//
-// RMAC - Reboot's Macro Assembler for all Atari computers
+// RMAC - Renamed Macro Assembler for all Atari computers
// TOKEN.C - Token Handling
-// Copyright (C) 199x Landon Dyer, 2011-2017 Reboot and Friends
+// Copyright (C) 199x Landon Dyer, 2011-2021 Reboot and Friends
// RMAC derived from MADMAC v1.07 Written by Landon Dyer, 1986
// Source utilised with the kind permission of Landon Dyer
//
#include "token.h"
+
+#include <errno.h>
#include "direct.h"
#include "error.h"
#include "macro.h"
int lnsave; // 1; strcpy() text of current line
-uint16_t curlineno; // Current line number (64K max currently)
+uint32_t curlineno; // Current line number (64K max currently)
int totlines; // Total # of lines
int mjump_align = 0; // mjump alignment flag
char lntag; // Line tag
TOKEN * tok; // Ptr to current token
TOKEN * etok; // Ptr past last token in tokbuf[]
TOKEN tokeol[1] = {EOL}; // Bailout end-of-line token
-char * string[TOKBUFSIZE*2]; // Token buffer string pointer storage
+char * string[TOKBUFSIZE*2];// Token buffer string pointer storage
int optimizeOff; // Optimization override flag
-// File record, used to maintain a list of every include file ever visited
-#define FILEREC struct _filerec
-FILEREC
-{
- FILEREC * frec_next;
- char * frec_name;
-};
FILEREC * filerec;
FILEREC * last_fr;
dotxtab['W'] = DOTW;
dotxtab['l'] = DOTL; // .l .L
dotxtab['L'] = DOTL;
- dotxtab['i'] = DOTI; // .i .I (???)
+ dotxtab['i'] = DOTI; // .i .I (WTF is this???)
dotxtab['I'] = DOTI;
- dotxtab['D'] = DOTD; // .d .D (quad word)
+ dotxtab['D'] = DOTD; // .d .D (double)
dotxtab['d'] = DOTD;
dotxtab['S'] = DOTS; // .s .S
dotxtab['s'] = DOTS;
- dotxtab['Q'] = DOTQ; // .q .Q
+ dotxtab['Q'] = DOTQ; // .q .Q (quad word)
dotxtab['q'] = DOTQ;
dotxtab['X'] = DOTX; // .x .x
dotxtab['x'] = DOTX;
*dst++ = *s++;
continue;
case '?': // \? <macro> set `questmark' flag
- ++s;
+ s++;
questmark = 1;
break;
case '#': // \#, number of arguments
switch ((int)*tk++)
{
case SYMBOL:
-#if 0
-// d = (char *)*tk++;
- d = string[*tk++];
-#else
- // This fix should be done for strings too
d = symbolString[*tk++];
DEBUG { printf("ExM: SYMBOL=\"%s\"", d); }
-#endif
break;
case STRING:
-#if 0
-// d = (char *)*tk++;
- d = string[*tk++];
-#else
d = symbolString[*tk++];
-#endif
+
if (dst >= edst)
goto overflow;
// to choke on legitimate code... Need to investigate this further
// before changing anything else here!
case CONST:
- sprintf(numbuf, "$%lx", (long unsigned int)*tk++);
+// sprintf(numbuf, "$%lx", (uint64_t)*tk++);
+ sprintf(numbuf, "$%" PRIX64, (uint64_t)*tk++);
+ tk++;
d = numbuf;
break;
case DEQUALS:
case CR_ABSCOUNT:
d = "^^abscount";
break;
+ case CR_FILESIZE:
+ d = "^^filesize";
+ break;
case CR_DATE:
d = "^^date";
break;
char * GetNextMacroLine(void)
{
IMACRO * imacro = cur_inobj->inobj.imacro;
-// LONG * strp = imacro->im_nextln;
LLIST * strp = imacro->im_nextln;
if (strp == NULL) // End-of-macro
DEBUG { printf("end-repeat-block\n"); }
return NULL;
}
-
+ reptuniq++;
// strp = irept->ir_nextln;
}
+ // Mark the current macro line in the irept object
+ // This is probably overkill - a global variable
+ // would suffice here (it only gets used during
+ // error reporting anyway)
+ irept->lineno = irept->ir_nextln->lineno;
+
+ // Copy the rept lines verbatim, unless we're in nest level 0.
+ // Then, expand any \~ labels to unique numbers (Rn)
+ if (rptlevel)
+ {
+ strcpy(irbuf, irept->ir_nextln->line);
+ }
+ else
+ {
+ uint32_t linelen = strlen(irept->ir_nextln->line);
+ uint8_t *p_line = irept->ir_nextln->line;
+ char *irbufwrite = irbuf;
+ for (int i = 0; i <= linelen; i++)
+ {
+ uint8_t c;
+ c = *p_line++;
+ if (c == '\\' && *p_line == '~')
+ {
+ p_line++;
+ irbufwrite += sprintf(irbufwrite, "R%u", reptuniq);
+ }
+ else
+ {
+ *irbufwrite++ = c;
+ }
+ }
+ }
-// strcpy(irbuf, (char *)(irept->ir_nextln + 1));
- strcpy(irbuf, irept->ir_nextln->line);
DEBUG { printf("repeat line='%s'\n", irbuf); }
// irept->ir_nextln = (LONG *)*strp;
irept->ir_nextln = irept->ir_nextln->next;
if (numUnmatched > 0)
warn("missing %d .endif(s)", numUnmatched);
- tok = inobj->in_otok; // Restore tok and otok
+ tok = inobj->in_otok; // Restore tok and etok
etok = inobj->in_etok;
switch (inobj->in_type)
{
uint8_t * ln = NULL; // Ptr to current position in line
uint8_t * p; // Random character ptr
- TOKEN * tk; // Token-deposit ptr
+ PTR tk; // Token-deposit ptr
int state = 0; // State for keyword detector
int j = 0; // Var for keyword detector
uint8_t c; // Random char
- VALUE v; // Random value
+ uint64_t v; // Random value
+ uint32_t cursize = 0; // Current line's size (.b, .w, .l, .s, .q, .d)
uint8_t * nullspot = NULL; // Spot to clobber for SYMBOL termination
int stuffnull; // 1:terminate SYMBOL '\0' at *nullspot
uint8_t c1;
retry:
- if (cur_inobj == NULL) // Return EOF if input stack is empty
+ if (cur_inobj == NULL) // Return EOF if input stack is empty
return TKEOF;
// Get another line of input from the current input source: a file, a
if ((ln = GetNextLine()) == NULL)
{
DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); }
- if (fpop() == 0) // Pop input level
- goto retry; // Try for more lines
+ if (fpop() == 0) // Pop input level
+ goto retry; // Try for more lines
else
{
ifent->if_prev = (IFENT *)-1; //Signal Assemble() that we have reached EOF with unbalanced if/endifs
}
}
- curlineno++; // Bump line number
+ curlineno++; // Bump line number
lntag = SPACE;
- if (as68_flag)
- {
- // AS68 compatibility, throw away all lines starting with
- // back-quotes, tildes, or '*'
- // On other lines, turn the first '*' into a semi-colon.
- if (*ln == '`' || *ln == '~' || *ln == '*')
- *ln = ';';
- else
- {
- for(p=ln; *p!=EOS; p++)
- {
- if (*p == '*')
- {
- *p = ';';
- break;
- }
- }
- }
- }
-
break;
// Macro-block:
// macro-type blocks, since it is expensive to unconditionally copy every
// line.
if (lnsave)
+ {
+ // Sanity check
+ if (strlen(ln) > LNSIZ)
+ return error("line too long (%d, max %d)", strlen(ln), LNSIZ);
+
strcpy(lnbuf, ln);
+ }
// General housekeeping
tok = tokeol; // Set "tok" to EOL in case of error
- tk = etok; // Reset token ptr
+ tk.u32 = etok; // Reset token ptr
stuffnull = 0; // Don't stuff nulls
totlines++; // Bump total #lines assembled
// o handle multiple-character tokens (constants, strings, etc.).
for(; *ln!=EOS;)
{
+ // Check to see if there's enough space in the token buffer
+ if (tk.cp >= ((uint8_t *)(&tokbuf[TOKBUFSIZE])) - 20)
+ {
+ return error("token buffer overrun");
+ }
+
// Skip whitespace, handle EOL
while (chrtab[*ln] & WHITE)
ln++;
// token stream:
ln++;
stuffnull = 0;
- *tk++ = (TOKEN)dotxtab[*ln++];
+ *tk.u32++ = (TOKEN)dotxtab[*ln++];
continue;
}
}
*ln++ = EOS; // Terminate symbol
stuffnull = 0; // And never try it again
- // Character following the `.' must have a DOT attribute, and
+ // Character following the '.' must have a DOT attribute, and
// the chararacter after THAT one must not have a start-symbol
// attribute (to prevent symbols that look like, for example,
// "zingo.barf", which might be a good idea anyway....)
if (((chrtab[*ln] & DOT) == 0) || (dotxtab[*ln] == 0))
return error("[bwsl] must follow '.' in symbol");
- v = (VALUE)dotxtab[*ln++];
+ v = (uint32_t)dotxtab[*ln++];
+ cursize = (uint32_t)v;
if (chrtab[*ln] & CTSYM)
return error("misuse of '.'; not allowed in symbols");
}
// Make j = -1 if user tries to use a RISC register while in 68K mode
- if (!(rgpu || rdsp) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31))
+ if (!(rgpu || rdsp || dsp56001) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31))
{
j = -1;
}
// If not tokenized keyword OR token was not found
if ((j < 0) || (state < 0))
{
- *tk++ = SYMBOL;
-//#warning
-//problem here: nullspot is a char * but TOKEN is a uint32_t. On a 64-bit
-//system, this will cause all kinds of mischief.
-#if 0
- *tk++ = (TOKEN)nullspot;
-#else
+ *tk.u32++ = SYMBOL;
string[stringNum] = nullspot;
- *tk++ = stringNum;
+ *tk.u32++ = stringNum;
stringNum++;
-#endif
}
else
{
- *tk++ = (TOKEN)j;
+ *tk.u32++ = (TOKEN)j;
stuffnull = 0;
}
- if (v) // Record attribute token (if any)
- *tk++ = (TOKEN)v;
+ if (v) // Record attribute token (if any)
+ *tk.u32++ = (TOKEN)v;
- if (stuffnull) // Arrange for string termination on next pass
+ if (stuffnull) // Arrange for string termination on next pass
nullspot = ln;
continue;
// Handle identity tokens
if (c & SELF)
{
- *tk++ = *ln++;
+ *tk.u32++ = *ln++;
continue;
}
case '!': // ! or !=
if (*ln == '=')
{
- *tk++ = NE;
- ++ln;
+ *tk.u32++ = NE;
+ ln++;
}
else
- *tk++ = '!';
+ *tk.u32++ = '!';
continue;
case '\'': // 'string'
if (m6502)
{
// Hardcoded for now, maybe this will change in the future
- *tk++ = STRINGA8;
+ *tk.u32++ = STRINGA8;
goto dostring;
}
// Fall through
case '\"': // "string"
- *tk++ = STRING;
+ *tk.u32++ = STRING;
dostring:
c1 = ln[-1];
string[stringNum] = ln;
- *tk++ = stringNum;
+ *tk.u32++ = stringNum;
stringNum++;
for(p=ln; *ln!=EOS && *ln!=c1;)
case '\\':
c = '\\';
break;
+ case '{':
+ // If we're evaluating a macro
+ // this is valid because it's
+ // a parameter expansion
+ case '!':
+ // If we're evaluating a macro
+ // this is valid and expands to
+ // "dot-size"
+ break;
default:
warn("bad backslash code in string");
ln--;
while (hextab[*ln] >= 0)
v = (v << 4) + (int)hextab[*ln++];
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
+
if (*ln == '.')
{
- if (obj_format == BSD)
+ if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
{
- if ((*(ln + 1) & 0xDF) == 'B')
- {
- v &= 0x000000FF;
- ln += 2;
- }
- else if ((*(ln + 1) & 0xDF) == 'W')
- {
- v &= 0x0000FFFF;
- ln += 2;
- }
- else if ((*(ln + 1) & 0xDF) == 'L')
- {
- ln += 2;
- }
+ *tk.u32++ = DOTW;
+ ln += 2;
}
- }
-
- *tk++ = CONST;
- *tk++ = v;
-
- if (obj_format == ALCYON)
- {
- if (*ln == '.')
+ else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
{
- if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
- {
- *tk++ = DOTW;
- ln += 2;
- }
- else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
- {
- *tk++ = DOTL;
- ln += 2;
- }
+ *tk.u32++ = DOTL;
+ ln += 2;
}
}
}
else
- *tk++ = '$';
+ *tk.u32++ = '$';
continue;
case '<': // < or << or <> or <=
switch (*ln)
{
case '<':
- *tk++ = SHL;
- ++ln;
+ *tk.u32++ = SHL;
+ ln++;
continue;
case '>':
- *tk++ = NE;
- ++ln;
+ *tk.u32++ = NE;
+ ln++;
continue;
case '=':
- *tk++ = LE;
- ++ln;
+ *tk.u32++ = LE;
+ ln++;
continue;
default:
- *tk++ = '<';
+ *tk.u32++ = '<';
continue;
}
case ':': // : or ::
if (*ln == ':')
{
- *tk++ = DCOLON;
- ++ln;
+ *tk.u32++ = DCOLON;
+ ln++;
}
else
- *tk++ = ':';
+ *tk.u32++ = ':';
continue;
case '=': // = or ==
if (*ln == '=')
{
- *tk++ = DEQUALS;
- ++ln;
+ *tk.u32++ = DEQUALS;
+ ln++;
}
else
- *tk++ = '=';
+ *tk.u32++ = '=';
continue;
case '>': // > or >> or >=
switch (*ln)
{
case '>':
- *tk++ = SHR;
+ *tk.u32++ = SHR;
ln++;
continue;
case '=':
- *tk++ = GE;
+ *tk.u32++ = GE;
ln++;
continue;
default:
- *tk++ = '>';
+ *tk.u32++ = '>';
continue;
}
case '%': // % or binary constant
if (*ln < '0' || *ln > '1')
{
- *tk++ = '%';
+ *tk.u32++ = '%';
continue;
}
if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
{
+ v &= 0xFFFFFFFF;
ln += 2;
}
}
- *tk++ = CONST;
- *tk++ = v;
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
continue;
case '@': // @ or octal constant
if (*ln < '0' || *ln > '7')
{
- *tk++ = '@';
+ *tk.u32++ = '@';
continue;
}
if (*ln == '.')
{
- if ((*(ln+1) == 'b') || (*(ln+1) == 'B'))
+ if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B'))
{
v &= 0x000000FF;
ln += 2;
}
- if ((*(ln+1) == 'w') || (*(ln+1) == 'W'))
+ if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
{
v &= 0x0000FFFF;
ln += 2;
}
- if ((*(ln+1) == 'l') || (*(ln+1) == 'L'))
+ if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
{
+ v &= 0xFFFFFFFF;
ln += 2;
}
}
- *tk++ = CONST;
- *tk++ = v;
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
continue;
case '^': // ^ or ^^ <operator-name>
if (*ln != '^')
{
- *tk++ = '^';
+ *tk.u32++ = '^';
continue;
}
continue;
}
- *tk++ = (TOKEN)j;
+ *tk.u32++ = (TOKEN)j;
continue;
default:
interror(2); // Bad MULTX entry in chrtab
// Handle decimal constant
if (c & DIGIT)
{
+ uint8_t * numStart = ln;
v = 0;
while ((int)chrtab[*ln] & DIGIT)
{
v &= 0x000000FF;
ln += 2;
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
+ *tk.u32++ = DOTB;
}
else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
{
v &= 0x0000FFFF;
ln += 2;
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
+ *tk.u32++ = DOTW;
}
else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
{
+ v &= 0xFFFFFFFF;
ln += 2;
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
+ *tk.u32++ = DOTL;
+ }
+ else if ((int)chrtab[*(ln + 1)] & DIGIT)
+ {
+ // Hey, more digits after the dot, so we assume it's a
+ // floating point number of some kind... numEnd will point
+ // to the first non-float character after it's done
+ char * numEnd;
+ errno = 0;
+ double f = strtod(numStart, &numEnd);
+ ln = (uint8_t *)numEnd;
+
+ if (errno != 0)
+ return error("floating point parse error");
+
+ // N.B.: We use the C compiler's internal double
+ // representation for all internal float calcs and
+ // are reasonably sure that the size of said double
+ // is 8 bytes long (which we check for in fltpoint.c)
+ *tk.u32++ = FCONST;
+ *tk.dp = f;
+ tk.u64++;
+ continue;
}
}
+ else
+ {
+ *tk.u32++ = CONST;
+ *tk.u64++ = v;
+ }
- *tk++ = CONST;
- *tk++ = v;
//printf("CONST: %i\n", v);
continue;
}
// Terminate line of tokens and return "success."
goteol:
- tok = etok; // Set tok to beginning of line
+ tok = etok; // Set tok to beginning of line
- if (stuffnull) // Terminate last SYMBOL
+ if (stuffnull) // Terminate last SYMBOL
*nullspot = EOS;
- *tk++ = EOL;
+ *tk.u32++ = EOL;
return OK;
}
{
// Compare names (sleazo string compare)
char * s1 = sym;
- char * s2 = defln->line;
+ char * s2 = defln->line + 1;
// Either we will match the strings to EOS on both, or we will
// match EOS on string 1 to whitespace on string 2. Otherwise, we
}
+void DumpToken(TOKEN t)
+{
+ if (t == COLON)
+ printf("[COLON]");
+ else if (t == CONST)
+ printf("[CONST]");
+ else if (t == FCONST)
+ printf("[FCONST]");
+ else if (t == ACONST)
+ printf("[ACONST]");
+ else if (t == STRING)
+ printf("[STRING]");
+ else if (t == SYMBOL)
+ printf("[SYMBOL]");
+ else if (t == EOS)
+ printf("[EOS]");
+ else if (t == TKEOF)
+ printf("[TKEOF]");
+ else if (t == DEQUALS)
+ printf("[DEQUALS]");
+ else if (t == SET)
+ printf("[SET]");
+ else if (t == REG)
+ printf("[REG]");
+ else if (t == DCOLON)
+ printf("[DCOLON]");
+ else if (t == GE)
+ printf("[GE]");
+ else if (t == LE)
+ printf("[LE]");
+ else if (t == NE)
+ printf("[NE]");
+ else if (t == SHR)
+ printf("[SHR]");
+ else if (t == SHL)
+ printf("[SHL]");
+ else if (t == UNMINUS)
+ printf("[UNMINUS]");
+ else if (t == DOTB)
+ printf("[DOTB]");
+ else if (t == DOTW)
+ printf("[DOTW]");
+ else if (t == DOTL)
+ printf("[DOTL]");
+ else if (t == DOTQ)
+ printf("[DOTQ]");
+ else if (t == DOTS)
+ printf("[DOTS]");
+ else if (t == DOTD)
+ printf("[DOTD]");
+ else if (t == DOTI)
+ printf("[DOTI]");
+ else if (t == ENDEXPR)
+ printf("[ENDEXPR]");
+ else if (t == CR_ABSCOUNT)
+ printf("[CR_ABSCOUNT]");
+ else if (t == CR_FILESIZE)
+ printf("[CR_FILESIZE]");
+ else if (t == CR_DEFINED)
+ printf("[CR_DEFINED]");
+ else if (t == CR_REFERENCED)
+ printf("[CR_REFERENCED]");
+ else if (t == CR_STREQ)
+ printf("[CR_STREQ]");
+ else if (t == CR_MACDEF)
+ printf("[CR_MACDEF]");
+ else if (t == CR_TIME)
+ printf("[CR_TIME]");
+ else if (t == CR_DATE)
+ printf("[CR_DATE]");
+ else if (t >= 0x20 && t <= 0x2F)
+ printf("[%c]", (char)t);
+ else if (t >= 0x3A && t <= 0x3F)
+ printf("[%c]", (char)t);
+ else if (t >= 0x80 && t <= 0x87)
+ printf("[D%u]", ((uint32_t)t) - 0x80);
+ else if (t >= 0x88 && t <= 0x8F)
+ printf("[A%u]", ((uint32_t)t) - 0x88);
+ else
+ printf("[%X:%c]", (uint32_t)t, (char)t);
+}
+
+
void DumpTokenBuffer(void)
{
printf("Tokens [%X]: ", sloc);
printf("[COLON]");
else if (*t == CONST)
{
- t++;
- printf("[CONST: $%X]", (uint32_t)*t);
+ PTR tp;
+ tp.u32 = t + 1;
+ printf("[CONST: $%lX]", *tp.u64);
+ t += 2;
+ }
+ else if (*t == FCONST)
+ {
+ PTR tp;
+ tp.u32 = t + 1;
+ printf("[FCONST: $%lX]", *tp.u64);
+ t += 2;
}
else if (*t == ACONST)
{
printf("[DOTW]");
else if (*t == DOTL)
printf("[DOTL]");
+ else if (*t == DOTQ)
+ printf("[DOTQ]");
+ else if (*t == DOTS)
+ printf("[DOTS]");
+ else if (*t == DOTD)
+ printf("[DOTD]");
else if (*t == DOTI)
printf("[DOTI]");
else if (*t == ENDEXPR)
printf("[ENDEXPR]");
else if (*t == CR_ABSCOUNT)
printf("[CR_ABSCOUNT]");
+ else if (*t == CR_FILESIZE)
+ printf("[CR_FILESIZE]");
else if (*t == CR_DEFINED)
printf("[CR_DEFINED]");
else if (*t == CR_REFERENCED)