More & names and numbers.

src/libhtml/lex.c

@@ -333,7 +333,9 @@ AsciiInt _chartab[] = {
 	{"kappa", 954},
 	{"lambda", 955},
 	{"laquo", 171},
+	{"ldquo", 8220},
 	{"ldots", 8230},
+	{"lsquo", 8216},
 	{"lt", 60},
 	{"macr", 175},
 	{"mdash", 8212},
@@ -364,8 +366,10 @@ AsciiInt _chartab[] = {
 	{"quad", 8193},
 	{"quot", 34},
 	{"raquo", 187},
+	{"rdquo", 8221},
 	{"reg", 174},
 	{"rho", 961},
+	{"rsquo", 8217},
 	{"sect", 167},
 	{"shy", 173},
 	{"sigma", 963},
@@ -809,7 +813,7 @@ gettag(TokenSource* ts, int starti, Token* a, int* pai)
 	// c starts a tagname
 	buf[0] = c;
 	i = 1;
-	while(1) {
+	for(;;){
 		c = getchar(ts);
 		if(c < 0)
 			goto eob_done;
@@ -826,7 +830,7 @@ gettag(TokenSource* ts, int starti, Token* a, int* pai)
 
 	// attribute gathering loop
 	al = nil;
-	while(1) {
+	for(;;){
 		// look for "ws name" or "ws name ws = ws val"  (ws=whitespace)
 		// skip whitespace
 attrloop_continue:
@@ -847,7 +851,7 @@ attrloop_continue:
 			if(warn)
 				fprint(2, "warning: expected attribute name\n");
 			// skipt to next attribute name
-			while(1) {
+			for(;;){
 				c = getchar(ts);
 				if(c < 0)
 					goto eob_done;
@@ -866,7 +870,7 @@ attrloop_continue:
 		// gather attribute name
 		buf[0] = c;
 		i = 1;
-		while(1) {
+		for(;;){
 			c = getchar(ts);
 			if(c < 0)
 				goto eob_done;
@@ -892,7 +896,7 @@ attrloop_continue:
 			goto attrloop_continue;
 		}
 		//# c is '=' here;  skip whitespace
-		while(1) {
+		for(;;){
 			c = getchar(ts);
 			if(c < 0)
 				goto eob_done;
@@ -908,7 +912,7 @@ attrloop_continue:
 		}
 		val = nil;
 		nv = 0;
-		while(1) {
+		for(;;){
 valloop_continue:
 			if(c < 0)
 				goto eob_done;
@@ -918,7 +922,7 @@ valloop_continue:
 					// but if line ends before close quote, assume
 					// there was an unmatched quote
 					ti = ts->i;
-					while(1) {
+					for(;;){
 						c = getchar(ts);
 						if(c < 0)
 							goto eob_done;
@@ -1053,7 +1057,7 @@ findstr(TokenSource* ts, Rune* s)
 
 	c0 = s[0];
 	n = runestrlen(s);
-	while(1) {
+	for(;;){
 		c = getchar(ts);
 		if(c < 0)
 			break;
@@ -1077,6 +1081,18 @@ mainloop_done:
 	return 0;
 }
 
+static int
+xdigit(int c)
+{
+	if('0' <= c && c <= '9')
+		return c-'0';
+	if('a' <= c && c <= 'f')
+		return c-'a'+10;
+	if('A' <= c && c <= 'F')
+		return c-'A'+10;
+	return -1;
+}
+
 // We've just read an '&'; look for an entity reference
 // name, and if found, return translated char.
 // if there is a complete entity name but it isn't known,
@@ -1103,11 +1119,17 @@ ampersand(TokenSource* ts)
 	if(c == '#'){
 		c = getchar(ts);
 		v = 0;
-		while(c >= 0) {
+		if(c == 'x'){
-			if(!(c < 256 && isdigit(c)))
-				break;
-			v = v*10 + c - 48;
 			c = getchar(ts);
+			while((i=xdigit(c)) != -1){
+				v = v*16 + i;
+				c = getchar(ts);
+			}
+		}else{
+			while('0' <= c && c <= '9'){
+				v = v*10 + c - '0';
+				c = getchar(ts);
+			}
 		}
 		if(c >= 0){
 			if(!(c == ';' || c == '\n' || c == '\r'))
@@ -1125,7 +1147,7 @@ ampersand(TokenSource* ts)
 	else if(c < 256 && isalpha(c)){
 		buf[0] = c;
 		k = 1;
-		while(1) {
+		for(;;){
 			c = getchar(ts);
 			if(c < 0)
 				break;