3 files changed, 472 insertions(+), 189 deletions(-)

M port/devcons.c
M port/lib.h
M port/print.c

@@ 383,9 383,9 @@ int
 readnum(ulong off, char *buf, ulong n, ulong val, int size)
 {
 	char tmp[64];
-	Op op = (Op){ tmp, tmp+sizeof(tmp), &val, size-1, 0, FUNSIGN|FLONG };
+	Fconv fconv = (Fconv){ tmp, tmp+sizeof(tmp), size-1, 0, 0, 'd' };
 
-	numbconv(&op, 10);
+	numbconv(&val, &fconv);
 	tmp[size-1] = ' ';
 	if(off >= size)
 		return 0;

@@ 24,12 24,19 @@ extern	int	strncmp(char*, char*, long);
 extern	long	strlen(char*);
 extern	int	atoi(char*);
 
+enum
+{
+	UTFmax		= 3,		/* maximum bytes per rune */
+	Runesync	= 0x80,		/* cannot represent part of a UTF sequence (<) */
+	Runeself	= 0x80,		/* rune and UTF sequences are the same (<) */
+	Runeerror	= 0x80,		/* decoding error in UTF */
+};
+
 /*
  * rune routines
  */
 extern	int	runetochar(char*, Rune*);
 extern	int	chartorune(Rune*, char*);
-extern	int	countrune(char*);
 extern	char*	utfrune(char*, long);
 extern	int	utflen(char*);
 


@@ 38,26 45,20 @@ extern	int	abs(int);
 /*
  * print routines
  */
-
-#define	FUNSIGN	4
-#define	FSHORT	2
-#define	FLONG	1
-
-typedef struct Op	Op;
-struct Op
+typedef
+struct
 {
-	char	*p;
-	char	*ep;
-	void	*argp;
+	char*	out;		/* pointer to next output */
+	char*	eout;		/* pointer to end */
 	int	f1;
 	int	f2;
 	int	f3;
-};
-typedef void*	Fconv;	/* to allow fcall.h to compile */
-extern	void	strconv(char*, Op*, int, int);
-extern	int	numbconv(Op*, int);
+	int	chr;
+} Fconv;
+extern	void	strconv(char*, Fconv*);
+extern	int	numbconv(void*, Fconv*);
 extern	char	*doprint(char*, char*, char*, void*);
-extern	int	fmtinstall(char, int (*)(Op*));
+extern	int	fmtinstall(int, int (*)(void*, Fconv*));
 extern	int	sprint(char*, char*, ...);
 extern	int	print(char*, ...);
 

@@ 1,292 1,574 @@
 #include	"u.h"
 #include	"../port/lib.h"
 
+enum
+{
+	SIZE	= 1024,
+	IDIGIT	= 30,
+	MAXCONV	= 40,
+	FDIGIT	= 30,
+	FDEFLT	= 6,
+	NONE	= -1000,
+	MAXFMT	= 512,
+
+	FPLUS	= (1<<0),
+	FMINUS	= (1<<1),
+	FSHARP	= (1<<2),
+	FLONG	= (1<<3),
+	FSHORT	= (1<<4),
+	FUNSIGN	= (1<<5),
+	FVLONG	= (1<<6),
+};
+
 #define	PTR	sizeof(char*)
 #define	SHORT	sizeof(int)
 #define	INT	sizeof(int)
 #define	LONG	sizeof(long)
-#define	IDIGIT	30
-#define	MAXCON	30
+#define	VLONG	sizeof(vlong)
 
-static	int	convcount  = { 10 };
+int	printcol;
 
-#define	PUT(o, c)	if((o)->p < (o)->ep) *(o)->p++ = c
+static	int	convcount;
+static	char	fmtindex[MAXFMT];
 
-static	int	noconv(Op*);
-static	int	cconv(Op*);
-static	int	dconv(Op*);
-static	int	hconv(Op*);
-static	int	lconv(Op*);
-static	int	oconv(Op*);
-static	int	sconv(Op*);
-static	int	uconv(Op*);
-static	int	xconv(Op*);
-static	int	percent(Op*);
+static	int	noconv(void*, Fconv*);
+static	int	flags(void*, Fconv*);
+
+static	int	cconv(void*, Fconv*);
+static	int	sconv(void*, Fconv*);
+static	int	percent(void*, Fconv*);
+
+int	numbconv(void*, Fconv*);
 
 static
-int	(*fmtconv[MAXCON])(Op*) =
+int	(*fmtconv[MAXCONV])(void*, Fconv*) =
 {
-	noconv,
-	cconv, dconv, hconv, lconv,
-	oconv, sconv, uconv, xconv,
-	percent,
+	noconv
 };
+
 static
-char	fmtindex[128] =
+void
+initfmt(void)
 {
-	['c'] 1,
-	['d'] 2,
-	['h'] 3,
-	['l'] 4,
-	['o'] 5,
-	['s'] 6,
-	['u'] 7,
-	['x'] 8,
-	['%'] 9,
-};
+	int cc;
+
+	cc = 0;
+	fmtconv[cc] = noconv;
+	cc++;
+
+	fmtconv[cc] = flags;
+	fmtindex['+'] = cc;
+	fmtindex['-'] = cc;
+	fmtindex['#'] = cc;
+	fmtindex['h'] = cc;
+	fmtindex['l'] = cc;
+	fmtindex['u'] = cc;
+	cc++;
+
+	fmtconv[cc] = numbconv;
+	fmtindex['d'] = cc;
+	fmtindex['o'] = cc;
+	fmtindex['x'] = cc;
+	fmtindex['X'] = cc;
+	cc++;
+
+	fmtconv[cc] = cconv;
+	fmtindex['c'] = cc;
+	fmtindex['C'] = cc;
+	cc++;
+
+	fmtconv[cc] = sconv;
+	fmtindex['s'] = cc;
+	fmtindex['S'] = cc;
+	cc++;
+
+	fmtconv[cc] = percent;
+	fmtindex['%'] = cc;
+	cc++;
+
+	convcount = cc;
+}
 
 int
-fmtinstall(char c, int (*f)(Op*))
+fmtinstall(int c, int (*f)(void*, Fconv*))
 {
 
-	c &= 0177;
-	if(fmtindex[c] == 0) {
-		if(convcount >= MAXCON)
-			return 1;
-		fmtindex[c] = convcount++;
-	}
-	fmtconv[fmtindex[c]] = f;
+	if(convcount == 0)
+		initfmt();
+	if(c < 0 || c >= MAXFMT)
+		return 1;
+	if(convcount >= MAXCONV)
+		return 1;
+	fmtconv[convcount] = f;
+	fmtindex[c] = convcount;
+	convcount++;
 	return 0;
 }
 
 char*
-doprint(char *p, char *ep, char *fmt, void *argp)
+doprint(char *s, char *es, char *fmt, void *argp)
 {
-	int sf1, c;
-	Op o;
+	int n, c;
+	Rune rune;
+	Fconv local;
 
-	o.p = p;
-	o.ep = ep;
-	o.argp = argp;
+	local.out = s;
+	local.eout = es-4;		/* room for multi-byte character and 0 */
 
 loop:
-	c = *fmt++;
-	if(c != '%') {
-		if(c == 0) {
-			if(o.p < o.ep)
-				*o.p = 0;
-			return o.p;
-		}
-		PUT(&o, c);
+	c = *fmt & 0xff;
+	if(c >= Runeself) {
+		n = chartorune(&rune, fmt);
+		fmt += n;
+		c = rune;
+	} else
+		fmt++;
+	switch(c) {
+	case 0:
+		*local.out = 0;
+		return local.out;
+	
+	default:
+		printcol++;
+		goto common;
+
+	case '\n':
+		printcol = 0;
+		goto common;
+
+	case '\t':
+		printcol = (printcol+8) & ~7;
+		goto common;
+
+	common:
+		if(local.out < local.eout)
+			if(c >= Runeself) {
+				rune = c;
+				n = runetochar(local.out, &rune);	/* BUG */
+				local.out += n;
+			} else
+				*local.out++ = c;
 		goto loop;
+
+	case '%':
+		break;
 	}
-	o.f1 = 0;
-	o.f2 = -1;
-	o.f3 = 0;
-	c = *fmt++;
-	sf1 = 0;
-	if(c == '-') {
-		sf1 = 1;
-		c = *fmt++;
+	local.f1 = NONE;
+	local.f2 = NONE;
+	local.f3 = 0;
+
+	/*
+	 * read one of the following
+	 *	1. number, => f1, f2 in order.
+	 *	2. '*' same as number (from args)
+	 *	3. '.' ignored (separates numbers)
+	 *	4. flag => f3
+	 *	5. verb and terminate
+	 */
+l0:
+	c = *fmt & 0xff;
+	if(c >= Runeself) {
+		n = chartorune(&rune, fmt);
+		fmt += n;
+		c = rune;
+	} else
+		fmt++;
+
+l1:
+	if(c == 0) {
+		fmt--;
+		goto loop;
 	}
-	while(c >= '0' && c <= '9') {
-		o.f1 = o.f1*10 + c-'0';
-		c = *fmt++;
+	if(c == '.') {
+		if(local.f1 == NONE)
+			local.f1 = 0;
+		local.f2 = 0;
+		goto l0;
 	}
-	if(sf1)
-		o.f1 = -o.f1;
-	if(c != '.')
+	if((c >= '1' && c <= '9') ||
+	   (c == '0' && local.f1 != NONE)) {	/* '0' is a digit for f2 */
+		n = 0;
+		while(c >= '0' && c <= '9') {
+			n = n*10 + c-'0';
+			c = *fmt++;
+		}
+		if(local.f1 == NONE)
+			local.f1 = n;
+		else
+			local.f2 = n;
 		goto l1;
-	c = *fmt++;
-	while(c >= '0' && c <= '9') {
-		if(o.f2 < 0)
-			o.f2 = 0;
-		o.f2 = o.f2*10 + c-'0';
-		c = *fmt++;
 	}
-l1:
-	if(c == 0)
-		fmt--;
-	c = (*fmtconv[fmtindex[c&0177]])(&o);
-	if(c < 0) {
-		o.f3 |= -c;
-		c = *fmt++;
-		goto l1;
+	if(c == '*') {
+		n = *(int*)argp;
+		argp = (char*)argp + INT;
+		if(local.f1 == NONE)
+			local.f1 = n;
+		else
+			local.f2 = n;
+		goto l0;
 	}
-	o.argp = (char*)o.argp + c;
+	n = 0;
+	if(c >= 0 && c < MAXFMT)
+		n = fmtindex[c];
+	local.chr = c;
+	n = (*fmtconv[n])(argp, &local);
+	if(n < 0) {
+		local.f3 |= -n;
+		goto l0;
+	}
+	argp = (char*)argp + n;
 	goto loop;
 }
 
 int
-numbconv(Op *op, int base)
+numbconv(void *o, Fconv *fp)
 {
-	char b[IDIGIT];
-	int i, f, n, r;
-	long v;
+	char s[IDIGIT];
+	int i, f, n, r, b, ucase;
 	short h;
+	long v;
+	vlong vl;
+
+	SET(v);
+	SET(vl);
+
+	ucase = 0;
+	b = fp->chr;
+	switch(fp->chr) {
+	case 'u':
+		fp->f3 |= FUNSIGN;
+	case 'd':
+		b = 10;
+		break;
+
+	case 'o':
+		b = 8;
+		break;
+
+	case 'X':
+		ucase = 1;
+	case 'x':
+		b = 16;
+		break;
+	}
 
 	f = 0;
-	switch(op->f3 & (FLONG|FSHORT|FUNSIGN)) {
+	switch(fp->f3 & (FVLONG|FLONG|FSHORT|FUNSIGN)) {
+	case FVLONG|FLONG:
+		vl = *(vlong*)o;
+		r = VLONG;
+		break;
+
 	case FLONG:
-		v = *(long*)op->argp;
+		v = *(long*)o;
 		r = LONG;
 		break;
 
 	case FUNSIGN|FLONG:
-		v = *(unsigned long*)op->argp;
+		v = *(ulong*)o;
 		r = LONG;
 		break;
 
 	case FSHORT:
-		h = *(int*)op->argp;
+		h = *(int*)o;
 		v = h;
 		r = SHORT;
 		break;
 
 	case FUNSIGN|FSHORT:
-		h = *(int*)op->argp;
-		v = (unsigned short)h;
+		h = *(int*)o;
+		v = (ushort)h;
 		r = SHORT;
 		break;
 
 	default:
-		v = *(int*)op->argp;
+		v = *(int*)o;
 		r = INT;
 		break;
 
 	case FUNSIGN:
-		v = *(unsigned*)op->argp;
+		v = *(unsigned*)o;
 		r = INT;
 		break;
 	}
-	if(!(op->f3 & FUNSIGN) && v < 0) {
+	if(!(fp->f3 & FUNSIGN) && v < 0) {
 		v = -v;
 		f = 1;
 	}
-	b[IDIGIT-1] = 0;
+	s[IDIGIT-1] = 0;
 	for(i = IDIGIT-2;; i--) {
-		n = (unsigned long)v % base;
+		if(fp->f3 & FVLONG)
+			n = vl % b;
+		else
+			n = (ulong)v % b;
 		n += '0';
-		if(n > '9')
+		if(n > '9') {
 			n += 'a' - ('9'+1);
-		b[i] = n;
+			if(ucase)
+				n += 'A'-'a';
+		}
+		s[i] = n;
 		if(i < 2)
 			break;
-		v = (unsigned long)v / base;
-		if(op->f2 >= 0 && i >= IDIGIT-op->f2)
+		if(fp->f3 & FVLONG)
+			vl = vl / b;
+		else
+			v = (ulong)v / b;
+		if(fp->f2 != NONE && i >= IDIGIT-fp->f2)
 			continue;
+		if(fp->f3 & FVLONG)
+			if(vl <= 0)
+				break;
 		if(v <= 0)
 			break;
 	}
-sout:
+	if(fp->f3 & FSHARP)
+	if(s[i] != '0') {
+		if(b == 8)
+			s[--i] = '0';
+		else
+		if(b == 16) {
+			if(ucase)
+				s[--i] = 'X';
+			else
+				s[--i] = 'x';
+			s[--i] = '0';
+		}
+	}
 	if(f)
-		b[--i] = '-';
-	strconv(b+i, op, op->f1, -1);
+		s[--i] = '-';
+	fp->f2 = NONE;
+	strconv(s+i, fp);
 	return r;
 }
 
 void
-strconv(char *o, Op *op, int f1, int f2)
+Strconv(Rune *s, Fconv *fp)
 {
-	int n, c;
-	char *p;
-
-	n = strlen(o);
-	if(f1 >= 0)
-		while(n < f1) {
-			PUT(op, ' ');
+	int n, c, i;
+	Rune rune;
+
+	if(fp->f3 & FMINUS)
+		fp->f1 = -fp->f1;
+	n = 0;
+	if(fp->f1 != NONE && fp->f1 >= 0) {
+		for(; s[n]; n++)
+			;
+		while(n < fp->f1) {
+			if(fp->out < fp->eout)
+				*fp->out++ = ' ';
+			printcol++;
 			n++;
 		}
-	for(p=o; c = *p++;)
-		if(f2 != 0) {
-			PUT(op, c);
-			f2--;
+	}
+	for(;;) {
+		c = *s++;
+		if(c == 0)
+			break;
+		n++;
+		if(fp->f2 == NONE || fp->f2 > 0) {
+			if(fp->out < fp->eout)
+				if(c >= Runeself) {
+					rune = c;
+					i = runetochar(fp->out, &rune);
+					fp->out += i;
+				} else
+					*fp->out++ = c;
+			if(fp->f2 != NONE)
+				fp->f2--;
+			switch(c) {
+			default:
+				printcol++;
+				break;
+			case '\n':
+				printcol = 0;
+				break;
+			case '\t':
+				printcol = (printcol+8) & ~7;
+				break;
+			}
 		}
-	if(f1 < 0) {
-		f1 = -f1;
-		while(n < f1) {
-			PUT(op, ' ');
+	}
+	if(fp->f1 != NONE && fp->f1 < 0) {
+		fp->f1 = -fp->f1;
+		while(n < fp->f1) {
+			if(fp->out < fp->eout)
+				*fp->out++ = ' ';
+			printcol++;
 			n++;
 		}
 	}
 }
 
-static	int
-noconv(Op *op)
+void
+strconv(char *s, Fconv *fp)
 {
-
-	strconv("***", op, 0, -1);
-	return 0;
+	int n, c, i;
+	Rune rune;
+
+	if(fp->f3 & FMINUS)
+		fp->f1 = -fp->f1;
+	n = 0;
+	if(fp->f1 != NONE && fp->f1 >= 0) {
+		n = utflen(s);
+		while(n < fp->f1) {
+			if(fp->out < fp->eout)
+				*fp->out++ = ' ';
+			printcol++;
+			n++;
+		}
+	}
+	for(;;) {
+		c = *s & 0xff;
+		if(c >= Runeself) {
+			i = chartorune(&rune, s);
+			s += i;
+			c = rune;
+		} else
+			s++;
+		if(c == 0)
+			break;
+		n++;
+		if(fp->f2 == NONE || fp->f2 > 0) {
+			if(fp->out < fp->eout)
+				if(c >= Runeself) {
+					rune = c;
+					i = runetochar(fp->out, &rune);
+					fp->out += i;
+				} else
+					*fp->out++ = c;
+			if(fp->f2 != NONE)
+				fp->f2--;
+			switch(c) {
+			default:
+				printcol++;
+				break;
+			case '\n':
+				printcol = 0;
+				break;
+			case '\t':
+				printcol = (printcol+8) & ~7;
+				break;
+			}
+		}
+	}
+	if(fp->f1 != NONE && fp->f1 < 0) {
+		fp->f1 = -fp->f1;
+		while(n < fp->f1) {
+			if(fp->out < fp->eout)
+				*fp->out++ = ' ';
+			printcol++;
+			n++;
+		}
+	}
 }
 
-static	int
-cconv(Op *op)
+static
+int
+noconv(void *o, Fconv *fp)
 {
-	char b[2];
-
-	b[0] = *(int*)op->argp;
-	b[1] = 0;
-	strconv(b, op, op->f1, -1);
-	return INT;
+	int n;
+	char s[10];
+
+	if(convcount == 0) {
+		initfmt();
+		n = 0;
+		if(fp->chr >= 0 && fp->chr < MAXFMT)
+			n = fmtindex[fp->chr];
+		return (*fmtconv[n])(o, fp);
+	}
+	sprint(s, "*%c*", fp->chr);
+	fp->f1 = 0;
+	fp->f2 = NONE;
+	fp->f3 = 0;
+	strconv(s, fp);
+	return 0;
 }
 
-static	int
-dconv(Op *op)
+static
+int
+cconv(void *o, Fconv *fp)
 {
+	char s[10];
+	Rune rune;
 
-	return numbconv(op, 10);
-}
+	rune = *(int*)o;
+	if(fp->chr == 'c')
+		rune &= 0xff;
+	s[runetochar(s, &rune)] = 0;
 
-static	int
-hconv(Op *op)
-{
-
-	USED(op);
-	return -FSHORT;
+	fp->f2 = NONE;
+	strconv(s, fp);
+	return INT;
 }
 
-static	int
-lconv(Op *op)
+static
+int
+sconv(void *o, Fconv *fp)
 {
-
-	USED(op);
-	return -FLONG;
+	char *s;
+	Rune *r;
+
+	if(fp->chr == 's') {
+		s = *(char**)o;
+		if(s == 0)
+			s = "<null>";
+		strconv(s, fp);
+	} else {
+		r = *(Rune**)o;
+		if(r == 0)
+			r = L"<null>";
+		Strconv(r, fp);
+	}
+	return PTR;
 }
 
-static	int
-oconv(Op *op)
+static
+int
+percent(void *o, Fconv *fp)
 {
 
-	return numbconv(op, 8);
+	USED(o);
+	if(fp->out < fp->eout)
+		*fp->out++ = '%';
+	return 0;
 }
 
-static	int
-sconv(Op *op)
+static
+int
+flags(void *o, Fconv *fp)
 {
+	int f;
 
-	strconv(*(char**)op->argp, op, op->f1, op->f2);
-	return PTR;
-}
+	USED(o);
 
-static	int
-uconv(Op *op)
-{
+	f = 0;
+	switch(fp->chr) {
+	case '+':
+		f = FPLUS;
+		break;
 
-	USED(op);
-	return -FUNSIGN;
-}
+	case '-':
+		f = FMINUS;
+		break;
 
-static	int
-xconv(Op *op)
-{
+	case '#':
+		f = FSHARP;
+		break;
 
-	return numbconv(op, 16);
-}
+	case 'h':
+		f = FSHORT;
+		break;
 
-static	int
-percent(Op *op)
-{
+	case 'l':
+		f = FLONG;
+		if(fp->f3 & FLONG)
+			f = FVLONG;
+		break;
 
-	PUT(op, '%');
-	return 0;
+	case 'u':
+		f = FUNSIGN;
+		break;
+	}
+	return -f;
 }