make all the commands agnostic about Rune width. (from sources)

3 files changed, 33 insertions, 28 deletions

diff --git a/sys/src/cmd/rc/glob.c b/sys/src/cmd/rc/glob.c
index 1c4983e40..295d7b6a6 100644
--- a/sys/src/cmd/rc/glob.c
+++ b/sys/src/cmd/rc/glob.c
@@ -118,18 +118,16 @@ glob(void *ap)
 int
 equtf(uchar *p, uchar *q)
 {
+	Rune pr, qr;
+ 
 	if(*p!=*q)
-		return 0;
-	if(twobyte(*p)) return p[1]==q[1];
-	if(threebyte(*p)){
-		if(p[1]!=q[1])
-			return 0;
-		if(p[1]=='\0')
-			return 1;	/* broken code at end of string! */
-		return p[2]==q[2];
-	}
-	return 1;
+ 		return 0;
+
+	chartorune(&pr, (char*)p);
+	chartorune(&qr, (char*)q);
+	return pr == qr;
 }
+
 /*
  * Return a pointer to the next utf code in the string,
  * not jumping past nuls in broken utf codes!
@@ -138,10 +136,11 @@ equtf(uchar *p, uchar *q)
 uchar*
 nextutf(uchar *p)
 {
-	if(twobyte(*p)) return p[1]=='\0'?p+1:p+2;
-	if(threebyte(*p)) return p[1]=='\0'?p+1:p[2]=='\0'?p+2:p+3;
-	return p+1;
+	Rune dummy;
+
+	return p + chartorune(&dummy, (char*)p);
 }
+
 /*
  * Convert the utf code at *p to a unicode value
  */
@@ -149,14 +148,12 @@ nextutf(uchar *p)
 int
 unicode(uchar *p)
 {
-	int u = *p;
+	Rune r;
 
-	if(twobyte(u))
-		return ((u&0x1f)<<6)|(p[1]&0x3f);
-	if(threebyte(u))
-		return (u<<12)|((p[1]&0x3f)<<6)|(p[2]&0x3f);
-	return u;
+	chartorune(&r, (char*)p);
+	return r;
 }
+
 /*
  * Does the string s match the pattern p
  * . and .. are only matched by patterns starting with .
diff --git a/sys/src/cmd/rc/lex.c b/sys/src/cmd/rc/lex.c
index 369348328..fecd0ec64 100644
--- a/sys/src/cmd/rc/lex.c
+++ b/sys/src/cmd/rc/lex.c
@@ -166,15 +166,25 @@ addtok(char *p, int val)
 char*
 addutf(char *p, int c)
 {
-	p = addtok(p, c);
-	if(twobyte(c))	 /* 2-byte escape */
-		return addtok(p, advance());
-	if(threebyte(c)){	/* 3-byte escape */
+	uchar b, m;
+	int i;
+
+	p = addtok(p, c);	/* 1-byte UTF runes are special */
+	if(onebyte(c))
+		return p;
+
+	m = 0xc0;
+	b = 0x80;
+	for(i=1; i < UTFmax; i++){
+		if((c&m) == b)
+			break;
 		p = addtok(p, advance());
-		return addtok(p, advance());
+		b = m;
+		m = (m >> 1)|0x80;
 	}
 	return p;
 }
+
 int lastdol;	/* was the last token read '$' or '$#' or '"'? */
 int lastword;	/* was the last token read a word or compound word terminator? */
 
diff --git a/sys/src/cmd/rc/rc.h b/sys/src/cmd/rc/rc.h
index 242a9b5ea..2e1d9ae59 100644
--- a/sys/src/cmd/rc/rc.h
+++ b/sys/src/cmd/rc/rc.h
@@ -123,12 +123,10 @@ int mypid;
  */
 #define	GLOB	((char)0x01)
 /*
- * onebyte(c), twobyte(c), threebyte(c)
- * Is c the first character of a one- two- or three-byte utf sequence?
+ * onebyte(c)
+ * Is c the first character of a one-byte utf sequence?
  */
 #define	onebyte(c)	((c&0x80)==0x00)
-#define	twobyte(c)	((c&0xe0)==0xc0)
-#define	threebyte(c)	((c&0xf0)==0xe0)
 
 char **argp;
 char **args;