOTP; // .p .P
226 void SetFilenameForErrorReporting(void)
230 // Check for absolute top filename (this should never happen)
233 curfname = "(*top*)";
237 FILEREC * fr = filerec;
239 // Advance to the correct record...
240 while (fr != NULL && fnum != 0)
246 // Check for file # record not found (this should never happen either)
249 curfname = "(*NOT FOUND*)";
253 curfname = fr->frec_name;
258 // Allocate an IFILE or IMACRO
260 INOBJ * a_inobj(int typ)
266 // Allocate and initialize INOBJ first
268 inobj = malloc(sizeof(INOBJ));
272 f_inobj = f_inobj->in_link;
277 case SRC_IFILE: // Alloc and init an IFILE
279 ifile = malloc(sizeof(IFILE));
283 f_ifile = f_ifile->if_link;
286 inobj->inobj.ifile = ifile;
289 case SRC_IMACRO: // Alloc and init an IMACRO
290 if (f_imacro == NULL)
291 imacro = malloc(sizeof(IMACRO));
295 f_imacro = f_imacro->im_link;
298 inobj->inobj.imacro = imacro;
301 case SRC_IREPT: // Alloc and init an IREPT
302 inobj->inobj.irept = malloc(sizeof(IREPT));
303 DEBUG { printf("alloc IREPT\n"); }
307 // Install INOBJ on top of input stack
308 inobj->in_ifent = ifent; // Record .if context on entry
309 inobj->in_type = (WORD)typ;
310 inobj->in_otok = tok;
311 inobj->in_etok = etok;
312 inobj->in_link = cur_inobj;
320 // Perform macro substitution from 'orig' to 'dest'. Return OK or some error.
321 // A macro reference is in one of two forms:
322 // \name <non-name-character>
324 // A doubled backslash (\\) is compressed to a single backslash (\).
325 // Argument definitions have been pre-tokenized, so we have to turn them back
326 // into text. This means that numbers, in particular, become hex, regardless of
327 // their representation when the macro was invoked. This is a hack.
328 // A label may appear at the beginning of the line:
329 // :<name><whitespace>
330 // (the colon must be in the first column). These labels are stripped before
331 // macro expansion takes place.
333 int ExpandMacro(char * src, char * dest, int destsiz)
336 int questmark; // \? for testing argument existence
337 char mname[128]; // Assume max size of a formal arg name
338 char numbuf[20]; // Buffer for text of CONSTs
341 char ** symbolString;
343 DEBUG { printf("ExM: src=\"%s\"\n", src); }
345 IMACRO * imacro = cur_inobj->inobj.imacro;
346 int macnum = (int)(imacro->im_macro->sattr);
348 char * dst = dest; // Next dest slot
349 char * edst = dest + destsiz - 1; // End + 1(?) of dest buffer
351 // Check for (and skip over) any "label" on the line
357 while (*s != EOS && !(chrtab[*s] & WHITE))
361 s++; // Skip first whitespace
364 // Expand the rest of the line
367 // Copy single character
373 // Skip comments in case a loose @ or \ is in there
374 // In that case the tokeniser was trying to expand it.
375 if ((*s == ';') || ((*s == '/') && (*(s + 1) == '/')))
380 // Do macro expansion
388 case '\\': // \\, \ (collapse to single backslash)
394 case '?': // \? <macro> set `questmark' flag
398 case '#': // \#, number of arguments
399 sprintf(numbuf, "%d", (int)imacro->im_nargs);
401 case '!': // \! size suffix supplied on invocation
402 switch ((int)imacro->im_siz)
404 case SIZN: d = ""; break;
405 case SIZB: d = ".b"; break;
406 case SIZW: d = ".w"; break;
407 case SIZL: d = ".l"; break;
411 case '~': // ==> unique label string Mnnnn...
412 sprintf(numbuf, "M%u", curuniq);
428 return error("missing argument name");
431 // \n ==> argument number 'n', 0..9
432 if (chrtab[*s] & DIGIT)
442 // Get argument name: \name, \{name}
452 while (chrtab[*s] & CTSYM);
457 for(++s; *s != EOS && *s != '}';)
461 return error("missing closing brace ('}')");
468 // Lookup the argument and copy its (string) value into the
469 // destination string
470 DEBUG { printf("argument='%s'\n", mname); }
472 if ((arg = lookup(mname, MACARG, macnum)) == NULL)
473 return error("undefined argument: '%s'", mname);
476 // Convert a string of tokens (terminated with EOL) back into
477 // text. If an argument is out of range (not specified in the
478 // macro invocation) then it is ignored.
479 i = (int)arg->svalue;
481 DEBUG { printf("~argnumber=%d\n", i); }
484 if (i < imacro->im_nargs)
486 tk = imacro->argument[i].token;
487 symbolString = imacro->argument[i].string;
490 // printf("ExM: Preparing to parse argument #%u...\n", i);
496 // 0 if the argument is empty or non-existant,
497 // 1 if the argument is not empty
500 if (tk == NULL || *tk == EOL)
506 *dst++ = (char)(questmark + '0');
510 // Argument # is in range, so expand it
515 // Reverse-translation from a token number to a string.
516 // This is a hack. It might be better table-driven.
519 if ((*tk >= KW_D0) && !rdsp && !rgpu)
521 d = regname[(int)*tk++ - KW_D0];
524 else if ((*tk >= KW_R0) && (*tk <= KW_R31))
526 d = riscregname[(int)*tk++ - KW_R0];
534 d = symbolString[*tk++];
535 DEBUG { printf("ExM: SYMBOL=\"%s\"", d); }
538 d = symbolString[*tk++];
559 // Shamus: Changing the format specifier from %lx to %ux caused the assembler
560 // to choke on legitimate code... Need to investigate this further
561 // before changing anything else here!
563 sprintf(numbuf, "$%lx", (uint64_t)*tk++);
628 *dst++ = (char)*(tk - 1);
633 // If 'd' != NULL, copy string to destination
637 DEBUG printf("d='%s'\n", d);
656 DEBUG { printf("ExM: dst=\"%s\"\n", dest); }
661 DEBUG { printf("*** OVERFLOW LINE ***\n%s\n", dest); }
662 return fatal("line too long as a result of macro expansion");
667 // Get next line of text from a macro
669 char * GetNextMacroLine(void)
671 IMACRO * imacro = cur_inobj->inobj.imacro;
672 // LONG * strp = imacro->im_nextln;
673 LLIST * strp = imacro->im_nextln;
675 if (strp == NULL) // End-of-macro
678 imacro->im_nextln = strp->next;
679 // ExpandMacro((char *)(strp + 1), imacro->im_lnbuf, LNSIZ);
680 ExpandMacro(strp->line, imacro->im_lnbuf, LNSIZ);
682 return imacro->im_lnbuf;
687 // Get next line of text from a repeat block
689 char * GetNextRepeatLine(void)
691 IREPT * irept = cur_inobj->inobj.irept;
692 // LONG * strp = irept->ir_nextln; // initial null
694 // Do repeat at end of .rept block's string list
696 if (irept->ir_nextln == NULL)
698 DEBUG { printf("back-to-top-of-repeat-block count=%d\n", (int)irept->ir_count); }
699 irept->ir_nextln = irept->ir_firstln; // copy first line
701 if (irept->ir_count-- == 0)
703 DEBUG { printf("end-repeat-block\n"); }
707 // strp = irept->ir_nextln;
709 // Mark the current macro line in the irept object
710 // This is probably overkill - a global variable
711 // would suffice here (it only gets used during
712 // error reporting anyway)
713 irept->lineno = irept->ir_nextln->lineno;
715 // strcpy(irbuf, (char *)(irept->ir_nextln + 1));
716 strcpy(irbuf, irept->ir_nextln->line);
717 DEBUG { printf("repeat line='%s'\n", irbuf); }
718 // irept->ir_nextln = (LONG *)*strp;
719 irept->ir_nextln = irept->ir_nextln->next;
726 // Include a source file used at the root, and for ".include" files
728 int include(int handle, char * fname)
731 DEBUG { printf("[include: %s, cfileno=%u]\n", fname, cfileno); }
733 // Alloc and initialize include-descriptors
734 INOBJ * inobj = a_inobj(SRC_IFILE);
735 IFILE * ifile = inobj->inobj.ifile;
737 ifile->ifhandle = handle; // Setup file handle
738 ifile->ifind = ifile->ifcnt = 0; // Setup buffer indices
739 ifile->ifoldlineno = curlineno; // Save old line number
740 ifile->ifoldfname = curfname; // Save old filename
741 ifile->ifno = cfileno; // Save old file number
743 // NB: This *must* be preincrement, we're adding one to the filecount here!
744 cfileno = ++filecount; // Compute NEW file number
745 curfname = strdup(fname); // Set current filename (alloc storage)
746 curlineno = 0; // Start on line zero
748 // Add another file to the file-record
749 FILEREC * fr = (FILEREC *)malloc(sizeof(FILEREC));
750 fr->frec_next = NULL;
751 fr->frec_name = curfname;
754 filerec = fr; // Add first filerec
756 last_fr->frec_next = fr; // Append to list of filerecs
759 DEBUG { printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno); }
766 // Pop the current input level
770 INOBJ * inobj = cur_inobj;
775 // Pop IFENT levels until we reach the conditional assembly context we
776 // were at when the input object was entered.
777 int numUnmatched = 0;
779 while (ifent != inobj->in_ifent)
781 if (d_endif() != 0) // Something bad happened during endif parsing?
782 return -1; // If yes, bail instead of getting stuck in a loop
787 // Give a warning to the user that we had to wipe their bum for them
788 if (numUnmatched > 0)
789 warn("missing %d .endif(s)", numUnmatched);
791 tok = inobj->in_otok; // Restore tok and otok
792 etok = inobj->in_etok;
794 switch (inobj->in_type)
796 case SRC_IFILE: // Pop and release an IFILE
798 DEBUG { printf("[Leaving: %s]\n", curfname); }
800 IFILE * ifile = inobj->inobj.ifile;
801 ifile->if_link = f_ifile;
803 close(ifile->ifhandle); // Close source file
804 DEBUG { printf("[fpop (pre): curfname=%s]\n", curfname); }
805 curfname = ifile->ifoldfname; // Set current filename
806 DEBUG { printf("[fpop (post): curfname=%s]\n", curfname); }
807 DEBUG { printf("[fpop: (pre) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); }
808 curlineno = ifile->ifoldlineno; // Set current line#
809 DEBUG { printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno); }
810 cfileno = ifile->ifno; // Restore current file number
811 DEBUG { printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); }
815 case SRC_IMACRO: // Pop and release an IMACRO
817 IMACRO * imacro = inobj->inobj.imacro;
818 imacro->im_link = f_imacro;
823 case SRC_IREPT: // Pop and release an IREPT
825 DEBUG { printf("dealloc IREPT\n"); }
826 LLIST * p = inobj->inobj.irept->ir_firstln;
828 // Deallocate repeat lines
839 cur_inobj = inobj->in_link;
840 inobj->in_link = f_inobj;
848 // Get line from file into buf, return NULL on EOF or ptr to the start of a
851 char * GetNextLine(void)
855 int readamt = -1; // 0 if last read() yeilded 0 bytes
856 IFILE * fl = cur_inobj->inobj.ifile;
860 // Scan for next end-of-line; handle stupid text formats by treating
861 // \r\n the same as \n. (lone '\r' at end of buffer means we have to
863 d = &fl->ifbuf[fl->ifind];
865 for(p=d, i=0, j=fl->ifcnt; i<j; i++, p++)
867 if (*p == '\r' || *p == '\n')
874 break; // Need to read more, then look for '\n' to eat
875 else if (p[1] == '\n')
879 // Cover up the newline with end-of-string sentinel
888 // Handle hanging lines by ignoring them (Input file is exhausted, no
889 // \r or \n on last line)
890 // Shamus: This is retarded. Never ignore any input!
891 if (!readamt && fl->ifcnt)
898 // Really should check to see if we're at the end of the buffer!
900 fl->ifbuf[fl->ifind + fl->ifcnt] = '\0';
902 return &fl->ifbuf[fl->ifind];
906 // Truncate and return absurdly long lines.
907 if (fl->ifcnt >= QUANTUM)
909 fl->ifbuf[fl->ifind + fl->ifcnt - 1] = '\0';
911 return &fl->ifbuf[fl->ifind];
914 // Relocate what's left of a line to the beginning of the buffer, and
915 // read some more of the file in; return NULL if the buffer's empty and
919 p = &fl->ifbuf[fl->ifind];
920 d = &fl->ifbuf[fl->ifcnt & 1];
922 for(i=0; i<fl->ifcnt; i++)
925 fl->ifind = fl->ifcnt & 1;
928 readamt = read(fl->ifhandle, &fl->ifbuf[fl->ifind + fl->ifcnt], QUANTUM);
933 if ((fl->ifcnt += readamt) == 0)
942 int TokenizeLine(void)
944 uint8_t * ln = NULL; // Ptr to current position in line
945 uint8_t * p; // Random character ptr
946 PTR tk; // Token-deposit ptr
947 int state = 0; // State for keyword detector
948 int j = 0; // Var for keyword detector
949 uint8_t c; // Random char
950 uint64_t v; // Random value
951 uint32_t cursize = 0; // Current line's size (.b, .w, .l, .s, .q, .d)
952 double f; // Random float
953 uint8_t * nullspot = NULL; // Spot to clobber for SYMBOL termination
954 int stuffnull; // 1:terminate SYMBOL '\0' at *nullspot
956 int stringNum = 0; // Pointer to string locations in tokenized line
960 if (cur_inobj == NULL) // Return EOF if input stack is empty
963 // Get another line of input from the current input source: a file, a
964 // macro, or a repeat-block
965 switch (cur_inobj->in_type)
969 // o bump source line number;
970 // o tag the listing-line with a space;
971 // o kludge lines generated by Alcyon C.
973 if ((ln = GetNextLine()) == NULL)
975 DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); }
976 if (fpop() == 0) // Pop input level
977 goto retry; // Try for more lines
980 ifent->if_prev = (IFENT *)-1; //Signal Assemble() that we have reached EOF with unbalanced if/endifs
985 curlineno++; // Bump line number
990 // AS68 compatibility, throw away all lines starting with
991 // back-quotes, tildes, or '*'
992 // On other lines, turn the first '*' into a semi-colon.
993 if (*ln == '`' || *ln == '~' || *ln == '*')
997 for(p=ln; *p!=EOS; p++)
1011 // o Handle end-of-macro;
1012 // o tag the listing-line with an at (@) sign.
1014 if ((ln = GetNextMacroLine()) == NULL)
1016 if (ExitMacro() == 0) // Exit macro (pop args, do fpop(), etc)
1017 goto retry; // Try for more lines...
1019 return TKEOF; // Oops, we got a non zero return code, signal EOF
1026 // o Handle end-of-repeat-block;
1027 // o tag the listing-line with a pound (#) sign.
1029 if ((ln = GetNextRepeatLine()) == NULL)
1031 DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); }
1040 // Save text of the line. We only do this during listings and within
1041 // macro-type blocks, since it is expensive to unconditionally copy every
1046 // General housekeeping
1047 tok = tokeol; // Set "tok" to EOL in case of error
1048 tk.u32 = etok; // Reset token ptr
1049 stuffnull = 0; // Don't stuff nulls
1050 totlines++; // Bump total #lines assembled
1052 // See if the entire line is a comment. This is a win if the programmer
1053 // puts in lots of comments
1054 if (*ln == '*' || *ln == ';' || ((*ln == '/') && (*(ln + 1) == '/')))
1057 // And here we have a very ugly hack for signalling a single line 'turn off
1058 // optimization'. There's really no nice way to do this, so hack it is!
1059 optimizeOff = 0; // Default is to take optimizations as they come
1063 optimizeOff = 1; // Signal that we don't want to optimize this line
1064 ln++; // & skip over the darned thing
1067 // Main tokenization loop;
1068 // o skip whitespace;
1069 // o handle end-of-line;
1070 // o handle symbols;
1071 // o handle single-character tokens (operators, etc.);
1072 // o handle multiple-character tokens (constants, strings, etc.).
1075 // Skip whitespace, handle EOL
1076 while (chrtab[*ln] & WHITE)
1079 // Handle EOL, comment with ';'
1080 if (*ln == EOS || *ln == ';'|| ((*ln == '/') && (*(ln + 1) == '/')))
1083 // Handle start of symbol. Symbols are null-terminated in place. The
1084 // termination is always one symbol behind, since there may be no place
1085 // for a null in the case that an operator immediately follows the name.
1090 if (stuffnull) // Terminate old symbol from previous pass
1093 v = 0; // Assume no DOT attrib follows symbol
1096 // In some cases, we need to check for a DOTx at the *beginning*
1097 // of a symbol, as the "start" of the line we're currently looking
1098 // at could be somewhere in the middle of that line!
1101 // Make sure that it's *only* a .[bwsl] following, and not the
1102 // start of a local symbol:
1103 if ((chrtab[*(ln + 1)] & DOT)
1104 && (dotxtab[*(ln + 1)] != 0)
1105 && !(chrtab[*(ln + 2)] & CTSYM))
1107 // We found a legitimate DOTx construct, so add it to the
1111 *tk.u32++ = (TOKEN)dotxtab[*ln++];
1116 p = nullspot = ln++; // Nullspot -> start of this symbol
1118 // Find end of symbol (and compute its length)
1119 for(j=1; (int)chrtab[*ln]&CTSYM; j++)
1122 // Handle "DOT" special forms (like ".b") that follow a normal
1123 // symbol or keyword:
1126 *ln++ = EOS; // Terminate symbol
1127 stuffnull = 0; // And never try it again
1129 // Character following the '.' must have a DOT attribute, and
1130 // the chararacter after THAT one must not have a start-symbol
1131 // attribute (to prevent symbols that look like, for example,
1132 // "zingo.barf", which might be a good idea anyway....)
1133 if (((chrtab[*ln] & DOT) == 0) || (dotxtab[*ln] == 0))
1134 return error("[bwsl] must follow '.' in symbol");
1136 v = (uint32_t)dotxtab[*ln++];
1137 cursize = (uint32_t)v;
1139 if (chrtab[*ln] & CTSYM)
1140 return error("misuse of '.'; not allowed in symbols");
1143 // If the symbol is small, check to see if it's really the name of
1147 for(state=0; state>=0;)
1149 j = (int)tolowertab[*p++];
1152 if (kwcheck[j] != state)
1158 if (*p == EOS || p == ln)
1172 // Make j = -1 if user tries to use a RISC register while in 68K mode
1173 if (!(rgpu || rdsp) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31))
1178 // Make j = -1 if time, date etc with no preceeding ^^
1179 // defined, referenced, streq, macdef, date and time
1182 case 112: // defined
1183 case 113: // referenced
1191 // If not tokenized keyword OR token was not found
1192 if ((j < 0) || (state < 0))
1195 string[stringNum] = nullspot;
1196 *tk.u32++ = stringNum;
1201 *tk.u32++ = (TOKEN)j;
1205 if (v) // Record attribute token (if any)
1206 *tk.u32++ = (TOKEN)v;
1208 if (stuffnull) // Arrange for string termination on next pass
1214 // Handle identity tokens
1221 // Handle multiple-character tokens
1226 case '!': // ! or !=
1236 case '\'': // 'string'
1239 // Hardcoded for now, maybe this will change in the future
1240 *tk.u32++ = STRINGA8;
1244 case '\"': // "string"
1248 string[stringNum] = ln;
1249 *tk.u32++ = stringNum;
1252 for(p=ln; *ln!=EOS && *ln!=c1;)
1261 return(error("unterminated string"));
1290 // If we're evaluating a macro
1291 // this is valid and expands to
1295 warn("bad backslash code in string");
1305 return error("unterminated string");
1309 case '$': // $, hex constant
1310 if (chrtab[*ln] & HDIGIT)
1314 // Parse the hex value
1315 while (hextab[*ln] >= 0)
1316 v = (v << 4) + (int)hextab[*ln++];
1323 if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1328 else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1339 case '<': // < or << or <> or <=
1358 case ':': // : or ::
1368 case '=': // = or ==
1371 *tk.u32++ = DEQUALS;
1378 case '>': // > or >> or >=
1393 case '%': // % or binary constant
1394 if (*ln < '0' || *ln > '1')
1402 while (*ln >= '0' && *ln <= '1')
1403 v = (v << 1) + *ln++ - '0';
1407 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1413 if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1419 if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1429 case '@': // @ or octal constant
1430 if (*ln < '0' || *ln > '7')
1438 while (*ln >= '0' && *ln <= '7')
1439 v = (v << 3) + *ln++ - '0';
1443 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1449 if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1455 if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1465 case '^': // ^ or ^^ <operator-name>
1472 if (((int)chrtab[*++ln] & STSYM) == 0)
1474 error("invalid symbol following ^^");
1480 while ((int)chrtab[*ln] & CTSYM)
1483 for(state=0; state>=0;)
1485 // Get char, convert to lowercase
1488 if (j >= 'A' && j <= 'Z')
1493 if (kwcheck[j] != state)
1499 if (*p == EOS || p == ln)
1508 if (j < 0 || state < 0)
1510 error("unknown symbol following ^^");
1514 *tk.u32++ = (TOKEN)j;
1517 interror(2); // Bad MULTX entry in chrtab
1522 // Handle decimal constant
1525 uint8_t * numStart = ln;
1528 while ((int)chrtab[*ln] & DIGIT)
1529 v = (v * 10) + *ln++ - '0';
1531 // See if there's a .[bwl] after the constant & deal with it if so
1534 if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
1542 else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
1550 else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
1558 else if ((int)chrtab[*(ln + 1)] & DIGIT)
1560 // Hey, more digits after the dot, so we assume it's a
1561 // floating point number of some kind... numEnd will point
1562 // to the first non-float character after it's done
1565 double f = strtod(numStart, &numEnd);
1566 ln = (uint8_t *)numEnd;
1569 return error("floating point parse error");
1571 // N.B.: We use the C compiler's internal double
1572 // representation for all internal float calcs and
1573 // are reasonably sure that the size of said double
1574 // is 8 bytes long (which we check for in fltpoint.c)
1587 //printf("CONST: %i\n", v);
1591 // Handle illegal character
1592 return error("illegal character $%02X found", *ln);
1595 // Terminate line of tokens and return "success."
1598 tok = etok; // Set tok to beginning of line
1600 if (stuffnull) // Terminate last SYMBOL
1610 // .GOTO <label> goto directive
1612 // The label is searched for starting from the first line of the current,
1613 // enclosing macro definition. If no enclosing macro exists, an error is
1616 // A label is of the form:
1618 // :<name><whitespace>
1620 // The colon must appear in column 1. The label is stripped prior to macro
1621 // expansion, and is NOT subject to macro expansion. The whitespace may also
1624 int d_goto(WORD unused)
1626 // Setup for the search
1628 return error("missing label");
1630 char * sym = string[tok[1]];
1633 if (cur_inobj->in_type != SRC_IMACRO)
1634 return error("goto not in macro");
1636 IMACRO * imacro = cur_inobj->inobj.imacro;
1637 LLIST * defln = imacro->im_macro->lineList;
1639 // Attempt to find the label, starting with the first line.
1640 for(; defln!=NULL; defln=defln->next)
1642 // Must start with a colon
1643 if (defln->line[0] == ':')
1645 // Compare names (sleazo string compare)
1647 char * s2 = defln->line;
1649 // Either we will match the strings to EOS on both, or we will
1650 // match EOS on string 1 to whitespace on string 2. Otherwise, we
1652 while ((*s1 == *s2) || ((*s1 == EOS) && (chrtab[*s2] & WHITE)))
1654 // If we reached the end of string 1 (sym), we're done.
1655 // Note that we're also checking for the end of string 2 as
1656 // well, since we've established they're equal above.
1659 // Found the label, set new macro next-line and return.
1660 imacro->im_nextln = defln;
1670 return error("goto label not found");
1674 void DumpToken(TOKEN t)
1678 else if (t == CONST)
1680 else if (t == FCONST)
1682 else if (t == ACONST)
1684 else if (t == STRING)
1686 else if (t == SYMBOL)
1690 else if (t == TKEOF)
1692 else if (t == DEQUALS)
1693 printf("[DEQUALS]");
1698 else if (t == DCOLON)
1710 else if (t == UNMINUS)
1711 printf("[UNMINUS]");
1726 else if (t == ENDEXPR)
1727 printf("[ENDEXPR]");
1728 else if (t == CR_ABSCOUNT)
1729 printf("[CR_ABSCOUNT]");
1730 else if (t == CR_DEFINED)
1731 printf("[CR_DEFINED]");
1732 else if (t == CR_REFERENCED)
1733 printf("[CR_REFERENCED]");
1734 else if (t == CR_STREQ)
1735 printf("[CR_STREQ]");
1736 else if (t == CR_MACDEF)
1737 printf("[CR_MACDEF]");
1738 else if (t == CR_TIME)
1739 printf("[CR_TIME]");
1740 else if (t == CR_DATE)
1741 printf("[CR_DATE]");
1742 else if (t >= 0x20 && t <= 0x2F)
1743 printf("[%c]", (char)t);
1744 else if (t >= 0x3A && t <= 0x3F)
1745 printf("[%c]", (char)t);
1746 else if (t >= 0x80 && t <= 0x87)
1747 printf("[D%u]", ((uint32_t)t) - 0x80);
1748 else if (t >= 0x88 && t <= 0x8F)
1749 printf("[A%u]", ((uint32_t)t) - 0x88);
1751 printf("[%X:%c]", (uint32_t)t, (char)t);
1755 void DumpTokenBuffer(void)
1757 printf("Tokens [%X]: ", sloc);
1759 for(TOKEN * t=tokbuf; *t!=EOL; t++)
1763 else if (*t == CONST)
1767 printf("[CONST: $%lX]", *tp.u64);
1770 else if (*t == FCONST)
1774 printf("[FCONST: $%lX]", *tp.u64);
1777 else if (*t == ACONST)
1779 printf("[ACONST: $%X, $%X]", (uint32_t)t[1], (uint32_t)t[2]);
1782 else if (*t == STRING)
1785 printf("[STRING:\"%s\"]", string[*t]);
1787 else if (*t == SYMBOL)
1790 printf("[SYMBOL:\"%s\"]", string[*t]);
1794 else if (*t == TKEOF)
1796 else if (*t == DEQUALS)
1797 printf("[DEQUALS]");
1802 else if (*t == DCOLON)
1814 else if (*t == UNMINUS)
1815 printf("[UNMINUS]");
1816 else if (*t == DOTB)
1818 else if (*t == DOTW)
1820 else if (*t == DOTL)
1822 else if (*t == DOTQ)
1824 else if (*t == DOTS)
1826 else if (*t == DOTD)
1828 else if (*t == DOTI)
1830 else if (*t == ENDEXPR)
1831 printf("[ENDEXPR]");
1832 else if (*t == CR_ABSCOUNT)
1833 printf("[CR_ABSCOUNT]");
1834 else if (*t == CR_DEFINED)
1835 printf("[CR_DEFINED]");
1836 else if (*t == CR_REFERENCED)
1837 printf("[CR_REFERENCED]");
1838 else if (*t == CR_STREQ)
1839 printf("[CR_STREQ]");
1840 else if (*t == CR_MACDEF)
1841 printf("[CR_MACDEF]");
1842 else if (*t == CR_TIME)
1843 printf("[CR_TIME]");
1844 else if (*t == CR_DATE)
1845 printf("[CR_DATE]");
1846 else if (*t >= 0x20 && *t <= 0x2F)
1847 printf("[%c]", (char)*t);
1848 else if (*t >= 0x3A && *t <= 0x3F)
1849 printf("[%c]", (char)*t);
1850 else if (*t >= 0x80 && *t <= 0x87)
1851 printf("[D%u]", ((uint32_t)*t) - 0x80);
1852 else if (*t >= 0x88 && *t <= 0x8F)
1853 printf("[A%u]", ((uint32_t)*t) - 0x88);
1855 printf("[%X:%c]", (uint32_t)*t, (char)*t);