Merge branch 'chris/display-idna'

This enables frontends to display international domain names in UTF-8, instead of the encoded versions.
It is disabled by default, as some frontends cannot display the full range of UTF-8 characters in their status or URL bar.

amiga/gui.c

@@ -4892,12 +4892,28 @@ static void gui_window_set_status(struct gui_window *g, const char *text)
 
 static nserror gui_window_set_url(struct gui_window *g, nsurl *url)
 {
+	size_t idn_url_l;
+	char *idn_url_s = NULL;
+	char *url_lc = NULL;
+
 	if(!g) return NSERROR_OK;
 
-	if (g == g->shared->gw) {
+	if(g == g->shared->gw) {
+		if(nsoption_bool(display_decoded_idn) == true) {
+			if (nsurl_access_utf8(url, &idn_url_s, &idn_url_l) == NSERROR_OK) {
+				url_lc = ami_utf8_easy(idn_url_s);
+			}
+		}
+
 		RefreshSetGadgetAttrs((struct Gadget *)g->shared->objects[GID_URL],
-				      g->shared->win, NULL, STRINGA_TextVal,
-				      nsurl_access(url), TAG_DONE);
+								g->shared->win, NULL,
+								STRINGA_TextVal, url_lc ? url_lc : nsurl_access(url),
+								TAG_DONE);
+
+		if(url_lc) {
+			ami_utf8_free(url_lc);
+			if(idn_url_s) free(idn_url_s);
+		}
 	}
 
 	ami_update_buttons(g->shared);

desktop/options.h

@@ -185,6 +185,9 @@ NSOPTION_UINT(min_reflow_period, DEFAULT_REFLOW_PERIOD)
 /* use core selection menu */
 NSOPTION_BOOL(core_select_menu, false)
 
+/* display decoded international domain names */
+NSOPTION_BOOL(display_decoded_idn, false)
+
 /******** Fetcher options ********/
 
 /** Maximum simultaneous active fetchers */

gtk/scaffolding.c

@@ -2343,10 +2343,21 @@ void nsgtk_window_set_title(struct gui_window *gw, const char *title)
 nserror gui_window_set_url(struct gui_window *gw, nsurl *url)
 {
 	struct nsgtk_scaffolding *g;
+	size_t idn_url_l;
+	char *idn_url_s = NULL;
 
 	g = nsgtk_get_scaffold(gw);
 	if (g->top_level == gw) {
-		gtk_entry_set_text(GTK_ENTRY(g->url_bar), nsurl_access(url));
+		if (nsoption_bool(display_decoded_idn) == true) {
+			if (nsurl_access_utf8(url, &idn_url_s, &idn_url_l) != NSERROR_OK)
+				idn_url_s = NULL;
+		}
+
+		gtk_entry_set_text(GTK_ENTRY(g->url_bar), idn_url_s ? idn_url_s : nsurl_access(url));
+
+		if(idn_url_s)
+			free(idn_url_s);
+
 		gtk_editable_set_position(GTK_EDITABLE(g->url_bar), -1);
 	}
 	return NSERROR_OK;

render/html_interaction.c

@@ -300,6 +300,8 @@ void html_mouse_action(struct content *c, struct browser_window *bw,
 	enum { ACTION_NONE, ACTION_SUBMIT, ACTION_GO } action = ACTION_NONE;
 	const char *title = 0;
 	nsurl *url = 0;
+	char *url_s = NULL;
+	size_t url_l = 0;
 	const char *target = 0;
 	char status_buffer[200];
 	const char *status = 0;
@@ -814,12 +816,26 @@ void html_mouse_action(struct content *c, struct browser_window *bw,
 					y - html_object_pos_y);
 		}
 	} else if (url) {
+		if (nsoption_bool(display_decoded_idn) == true) {
+			if (nsurl_access_utf8(url, &url_s, &url_l) != NSERROR_OK) {
+				/* Unable to obtain a decoded IDN.  This is not a fatal error.
+				 * Ensure the string pointer is NULL so we use the encoded version. */
+				url_s = NULL;
+			}
+		}
+
 		if (title) {
 			snprintf(status_buffer, sizeof status_buffer, "%s: %s",
-					nsurl_access(url), title);
+					url_s ? url_s : nsurl_access(url), title);
+		} else {
+			snprintf(status_buffer, sizeof status_buffer, "%s",
+					url_s ? url_s : nsurl_access(url));
+		}
+
 		status = status_buffer;
-		} else
-			status = nsurl_access(url);
+
+		if (url_s != NULL)
+			free(url_s);
 
 		pointer = get_pointer_shape(url_box, imagemap);
 

riscos/window.c

@@ -1038,8 +1038,20 @@ void gui_window_set_pointer(struct gui_window *g, gui_pointer_shape shape)
 /* exported function documented in riscos/window.h */
 nserror ro_gui_window_set_url(struct gui_window *g, nsurl *url)
 {
+	size_t idn_url_l;
+	char *idn_url_s = NULL;
+
 	if (g->toolbar) {
-		ro_toolbar_set_url(g->toolbar, nsurl_access(url), true, false);
+		if (nsoption_bool(display_decoded_idn) == true) {
+			if (nsurl_access_utf8(url, &idn_url_s, &idn_url_l) != NSERROR_OK)
+				idn_url_s = NULL;
+		}
+
+		ro_toolbar_set_url(g->toolbar, idn_url_s ? idn_url_s : nsurl_access(url), true, false);
+
+		if (idn_url_s)
+			free(idn_url_s);
+
 		ro_gui_url_complete_start(g->toolbar);
 	}
 

utils/nsurl.c

@@ -1699,6 +1699,57 @@ const char *nsurl_access(const nsurl *url)
 }
 
 
+/* exported interface, documented in nsurl.h */
+nserror nsurl_access_utf8(const nsurl *url, char **url_s, size_t *url_l)
+{
+	nserror err;
+	lwc_string *host;
+	char *idna_host;
+	size_t idna_host_len;
+	char *scheme;
+	size_t scheme_len;
+	char *path;
+	size_t path_len;
+
+	assert(url != NULL);
+
+	host = nsurl_get_component(url, NSURL_HOST);
+
+	if(host == NULL)
+		return NSERROR_BAD_URL;
+
+	err = idna_decode(lwc_string_data(host), lwc_string_length(host),
+						&idna_host, &idna_host_len);
+
+	lwc_string_unref(host);
+
+	if (err != NSERROR_OK)
+		return err;
+
+	err = nsurl_get(url, NSURL_SCHEME | NSURL_CREDENTIALS,
+					&scheme, &scheme_len);
+
+	if (err != NSERROR_OK)
+		return err;
+
+	err = nsurl_get(url, NSURL_PORT | NSURL_PATH | NSURL_QUERY | NSURL_FRAGMENT,
+					&path, &path_len);
+
+	if (err != NSERROR_OK)
+		return err;
+
+	*url_l = scheme_len + idna_host_len + path_len + 1; /* +1 for \0 */
+	*url_s = malloc(*url_l); 
+
+	if (*url_s == NULL)
+		return NSERROR_NOMEM;
+
+	snprintf(*url_s, *url_l, "%s%s%s", scheme, idna_host, path);
+
+	return NSERROR_OK;
+}
+
+
 /* exported interface, documented in nsurl.h */
 const char *nsurl_access_leaf(const nsurl *url)
 {

utils/nsurl.h

@@ -180,6 +180,24 @@ bool nsurl_has_component(const nsurl *url, nsurl_component part);
 const char *nsurl_access(const nsurl *url);
 
 
+/**
+ * Access a NetSurf URL object as a UTF-8 string (for human readable IDNs)
+ *
+ * \param url	  NetSurf URL object
+ * \param url_s	  Returns a url string
+ * \param url_l	  Returns length of url_s
+ * \return NSERROR_OK on success, appropriate error otherwise
+ *
+ * If return value != NSERROR_OK, nothing will be returned in url_s or url_l.
+ *
+ * The string returned in url_s is owned by the client and it is up to them
+ * to free it.  It includes a trailing '\0'.
+ *
+ * The length returned in url_l excludes the trailing '\0'.
+ */
+nserror nsurl_access_utf8(const nsurl *url, char **url_s, size_t *url_l);
+
+
 /**
  * Access a URL's path leaf as a string
  *