From c0de4379d2a97a61afe862d2d8b246951f44b1c5 Mon Sep 17 00:00:00 2001
From: Kaz Kylheku <kaz@kylheku.com>
Date: Sat, 13 Sep 2014 21:47:02 -0700
Subject: Fixes: font escape, macros, crash.

Macros can have up to four letter names, not just two.
The \f[..] parsing is fixed so inline font changes work.
The CR font is supported.
Crash fixed when the section table is empty.
---
 man2html/man2html.c | 76 +++++++++++++++++++++++++++++++++++------------------
 1 file changed, 51 insertions(+), 25 deletions(-)

diff --git a/man2html/man2html.c b/man2html/man2html.c
index b37eef6..6565b30 100644
--- a/man2html/man2html.c
+++ b/man2html/man2html.c
@@ -53,12 +53,13 @@ static char *scan_troff_mandoc(char *c, int san, char **result);
 
 static char **argument=NULL;
 
-static char charb[3];
+static char charb[5];
 
 static char *
 expand_char(int nr)
 {
 	STRDEF *h;
+	int i;
 
 	if (!nr)
 		return NULL;
@@ -74,10 +75,15 @@ expand_char(int nr)
 			curpos += h->slen;
 			return h->st;
 		}
-	charb[0] = nr/256;
-	charb[1] = nr%256;
-	charb[2] = 0;
-	curpos += 2;
+
+	charb[0] = 0;
+
+	for (i = 0; nr && i < 4; i++) {
+		memmove(charb+1, charb, 4);
+		charb[0] = nr%256;
+		nr /= 256;
+		curpos++;
+	}
 	return charb;
 }
 
@@ -96,6 +102,15 @@ expand_string(int nr)
 	return NULL;
 }
 
+static int str_to_code(char *str)
+{
+	int code = 0;
+	while (*str && !isspace(*str)) {
+		code *= 256;
+		code += *str++;
+	}
+	return code;
+}
 
 static char outbuffer[1024];
 static int obp=0;
@@ -452,7 +467,7 @@ change_to_font(int nr)
   case '0': nr++;
   case '1': case '2': case '3': case '4':
        nr = nr-'1'; break;
-  case V('C','W'): break;
+  case V('C','W'): case V('C','R'): nr=3; break;
   case 'L': nr=3; break;
   case 'B': nr=2; break;
   case 'I': nr=1; break;
@@ -570,12 +585,20 @@ scan_escape(char *c) {
 	    c=scan_escape(c);
 	    c--;
 	    i=intresult;
-	} else 	if (*c != '(')
-	    i=*c;
-	else {
-	    c++;
-	    i=c[0]*256+c[1];
+	} else if (*c == '[' || *c == '(') {
 	    c++;
+		if (*c == ']' || *c == ')') {
+			i = 0;
+		} else if (c[1] == ']' || c[1] == ')') {
+			i = *c++;
+		} else if (c[2] == ']' || c[2] == ')') {
+			i=c[0]*256+c[1];
+			c += 2;
+		} else {
+			abort();
+		}
+	} else {
+	    i=*c;
 	}
 	if (!skip_escape) h=change_to_font(i); else h="";
 	break;
@@ -1535,12 +1558,10 @@ scan_request(char *c) {
 	    c++;
     if (c[0] == '\n')
 	    return c+1;
-    if (c[1] == '\n')
-	    j = 1;
-    else
-	    j = 2;
-    while (c[j] == ' ' || c[j] == '\t')
-	    j++;
+
+    j = strcspn(c, " \n\t");
+    j += strspn(c + j, " \t");
+
     if (c[0] == escapesym) {
 	/* some pages use .\" .\$1 .\} */
 	/* .\$1 is too difficult/stupid */
@@ -1549,7 +1570,7 @@ scan_request(char *c) {
 	else
 	    c = scan_escape(c+1);
     } else {
-	i=V(c[0],c[1]);
+	i=str_to_code(c);
 	switch (i) {
 	case V('a','b'):
 	    h=c+j;
@@ -1566,7 +1587,7 @@ scan_request(char *c) {
 	    {
 		STRDEF *de;
 		c=c+j;
-		i=V(c[0],c[1]);
+		i=str_to_code(c);
 		if (*c == '\n') { c++;break; }
 		while (*c && *c!='\n') c++;
 		c++;
@@ -1600,7 +1621,7 @@ scan_request(char *c) {
 		int oldcurpos=curpos;
 		c=c+j;
 		while (*c == ' ') c++;
-		i=V(c[0],c[1]);
+		i=str_to_code(c);
 		j=0;
 		while (c[j] && c[j]!='\n') j++;
 		if (j<3) { c=c+j; break; }
@@ -2212,11 +2233,14 @@ scan_request(char *c) {
 		int olen=0;
 		c=c+j;
 		sl=fill_words(c, wordlist, SIZE(wordlist), &words, '\n');
-		i=V(c[0],c[1]);j=2;
-		if (words == 1) wordlist[1]=".."; else {
+		i=str_to_code(c);
+		if (words == 1) {
+			wordlist[1]="..";
+			j = 2;
+		} else {
 		    wordlist[1]--;
 		    wordlist[1][0]='.';
-		    j=3;
+		    j+=1;
 		}
 		c=sl+1;
 		sl=c;
@@ -3212,8 +3236,10 @@ main(int argc, char **argv) {
     if (output_possible) {
 	/* &nbsp; for mosaic users */
 	printf("<HR>\n<A NAME=\"index\">&nbsp;</A><H2>Index</H2>\n<DL>\n");
-	manidx[mip]=0;
-	printf("%s", manidx);
+	if (manidx) {
+		manidx[mip]=0;
+		printf("%s", manidx);
+	}
 	if (subs) printf("</DL>\n");
 	printf("</DL>\n");
 	print_sig();
-- 
cgit v1.2.3