===================================================================
RCS file: /cvs/pod2mdoc/pod2mdoc.c,v
retrieving revision 1.7
retrieving revision 1.9
diff -u -p -r1.7 -r1.9
--- pod2mdoc/pod2mdoc.c	2014/03/23 23:35:59	1.7
+++ pod2mdoc/pod2mdoc.c	2014/03/24 01:07:30	1.9
@@ -1,4 +1,4 @@
-/*	$Id: pod2mdoc.c,v 1.7 2014/03/23 23:35:59 kristaps Exp $ */
+/*	$Id: pod2mdoc.c,v 1.9 2014/03/24 01:07:30 kristaps Exp $ */
 /*
  * Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -158,17 +158,74 @@ formatescape(const char *buf, size_t *start, size_t en
 }
 
 /*
- * Skip space characters.
+ * 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
+ * that the caller can safely just continue processing.
+ * Otherwise, I don't touch "start".
  */
 static int
-skipspace(const char *buf, size_t *start, size_t end)
+trylink(const char *buf, size_t *start, size_t end, size_t dsz)
 {
-	size_t		 sv = *start;
+	size_t		sv, nstart, nend, i, j;
+	int		hasdouble;
 
-	while (*start < end && ' ' == buf[*start])
-		(*start)++;
+	/* 
+	 * Scan to the start of the terminus. 
+	 * This function is more or less replicated in the formatcode()
+	 * for null or index formatting codes.
+	 */
+	hasdouble = 0;
+	for (sv = nstart = *start; nstart < end; nstart++) {
+		/* Do we have a double-colon? */
+		if (':' == buf[nstart] && 
+			nstart > sv &&
+			':' == buf[nstart - 1])
+			hasdouble = 1;
+		if ('>' != buf[nstart])
+			continue;
+		else if (dsz == 1)
+			break;
+		assert(nstart > 0);
+		if (' ' != buf[nstart - 1])
+			continue;
+		i = nstart;
+		for (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)
+		return(0);
 
-	return(*start > sv);
+	/* Set nend to the end of content. */
+	nend = nstart;
+	if (dsz > 1)
+		nend--;
+
+	/*
+	 * Provide for some common invocations of the link primitive.
+	 * First, allow us to link to other Perl manuals.
+	 */
+	if (hasdouble)
+		printf("Xr %.*s 3p", 
+			(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]);
+	else
+		return(0);
+
+	*start = nstart;
+	return(1);
 }
 
 /*
@@ -204,6 +261,8 @@ formatcode(const char *buf, size_t *start, 
 
 	if (FMT__MAX == fmt) {
 		putchar(last = buf[(*start)++]);
+		if ('\\' == last)
+			putchar('e');
 		return(0);
 	}
 
@@ -296,13 +355,13 @@ formatcode(const char *buf, size_t *start, 
 			printf("Qo Li ");
 			break;
 		case (FMT_LINK):
-			printf("Lk ");
+			if ( ! trylink(buf, start, end, dsz))
+				printf("No ");
 			break;
 		case (FMT_FILE):
 			printf("Pa ");
 			break;
 		case (FMT_NBSP):
-			/* TODO. */
 			printf("No ");
 			break;
 		default:
@@ -363,6 +422,10 @@ formatcode(const char *buf, size_t *start, 
 		else
 			putchar(last = buf[*start]);
 
+		/* Protect against character escapes. */
+		if ('\\' == last)
+			putchar('e');
+
 		(*start)++;
 
 		if (' ' == last)
@@ -412,10 +475,15 @@ formatcodeln(const char *buf, size_t *start, size_t en
 				 ' ' == buf[*start + 2]))
 			printf("\\&");
 
-		if ('\n' != buf[*start])
-			putchar(last = buf[*start]);
-		else
+		if ('\n' == buf[*start])
 			putchar(last = ' ');
+		else
+			putchar(last = buf[*start]);
+
+		/* Protect against character escapes. */
+		if ('\\' == last)
+			putchar('e');
+
 		(*start)++;
 	}
 }
@@ -471,7 +539,9 @@ command(struct state *st, const char *buf, size_t star
 		return;
 
 	start += csz;
-	skipspace(buf, &start, end);
+	while (start < end && ' ' == buf[start])
+		start++;
+
 	len = end - start;
 
 	if (st->paused) {
@@ -624,17 +694,23 @@ command(struct state *st, const char *buf, size_t star
 static void
 verbatim(struct state *st, const char *buf, size_t start, size_t end)
 {
-	size_t		 sv = start;
+	int		 last;
 
 	if ( ! st->parsing || st->paused)
 		return;
 
 	puts(".Bd -literal");
-	while (start < end) {
-		if (start > sv && '\n' == buf[start - 1])
+	for (last = ' '; start < end; start++) {
+		/*
+		 * Handle accidental macros (newline starting with
+		 * control character) and escapes.
+		 */
+		if ('\n' == last)
 			if ('.' == buf[start] || '\'' == buf[start])
 				printf("\\&");
-		putchar(buf[start++]);
+		putchar(last = buf[start]);
+		if ('\\' == buf[start])
+			printf("e");
 	}
 	putchar('\n');
 	puts(".Ed");
@@ -704,6 +780,8 @@ ordinary(struct state *st, const char *buf, size_t sta
 			else if ('\n' == last && '\'' == buf[start])
 				printf("\\&");
 			putchar(last = buf[start++]);
+			if ('\\' == last)
+				putchar('e');
 		}
 
 		if (start < end - 1 && '<' == buf[start + 1]) {
@@ -713,6 +791,8 @@ ordinary(struct state *st, const char *buf, size_t sta
 			 * what, so print a newline now.
 			 * Then print the (possibly nested) macros and
 			 * following that, a newline.
+			 * Consume all whitespace so we don't
+			 * accidentally start an implicit literal line.
 			 */
 			if (formatcode(buf, &start, end, 0, 0)) {
 				putchar(last = '\n');
@@ -753,6 +833,8 @@ ordinary(struct state *st, const char *buf, size_t sta
 			else if ('\n' == last && '\'' == buf[start])
 				printf("\\&");
 			putchar(last = buf[start++]);
+			if ('\\' == last)
+				putchar('e');
 		}
 	}