scc

commit f53a0d9b613467cdfaf87c868f25c6f79f742f56
parent 6d702ec165be91c82ccd3940c94625967957ae8b
Author: Roberto E. Vargas Caballero <k0ga@shike2.net>
Date:   Wed,  8 Apr 2026 15:04:29 +0200

cc1: Add support for wchat_t strings

Diffstat:
M
src/cmd/scc-cc/cc1/cc1.h
 | 
3
+++
M
src/cmd/scc-cc/cc1/code.c
 | 
21
+++++++++++++++++++++
M
src/cmd/scc-cc/cc1/expr.c
 | 
33
+++++++++++++++++----------------
M
src/cmd/scc-cc/cc1/lex.c
 | 
77
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++----------
M
src/cmd/scc-cc/cc1/symbol.c
 | 
16
++++++++++++++++

5 files changed, 124 insertions(+), 26 deletions(-)
diff --git a/src/cmd/scc-cc/cc1/cc1.h b/src/cmd/scc-cc/cc1/cc1.h
@@ -116,6 +116,7 @@ enum {
 	SLOCAL    = 1 << 9,
 	SEMITTED  = 1 << 10,
 	SDEFINED  = 1 << 11,
+	SRSTRING  = 1 << 11,
 	SSTRING   = 1 << 12,
 	STYPEDEF  = 1 << 13,
 	SINITLST  = 1 << 14,
@@ -356,6 +357,7 @@ struct symbol {
 		double d;
 		long double ld;
 		char *s;
+		Rune *rs;
 		unsigned char token;
 		Node **init;
 		Symbol **pars;
@@ -466,6 +468,7 @@ void killsym(Symbol *sym);
 Symbol *newlabel(void);
 void builtins(struct builtin *builts);
 Symbol *newstring(char *s, size_t len);
+Symbol *newrstring(Rune *, size_t);
 unsigned newid(void);
 void isyms(void);
 
diff --git a/src/cmd/scc-cc/cc1/code.c b/src/cmd/scc-cc/cc1/code.c
@@ -1,4 +1,5 @@
 #include <assert.h>
+#include <inttypes.h>
 #include <ctype.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -322,6 +323,20 @@ emittype(int op, void *arg)
 }
 
 static void
+emitrstring(Symbol *sym, Type *tp)
+{
+	Rune *bp, *lim;
+
+	bp = sym->u.rs;
+	lim = &sym->u.rs[tp->n.elem];
+	while (bp < lim) {
+		fprintf(outfp,
+			"\t#%c%" PRIX32 "\n",
+			wchartype->letter, *bp++);
+	}
+}
+
+static void
 emitstring(Symbol *sym, Type *tp)
 {
 	char *bp, *s, *lim;
@@ -389,6 +404,12 @@ emitdesig(Node *np, Type *tp, unsigned long long *addr)
 			*addr += tp->n.elem;
 			return;
 		}
+		if (sym->flags & SRSTRING) {
+			emitrstring(sym, tp);
+			*addr += tp->n.elem * wchartype->size;
+			return;
+		}
+		
 		if ((sym->flags & SINITLST) == 0)
 			goto emit_expression;
 	}
diff --git a/src/cmd/scc-cc/cc1/expr.c b/src/cmd/scc-cc/cc1/expr.c
@@ -652,28 +652,29 @@ negation(int op, Node *np)
 static Symbol *
 adjstrings(Symbol *sym)
 {
-	char *s, *t;
-	size_t len, n;
-	Type *tp;
+	char *d, *s;
+	Type *tp, *base;
+	size_t len, n, osiz, nsiz;
 
 	tp = sym->type;
-	s = sym->u.s;
-	for (len = tp->n.elem;; len += n) {
-		next();
-		if (yytoken != STRING)
+	base = tp->type;
+	for (len = tp->n.elem; next() == STRING; len += n - 1) {
+		if (yylval.sym->type->type != base)
 			break;
-		t = yylval.sym->u.s;
-		n = yylval.sym->type->n.elem - 1;
 
-		s = xrealloc(s, len + n);
-		memcpy(s + len - 1, t, n);
-		s[len + n - 1] = '\0';
-	}
+		d = sym->u.s;
+		s = yylval.sym->u.s;
+		n = yylval.sym->type->n.elem;
+		osiz = (len-1) * base->size;
+		nsiz = n * base->size;
 
-	if (tp->n.elem != len) {
-		sym->type = mktype(chartype, ARY, len, NULL);
-		sym->u.s = s;
+		sym->u.s = d = xrealloc(d, osiz + nsiz);
+		memcpy(d + osiz, s, nsiz);
 	}
+
+	if (tp->n.elem != len)
+		sym->type = mktype(base, ARY, len, NULL);
+
 	return sym;
 }
 
diff --git a/src/cmd/scc-cc/cc1/lex.c b/src/cmd/scc-cc/cc1/lex.c
@@ -12,6 +12,8 @@
 #include "cc1.h"
 
 #define REPLACECHAR 0xFFFD
+#define NOMULTICHAR 0
+#define MULTICHAR   1
 
 int yytoken;
 struct yystype yylval;
@@ -712,24 +714,66 @@ character(void)
 }
 
 static int
-string(void)
+rstring(void)
 {
-	int c, multi = 0;
-	char buff[STRINGSIZ+1], *bp = buff, *beg = input->p, *end;
+	char *beg = input->p++;
+	Rune c, buff[STRINGSIZ+1], *bp = buff;
 
 	for (++input->p; ; ++input->p) {
-		c = *input->p;
-
-		if (c == '"')
-			break;
+		c = *(unsigned char *) input->p;
 
-		if (c == '\0') {
+		switch (c) {
+		case '\0':
 			errorp("missing terminating '\"' character");
+		case '"':
+			goto end_loop;
+		case '\\':
+			c = escape(MULTICHAR);
+		}
+
+		if (input->p - beg == STRINGSIZ + 1) {
+			/* too long, ignore everything until next quote */
+			for (++input->p; *input->p != '"'; ++input->p) {
+				if (*input->p == '\\')
+					++input->p;
+				if (*input->p == '\0')
+					break;
+			}
+			--bp;
+			errorp("string too long");
 			break;
 		}
+		*bp++ = c;
+	}
 
-		if (c == '\\')
-			c = escape(multi);
+end_loop:
+	*bp++ = '\0';
+	input->p++;
+
+	yylval.sym = newrstring(buff, bp - buff);
+	tok2str();
+
+	return STRING;
+}
+
+static int
+sstring(void)
+{
+	int c;
+	struct limits *lim;
+	char buff[STRINGSIZ+1], *bp = buff, *beg = input->p;
+
+	for (++input->p; ; ++input->p) {
+		c = *input->p;
+
+		switch (c) {
+		case '\0':
+			errorp("missing terminating '\"' character");
+		case '"':
+			goto end_loop;
+		case '\\':
+			c = escape(NOMULTICHAR);
+		}
 
 		if (input->p - beg == STRINGSIZ + 1) {
 			/* too long, ignore everything until next quote */
@@ -745,6 +789,8 @@ string(void)
 		}
 		*bp++ = c;
 	}
+
+end_loop:
 	*bp++ = '\0';
 	input->p++;
 
@@ -755,6 +801,14 @@ string(void)
 }
 
 static int
+string(void)
+{
+	if (*input->p == 'L')
+		return rstring();
+	return sstring();
+}
+
+static int
 iden(void)
 {
 	Symbol *sym;
@@ -763,6 +817,9 @@ iden(void)
 	if (input->p[0] == 'L' && input->p[1] == '\'')
 		return character();
 
+	if (input->p[0] == 'L' && input->p[1] == '"')
+		return string();
+
 	begin = input->p;
 	for (p = begin; isalnum(*p) || *p == '_'; ++p)
 		;
diff --git a/src/cmd/scc-cc/cc1/symbol.c b/src/cmd/scc-cc/cc1/symbol.c
@@ -252,6 +252,22 @@ linkhash(Symbol *sym)
 }
 
 Symbol *
+newrstring(Rune *rs, size_t len)
+{
+	Symbol *sym = newsym(NS_IDEN, NULL);
+
+	if (lexmode != CPPMODE)
+		sym->type = mktype(wchartype, ARY, len, NULL);
+	sym->id = newid();
+	sym->flags |= SRSTRING | SCONSTANT | SPRIVATE;
+	sym->u.rs = xcalloc(len, sizeof(Rune));
+	if (rs)
+		memcpy(sym->u.rs, rs, len * sizeof(Rune));
+
+	return sym;
+}
+
+Symbol *
 newstring(char *s, size_t len)
 {
 	Symbol *sym = newsym(NS_IDEN, NULL);