handle 16-bit unicode from Windows in clipboard and fs names

4 files changed, 61 insertions, 16 deletions

diff --git a/emu/Nt/devfs.c b/emu/Nt/devfs.c
index 6d79c35b..2e61e760 100644
--- a/emu/Nt/devfs.c
+++ b/emu/Nt/devfs.c
@@ -1659,10 +1659,10 @@ secstat(Dir *dir, char *file, Rune16 *srv)
 		free(sd);
 	if(ok){
 		dir->mode = st.mode;
-		n = runenlen(st.owner->name, runes16len(st.owner->name));
+		n = rune16nlen(st.owner->name, runes16len(st.owner->name));
 		dir->uid = smalloc(n+1);
 		runes16toutf(dir->uid, st.owner->name, n+1);
-		n = runenlen(st.group->name, runes16len(st.group->name));
+		n = rune16nlen(st.group->name, runes16len(st.group->name));
 		dir->gid = smalloc(n+1);
 		runes16toutf(dir->gid, st.group->name, n+1);
 	}
@@ -1688,7 +1688,7 @@ secsize(char *file, Rune16 *srv)
 	if(sd != (void*)sdrock)
 		free(sd);
 	if(ok)
-		return runenlen(st.owner->name, runes16len(st.owner->name))+runenlen(st.group->name, runes16len(st.group->name));
+		return rune16nlen(st.owner->name, runes16len(st.owner->name))+rune16nlen(st.group->name, runes16len(st.group->name));
 	return -1;
 }
 
diff --git a/emu/Nt/r16.c b/emu/Nt/r16.c
index cdbc13ac..92a36521 100644
--- a/emu/Nt/r16.c
+++ b/emu/Nt/r16.c
@@ -9,6 +9,28 @@
 #include	"error.h"
 #include	"r16.h"
 
+#define Bit(i) (7-(i))
+/* N 0's preceded by i 1's, T(Bit(2)) is 1100 0000 */
+#define T(i) (((1 << (Bit(i)+1))-1) ^ 0xFF)
+/* 0000 0000 0000 0111 1111 1111 */
+#define	RuneX(i) ((1 << (Bit(i) + ((i)-1)*Bitx))-1)
+
+enum
+{
+	Bitx	= Bit(1),
+
+	Tx	= T(1),			/* 1000 0000 */
+	Rune1 = (1<<(Bit(0)+0*Bitx))-1,	/* 0000 0000 0000 0000 0111 1111 */
+
+	Maskx	= (1<<Bitx)-1,		/* 0011 1111 */
+	Testx	= Maskx ^ 0xFF,		/* 1100 0000 */
+
+	SurrogateMin	= 0xD800,
+	SurrogateMax	= 0xDFFF,
+
+	Bad	= Runeerror,
+};
+
 Rune16*
 runes16dup(Rune16 *r)
 {
@@ -59,6 +81,28 @@ runes16toutf(char *p, Rune16 *r, int nc)
 	return op;
 }
 
+int
+rune16nlen(Rune16 *r, int nrune)
+{
+	int nb, i;
+	Rune c;
+
+	nb = 0;
+	while(nrune--) {
+		c = *r++;
+		if(c <= Rune1){
+			nb++;
+		} else {
+			for(i = 2; i < UTFmax + 1; i++)
+				if(c <= RuneX(i) || i == UTFmax){
+					nb += i;
+					break;
+				}
+		}
+	}
+	return nb;
+}
+
 Rune16*
 utftorunes16(Rune16 *r, char *p, int nc)
 {
diff --git a/emu/Nt/r16.h b/emu/Nt/r16.h
index de09eacc..ea4ee286 100644
--- a/emu/Nt/r16.h
+++ b/emu/Nt/r16.h
@@ -1,10 +1,11 @@
 typedef unsigned short Rune16;
 
-	wchar_t	*widen(char *s);
-	char		*narrowen(wchar_t *ws);
-	int		widebytes(wchar_t *ws);
-	int		runes16len(Rune16*);
-	Rune16*	runes16dup(Rune16*);
-	Rune16*	utftorunes16(Rune16*, char*, int);
-	char*	runes16toutf(char*, Rune16*, int);
-	int		runes16cmp(Rune16*, Rune16*);
+wchar_t	*widen(char *s);
+char		*narrowen(wchar_t *ws);
+int		widebytes(wchar_t *ws);
+int		runes16len(Rune16*);
+int		rune16nlen(Rune16*, int);
+Rune16*	runes16dup(Rune16*);
+Rune16*	utftorunes16(Rune16*, char*, int);
+char*	runes16toutf(char*, Rune16*, int);
+int		runes16cmp(Rune16*, Rune16*);
diff --git a/emu/Nt/win.c b/emu/Nt/win.c
index b6330b16..c1bccfa7 100644
--- a/emu/Nt/win.c
+++ b/emu/Nt/win.c
@@ -32,10 +32,10 @@
 #include	<draw.h>
 #include	"keyboard.h"
 #include	"cursor.h"
+#include	"r16.h"
 
 extern ulong displaychan;
 
-extern	char*	runes16toutf(char*, Rune*, int);
 extern	int	bytesperline(Rectangle, int);
 extern	int	main(int argc, char **argv);
 static	void	dprint(char*, ...);
@@ -698,12 +698,12 @@ drawcursor(Drawcursor* c)
 static char*
 clipreadunicode(HANDLE h)
 {
-	Rune *p;
+	Rune16 *p;
 	int n;
 	char *q;
 	
 	p = GlobalLock(h);
-	n = runenlen(p, runestrlen(p)+1);
+	n = rune16nlen(p, runes16len(p)+1);
 	q = malloc(n);
 	if(q != nil)
 		runes16toutf(q, p, n);
@@ -753,7 +753,7 @@ clipwrite(char *buf)
 {
 	HANDLE h;
 	char *p;
-	Rune *rp;
+	Rune16 *rp;
 	int n;
 
 	n = 0;
@@ -767,7 +767,7 @@ clipwrite(char *buf)
 		return -1;
 	}
 
-	h = GlobalAlloc(GMEM_MOVEABLE|GMEM_DDESHARE, (n+1)*sizeof(Rune));
+	h = GlobalAlloc(GMEM_MOVEABLE|GMEM_DDESHARE, (n+1)*sizeof(Rune16));
 	if(h == NULL)
 		error(Enovmem);
 	rp = GlobalLock(h);