support for long strings

mach/proto/as/comm1.h

@@ -58,8 +58,8 @@ extern char	temppath[50];
 extern FILE	*input;
 extern FILE	*tempfile;
 
-extern char	stringbuf[STRINGMAX];
-				/* contains last string value */
+extern char	*stringbuf;	/* contains last string value */
+extern int	stringlen;	/* contains length of last string value */
 
 extern sect_t	sect[SECTMAX];
 
@@ -116,6 +116,7 @@ extern FILE	*fftemp();
 extern long	atol();
 extern char	*mktemp();
 extern char	*malloc();
+extern char	*realloc();
 extern char	*calloc();
 extern char	*getenv();
 extern char	*strncpy();

mach/proto/as/comm2.y

@@ -128,7 +128,7 @@ program	:	/* empty */
 			{	lineno++; LISTLINE(1); RELODONE;}
 	|	program '#' NUMBER STRING '\n'
 			{	lineno = $3;
-				if (modulename) strncpy(modulename, &stringbuf[1], STRINGMAX-1);
+				if (modulename) strncpy(modulename, stringbuf, STRINGMAX-1);
 				LISTLINE(1); RELODONE;
 			}
 	|	program error '\n'
@@ -196,7 +196,7 @@ operation
 #endif
 					    
 					newsymb(
-						*(stringbuf+1) ? stringbuf+1 : (char *) 0,
+						*stringbuf ? stringbuf : (char *) 0,
 						(short)(
 							($4.typ & (S_EXT|S_TYP))
 							|
@@ -210,7 +210,7 @@ operation
 	|	SYMD STRING ','  absexp ',' absexp
 			{	if ((sflag & SYM_SMB) && PASS_SYMB) {
 					newsymb(
-						*(stringbuf+1) ? stringbuf+1 : (char *) 0,
+						*stringbuf ? stringbuf : (char *) 0,
 						(short)(
 							(DOTTYP & (S_EXT|S_TYP))
 							|
@@ -239,7 +239,7 @@ operation
 			{	if ((sflag & SYM_LIN) && PASS_SYMB) {
 					hllino = 0;
 					newsymb(
-						stringbuf+1,
+						stringbuf,
 						(short)(DOTTYP | S_FIL),
 						(short)0,
 						(valu_t)DOTVAL
@@ -295,9 +295,9 @@ expr	:	error
 				$$.typ = $1->i_type & ~S_EXT;
 			}
 	|	STRING
-			{	if (stringbuf[0] != 1)
+			{	if (stringlen != 1)
 					serror("too many chars");
-				$$.val = stringbuf[1];
+				$$.val = stringbuf[0];
 				$$.typ = S_ABS;
 			}
 	|	ASC_LPAR expr ASC_RPAR

mach/proto/as/comm5.c

@@ -105,8 +105,23 @@ putval(c)
 		p = (char *) &yylval.y_strp; break;
 #endif
 	case STRING:
+		v = stringlen;
+		putc(c-128, tempfile);
+		for (n = 0; n < sizeof(v); n++) {
+			if (v == 0)
+				break;
+			v >>= 8;
+		}
+		c = NUMBER0 + n;
+		putc(c-128, tempfile);
+		v = stringlen;
+		while (--n >= 0)
+			putc((int) (v >> (n*8)), tempfile);
 		p = stringbuf;
-		n = (*p & 0377) + 1; break;
+		n = stringlen;
+		while (--n >= 0)
+			putc(*p++, tempfile);
+		return;
 	case OP_EQ:
 	case OP_NE:
 	case OP_LE:
@@ -166,8 +181,10 @@ getval(c)
 		p = (char *) &yylval.y_strp; break;
 #endif
 	case STRING:
+		getval(getc(tempfile)+128);
+		stringlen = n = yylval.y_valu;
 		p = stringbuf;
-		*p++ = n = getc(tempfile); p[n] = '\0'; break;
+		p[n] = '\0'; break;
 	case OP_EQ:
 	case OP_NE:
 	case OP_LE:
@@ -346,8 +363,15 @@ instring(termc)
 {
 	register char *p;
 	register c;
+	static int maxstring = 0;
 
-	p = stringbuf+1;
+	if (! maxstring) {
+		maxstring = STRINGMAX;
+		if ((stringbuf = malloc(maxstring)) == 0) {
+			fatal("out of memory");
+		}
+	}
+	p = stringbuf;
 	for (;;) {
 		c = nextchar();
 		if (c == '\n' || c == '\0') {
@@ -359,11 +383,17 @@ instring(termc)
 			break;
 		if (c == '\\')
 			c = inescape();
-		if (p >= &stringbuf[STRINGMAX - 1])
-			fatal("string buffer overflow");
+		if (p >= &stringbuf[maxstring - 1]) {
+			int cnt = p - stringbuf;
+
+			if ((stringbuf = realloc(stringbuf, maxstring += 256)) == 0) {
+				fatal("out of memory");
+			}
+			p = stringbuf + cnt;
+		}
 		*p++ = c;
 	}
-	stringbuf[0] = p - stringbuf - 1;
+	stringlen = p - stringbuf;
 	*p = '\0';
 	return(STRING);
 }

mach/proto/as/comm7.c

@@ -306,7 +306,7 @@ emitstr(zero)
 	register char *p;
 
 	p = stringbuf;
-	i = *p++ & 0377;
+	i = stringlen;
 	while (--i >= 0)
 		emit1(*p++);
 	if (zero)

modules/src/read_em/read_em.c

@@ -18,6 +18,7 @@
 */
 
 #include <assert.h>
+#include <alloc.h>
 #include <system.h>
 #include <em_label.h>
 #include <em_arith.h>
@@ -60,29 +61,21 @@ _fill()
 	}
 }
 
-#define STRSIZ	256		/* Maximum length of strings */
-
 static struct e_instr *emhead;	/* Where we put the head */
 static struct e_instr aheads[3];
 static int ahead;
 
 static struct string {
 	int length;
-	char str[STRSIZ + 1];
+	unsigned int maxlen;
+	char *str;
 } string;
 
-#ifdef COMPACT
-static arith strleft;		/* count # of chars left to read
-				   in a string
-				*/
-#endif
-
 static int state;		/* What state are we in? */
 #define CON	01		/* Reading a CON */
 #define ROM	02		/* Reading a ROM */
 #define MES	03		/* Reading a MES */
 #define PSEUMASK 03
-#define INSTRING 010		/* Reading a string */
 
 static int EM_initialized;	/* EM_open called? */
 
@@ -399,35 +392,6 @@ EM_getinstr(p)
 		return EM_error == 0;
 	}
 
-	if (state & INSTRING) {	/* We already delivered part of a string.
-				   Deliver the next part
-				*/
-		register struct string *s;
-		
-		s = getstring(0);
-		p->em_argtype = str_ptyp;
-		p->em_string = s->str;
-		p->em_size = s->length;
-		switch(state & PSEUMASK) {
-		default:
-			assert(0);
-		case MES:
-			if (!EM_error)
-				EM_error = "String too long in message";
-			p->em_type = EM_MESARG;
-			break;
-		case CON:
-			p->em_type = EM_PSEU;
-			p->em_opcode = ps_con;
-			break;
-		case ROM:
-			p->em_type = EM_PSEU;
-			p->em_opcode = ps_rom;
-			break;
-		}
-		return EM_error == 0;
-	}
-
 	/* Here, we are in a state reading arguments */
 	getarg(any_ptyp, &(p->em_arg));
 	if (EM_error && p->em_type != EM_FATAL) {

modules/src/read_em/reade.c

@@ -3,7 +3,7 @@
  * (c) copyright 1987 by the Vrije Universiteit, Amsterdam, The Netherlands.
  * See the copyright notice in the ACK home directory, in the file "Copyright".
  */
-/*	This file is ment to be included in the file read_emeV.c.
+/*	This file is ment to be included in the file read_em.c.
 	It contains the part that takes care of the reading of human readable
 	EM-code.
 */
@@ -171,6 +171,10 @@ getname()
 
 	s = &string;
 	p = s->str;
+	if (!p) {
+		s->maxlen = 256;
+		s->str = p = Malloc(256);
+	}
 	c = getbyte();
 
 	if (!(isalpha(c) || c == '_')) {
@@ -180,7 +184,12 @@ getname()
 	}
 
 	while (isalnum(c) || c == '_') {
-		if (p < &(s->str[STRSIZ])) *p++ = c;
+		if (p >= &(s->str[s->maxlen])) {
+			int df = p - s->str;
+			s->str = Realloc(s->str, (s->maxlen += 256));
+			p = s->str + df;
+		}
+		*p++ = c;
 		c = getbyte();
 	}
 
@@ -202,15 +211,17 @@ getstring()
 
 	s = &string;
 	p = s->str;
+	if (!p) {
+		s->maxlen = 256;
+		s->str = p = Malloc(256);
+	}
 
-	if (!(state & INSTRING)) {	/* Not reading a string yet */
 	termc = getbyte();
 	/* assert(termc == '"' || termc == '\''); */
 	/* This assertion does not work. Compiler error messages.
 	   The trouble lies in the ", it terminates the string
 	   created in the assertion macro
 	*/
-	}
 
 	for (;;) {
 		if ((c = getbyte()) == '\n' || c == EOF) {
@@ -221,16 +232,15 @@ getstring()
 
 		if (c == termc) {
 			if (termc == '"') *p++ = '\0';
-			state &= ~INSTRING;
 			break;
 		}
 
 		if (c == '\\') c = getescape();
 
-		if (p >= &(s->str[STRSIZ])) {
-			state |= INSTRING;
-			ungetbyte(c);
-			break;
+		if (p >= &(s->str[s->maxlen])) {
+			int df = p - s->str;
+			s->str = Realloc(s->str, (s->maxlen += 256));
+			p = s->str + df;
 		}
 
 		*p++ = c;	
@@ -268,7 +278,7 @@ getnumber(c, ap)
 	register int c;
 	register struct e_arg *ap;
 {
-	char str[STRSIZ + 1];
+	char str[256 + 1];
 	register char *p = str;
 	int n;
 	int expsign;
@@ -291,7 +301,7 @@ getnumber(c, ap)
 	n = sp_cst4;
 
 	for (;;) {
-		if (p >= &(str[STRSIZ])) {
+		if (p >= &(str[256])) {
 			syntax("number too long");
 			return sp_cst4;
 		}
@@ -580,7 +590,7 @@ getmnem(c, p)
 PRIVATE
 line_line()
 {
-	static char filebuf[STRSIZ + 1];
+	static char filebuf[256 + 1];
 	char *btscpy();
 	struct e_arg dummy;
 

modules/src/read_em/readk.c

@@ -109,11 +109,6 @@ getarg(typset, ap)
 		register struct string *p;
 
 		p = getstring(1);
-#ifdef CHECKING
-		if (state & INSTRING) {
-			xerror("Procedure name too long");
-		}
-#endif CHECKING
 		ap->ema_pnam = p->str;
 		ap->ema_argtype = pro_ptyp;
 		break;
@@ -124,11 +119,6 @@ getarg(typset, ap)
 		register struct string *p;
 
 		p = getstring(1);
-#ifdef CHECKING
-		if (state & INSTRING) {
-			xerror("Data label too long");
-		}
-#endif CHECKING
 		ap->ema_dnam = p->str;
 		ap->ema_szoroff = 0;
 		ap->ema_argtype = sof_ptyp;
@@ -154,11 +144,6 @@ getarg(typset, ap)
 		getarg(cst_ptyp, ap);
 		ap->ema_szoroff = ap->ema_cst;
 		p = getstring(0);
-#ifdef CHECKING
-		if (state & INSTRING) {
-			xerror("Numeric constant too long");
-		}
-#endif CHECKING
 		ap->ema_argtype = ptyp(i);
 		ap->ema_string = p->str;
 		break;
@@ -226,8 +211,7 @@ checkident(s)
 }
 #endif CHECKING
 
-/* getstring: read a string from the input, but at most STRSIZ characters
-	of it. The next characters will be read another time around
+/* getstring: read a string from the input
 */
 /*ARGSUSED*/
 PRIVATE struct string *
@@ -236,32 +220,24 @@ getstring(isident)
 	register char *p;
 	register int n;
 	register struct string *s = &string;
-
-	if (!(state & INSTRING)) {	/* Not reading a string yet */
 	struct e_arg dummy;
 
 	getarg(cst_ptyp, &dummy);
 					/* Read length of string */
-		strleft = dummy.ema_cst;
+	n = dummy.ema_cst;
 #ifdef CHECKING
-		if (strleft < 0) {
+	if (n < 0) {
 		xerror("Negative length in string");
 		s->length = 0;
 		return s;
 	}
 #endif CHECKING
-	}
 
-	if (strleft <= STRSIZ) {	/* Handle the whole string */
-		n = strleft;
-		state &= ~INSTRING;
+	if (n > s->maxlen) {
+		if (! s->maxlen) {
+			s->str = Malloc(s->maxlen = 256);
 		}
-	else {				/* Handle STRSIZ characters of it, and
-					   indicate that there is more
-					*/
-	 	n = STRSIZ;
-		strleft -= STRSIZ;
-		state |= INSTRING;
+		else	s->str = Realloc(s->str, (s->maxlen = (n+255)&~255));
 	}
 
 	s->length = n;