//
// RMAC - Reboot's Macro Assembler for the Atari Jaguar Console System
// TOKEN.C - Token Handling
-// Copyright (C) 199x Landon Dyer, 2011-2012 Reboot and Friends
+// Copyright (C) 199x Landon Dyer, 2011-2017 Reboot and Friends
// RMAC derived from MADMAC v1.07 Written by Landon Dyer, 1986
// Source utilised with the kind permission of Landon Dyer
//
#include "symbol.h"
#define DECL_KW // Declare keyword arrays
-#define DEF_KW // Declare keyword values
+#define DEF_KW // Declare keyword values
#include "kwtab.h" // Incl generated keyword tables & defs
int mjump_align = 0; // mjump alignment flag
char lntag; // Line tag
char * curfname; // Current filename
-char tolowertab[128]; // Uppercase ==> lowercase
-char hextab[128]; // Table of hex values
+char tolowertab[128]; // Uppercase ==> lowercase
+int8_t hextab[128]; // Table of hex values
char dotxtab[128]; // Table for ".b", ".s", etc.
char irbuf[LNSIZ]; // Text for .rept block line
char lnbuf[LNSIZ]; // Text of current line
static TOKEN tokbuf[TOKBUFSIZE]; // Token buffer (stack-like, all files)
char chrtab[] = {
- ILLEG, ILLEG, ILLEG, ILLEG, // NUL SOH STX ETX
- ILLEG, ILLEG, ILLEG, ILLEG, // EOT ENQ ACK BEL
- ILLEG, WHITE, ILLEG, ILLEG, // BS HT LF VT
- WHITE, ILLEG, ILLEG, ILLEG, // FF CR SO SI
+ ILLEG, ILLEG, ILLEG, ILLEG, // NUL SOH STX ETX
+ ILLEG, ILLEG, ILLEG, ILLEG, // EOT ENQ ACK BEL
+ ILLEG, WHITE, ILLEG, ILLEG, // BS HT LF VT
+ WHITE, ILLEG, ILLEG, ILLEG, // FF CR SO SI
- ILLEG, ILLEG, ILLEG, ILLEG, // DLE DC1 DC2 DC3
- ILLEG, ILLEG, ILLEG, ILLEG, // DC4 NAK SYN ETB
- ILLEG, ILLEG, ILLEG, ILLEG, // CAN EM SUB ESC
- ILLEG, ILLEG, ILLEG, ILLEG, // FS GS RS US
+ ILLEG, ILLEG, ILLEG, ILLEG, // DLE DC1 DC2 DC3
+ ILLEG, ILLEG, ILLEG, ILLEG, // DC4 NAK SYN ETB
+ ILLEG, ILLEG, ILLEG, ILLEG, // CAN EM SUB ESC
+ ILLEG, ILLEG, ILLEG, ILLEG, // FS GS RS US
WHITE, MULTX, MULTX, SELF, // SP ! " #
MULTX+CTSYM, MULTX, SELF, MULTX, // $ % & '
SELF, SELF, SELF, SELF, // ( ) * +
SELF, SELF, STSYM, SELF, // , - . /
- DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 0 1
- DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 2 3
- DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 4 5
- DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 6 7
- DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 8 9
- MULTX, MULTX, // : ;
- MULTX, MULTX, MULTX, STSYM+CTSYM, // < = > ?
+ DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 0 1
+ DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 2 3
+ DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 4 5
+ DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 6 7
+ DIGIT+HDIGIT+CTSYM, DIGIT+HDIGIT+CTSYM, // 8 9
+ MULTX, MULTX, // : ;
+ MULTX, MULTX, MULTX, STSYM+CTSYM, // < = > ?
MULTX, STSYM+CTSYM+HDIGIT, // @ A
(char)((BYTE)DOT)+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // B C
STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // h i j k
(char)((BYTE)DOT)+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // l m n o
- STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // p q r s
- STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // t u v w
- STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // x y z {
- SELF, SELF, SELF, ILLEG // | } ~ DEL
+ STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // p q r s
+ STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // t u v w
+ STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // x y z {
+ SELF, SELF, SELF, ILLEG // | } ~ DEL
};
// Names of registers
};
static char * riscregname[] = {
- "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
+ "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31"
tolowertab[i] |= 0x20;
// These characters are legal immediately after a period
- dotxtab['b'] = DOTB; // .b .B .s .S
+ dotxtab['b'] = DOTB; // .b .B .s .S
dotxtab['B'] = DOTB;
dotxtab['s'] = DOTB;
dotxtab['S'] = DOTB;
- dotxtab['w'] = DOTW; // .w .W
+ dotxtab['w'] = DOTW; // .w .W
dotxtab['W'] = DOTW;
- dotxtab['l'] = DOTL; // .l .L
+ dotxtab['l'] = DOTL; // .l .L
dotxtab['L'] = DOTL;
- dotxtab['i'] = DOTI; // .i .I (???)
+ dotxtab['i'] = DOTI; // .i .I (???)
dotxtab['I'] = DOTI;
}
inobj->inobj.ifile = ifile;
break;
- case SRC_IMACRO: // Alloc and init an IMACRO
+ case SRC_IMACRO: // Alloc and init an IMACRO
if (f_imacro == NULL)
imacro = malloc(sizeof(IMACRO));
else
IMACRO * imacro = cur_inobj->inobj.imacro;
int macnum = (int)(imacro->im_macro->sattr);
-// destsiz--;
char * dst = dest; // Next dest slot
char * edst = dest + destsiz - 1; // End + 1(?) of dest buffer
if (dst >= edst)
goto overflow;
+ // Skip comments in case a loose @ or \ is in there
+ // In that case the tokeniser was trying to expand it.
+ if ((*s == ';') || ((*s == '/') && (*(s + 1) == '/')))
+ goto skipcomments;
+
*dst++ = *s++;
}
// Do macro expansion
*dst++ = *s++;
continue;
- case '?': // \? <macro> set `questmark' flag
+ case '?': // \? <macro> set `questmark' flag
++s;
questmark = 1;
break;
- case '#': // \#, number of arguments
+ case '#': // \#, number of arguments
sprintf(numbuf, "%d", (int)imacro->im_nargs);
goto copystr;
case '!': // \! size suffix supplied on invocation
}
goto copy_d;
- case '~': // ==> unique label string Mnnnn...
+ case '~': // ==> unique label string Mnnnn...
sprintf(numbuf, "M%u", curuniq);
copystr:
d = numbuf;
}
}
+skipcomments:
+
*dst = EOS;
DEBUG { printf("ExM: dst=\"%s\"\n", dest); }
return OK;
//
char * GetNextMacroLine(void)
{
-// unsigned source_addr;
-
IMACRO * imacro = cur_inobj->inobj.imacro;
// LONG * strp = imacro->im_nextln;
struct LineList * strp = imacro->im_nextln;
if (strp == NULL) // End-of-macro
return NULL;
-// imacro->im_nextln = (LONG *)*strp;
imacro->im_nextln = strp->next;
// ExpandMacro((char *)(strp + 1), imacro->im_lnbuf, LNSIZ);
ExpandMacro(strp->line, imacro->im_lnbuf, LNSIZ);
//
int include(int handle, char * fname)
{
- IFILE * ifile;
- INOBJ * inobj;
- FILEREC * fr;
-
- // Verbose mode
- if (verb_flag)
+ // Debug mode
+ if (debug)
printf("[include: %s, cfileno=%u]\n", fname, cfileno);
// Alloc and initialize include-descriptors
- inobj = a_inobj(SRC_IFILE);
- ifile = inobj->inobj.ifile;
+ INOBJ * inobj = a_inobj(SRC_IFILE);
+ IFILE * ifile = inobj->inobj.ifile;
ifile->ifhandle = handle; // Setup file handle
ifile->ifind = ifile->ifcnt = 0; // Setup buffer indices
ifile->ifoldfname = curfname; // Save old filename
ifile->ifno = cfileno; // Save old file number
-// cfileno = filecount++; // Compute new file number
// NB: This *must* be preincrement, we're adding one to the filecount here!
cfileno = ++filecount; // Compute NEW file number
curfname = strdup(fname); // Set current filename (alloc storage)
curlineno = 0; // Start on line zero
// Add another file to the file-record
- fr = (FILEREC *)malloc(sizeof(FILEREC));
+ FILEREC * fr = (FILEREC *)malloc(sizeof(FILEREC));
fr->frec_next = NULL;
fr->frec_name = curfname;
if (last_fr == NULL)
- filerec = fr; // Add first filerec
+ filerec = fr; // Add first filerec
else
- last_fr->frec_next = fr; // Append to list of filerecs
+ last_fr->frec_next = fr; // Append to list of filerecs
last_fr = fr;
DEBUG printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno);
{
// Pop IFENT levels until we reach the conditional assembly context we
// were at when the input object was entered.
+ int numUnmatched = 0;
+
while (ifent != inobj->in_ifent)
{
- if (d_endif() != 0) // Something bad happened during endif parsing?
- return -1; // If yes, bail instead of getting stuck in a loop
+ if (d_endif() != 0) // Something bad happened during endif parsing?
+ return -1; // If yes, bail instead of getting stuck in a loop
+
+ numUnmatched++;
}
- tok = inobj->in_otok; // Restore tok and otok
+ // Give a warning to the user that we had to wipe their bum for them
+ if (numUnmatched > 0)
+ warni("missing %d .endif(s)", numUnmatched);
+
+ tok = inobj->in_otok; // Restore tok and otok
etok = inobj->in_etok;
switch (inobj->in_type)
{
- case SRC_IFILE: // Pop and release an IFILE
- if (verb_flag)
+ case SRC_IFILE: // Pop and release an IFILE
+ if (debug)
printf("[Leaving: %s]\n", curfname);
ifile = inobj->inobj.ifile;
ifile->if_link = f_ifile;
f_ifile = ifile;
close(ifile->ifhandle); // Close source file
-if (verb_flag) printf("[fpop (pre): curfname=%s]\n", curfname);
+if (debug) printf("[fpop (pre): curfname=%s]\n", curfname);
curfname = ifile->ifoldfname; // Set current filename
-if (verb_flag) printf("[fpop (post): curfname=%s]\n", curfname);
-if (verb_flag) printf("[fpop: (pre) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
- curlineno = ifile->ifoldlineno; // Set current line#
+if (debug) printf("[fpop (post): curfname=%s]\n", curfname);
+if (debug) printf("[fpop: (pre) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
+ curlineno = ifile->ifoldlineno; // Set current line#
DEBUG printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno);
cfileno = ifile->ifno; // Restore current file number
-if (verb_flag) printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
+if (debug) printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno);
break;
case SRC_IMACRO: // Pop and release an IMACRO
imacro = inobj->inobj.imacro;
if (*p == '\r')
{
if (i >= j)
- break; // Need to read more, then look for '\n' to eat
+ break; // Need to read more, then look for '\n' to eat
else if (p[1] == '\n')
i++;
}
case SRC_IFILE:
if ((ln = GetNextLine()) == NULL)
{
-if (verb_flag) printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n");
- fpop(); // Pop input level
- goto retry; // Try for more lines
+if (debug) printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n");
+ 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
+ return TKEOF;
+ }
}
curlineno++; // Bump line number
case SRC_IREPT:
if ((ln = GetNextRepeatLine()) == NULL)
{
-if (verb_flag) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n");
+if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n");
fpop();
goto retry;
}
ln++;
// Handle EOL, comment with ';'
- if (*ln == EOS || *ln == ';'|| ((*ln == '/') && (*(ln + 1) == '/')))
+ if (*ln == EOS || *ln == ';'|| ((*ln == '/') && (*(ln + 1) == '/')))
break;
// Handle start of symbol. Symbols are null-terminated in place. The
v = 0; // Assume no DOT attrib follows symbol
stuffnull = 1;
+
+ // In some cases, we need to check for a DOTx at the *beginning*
+ // of a symbol, as the "start" of the line we're currently looking
+ // at could be somewhere in the middle of that line!
+ if (*ln == '.')
+ {
+ // Make sure that it's *only* a .[bwsl] following, and not the
+ // start of a local symbol:
+ if ((chrtab[*(ln + 1)] & DOT)
+ && (dotxtab[*(ln + 1)] != 0)
+ && !(chrtab[*(ln + 2)] & CTSYM))
+ {
+ // We found a legitimate DOTx construct, so add it to the
+ // token stream:
+ ln++;
+ stuffnull = 0;
+ *tk++ = (TOKEN)dotxtab[*ln++];
+ continue;
+ }
+ }
+
p = nullspot = ln++; // Nullspot -> start of this symbol
// Find end of symbol (and compute its length)
if (*ln == '.')
{
*ln++ = EOS; // Terminate symbol
- stuffnull = 0; // And never try it again
+ stuffnull = 0; // And never try it again
// 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 ((((int)chrtab[*ln] & DOT) == 0) || ((int)dotxtab[*ln] <= 0))
- return error("[bwsl] must follow `.' in symbol");
+ if (((chrtab[*ln] & DOT) == 0) || (dotxtab[*ln] == 0))
+ return error("[bwsl] must follow '.' in symbol");
v = (VALUE)dotxtab[*ln++];
- if ((int)chrtab[*ln] & CTSYM)
- return error("misuse of `.', not allowed in symbols");
+ if (chrtab[*ln] & CTSYM)
+ return error("misuse of '.', not allowed in symbols");
}
// If the symbol is small, check to see if it's really the name of
}
// If not tokenized keyword OR token was not found
- if (j < 0 || state < 0)
+ if ((j < 0) || (state < 0))
{
*tk++ = SYMBOL;
//#warning
{
switch (*ln++)
{
- case '!': // ! or !=
+ case '!': // ! or !=
if (*ln == '=')
{
*tk++ = NE;
*tk++ = '!';
continue;
- case '\'': // 'string'
- case '\"': // "string"
+ case '\'': // 'string'
+ case '\"': // "string"
c1 = ln[-1];
*tk++ = STRING;
//#warning
*p++ = EOS;
continue;
case '$': // $, hex constant
- if ((int)chrtab[*ln] & HDIGIT)
+ if (chrtab[*ln] & HDIGIT)
{
v = 0;
// Parse the hex value
- while ((int)hextab[*ln] >= 0)
+ while (hextab[*ln] >= 0)
v = (v << 4) + (int)hextab[*ln++];
// ggn: Okay, some comments here are in order I think....
// auto-optimise? I think it's ok for now...
if (*ln == '.')
{
- if (obj_format == ALCYON)
- {
- if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B') || (*(ln + 1) == 'w') || (*(ln + 1) == 'W') || (*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
- {
- ln += 2;
- }
- }
- else
+ if (obj_format == BSD)
{
if ((*(ln + 1) & 0xDF) == 'B')
{
*tk++ = CONST;
*tk++ = v;
+
+ if (obj_format == ALCYON)
+ {
+ if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W'))
+ {
+ *tk++ = DOTW;
+ ln += 2;
+ }
+ else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L'))
+ {
+ *tk++ = DOTL;
+ ln += 2;
+ }
+ }
}
else
*tk++ = '$';
continue;
- case '<': // < or << or <> or <=
+ case '<': // < or << or <> or <=
switch (*ln)
{
case '<':
*tk++ = ':';
continue;
- case '=': // = or ==
+ case '=': // = or ==
if (*ln == '=')
{
*tk++ = DEQUALS;
*tk++ = '=';
continue;
- case '>': // > or >> or >=
+ case '>': // > or >> or >=
switch (*ln)
{
case '>':
*tk++ = '>';
continue;
}
- case '%': // % or binary constant
+ case '%': // % or binary constant
if (*ln < '0' || *ln > '1')
{
*tk++ = '%';
*tk++ = CONST;
*tk++ = v;
continue;
- case '@': // @ or octal constant
+ case '@': // @ or octal constant
if (*ln < '0' || *ln > '7')
{
*tk++ = '@';
for(state=0; state>=0;)
{
- // Get char, convert to lowercase
+ // Get char, convert to lowercase
j = *p++;
if (j >= 'A' && j <= 'Z')
//
// .GOTO <label> goto directive
-//
+//
// The label is searched for starting from the first line of the current,
// enclosing macro definition. If no enclosing macro exists, an error is
// generated.
-//
+//
// A label is of the form:
-//
+//
// :<name><whitespace>
-//
+//
// The colon must appear in column 1. The label is stripped prior to macro
// expansion, and is NOT subject to macro expansion. The whitespace may also
// be EOL.
struct LineList * defln = imacro->im_macro->lineList;
// Find the label, starting with the first line.
-// for(; defln!=NULL; defln=(LONG *)*defln)
for(; defln!=NULL; defln=defln->next)
{
// if (*(char *)(defln + 1) == ':')