Fix possibly invalid utf-8 data before showing it to the user (GscTextWindow,

GscExecutorLogWindow).

gsmartcontrol/src/applib/app_gtkmm_utils.cpp

@@ -12,6 +12,7 @@
 #include <gtkmm.h>
 #include <gtk/gtk.h>  // gtk_* stuff
 #include <vector>
+#include <glibmm.h>
 
 #include "app_gtkmm_features.h"  // APP_GTKMM_OLD_TOOLTIPS
 
@@ -193,7 +194,6 @@ void gtkmm_set_iconview_tooltip_column(Gtk::IconView* iconview,
 
 
 
-
 bool app_gtkmm_icon_theme_has_icon(Glib::RefPtr<Gtk::IconTheme> theme,
 		const Glib::ustring& icon_name, int size)
 {
@@ -212,6 +212,81 @@ bool app_gtkmm_icon_theme_has_icon(Glib::RefPtr<Gtk::IconTheme> theme,
 
 
 
+namespace {
+
+	/// This has been copied from _g_utf8_make_valid() (glib-2.20.4).
+	/// _g_utf8_make_valid() is GLib's private function for auto-correcting
+	/// the potentially invalid utf-8 data.
+	inline gchar* gsc_g_utf8_make_valid (const gchar* name)
+	{
+		GString* string;
+		const gchar* remainder, *invalid;
+		gint remaining_bytes, valid_bytes;
+
+		g_return_val_if_fail (name != NULL, NULL);
+
+		string = NULL;
+		remainder = name;
+		remaining_bytes = gint(strlen(name));
+
+		while (remaining_bytes != 0) {
+			if (g_utf8_validate (remainder, remaining_bytes, &invalid))
+				break;
+
+			valid_bytes = gint(invalid - remainder);
+
+			if (string == NULL)
+				string = g_string_sized_new (remaining_bytes);
+
+			g_string_append_len (string, remainder, valid_bytes);
+			/* append U+FFFD REPLACEMENT CHARACTER */
+			g_string_append (string, "\357\277\275");
+
+			remaining_bytes -= valid_bytes + 1;
+			remainder = invalid + 1;
+		}
+
+		if (string == NULL)
+			return g_strdup (name);
+
+		g_string_append (string, remainder);
+
+		g_assert (g_utf8_validate (string->str, -1, NULL));
+
+		return g_string_free (string, FALSE);
+	}
+
+}
+
+
+
+Glib::ustring app_utf8_make_valid(const Glib::ustring& str)
+{
+	char* s = gsc_g_utf8_make_valid(str.c_str());
+	if (!s) {
+		return Glib::ustring();
+	}
+	Glib::ustring res(s);
+	g_free(s);
+	return res;
+}
+
+
+
+Glib::ustring app_output_make_valid(const Glib::ustring& str)
+{
+	#ifdef _WIN32
+	try {
+		return app_utf8_make_valid(Glib::locale_to_utf8(str));
+	} catch (Glib::ConvertError& e) {
+		// nothing, try to fix as it is
+	}
+	#endif
+	return app_utf8_make_valid(str);
+}
+
+
+
 
 
 

gsmartcontrol/src/applib/app_gtkmm_utils.h

@@ -99,6 +99,17 @@ bool app_gtkmm_icon_theme_has_icon(Glib::RefPtr<Gtk::IconTheme> theme,
 		const Glib::ustring& icon_name, int size);
 
 
+/// Convert a possibly invalid utf-8 string to valid utf-8.
+/// \param str string to test and fix.
+/// \param in_locale If true, \c str will be converted from locale charset to utf-8 first.
+Glib::ustring app_utf8_make_valid(const Glib::ustring& str);
+
+
+/// Make command output a valid utf-8 string. Essentially, this calls app_utf8_make_valid(),
+/// supplying true for \c in_locale under Win32, and false under other systems.
+/// The reason for this is that in Win32 we can't execute commands under C locale,
+/// but we do execute them under C in other systems.
+Glib::ustring app_output_make_valid(const Glib::ustring& str);
 
 
 

gsmartcontrol/src/gsc_executor_error_dialog.cpp

@@ -86,18 +86,7 @@ void gsc_no_info_dialog_show(const std::string& message, const std::string& sec_
 
 	if (response == Gtk::RESPONSE_HELP) {
 		GscTextWindow<SmartctlOutputInstance>* win = GscTextWindow<SmartctlOutputInstance>::create();
-		// make save visible and enable monospace font
-
-		std::string buf_text = output;
-		// We receive locale'd thousands separators in win32, so convert them.
-		#ifdef _WIN32
-		try {
-			buf_text = Glib::locale_to_utf8(buf_text);
-		} catch (Glib::ConvertError& e) {
-			buf_text = "";  // inserting invalid utf8 may trigger a segfault, so empty is better.
-		}
-		#endif
-		win->set_text(output_window_title, buf_text, true, true);
+		win->set_text(output_window_title, output, true, true);
 
 		if (!default_save_filename.empty())
 			win->set_save_filename(default_save_filename);

gsmartcontrol/src/gsc_executor_log_window.cpp

@@ -322,25 +322,7 @@ void GscExecutorLogWindow::on_tree_selection_changed()
 		if (output_textview) {
 			Glib::RefPtr<Gtk::TextBuffer> buffer = output_textview->get_buffer();
 			if (buffer) {
-
-				// Under win32, we can't execute smartctl under C locale. Smartctl
-				// uses locale information only for thousands separator in User Capacity.
-				// We can parse that, but we need to insert that text into a textarea
-				// widget, which may error out on invalid utf8 char.
-				// So, we convert the whole output to utf8, hoping that the result is ok.
-				// Note that a separator converted to utf8 may be a different sequence
-				// of chars, so we parse it as original charset, but insert into a textarea
-				// as utf8.
-				std::string buf_text = entry->std_output;
-				#ifdef _WIN32
-				try {
-					buf_text = Glib::locale_to_utf8(buf_text);
-				} catch (Glib::ConvertError& e) {
-					buf_text = "";  // inserting invalid utf8 may trigger a segfault, so empty better.
-				}
-				#endif
-
-				buffer->set_text(buf_text);
+				buffer->set_text(app_output_make_valid(entry->std_output));
 
 				Glib::RefPtr<Gtk::TextTag> tag;
 				Glib::RefPtr<Gtk::TextTagTable> table = buffer->get_tag_table();
@@ -354,15 +336,11 @@ void GscExecutorLogWindow::on_tree_selection_changed()
 			}
 		}
 
-		// Hide in win32 because it is known to cause segfaults there (see above).
-		// Not anymore...
-// #ifndef _WIN32
 		Gtk::Entry* command_entry = this->lookup_widget<Gtk::Entry*>("command_entry");
 		if (command_entry) {
 			std::string cmd_text = entry->command + " " + entry->parameters;
-			command_entry->set_text(cmd_text);
+			command_entry->set_text(app_output_make_valid(cmd_text));
 		}
-// #endif
 
 		Gtk::Button* window_save_current_button = this->lookup_widget<Gtk::Button*>("window_save_current_button");
 		if (window_save_current_button)

gsmartcontrol/src/gsc_info_window.cpp

@@ -1261,22 +1261,12 @@ bool GscInfoWindow::on_delete_event_before(GdkEventAny* e)
 
 
 
-
 void GscInfoWindow::on_view_output_button_clicked()
 {
 	GscTextWindow<SmartctlOutputInstance>* win = GscTextWindow<SmartctlOutputInstance>::create();
 	// make save visible and enable monospace font
 
-	std::string buf_text = this->drive->get_full_output();
-	// We receive locale'd thousands separators in win32, so convert them.
-	#ifdef _WIN32
-	try {
-		buf_text = Glib::locale_to_utf8(buf_text);
-	} catch (Glib::ConvertError& e) {
-		buf_text = "";  // inserting invalid utf8 may trigger a segfault, so empty better.
-	}
-	#endif
-	win->set_text("Smartctl Output", buf_text, true, true);
+	win->set_text("Smartctl Output", this->drive->get_full_output(), true, true);
 
 	std::string filename = drive->get_save_filename();
 	if (!filename.empty())

gsmartcontrol/src/gsc_text_window.h

@@ -18,6 +18,7 @@
 #include "hz/fs_file.h"
 
 #include "applib/app_ui_res_utils.h"
+#include "applib/app_gtkmm_utils.h"
 
 
 
@@ -87,7 +88,7 @@ class GscTextWindow : public AppUIResWidget<GscTextWindow<InstanceSwitch>, Insta
 			Gtk::TextView* textview = this->template lookup_widget<Gtk::TextView*>("main_textview");
 			if (textview) {
 				Glib::RefPtr<Gtk::TextBuffer> buffer = textview->get_buffer();
-				buffer->set_text(contents);
+				buffer->set_text(app_output_make_valid(contents));
 
 				if (use_monospace) {
 					Glib::RefPtr<Gtk::TextTag> tag = buffer->create_tag();