X-Git-Url: http://shamusworld.gotdns.org/cgi-bin/gitweb.cgi?p=rmac;a=blobdiff_plain;f=token.c;h=6a47bdb3fd04daab8e8c17b8364e83cc88712cfb;hp=f719dc356791e84142046f6e6251afe62e2ad345;hb=4205233c8397c581b4d27ab36ab81ec896ef3dd0;hpb=582df8950c285e1746d0c4a9e3ead6545c962dc8 diff --git a/token.c b/token.c index f719dc3..6a47bdb 100644 --- a/token.c +++ b/token.c @@ -1,12 +1,14 @@ // // RMAC - Reboot's 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, 2011-2020 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 #include "direct.h" #include "error.h" #include "macro.h" @@ -35,7 +37,7 @@ WORD cfileno; // Current file number 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 @@ -206,13 +208,13 @@ void InitTokenizer(void) 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; @@ -390,7 +392,7 @@ int ExpandMacro(char * src, char * dest, int destsiz) *dst++ = *s++; continue; case '?': // \? set `questmark' flag - ++s; + s++; questmark = 1; break; case '#': // \#, number of arguments @@ -529,22 +531,12 @@ arg_num: 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; @@ -568,7 +560,8 @@ DEBUG { printf("ExM: SYMBOL=\"%s\"", d); } // to choke on legitimate code... Need to investigate this further // before changing anything else here! case CONST: - sprintf(numbuf, "$%lx", (uint64_t)*tk++); +// sprintf(numbuf, "$%lx", (uint64_t)*tk++); + sprintf(numbuf, "$%" PRIX64, (uint64_t)*tk++); tk++; d = numbuf; break; @@ -714,6 +707,11 @@ char * GetNextRepeatLine(void) // 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; // strcpy(irbuf, (char *)(irept->ir_nextln + 1)); strcpy(irbuf, irept->ir_nextln->line); @@ -946,7 +944,7 @@ int TokenizeLine(void) { 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 @@ -957,11 +955,10 @@ int TokenizeLine(void) int stuffnull; // 1:terminate SYMBOL '\0' at *nullspot uint8_t c1; int stringNum = 0; // Pointer to string locations in tokenized line - uint64_t * tk64; 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 @@ -977,8 +974,8 @@ retry: 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 @@ -986,7 +983,7 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } } } - curlineno++; // Bump line number + curlineno++; // Bump line number lntag = SPACE; if (as68_flag) @@ -1049,7 +1046,7 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } // 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 @@ -1112,7 +1109,7 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } // token stream: ln++; stuffnull = 0; - *tk++ = (TOKEN)dotxtab[*ln++]; + *tk.u32++ = (TOKEN)dotxtab[*ln++]; continue; } } @@ -1130,7 +1127,7 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } *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....) @@ -1174,7 +1171,7 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } } // 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; } @@ -1195,28 +1192,21 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } // 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; @@ -1225,7 +1215,7 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } // Handle identity tokens if (c & SELF) { - *tk++ = *ln++; + *tk.u32++ = *ln++; continue; } @@ -1237,27 +1227,27 @@ DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } 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;) @@ -1297,6 +1287,10 @@ dostring: 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 @@ -1320,131 +1314,91 @@ dostring: case '$': // $, hex constant if (chrtab[*ln] & HDIGIT) { - if (cursize == 'q' || cursize == 'Q') - { - // Parse 64-bit integer - uint64_t v64 = 0; - - while (hextab[*ln] >= 0) - v64 = (v64 << 4) + (int)hextab[*ln++]; - - *(uint64_t *)tk = v64; - tk = tk + 2; - - continue; - } v = 0; // Parse the hex value while (hextab[*ln] >= 0) v = (v << 4) + (int)hextab[*ln++]; + *tk.u32++ = CONST; + *tk.u64++ = v; + if (*ln == '.') { - if (obj_format == BSD) + if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W')) { - if ((*(ln + 1) & 0xDF) == 'B') - { - v &= 0x000000FF; - ln += 2; - } - else if ((*(ln + 1) & 0xDF) == 'W') - { - v &= 0x0000FFFF; - ln += 2; - } - else if ((*(ln + 1) & 0xDF) == 'L') - { - v &= 0xFFFFFFFF; - ln += 2; - } + *tk.u32++ = DOTW; + ln += 2; } - } - - *tk++ = CONST; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (TOKEN *)tk64; - - if (obj_format == ALCYON) - { - if (*ln == '.') + else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) { - if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W')) - { - *tk++ = DOTW; - ln += 2; - } - else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) - { - *tk++ = DOTL; - ln += 2; - } + *tk.u32++ = DOTL; + ln += 2; } } } else - *tk++ = '$'; + *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; } @@ -1474,15 +1428,13 @@ dostring: } } - *tk++ = CONST; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (TOKEN *)tk64; + *tk.u32++ = CONST; + *tk.u64++ = v; continue; case '@': // @ or octal constant if (*ln < '0' || *ln > '7') { - *tk++ = '@'; + *tk.u32++ = '@'; continue; } @@ -1493,34 +1445,32 @@ dostring: 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; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (TOKEN *)tk64; + *tk.u32++ = CONST; + *tk.u64++ = v; continue; case '^': // ^ or ^^ if (*ln != '^') { - *tk++ = '^'; + *tk.u32++ = '^'; continue; } @@ -1566,7 +1516,7 @@ dostring: continue; } - *tk++ = (TOKEN)j; + *tk.u32++ = (TOKEN)j; continue; default: interror(2); // Bad MULTX entry in chrtab @@ -1577,6 +1527,7 @@ dostring: // Handle decimal constant if (c & DIGIT) { + uint8_t * numStart = ln; v = 0; while ((int)chrtab[*ln] & DIGIT) @@ -1589,60 +1540,53 @@ dostring: { v &= 0x000000FF; ln += 2; - *tk++ = CONST; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (uint32_t *)tk64; - *tk++ = DOTB; + *tk.u32++ = CONST; + *tk.u64++ = v; + *tk.u32++ = DOTB; } else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W')) { v &= 0x0000FFFF; ln += 2; - *tk++ = CONST; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (uint32_t *)tk64; - - *tk++ = DOTW; + *tk.u32++ = CONST; + *tk.u64++ = v; + *tk.u32++ = DOTW; } else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) { v &= 0xFFFFFFFF; ln += 2; - *tk++ = CONST; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (uint32_t *)tk64; - - *tk++ = DOTL; + *tk.u32++ = CONST; + *tk.u64++ = v; + *tk.u32++ = DOTL; } else if ((int)chrtab[*(ln + 1)] & DIGIT) { - // Hey, more digits after the dot, so assume it's a - // fractional number - double fract = 10; - ln++; - f = (double)v; - - while ((int)chrtab[*ln] & DIGIT) - { - f = f + (double)(*ln++ - '0') / fract; - fract *= 10; - } - - *tk++ = FCONST; - *((double *)tk) = f; - tk += 2; + // 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++ = CONST; - tk64 = (uint64_t *)tk; - *tk64++ = v; - tk = (TOKEN *)tk64; + *tk.u32++ = CONST; + *tk.u64++ = v; } //printf("CONST: %i\n", v); @@ -1656,12 +1600,12 @@ dostring: // 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; } @@ -1732,6 +1676,87 @@ int d_goto(WORD unused) } +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_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); @@ -1742,7 +1767,16 @@ void DumpTokenBuffer(void) printf("[COLON]"); else if (*t == CONST) { - printf("[CONST: $%lX]", ((uint64_t)t[1] << 32) | (uint64_t)t[2]); + 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) @@ -1790,6 +1824,12 @@ void DumpTokenBuffer(void) 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)