Modified version of Ingo Schwarze's patch for hyphen-breaking.

Breakable hyphens are cued in the back-ends (with ASCII_HYPH) and acted
upon in term.c or ignored in html.c.

Also cleaned up XML decl printing (no need for extra vars).

1 files changed, 14 insertions, 15 deletions

diff --git a/html.c b/html.c
index 292e503d..5dca90a1 100644
--- a/html.c
+++ b/html.c
@@ -29,6 +29,7 @@
 #include <string.h>
 #include <unistd.h>
 
+#include "mandoc.h"
 #include "out.h"
 #include "chars.h"
 #include "html.h"
@@ -296,11 +297,12 @@ print_encode(struct html *h, const char *p, int norecurse)
 	int		 len, nospace;
 	const char	*seq;
 	enum roffdeco	 deco;
+	static const char rejs[6] = { '\\', '<', '>', '&', ASCII_HYPH, '\0' };
 
 	nospace = 0;
 
 	for (; *p; p++) {
-		sz = strcspn(p, "\\<>&");
+		sz = strcspn(p, rejs);
 
 		fwrite(p, 1, sz, stdout);
 		p += /* LINTED */
@@ -315,6 +317,15 @@ print_encode(struct html *h, const char *p, int norecurse)
 		} else if ('&' == *p) {
 			printf("&amp;");
 			continue;
+		} else if (ASCII_HYPH == *p) {
+			/*
+			 * Note: "soft hyphens" aren't graphically
+			 * displayed when not breaking the text; we want
+			 * them to be displayed.
+			 */
+			/*printf("&#173;");*/
+			putchar('-');
+			continue;
 		} else if ('\0' == *p)
 			break;
 
@@ -443,21 +454,9 @@ print_gen_decls(struct html *h)
 static void
 print_xmltype(struct html *h)
 {
-	const char	*decl;
-
-	switch (h->type) {
-	case (HTML_XHTML_1_0_STRICT):
-		decl = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
-		break;
-	default:
-		decl = NULL;
-		break;
-	}
-
-	if (NULL == decl)
-		return;
 
-	printf("%s\n", decl);
+	if (HTML_XHTML_1_0_STRICT == h->type)
+		printf("<?xml version=\"1.0\" encoding=\"UTF-8\"?>");
 }