range over strings

R=r
OCL=27332
CL=27332

src/cmd/gc/builtin.c.boot

@@ -20,6 +20,8 @@ char *sysimport =
 	"func sys.intstring (? int64) (? string)\n"
 	"func sys.byteastring (? *uint8, ? int) (? string)\n"
 	"func sys.arraystring (? []uint8) (? string)\n"
+	"func sys.stringiter (? string, ? int) (? int)\n"
+	"func sys.stringiter2 (? string, ? int) (retk int, retv int)\n"
 	"func sys.ifaceT2I (sigi *uint8, sigt *uint8, elem any) (ret any)\n"
 	"func sys.ifaceI2T (sigt *uint8, iface any) (ret any)\n"
 	"func sys.ifaceI2T2 (sigt *uint8, iface any) (ret any, ok bool)\n"

src/cmd/gc/sys.go

@@ -29,6 +29,8 @@ func	indexstring(string, int) byte;
 func	intstring(int64) string;
 func	byteastring(*byte, int) string;
 func	arraystring([]byte) string;
+func	stringiter(string, int) int;
+func	stringiter2(string, int) (retk int, retv int);
 
 func	ifaceT2I(sigi *byte, sigt *byte, elem any) (ret any);
 func	ifaceI2T(sigt *byte, iface any) (ret any);

src/cmd/gc/walk.c

@@ -3191,7 +3191,7 @@ badt:
  * rewrite a range statement
  * k and v are names/new_names
  * m is an array or map
- * local is =/0 or :=/1
+ * local is 0 (meaning =) or 1 (meaning :=)
  */
 Node*
 dorange(Node *nn)
@@ -3227,8 +3227,10 @@ dorange(Node *nn)
 		goto map;
 	if(t->etype == TCHAN)
 		goto chan;
+	if(t->etype == TSTRING)
+		goto strng;
 
-	yyerror("range must be over map/array");
+	yyerror("range must be over map/array/chan/string");
 	goto out;
 
 ary:
@@ -3333,6 +3335,77 @@ chan:
 	addtotop(n);
 	goto out;
 
+strng:
+	hk = nod(OXXX, N, N);		// hidden key
+	tempname(hk, types[TINT]);
+
+	ha = nod(OXXX, N, N);		// hidden string
+	tempname(ha, t);
+
+
+	if(local) {
+		k = old2new(k, types[TINT]);
+		if(v != N)
+			v = old2new(v, types[TINT]);
+	}
+
+	// ha = s
+	a = nod(OAS, ha, m);
+	n->ninit = a;
+
+	// kh = 0
+	a = nod(OAS, hk, nodintconst(0));
+	n->ninit = list(n->ninit, a);
+
+	// k = hk
+	a = nod(OAS, k, hk);
+	n->ninit = list(n->ninit, a);
+
+
+	// hk[,v] = stringiter(ha,hk)
+	if(v != N) {
+		// hk,v = stringiter2(ha, hk)
+		on = syslook("stringiter2", 0);
+//		argtype(on, v->type);
+		a = list(ha, hk);
+		a = nod(OCALL, on, a);
+		a = nod(OAS, list(hk, v), a);
+	} else {
+		// hk = stringiter(ha, hk)
+		on = syslook("stringiter", 0);
+		a = list(ha, hk);
+		a = nod(OCALL, on, a);
+		a = nod(OAS, hk, a);
+	}
+	n->ninit = list(n->ninit, a);
+
+	// while(hk != 0)
+	n->ntest = nod(ONE, hk, nodintconst(0));
+
+	// k = hk
+	a = nod(OAS, k, hk);
+	n->nincr = a;
+
+	// hk[,v] = stringiter(ha,hk)
+	if(v != N) {
+		// hk,v = stringiter2(ha, hk)
+		on = syslook("stringiter2", 0);
+//		argtype(on, v->type);
+		a = list(ha, hk);
+		a = nod(OCALL, on, a);
+		a = nod(OAS, list(hk, v), a);
+	} else {
+		// hk = stringiter(ha, hk)
+		on = syslook("stringiter", 0);
+		a = list(ha, hk);
+		a = nod(OCALL, on, a);
+		a = nod(OAS, hk, a);
+	}
+	n->nincr = list(n->nincr, a);
+
+	addtotop(n);
+	goto out;
+
 out:
 	return n;
 }

src/runtime/string.c

@@ -188,3 +188,159 @@ sys·arraystring(Array b, String s)
 	mcpy(s.str, b.array, s.len);
 	FLUSH(&s);
 }
+
+static	int32	chartorune(int32 *rune, byte *str);
+enum
+{
+	Runeself	= 0x80,
+	Runeerror	= 0xfff8,	// fffd in plan9
+};
+
+// func	stringiter(string, int) (retk int);
+void
+sys·stringiter(String s, int32 k, int32 retk)
+{
+	int32 l, n;
+
+	if(k >= s.len) {
+		// retk=0 is end of iteration
+		retk = 0;
+		goto out;
+	}
+
+	l = s.str[k];
+	n = 1;
+
+	if(l >= Runeself) {
+		// multi-char rune
+		n = chartorune(&l, s.str+k);
+		if(k+n > s.len) {
+			// special case of multi-char rune
+			// that ran off end of string
+			l = Runeerror;
+			n = 1;
+		}
+	}
+
+	retk = k+n;
+
+out:
+	FLUSH(&retk);
+}
+
+// func	stringiter2(string, int) (retk int, retv any);
+void
+sys·stringiter2(String s, int32 k, int32 retk, int32 retv)
+{
+	int32 l, n;
+
+	if(k >= s.len) {
+		// retk=0 is end of iteration
+		retk = 0;
+		retv = 0;
+		goto out;
+	}
+
+	l = s.str[k];
+	n = 1;
+
+	if(l >= Runeself) {
+		// multi-char rune
+		n = chartorune(&l, s.str+k);
+		if(k+n > s.len) {
+			// special case of multi-char rune
+			// that ran off end of string
+			l = Runeerror;
+			n = 1;
+		}
+	}
+
+	retk = k+n;
+	retv = l;
+
+out:
+	FLUSH(&retk);
+	FLUSH(&retv);
+}
+
+//
+// copied from plan9 library
+//
+
+enum
+{
+	Bit1	= 7,
+	Bitx	= 6,
+	Bit2	= 5,
+	Bit3	= 4,
+	Bit4	= 3,
+
+	T1	= ((1<<(Bit1+1))-1) ^ 0xFF,	/* 0000 0000 */
+	Tx	= ((1<<(Bitx+1))-1) ^ 0xFF,	/* 1000 0000 */
+	T2	= ((1<<(Bit2+1))-1) ^ 0xFF,	/* 1100 0000 */
+	T3	= ((1<<(Bit3+1))-1) ^ 0xFF,	/* 1110 0000 */
+	T4	= ((1<<(Bit4+1))-1) ^ 0xFF,	/* 1111 0000 */
+
+	Rune1	= (1<<(Bit1+0*Bitx))-1,		/* 0000 0000 0111 1111 */
+	Rune2	= (1<<(Bit2+1*Bitx))-1,		/* 0000 0111 1111 1111 */
+	Rune3	= (1<<(Bit3+2*Bitx))-1,		/* 1111 1111 1111 1111 */
+
+	Maskx	= (1<<Bitx)-1,			/* 0011 1111 */
+	Testx	= Maskx ^ 0xFF,			/* 1100 0000 */
+};
+
+static int32
+chartorune(int32 *rune, byte *str)
+{
+	int32 c, c1, c2;
+	int32 l;
+
+	/*
+	 * one character sequence
+	 *	00000-0007F => T1
+	 */
+	c = str[0];
+	if(c < Tx) {
+		*rune = c;
+		return 1;
+	}
+
+	/*
+	 * two character sequence
+	 *	0080-07FF => T2 Tx
+	 */
+	c1 = str[1] ^ Tx;
+	if(c1 & Testx)
+		goto bad;
+	if(c < T3) {
+		if(c < T2)
+			goto bad;
+		l = ((c << Bitx) | c1) & Rune2;
+		if(l <= Rune1)
+			goto bad;
+		*rune = l;
+		return 2;
+	}
+
+	/*
+	 * three character sequence
+	 *	0800-FFFF => T3 Tx Tx
+	 */
+	c2 = str[2] ^ Tx;
+	if(c2 & Testx)
+		goto bad;
+	if(c < T4) {
+		l = ((((c << Bitx) | c1) << Bitx) | c2) & Rune3;
+		if(l <= Rune2)
+			goto bad;
+		*rune = l;
+		return 3;
+	}
+
+	/*
+	 * bad decoding
+	 */
+bad:
+	*rune = Runeerror;
+	return 1;
+}