===================================================================
RCS file: /cvs/pod2mdoc/pod2mdoc.c,v
retrieving revision 1.18
retrieving revision 1.19
diff -u -p -r1.18 -r1.19
--- pod2mdoc/pod2mdoc.c	2014/04/02 22:36:56	1.18
+++ pod2mdoc/pod2mdoc.c	2014/04/03 10:17:14	1.19
@@ -1,4 +1,4 @@
-/*	$Id: pod2mdoc.c,v 1.18 2014/04/02 22:36:56 kristaps Exp $ */
+/*	$Id: pod2mdoc.c,v 1.19 2014/04/03 10:17:14 kristaps Exp $ */
 /*
  * Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -27,7 +27,9 @@
 #include <unistd.h>
 
 /*
- * In what section can we find Perl manuals?
+ * In what section can we find Perl module manuals?
+ * Sometimes (Mac OS X) it's 3pm, sometimes (OpenBSD, etc.) 3p.
+ * XXX IF YOU CHANGE THIS, CHANGE POD2MDOC.1 AS WELL.
  */
 #define	PERL_SECTION	"3p"
 
@@ -170,85 +172,117 @@ formatescape(const char *buf, size_t *start, size_t en
 
 /*
  * Run some heuristics to intuit a link format.
- * I recognise L<foo::bar> as a Perl manpage, printing it in section 3p;
- * or a general UNIX foo(5) manpage.
- * If I recognise one, I set "start" to be the end of the sequence so
+ * I set "start" to be the end of the sequence (last right-carrot) so
  * that the caller can safely just continue processing.
- * Otherwise, I don't touch "start".
+ * If this is just an empty tag, I'll return 0.
  */
 static int
 trylink(const char *buf, size_t *start, size_t end, size_t dsz)
 {
-	size_t		 sv, nstart, nend, i, j, textsz;
+	size_t		 linkstart, realend, linkend, i, j, textsz;
 	const char	*text;
-	int		 hasdouble;
 
 	/* 
 	 * Scan to the start of the terminus. 
 	 * This function is more or less replicated in the formatcode()
 	 * for null or index formatting codes.
 	 */
-	for (sv = nstart = *start; nstart < end; nstart++) {
-		if ('>' != buf[nstart])
+	for (linkstart = realend = *start; realend < end; realend++) {
+		if ('>' != buf[realend])
 			continue;
 		else if (dsz == 1)
 			break;
-		assert(nstart > 0);
-		if (' ' != buf[nstart - 1])
+		assert(realend > 0);
+		if (' ' != buf[realend - 1])
 			continue;
-		for (i = nstart, j = 0; i < end && j < dsz; j++) 
+		for (i = realend, j = 0; i < end && j < dsz; j++) 
 			if ('>' != buf[i++])
 				break;
 		if (dsz == j) 
 			break;
 	}
-	
-	/* We don't care about stubs. */
-	if (nstart == end || nstart == *start)
+
+	/* Ignore stubs. */
+	if (realend == end || realend == *start)
 		return(0);
 
-	/* Set nend to the end of content. */
-	nend = nstart;
-	if (dsz > 1)
-		nend--;
+	/* Set linkend to the end of content. */
+	linkend = dsz > 1 ? realend - 1 : realend;
 
-	/* Re-scan to see if we have a title. */
+	/* Re-scan to see if we have a title or section. */
 	text = &buf[*start];
-	for (textsz = *start; textsz < nend; textsz++)
-		if ('|' == buf[textsz])
+	for (textsz = *start; textsz < linkend; textsz++)
+		if ('|' == buf[textsz] || '/' == buf[textsz])
 			break;
 
-	if (textsz < nend) {
-		sv = textsz + 1;
+	/* If we have a title, find the section. */
+	if (textsz < linkend && '|' == buf[textsz]) {
+		linkstart = textsz + 1;
 		textsz = textsz - *start;
-	} else 
+		for (i = linkstart; i < linkend; i++)
+			if ('/' == buf[i])
+				break;
+		if (i < linkend)
+			linkend = i;
+	} else {
 		textsz = 0;
+		if (textsz < linkend && '/' == buf[textsz])
+			linkend = textsz;
+	}
 
-	/* Now see if we're a Perl manual. */
-	for (hasdouble = 0, i = sv + 1; i < end; i++) 
-		if (':' == buf[i] && ':' == buf[i - 1]) {
-			hasdouble = 1;
+	*start = realend;
+
+	j = linkend - linkstart;
+
+	if (0 == j)
+		return(0);
+
+	/* See if we qualify as being a link or not. */
+	if ((j > 5 && 0 == memcmp("http:", &buf[linkstart], j)) ||
+		(j > 6 && 0 == memcmp("https:", &buf[linkstart], j)) ||
+		(j > 4 && 0 == memcmp("ftp:", &buf[linkstart], j)) ||
+		(j > 5 && 0 == memcmp("sftp:", &buf[linkstart], j)) ||
+		(j > 4 && 0 == memcmp("smb:", &buf[linkstart], j)) ||
+		(j > 4 && 0 == memcmp("afs:", &buf[linkstart], j))) {
+		printf("Lk %.*s", (int)j, &buf[linkstart]);
+		return(1);
+	} 
+	
+	/* See if we qualify as a mailto. */
+	if (j > 7 && 0 == memcmp("mailto:", &buf[linkstart], j)) {
+		printf("Mt %.*s", (int)j, &buf[linkstart]);
+		return(1);
+	}
+
+	/* See if we're a foo(5), foo(5x), or foo(5xx) manpage. */
+	if ((j > 3 && ')' == buf[linkend - 1]) && 
+		('(' == buf[linkend - 3])) {
+		printf("Xr %.*s %c", (int)(j - 3), 
+			&buf[linkstart], buf[linkend - 2]);
+		return(1);
+	} else if ((j > 4 && ')' == buf[linkend - 1]) &&
+		('(' == buf[linkend - 4])) {
+		printf("Xr %.*s %.*s", (int)(j - 4), 
+			&buf[linkstart], 2, &buf[linkend - 3]);
+		return(1);
+	} else if ((j > 5 && ')' == buf[linkend - 1]) &&
+		('(' == buf[linkend - 5])) {
+		printf("Xr %.*s %.*s", (int)(j - 5), 
+			&buf[linkstart], 3, &buf[linkend - 4]);
+		return(1);
+	}
+
+	/* Last try: do we have a double-colon? */
+	for (i = linkstart + 1; i < linkend; i++)
+		if (':' == buf[i] && ':' == buf[i - 1])
 			break;
-		}
 
-	/*
-	 * Provide for some common invocations of the link primitive.
-	 * First, allow us to link to other Perl manuals.
-	 */
-	if (hasdouble)
+	if (i < linkend)
 		printf("Xr %.*s " PERL_SECTION, 
-			(int)(nend - sv), &buf[sv]);
-	else if (nend - sv > 3 && isalnum(buf[sv]) && 
-			')' == buf[nend - 1] && 
-			isdigit((int)buf[nend - 2]) &&
-			'(' == buf[nend - 3]) 
-		printf("Xr %.*s %c", 
-			(int)(nend - 3 - sv),
-			&buf[sv], buf[nend - 2]);
+			(int)j, &buf[linkstart]);
 	else
-		return(0);
+		printf("Xr %.*s 1", (int)j, &buf[linkstart]);
 
-	*start = nstart;
 	return(1);
 }
 
@@ -457,6 +491,7 @@ formatcode(struct state *st, const char *buf, size_t *
 			printf("Qo Li ");
 			break;
 		case (FMT_LINK):
+			/* Try to link; use "No" if it's empty. */
 			if ( ! trylink(buf, start, end, dsz))
 				printf("No ");
 			break;