major internal changes in representation of character cells.

we now use 12 bytes per cell instead of 10. however, this allows us to
support 256-color mode (not yet implemented but the framework is in
place) and to mix scripts when using combining characters. while the
latter sounds ridiculous at first, being able to visibly see a
combining letter attached to a [, ", or ' is extremely useful in
scripting and regular expressions with some languages.

some code is left slightly messy, but overall it's much cleaner now
since struct uucell is now properly encapsulated.

7 files changed, 52 insertions, 264 deletions

diff --git a/Makefile b/Makefile
index 9dfc429..e3d3bb9 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
 # uuterm, Copyright (C) 2006 Rich Felker; licensed under GNU GPL v2 only
 
 
-SRCS = main.c term.c comb.c decomp.c tty.c alloc.c refresh.c ascii.c ucf.c font_load.c
+SRCS = main.c term.c cell.c decomp.c tty.c alloc.c refresh.c ascii.c ucf.c font_load.c
 
 OBJS_FB = fbcon.o dblbuf.o
 OBJS_X11 = xlib.o
diff --git a/comb.c b/comb.c
deleted file mode 100644
index 6bf7f7d..0000000
--- a/comb.c
+++ /dev/null
@@ -1,198 +0,0 @@
-/* uuterm, Copyright (C) 2006 Rich Felker; licensed under GNU GPL v2 only */
-
-#define R(a,b) { (a), (b)-(a) }
-
-static const unsigned short common[][2] = {
-	R( 0x300, 0x341 ),
-	R( 0x346, 0x362 ),
-	R( 0x200B, 0x200F ),
-	R( 0x202A, 0x202E ),
-	R( 0x2060, 0x206F ),
-	R( 0x20D0, 0x20EA ),
-	{ 0, 0 }
-};
-
-static const unsigned short latin[][2] = {
-	R( 0x363, 0x36F ),
-	{ 0, 0 }
-};
-
-static const unsigned short greek[][2] = {
-	R( 0x342, 0x345 ),
-	{ 0, 0 }
-};
-
-static const unsigned short cyrillic[][2] = {
-	R( 0x483, 0x489 ),
-	{ 0, 0 }
-};
-
-static const unsigned short hebrew[][2] = {
-	R( 0x591, 0x5C4 ),
-	{ 0, 0 }
-};
-
-static const unsigned short arabic[][2] = {
-	R( 0x600, 0x603 ),
-	R( 0x610, 0x615 ),
-	R( 0x64B, 0x658 ),
-	R( 0x670, 0x670 ),
-	R( 0x6D6, 0x6ED ),
-	{ 0, 0 }
-};
-
-static const unsigned short syriac[][2] = {
-	R( 0x70F, 0x711 ),
-	R( 0x730, 0x74A ),
-	{ 0, 0 }
-};
-
-static const unsigned short thaana[][2] = {
-	R( 0x7A6, 0x7B0 ),
-	{ 0, 0 }
-};
-
-static const unsigned short devanagari[][2] = {
-	R( 0x901, 0x902 ),
-	R( 0x93C, 0x963 ),
-	{ 0, 0 }
-};
-
-static const unsigned short bengali[][2] = {
-	R( 0x981, 0x981 ),
-	R( 0x9BC, 0x9E3 ),
-	{ 0, 0 }
-};
-
-static const unsigned short gurmukhi[][2] = {
-	R( 0xA01, 0xA02 ),
-	R( 0xA3C, 0xA4D ),
-	R( 0xA70, 0xA71 ),
-	{ 0, 0 }
-};
-
-static const unsigned short gujarati[][2] = {
-	R( 0xA81, 0xA82 ),
-	R( 0xABC, 0xAE3 ),
-	{ 0, 0 }
-};
-
-static const unsigned short oriya[][2] = {
-	R( 0xB01, 0xB01 ),
-	R( 0xB3C, 0xB4D ),
-	R( 0xB56, 0xB56 ),
-	{ 0, 0 }
-};
-
-static const unsigned short tamil[][2] = {
-	R( 0xB82, 0xB82 ),
-	R( 0xBC0, 0xBCD ),
-	{ 0, 0 }
-};
-
-static const unsigned short telugu[][2] = {
-	R( 0xC3E, 0xC56 ),
-	{ 0, 0 }
-};
-
-static const unsigned short kannada[][2] = {
-	R( 0xCBC, 0xCCD ),
-	{ 0, 0 }
-};
-
-static const unsigned short malayalam[][2] = {
-	R( 0xD41, 0xD4D ),
-	{ 0, 0 }
-};
-
-static const unsigned short sinhala[][2] = {
-	R( 0xDCA, 0xDD6 ),
-	{ 0, 0 }
-};
-
-static const unsigned short thai[][2] = {
-	R( 0xE31, 0xE3A ),
-	R( 0xE47, 0xE4E ),
-	{ 0, 0 }
-};
-
-static const unsigned short lao[][2] = {
-	R( 0xEB1, 0xECD ),
-	{ 0, 0 }
-};
-
-static const unsigned short tibetan[][2] = {
-	R( 0xF18, 0xF19 ),
-	R( 0xF35, 0xF35 ),
-	R( 0xF39, 0xF39 ),
-	R( 0xF71, 0xF84 ),
-	R( 0xF90, 0xFBC ),
-	R( 0xFC6, 0xFC6 ),
-	{ 0, 0 }
-};
-
-static const unsigned short burmese[][2] = {
-	R( 0x102D, 0x1039 ),
-	R( 0x1058, 0x1059 ),
-	{ 0, 0 }
-};
-
-static const unsigned short misc_scripts[][2] = {
-	R( 0x1732, 0x1734 ), /* hanunoo */
-	R( 0x1752, 0x1753 ), /* buhid */
-	R( 0x17B4, 0x17BD ), /* khmer */
-	R( 0x17C6, 0x17D3 ),
-	R( 0x17DD, 0x17DD ),
-	R( 0x18A9, 0x18A9 ), /* mongolian */
-	R( 0x1920, 0x193B ), /* limbu (can be broken down more) */
-	{ 0, 0 }
-};
-
-#undef R
-#define R(a,b,s) { (a), (b)-(a), (s) }
-
-static const struct {
-	unsigned a, l;
-	const unsigned short (*r)[2];
-} scripts[] = {
-	R( 0x400, 0x52F, cyrillic ),
-	R( 0x590, 0x5FF, hebrew ),
-	R( 0x600, 0x6FF, arabic ),
-	R( 0x700, 0x74F, syriac ),
-	R( 0x780, 0x7B1, thaana ),
-	R( 0x900, 0x97F, devanagari ),
-	R( 0x980, 0x9FF, bengali ),
-	R( 0xA00, 0xA7F, gurmukhi ),
-	R( 0xA80, 0xAFF, gujarati ),
-	R( 0xB00, 0xB7F, oriya ),
-	R( 0xB80, 0xBFF, tamil ),
-	R( 0xC00, 0xC7F, telugu ),
-	R( 0xC80, 0xCFF, kannada ),
-	R( 0xD00, 0xD7F, malayalam ),
-	R( 0xD80, 0xDFF, sinhala ),
-	R( 0xE00, 0xE7F, thai ),
-	R( 0xE80, 0xEFF, lao ),
-	R( 0xF00, 0xFFF, tibetan ),
-	R( 0x1000, 0x108F, burmese ),
-	R( 0x1720, 0x19FF, misc_scripts ),
-	R( 0x1D2B, 0x1D2B, cyrillic ),
-	R( 0x0000, 0x10FFFF, common ),
-	{ }
-};
-
-#undef R
-
-int uu_combine_involution(unsigned b, unsigned c)
-{
-	int i;
-	unsigned code = 1;
-	const unsigned short (*r)[2];
-	for (i=0; scripts[i].l; i++)
-		if (b - scripts[i].a <= scripts[i].l)
-			for (r = scripts[i].r; r[0][0]; code += r++[0][1]+1)
-				if (c - r[0][0] <= r[0][1])
-					return c - r[0][0] + code;
-				else if (c - code <= r[0][1])
-					return c + r[0][0] - code;
-	return 0;
-}
diff --git a/dblbuf.c b/dblbuf.c
index 32ff996..18e2cd5 100644
--- a/dblbuf.c
+++ b/dblbuf.c
@@ -117,8 +117,8 @@ void uudisp_predraw_cell(struct uudisp *d, int idx, int x, int color)
 {
 	struct dblbuf *b = (void *)&d->priv;
 
-	b->slices[idx].colors[2*x] = expand_color(d, color&15);
-	b->slices[idx].colors[2*x+1] = expand_color(d, color>>4) ^ b->slices[idx].colors[2*x];
+	b->slices[idx].colors[2*x] = expand_color(d, color&255);
+	b->slices[idx].colors[2*x+1] = expand_color(d, color>>8) ^ b->slices[idx].colors[2*x];
 }
 
 void uudisp_draw_glyph(struct uudisp *d, int idx, int x, const void *glyph)
diff --git a/refresh.c b/refresh.c
index 9e3ab57..06e6f69 100644
--- a/refresh.c
+++ b/refresh.c
@@ -7,16 +7,17 @@
 
 static void extract_cell(unsigned *ch, size_t max, struct uucell *cell)
 {
-	int i, l;
-	unsigned b;
-	for (b=i=0; i<3; i++) b |= cell->c[i] << 8*i;
-	l = uu_decompose_char(b, ch, max);
-	ch += l; max -= l;
-	for (; i<sizeof(cell->c) && cell->c[i]; i++) {
-		l = uu_decompose_char(uu_combine_involution(b, cell->c[i]), ch, max);
+	int i, j, l;
+	wchar_t ws[8];
+	int attr = uucell_get_attr(cell);
+
+	uucell_get_wcs(cell, ws, 8);
+
+	for (i=j=0; ws[i]; i++) {
+		l = uu_decompose_char(ws[i], ch, max);
 		ch += l; max -= l;
 	}
-	if ((cell->a & UU_ATTR_UL) && max)
+	if ((attr & UU_ATTR_UL) && max)
 		max--, *ch++ = '_'; //0x0332;
 	for (; max; max--) *ch++ = 0;
 	ch[-1] = 0;
@@ -54,7 +55,7 @@ void uuterm_refresh_row(struct uudisp *d, struct uurow *row, int x1, int x2)
 		} else {
 			width = 1; part = 0; chp = ch[x&3];
 		}
-		uudisp_predraw_cell(d, row->idx, x, row->cells[x].a & 0xff);
+		uudisp_predraw_cell(d, row->idx, x, uucell_get_color(&row->cells[x]));
 		for (i=0; i<sizeof(ch[0]) && chp[i]; i++) {
 			const void *glyph = lookup_glyph(d->font, i, chp, ch[(x+3)&3], ch[(x+1)&3], width, part);
 			if (glyph) uudisp_draw_glyph(d, row->idx, x, glyph);
diff --git a/term.c b/term.c
index 83a7c42..1bfc4c7 100644
--- a/term.c
+++ b/term.c
@@ -61,7 +61,8 @@ static void reset(struct uuterm *t)
 {
 	/* cheap trick */
 	memset(&t->reset, 0, sizeof *t - offsetof(struct uuterm, reset));
-	t->attr = 7;
+	t->attr = 0;
+	t->color = 7;
 	t->sr_y2 = t->h - 1;
 	erase_display(t, 2);
 }
@@ -198,19 +199,30 @@ static void csi(struct uuterm *t, unsigned c)
 			if (t->param[i] == 39) t->param[i] = 37;
 			if (t->param[i] == 49) t->param[i] = 40;
 			if (!t->param[i]) {
-				t->attr = 7;
+				t->attr = 0;
+				t->color = 7;
 			} else if (t->param[i] < 8) {
 				t->attr |= attr[t->param[i]-1];
 			} else if (t->param[i]-21 < 7) {
 				t->attr &= ~attr[t->param[i]-21];
 			} else if (t->param[i]-30 < 8) {
-				t->attr &= ~UU_ATTR_FG;
-				t->attr |= t->param[i] - 30;
+				t->color &= ~255;
+				t->color |= t->param[i] - 30;
 			} else if (t->param[i]-40 < 8) {
-				t->attr &= ~UU_ATTR_BG;
-				t->attr |= t->param[i] - 40 << 4;
+				t->color &= ~(255<<8);
+				t->color |= t->param[i] - 40 << 8;
 			}
 		}
+		if ((t->color&255) < 16)
+			if (t->attr & UU_ATTR_BOLD)
+				t->color |= 8;
+			else
+				t->color &= ~8;
+		if ((t->color>>8) < 16)
+			if (t->attr & UU_ATTR_BLINK)
+				t->color |= 8<<8;
+			else
+				t->color &= ~(8<<8);
 		break;
 	case 'n':
 		switch (t->param[0]) {
@@ -328,38 +340,6 @@ static void escape(struct uuterm *t, unsigned c)
 	}
 }
 
-static void setchar(struct uucell *cell, unsigned c, unsigned a)
-{
-	int i;
-	if (a & UU_ATTR_REV)
-		cell->a = (a & ~(0xff))
-			| ((a & 0x0f) << 4)
-			| ((a & 0xf0) >> 4);
-	else
-		cell->a = a;
-	for (i=0; i<sizeof(cell->c); i++, c>>=8)
-		cell->c[i] = c;
-}
-
-static void addchar(struct uucell *cell, unsigned c)
-{
-	int i;
-	unsigned b;
-
-	for (i=b=0; i<3; i++) b |= cell->c[i] << 8*i;
-	if (b == 0 || b == ' ' || b == 0xa0) {
-		setchar(cell, c, cell->a);
-		return;
-	}
-	if (!(c = uu_combine_involution(b, c))) return;
-	for (; i<sizeof(cell->c); i++) {
-		if (!cell->c[i]) {
-			cell->c[i] = c;
-			break;
-		}
-	}
-}
-
 static void process_char(struct uuterm *t, unsigned c)
 {
 	int x, y, w;
@@ -415,7 +395,7 @@ static void process_char(struct uuterm *t, unsigned c)
 			x = 0;
 		}
 
-		addchar(&t->rows[y]->cells[x], c);
+		uucell_append(&t->rows[y]->cells[x], c);
 		dirty(t, y, x, 1);
 		break;
 	case 1:
@@ -424,7 +404,8 @@ static void process_char(struct uuterm *t, unsigned c)
 		while (w--) {
 			if (t->am) newline(t); // kills am flag
 			dirty(t, t->y, t->x, 1);
-			setchar(&t->rows[t->y]->cells[t->x++], w?UU_FULLWIDTH:c, t->attr);
+			uucell_set(&t->rows[t->y]->cells[t->x++],
+				w?UU_FULLWIDTH:c, t->attr, t->color);
 			if (t->x == t->w) {
 				t->x--;
 				t->am=1;
diff --git a/uuterm.h b/uuterm.h
index 3ce390e..defe16f 100644
--- a/uuterm.h
+++ b/uuterm.h
@@ -1,12 +1,12 @@
 /* uuterm, Copyright (C) 2006 Rich Felker; licensed under GNU GPL v2 only */
 
 #include <stddef.h>
-#include <wchar.h> /* for mbstate_t... */
+#include <stdint.h>
+#include <wchar.h>
 
 struct uucell
 {
-	unsigned short a;
-	unsigned char c[8];
+	uint32_t x[3];
 };
 
 struct uurow
@@ -16,15 +16,13 @@ struct uurow
 	struct uucell cells[1];
 };
 
-#define UU_ATTR_DIM     0x0100
-#define UU_ATTR_UL      0x0200
-#define UU_ATTR_REV     0x0400
+/* Cell-preserved attributes must lie in 0xC000FC01, see cell.c */
 
-#define UU_ATTR_BOLD    0x0008
-#define UU_ATTR_FG      0x0007
-
-#define UU_ATTR_BLINK   0x0080
-#define UU_ATTR_BG      0x0070
+#define UU_ATTR_UL      0x00000001
+#define UU_ATTR_BOLD    0x00000002
+#define UU_ATTR_BLINK   0x00000004
+#define UU_ATTR_DIM     0x00000008
+#define UU_ATTR_REV     0x00000010
 
 struct uuterm
 {
@@ -37,6 +35,7 @@ struct uuterm
 	// output state
 	int x, y;
 	int attr;
+	int color;
 	int sr_y1, sr_y2;
 	int ins  :1;
 	int am   :1;
@@ -84,7 +83,12 @@ void uuterm_stuff_byte(struct uuterm *, unsigned char);
 
 void uuterm_refresh_row(struct uudisp *, struct uurow *, int, int);
 
-int uu_combine_involution(unsigned, unsigned);
+void uucell_set(struct uucell *, wchar_t, int, int);
+void uucell_append(struct uucell *, wchar_t);
+int uucell_get_attr(struct uucell *);
+int uucell_get_color(struct uucell *);
+int uucell_get_wcs(struct uucell *, wchar_t *, size_t);
+
 int uu_decompose_char(unsigned, unsigned *, unsigned);
 
 int uudisp_open(struct uudisp *);
diff --git a/xlib.c b/xlib.c
index 99c7573..4dce890 100644
--- a/xlib.c
+++ b/xlib.c
@@ -361,8 +361,8 @@ void uudisp_predraw_cell(struct uudisp *d, int idx, int x, int color)
 		commit_cells(d, idx);
 		p->x1 = x;
 		p->color = color;
-		XSetForeground(p->display, p->gc, p->colors[color&15]);
-		XSetBackground(p->display, p->gc, p->colors[color>>4]);
+		XSetForeground(p->display, p->gc, p->colors[color&255]);
+		XSetBackground(p->display, p->gc, p->colors[color>>8]);
 	}
 
 	p->x2 = x;