gc: implement []int(string) and []byte(string)

R=ken2
CC=golang-dev
https://golang.org/cl/224060

8 files changed, 175 insertions, 14 deletions

diff --git a/src/cmd/gc/builtin.c.boot b/src/cmd/gc/builtin.c.boot
index 6eed40230d..3114d75807 100644
--- a/src/cmd/gc/builtin.c.boot
+++ b/src/cmd/gc/builtin.c.boot
@@ -26,6 +26,8 @@ char *runtimeimport =
 	"func \"\".intstring (? int64) string\n"
 	"func \"\".slicebytetostring (? []uint8) string\n"
 	"func \"\".sliceinttostring (? []int) string\n"
+	"func \"\".stringtoslicebyte (? string) []uint8\n"
+	"func \"\".stringtosliceint (? string) []int\n"
 	"func \"\".stringiter (? string, ? int) int\n"
 	"func \"\".stringiter2 (? string, ? int) (retk int, retv int)\n"
 	"func \"\".slicecopy (to any, fr any, wid uint32) int\n"
diff --git a/src/cmd/gc/go.h b/src/cmd/gc/go.h
index 753de0399a..cbcdc9c39d 100644
--- a/src/cmd/gc/go.h
+++ b/src/cmd/gc/go.h
@@ -351,6 +351,7 @@ enum
 	OAPPENDSTR,
 	OARRAY,
 	OARRAYBYTESTR, OARRAYRUNESTR,
+	OSTRARRAYBYTE, OSTRARRAYRUNE,
 	OAS, OAS2, OAS2MAPW, OAS2FUNC, OAS2RECV, OAS2MAPR, OAS2DOTTYPE, OASOP,
 	OBAD,
 	OCALL, OCALLFUNC, OCALLMETH, OCALLINTER,
@@ -411,7 +412,7 @@ enum
 	OTINTER,
 	OTFUNC,
 	OTARRAY,
-	
+
 	// misc
 	ODDD,
 
@@ -458,7 +459,7 @@ enum
 	TIDEAL,			// 32
 	TNIL,
 	TBLANK,
-	
+
 	// pseudo-type for frame layout
 	TFUNCARGS,
 	TCHANARGS,
diff --git a/src/cmd/gc/runtime.go b/src/cmd/gc/runtime.go
index e350c282dd..ca3b6a1bc1 100644
--- a/src/cmd/gc/runtime.go
+++ b/src/cmd/gc/runtime.go
@@ -38,6 +38,8 @@ func indexstring(string, int) byte
 func intstring(int64) string
 func slicebytetostring([]byte) string
 func sliceinttostring([]int) string
+func stringtoslicebyte(string) []byte
+func stringtosliceint(string) []int
 func stringiter(string, int) int
 func stringiter2(string, int) (retk int, retv int)
 func slicecopy(to any, fr any, wid uint32) int
diff --git a/src/cmd/gc/typecheck.c b/src/cmd/gc/typecheck.c
index 158dee6738..4c4c928338 100644
--- a/src/cmd/gc/typecheck.c
+++ b/src/cmd/gc/typecheck.c
@@ -32,6 +32,7 @@ static void	checklvalue(Node*, char*);
 static void	checkassign(Node*);
 static void	checkassignlist(NodeList*);
 static void	toslice(Node**);
+static void stringtoarraylit(Node**);
 
 void
 typechecklist(NodeList *l, int top)
@@ -835,6 +836,13 @@ reswitch:
 		n = typecheckconv(n, n->left, n->type, 1, "conversion");
 		if(n->type == T)
 			goto error;
+		switch(n->op) {
+		case OSTRARRAYBYTE:
+		case OSTRARRAYRUNE:
+			if(n->left->op == OLITERAL)
+				stringtoarraylit(&n);
+			break;
+		}
 		goto ret;
 
 	case OMAKE:
@@ -1406,6 +1414,18 @@ checkconv(Type *nt, Type *t, int explicit, int *op, int *et, char *desc)
 		}
 	}
 
+	// from string
+	if(istype(nt, TSTRING) && isslice(t) && t->sym == S) {
+		switch(t->type->etype) {
+		case TUINT8:
+			*op = OSTRARRAYBYTE;
+			return 1;
+		case TINT:
+			*op = OSTRARRAYRUNE;
+			return 1;
+		}
+	}
+
 	// convert to unsafe pointer
 	if(isptrto(t, TANY)
 	&& (isptr[nt->etype] || nt->etype == TUINTPTR))
@@ -1534,7 +1554,7 @@ typecheckaste(int op, Type *tstruct, NodeList *nl, char *desc)
 				// TODO(rsc): drop first if in DDD cleanup
 				if(t->etype != TINTER)
 				if(checkconv(nl->n->type, t->type, 0, &xx, &yy, desc) < 0)
-					yyerror("cannot use %+N as type %T in %s", nl->n, t->type, desc);					
+					yyerror("cannot use %+N as type %T in %s", nl->n, t->type, desc);
 			}
 			goto out;
 		}
@@ -1587,7 +1607,7 @@ exportassignok(Type *t, char *desc)
 			// it only happens for fields in a ... struct.
 			if(s != nil && !exportname(s->name) && s->pkg != localpkg) {
 				char *prefix;
-				
+
 				prefix = "";
 				if(desc != nil)
 					prefix = " in ";
@@ -2164,3 +2184,39 @@ typecheckfunc(Node *n)
 	if(rcvr != nil && n->shortname != N && !isblank(n->shortname))
 		addmethod(n->shortname->sym, t, 1);
 }
+
+static void
+stringtoarraylit(Node **np)
+{
+	int32 i;
+	NodeList *l;
+	Strlit *s;
+	char *p, *ep;
+	Rune r;
+	Node *nn, *n;
+
+	n = *np;
+	if(n->left->op != OLITERAL || n->left->val.ctype != CTSTR)
+		fatal("stringtoarraylit %N", n);
+
+	s = n->left->val.u.sval;
+	l = nil;
+	p = s->s;
+	ep = s->s + s->len;
+	i = 0;
+	if(n->type->type->etype == TUINT8) {
+		// raw []byte
+		while(p < ep)
+			l = list(l, nod(OKEY, nodintconst(i++), nodintconst((uchar)*p++)));
+	} else {
+		// utf-8 []int
+		while(p < ep) {
+			p += chartorune(&r, p);
+			l = list(l, nod(OKEY, nodintconst(i++), nodintconst(r)));
+		}
+	}
+	nn = nod(OCOMPLIT, N, typenod(n->type));
+	nn->list = l;
+	typecheck(&nn, Erv);
+	*np = nn;
+}
diff --git a/src/cmd/gc/walk.c b/src/cmd/gc/walk.c
index e902600ba0..fa63646c50 100644
--- a/src/cmd/gc/walk.c
+++ b/src/cmd/gc/walk.c
@@ -8,6 +8,7 @@ static	Node*	walkprint(Node*, NodeList**, int);
 static	Node*	conv(Node*, Type*);
 static	Node*	mapfn(char*, Type*);
 static	Node*	makenewvar(Type*, NodeList**, Node**);
+
 enum
 {
 	Inone,
@@ -122,7 +123,7 @@ static void
 domethod(Node *n)
 {
 	Node *nt;
-	
+
 	nt = n->type->nname;
 	typecheck(&nt, Etype);
 	if(nt->type == T) {
@@ -142,7 +143,7 @@ walkdeftype(Node *n)
 	int maplineno, embedlineno, lno;
 	Type *t;
 	NodeList *l;
-	
+
 	nwalkdeftype++;
 	lno = lineno;
 	setlineno(n);
@@ -183,7 +184,7 @@ walkdeftype(Node *n)
 
 ret:
 	lineno = lno;
-	
+
 	// if there are no type definitions going on, it's safe to
 	// try to resolve the method types for the interfaces
 	// we just read.
@@ -868,7 +869,7 @@ walkexpr(Node **np, NodeList **init)
 	case OINDEX:
 		walkexpr(&n->left, init);
 		walkexpr(&n->right, init);
-		
+
 		// if range of type cannot exceed static array bound,
 		// disable bounds check
 		if(!isslice(n->left->type))
@@ -1092,10 +1093,20 @@ walkexpr(Node **np, NodeList **init)
 		goto ret;
 
 	case OARRAYRUNESTR:
-		// sliceinttostring([]byte) string;
+		// sliceinttostring([]int) string;
 		n = mkcall("sliceinttostring", n->type, init, n->left);
 		goto ret;
 
+	case OSTRARRAYBYTE:
+		// stringtoslicebyte(string) []byte;
+		n = mkcall("stringtoslicebyte", n->type, init, n->left);
+		goto ret;
+
+	case OSTRARRAYRUNE:
+		// stringtosliceint(string) []int
+		n = mkcall("stringtosliceint", n->type, init, n->left);
+		goto ret;
+
 	case OCMPIFACE:
 		// ifaceeq(i1 any-1, i2 any-2) (ret bool);
 		if(!eqtype(n->left->type, n->right->type))
@@ -1117,6 +1128,7 @@ walkexpr(Node **np, NodeList **init)
 	case OARRAYLIT:
 	case OMAPLIT:
 	case OSTRUCTLIT:
+	arraylit:
 		nvar = nod(OXXX, N, N);
 		tempname(nvar, n->type);
 		anylit(n, nvar, init);
@@ -1448,18 +1460,18 @@ mkdotargslice(NodeList *lr0, NodeList *nn, Type *l, int fp, NodeList **init)
 {
 	Node *a, *n;
 	Type *tslice;
-	
+
 	tslice = typ(TARRAY);
 	tslice->type = l->type->type;
 	tslice->bound = -1;
-	
+
 	n = nod(OCOMPLIT, N, typenod(tslice));
 	n->list = lr0;
 	typecheck(&n, Erv);
 	if(n->type == T)
 		fatal("mkdotargslice: typecheck failed");
 	walkexpr(&n, init);
-	
+
 	a = nod(OAS, nodarg(l, fp), n);
 	nn = list(nn, convas(a, init));
 	return nn;
@@ -1758,7 +1770,7 @@ walkprint(Node *nn, NodeList **init, int defer)
 			n = nod(OCONV, n, N);
 			n->type = t;
 		}
-		
+
 		if(defer) {
 			intypes = list(intypes, nod(ODCLFIELD, N, typenod(t)));
 			args = list(args, n);
@@ -1788,7 +1800,7 @@ walkprint(Node *nn, NodeList **init, int defer)
 			calls = list(calls, mkcall("printnl", T, nil));
 		typechecklist(calls, Etop);
 		walkexprlist(calls, init);
-	
+
 		if(op == OPANIC || op == OPANICN)
 			r = mkcall("panicl", T, nil);
 		else
diff --git a/src/pkg/runtime/string.cgo b/src/pkg/runtime/string.cgo
index 4c85766c2f..2cb518c6f8 100644
--- a/src/pkg/runtime/string.cgo
+++ b/src/pkg/runtime/string.cgo
@@ -4,6 +4,7 @@
 
 package runtime
 #include "runtime.h"
+#include "malloc.h"
 
 String	emptystring;
 
@@ -210,6 +211,12 @@ func slicebytetostring(b Slice) (s String) {
 	mcpy(s.str, b.array, s.len);
 }
 
+func stringtoslicebyte(s String) (b Slice) {
+	b.array = mallocgc(s.len, RefNoPointers, 1, 1);
+	b.len = s.len;
+	b.cap = s.len;
+	mcpy(b.array, s.str, s.len);
+}
 
 func sliceinttostring(b Slice) (s String) {
 	int32 siz1, siz2, i;
@@ -233,6 +240,30 @@ func sliceinttostring(b Slice) (s String) {
 	s.len = siz2;
 }
 
+func stringtosliceint(s String) (b Slice) {
+	int32 n;
+	int32 dum, *r;
+	uint8 *p, *ep;
+
+	// two passes.
+	// unlike sliceinttostring, no race because strings are immutable.
+	p = s.str;
+	ep = s.str+s.len;
+	n = 0;
+	while(p < ep) {
+		p += charntorune(&dum, p, ep-p);
+		n++;
+	}
+
+	b.array = mallocgc(n*sizeof(r[0]), RefNoPointers, 1, 1);
+	b.len = n;
+	b.cap = n;
+	p = s.str;
+	r = (int32*)b.array;
+	while(p < ep)
+		p += charntorune(r++, p, ep-p);
+}
+
 enum
 {
 	Runeself	= 0x80,
diff --git a/test/convlit.go b/test/convlit.go
index e65dad3df6..22415bb324 100644
--- a/test/convlit.go
+++ b/test/convlit.go
@@ -35,3 +35,30 @@ var good2 int = 1.0;
 var good3 int = 1e9;
 var good4 float = 1e20;
 
+// explicit conversion of string is okay
+var _ = []int("abc")
+var _ = []byte("abc")
+
+// implicit is not
+var _ []int = "abc"	// ERROR "cannot use|incompatible|invalid"
+var _ []byte = "abc"	// ERROR "cannot use|incompatible|invalid"
+
+// named string is okay
+type Tstring string
+var ss Tstring = "abc"
+var _ = []int(ss)
+var _ = []byte(ss)
+
+// implicit is still not
+var _ []int = ss	// ERROR "cannot use|incompatible|invalid"
+var _ []byte = ss	// ERROR "cannot use|incompatible|invalid"
+
+// named slice is not
+type Tint []int
+type Tbyte []byte
+var _ = Tint("abc")	// ERROR "convert|incompatible|invalid"
+var _ = Tbyte("abc")	// ERROR "convert|incompatible|invalid"
+
+// implicit is still not
+var _ Tint = "abc"	// ERROR "cannot use|incompatible|invalid"
+var _ Tbyte = "abc"	// ERROR "cannot use|incompatible|invalid"
diff --git a/test/string_lit.go b/test/string_lit.go
index 547be8003a..88b5d251ff 100644
--- a/test/string_lit.go
+++ b/test/string_lit.go
@@ -34,6 +34,19 @@ func assert(a, b, c string) {
 	}
 }
 
+const (
+	gx1 = "aä本☺"
+	gx2 = "aä\xFF\xFF本☺"
+	gx2fix = "aä\uFFFD\uFFFD本☺"
+)
+
+var (
+	gr1 = []int(gx1)
+	gr2 = []int(gx2)
+	gb1 = []byte(gx1)
+	gb2 = []byte(gx2)
+)
+
 func main() {
 	ecode = 0;
 	s :=
@@ -86,5 +99,22 @@ func main() {
 	r = 0x10ffff + 1;
 	s = string(r);
 	assert(s, "\xef\xbf\xbd", "too-large rune");
+
+	assert(string(gr1), gx1, "global ->[]int")
+	assert(string(gr2), gx2fix, "global invalid ->[]int")
+	assert(string(gb1), gx1, "->[]byte")
+	assert(string(gb2), gx2, "global invalid ->[]byte")
+
+	var (
+		r1 = []int(gx1)
+		r2 = []int(gx2)
+		b1 = []byte(gx1)
+		b2 = []byte(gx2)
+	)
+	assert(string(r1), gx1, "->[]int")
+	assert(string(r2), gx2fix, "invalid ->[]int")
+	assert(string(b1), gx1, "->[]byte")
+	assert(string(b2), gx2, "invalid ->[]byte")
+
 	os.Exit(ecode);
 }