fixed crash when embedded data contains invalid utf8 values

extensions/comments/gth-comment.c

@@ -638,8 +638,12 @@ gth_comment_update_from_general_attributes (GthFileData *file_data)
 	if (metadata != NULL) {
 		text = g_file_info_get_attribute_string (file_data->info, "comment::note");
 		if (! dom_str_equal (gth_metadata_get_formatted (metadata), text)) {
-			gth_comment_set_note (comment, gth_metadata_get_formatted (metadata));
-			write_comment = TRUE;
+			char *value = _g_utf8_try_from_any (gth_metadata_get_formatted (metadata));
+			if (value != NULL) {
+				gth_comment_set_note (comment, value);
+				g_free (value);
+				write_comment = TRUE;
+			}
 		}
 	}
 
@@ -647,8 +651,12 @@ gth_comment_update_from_general_attributes (GthFileData *file_data)
 	if (metadata != NULL) {
 		text = g_file_info_get_attribute_string (file_data->info, "comment::caption");
 		if (! dom_str_equal (gth_metadata_get_formatted (metadata), text)) {
-			gth_comment_set_caption (comment, gth_metadata_get_formatted (metadata));
-			write_comment = TRUE;
+			char *value = _g_utf8_try_from_any (gth_metadata_get_formatted (metadata));
+			if (value != NULL) {
+				gth_comment_set_caption (comment, value);
+				g_free (value);
+				write_comment = TRUE;
+			}
 		}
 	}
 
@@ -656,8 +664,12 @@ gth_comment_update_from_general_attributes (GthFileData *file_data)
 	if (metadata != NULL) {
 		text = g_file_info_get_attribute_string (file_data->info, "comment::place");
 		if (! dom_str_equal (gth_metadata_get_formatted (metadata), text)) {
-			gth_comment_set_place (comment, gth_metadata_get_formatted (metadata));
-			write_comment = TRUE;
+			char *value = _g_utf8_try_from_any (gth_metadata_get_formatted (metadata));
+			if (value != NULL) {
+				gth_comment_set_place (comment, value);
+				g_free (value);
+				write_comment = TRUE;
+			}
 		}
 	}
 
@@ -680,12 +692,17 @@ gth_comment_update_from_general_attributes (GthFileData *file_data)
 	if (categories != NULL) {
 		metadata = (GthMetadata *) g_file_info_get_attribute_object (file_data->info, "comment::categories");
 		comment_categories = gth_metadata_get_string_list (metadata);
-		if (! gth_string_list_equal (categories, comment_categories)) {
+		if (! gth_string_list_equal_custom (categories, comment_categories, (GCompareFunc) dom_str_find)) {
 			GList *scan;
 
 			gth_comment_clear_categories (comment);
-			for (scan = gth_string_list_get_list (categories); scan; scan = scan->next)
-				gth_comment_add_category (comment, scan->data);
+			for (scan = gth_string_list_get_list (categories); scan; scan = scan->next) {
+				char *value = _g_utf8_try_from_any (scan->data);
+				if (value != NULL) {
+					gth_comment_add_category (comment, value);
+					g_free (value);
+				}
+			}
 			write_comment = TRUE;
 		}
 	}

extensions/exiv2_tools/exiv2-utils.cpp

@@ -39,7 +39,7 @@
 using namespace std;
 
 
-#define INVALID_VALUE "---"
+#define INVALID_VALUE N_("(invalid value)")
 
 
 /* Some bits of information may be contained in more than one metadata tag.
@@ -451,6 +451,8 @@ set_attribute_from_metadata (GFileInfo  *info,
 	char *formatted_value;
 	char *raw_value;
 	char *type_name;
+	char *formatted_value_utf8;
+	char *raw_value_utf8;
 
 	if (metadata == NULL)
 		return;
@@ -462,14 +464,19 @@ set_attribute_from_metadata (GFileInfo  *info,
 		      "value-type", &type_name,
 		      NULL);
 
+	formatted_value_utf8 = _g_utf8_try_from_any (formatted_value);
+	raw_value_utf8 = _g_utf8_try_from_any (raw_value);
+
 	set_file_info (info,
 		       attribute,
 		       description,
-		       formatted_value,
-		       raw_value,
+		       formatted_value_utf8,
+		       raw_value_utf8,
 		       NULL,
 		       type_name);
 
+	g_free (raw_value_utf8);
+	g_free (formatted_value_utf8);
 	g_free (description);
 	g_free (formatted_value);
 	g_free (raw_value);
@@ -528,11 +535,16 @@ set_string_list_attribute_from_tagset (GFileInfo  *info,
 
 	if (GTH_IS_METADATA (metadata) && (gth_metadata_get_data_type (GTH_METADATA (metadata)) != GTH_METADATA_TYPE_STRING_LIST)) {
 		char           *raw;
+		char           *utf8_raw;
 		char          **keywords;
 		GthStringList  *string_list;
 
 		g_object_get (metadata, "raw", &raw, NULL);
-		keywords = g_strsplit (raw, ",", -1);
+		utf8_raw = _g_utf8_try_from_any (raw);
+		if (utf8_raw == NULL)
+			return;
+
+		keywords = g_strsplit (utf8_raw, ",", -1);
 		string_list = gth_string_list_new_from_strv (keywords);
 		metadata = (GObject *) gth_metadata_new_for_string_list (string_list);
 		g_file_info_set_attribute_object (info, attribute, metadata);
@@ -541,6 +553,7 @@ set_string_list_attribute_from_tagset (GFileInfo  *info,
 		g_object_unref (string_list);
 		g_strfreev (keywords);
 		g_free (raw);
+		g_free (utf8_raw);
 	}
 	else
 		g_file_info_set_attribute_object (info, attribute, metadata);

gthumb/dom.c

@@ -71,7 +71,22 @@ _g_xml_attribute_quote (char *value)
 
 	g_return_val_if_fail (value != NULL, NULL);
 
-  	escaped = g_markup_escape_text (value, -1);
+	if (! g_utf8_validate (value, -1, NULL)) {
+		char *temp;
+
+		temp = g_locale_to_utf8 (value, -1, NULL, NULL, NULL);
+		if (temp == NULL)
+			temp = g_utf16_to_utf8 ((gunichar2 *) value, -1, NULL, NULL, NULL);
+
+		if (temp != NULL) {
+			escaped = g_markup_escape_text (temp, -1);
+			g_free (temp);
+		}
+		else
+			escaped = g_strdup (value);
+	}
+	else
+		escaped = g_markup_escape_text (value, -1);
 
   	dest = g_string_new ("\"");
 	for (p = escaped; (*p); p++) {
@@ -700,12 +715,20 @@ dom_domizable_load_from_element (DomDomizable *self,
 
 
 /* GMarkupParser converts \r into \n, this function compares two strings
- * treating \r characters as they were equal to \n */
+ * treating \r characters as they were equal to \n.  Furthermore treats invalid
+ * utf8 strings as null values. */
 gboolean
 dom_str_equal (const char *a,
 	       const char *b)
 {
 	const char *ai, *bi;
+	gboolean    a_valid, b_valid;
+
+	if ((a != NULL) && ! g_utf8_validate (a, -1, NULL))
+		a = NULL;
+
+	if ((b != NULL) && ! g_utf8_validate (b, -1, NULL))
+		b = NULL;
 
 	if ((a == NULL) && (b == NULL))
 		return TRUE;
@@ -713,6 +736,9 @@ dom_str_equal (const char *a,
 	if ((a == NULL) || (b == NULL))
 		return FALSE;
 
+	if (g_utf8_collate (a, b) == 0)
+		return TRUE;
+
 	ai = a;
 	bi = b;
 	while ((*ai != '\0') && (*bi != '\0')) {
@@ -728,12 +754,12 @@ dom_str_equal (const char *a,
 			/* \r\n equal to \n */
 
 			if ((*ai == '\r') && (*(ai + 1) == '\n') && (*bi == '\n') && (*(bi + 1) != '\n'))
-				ai++;
+				ai = g_utf8_next_char (ai);
 			if ((*bi == '\r') && (*(bi + 1) == '\n') && (*ai == '\n') && (*(ai + 1) != '\n'))
-				ai++;
+				ai = g_utf8_next_char (ai);
 		}
-		ai++;
-		bi++;
+		ai = g_utf8_next_char (ai);
+		bi = g_utf8_next_char (bi);
 	}
 
 	/* 'end of string' equal to \n and to \r\n */
@@ -744,3 +770,11 @@ dom_str_equal (const char *a,
 		|| ((*ai == '\0') && (*bi == '\r') && (*(bi + 1) == '\n') && (*(bi + 2) == '\0'))
 		|| ((*bi == '\0') && (*ai == '\r') && (*(ai + 1) == '\n') && (*(ai + 2) == '\0')));
 }
+
+
+int
+dom_str_find (const char *a,
+              const char *b)
+{
+	return dom_str_equal (a, b) ? 0 : -1;
+}

gthumb/dom.h

@@ -187,6 +187,8 @@ void          dom_domizable_load_from_element       (DomDomizable  *self,
 
 gboolean      dom_str_equal                         (const char    *a,
 						     const char    *b);
+int           dom_str_find			    (const char    *a,
+						     const char    *b);
 
 G_END_DECLS
 

gthumb/glib-utils.c

@@ -1029,6 +1029,42 @@ _g_utf8_remove_extension (const char *str)
 }
 
 
+char *
+_g_utf8_try_from_any (const char *str)
+{
+	char *utf8_str;
+
+	if (str == NULL)
+		return NULL;
+
+	if (! g_utf8_validate (str, -1, NULL)) {
+		utf8_str = g_locale_to_utf8 (str, -1, NULL, NULL, NULL);
+		if (utf8_str == NULL)
+			utf8_str = g_utf16_to_utf8 ((gunichar2 *) str, -1, NULL, NULL, NULL);
+	}
+	else
+		utf8_str = g_strdup (str);
+
+	return utf8_str;
+}
+
+
+char *
+_g_utf8_from_any (const char *str)
+{
+	char *utf8_str;
+
+	if (str == NULL)
+		return NULL;
+
+	utf8_str = _g_utf8_try_from_any (str);
+	if (utf8_str == NULL)
+		utf8_str = g_strdup (_("(invalid value)"));
+
+	return utf8_str;
+}
+
+
 static int
 remove_from_file_list_and_get_position (GList **file_list,
 					GFile  *file)

gthumb/glib-utils.h

@@ -191,6 +191,8 @@ char **         _g_utf8_strsplit                 (const char  *string,
 char *          _g_utf8_strstrip                 (const char  *str);
 gboolean        _g_utf8_all_spaces               (const char  *utf8_string);
 char *          _g_utf8_remove_extension         (const char  *str);
+char *          _g_utf8_try_from_any             (const char  *str);
+char *          _g_utf8_from_any                 (const char  *str);
 GList *         _g_list_insert_list_before       (GList       *list1,
 						  GList       *sibling,
 						  GList       *list2);

gthumb/gth-file-data.c

@@ -526,8 +526,9 @@ char *
 gth_file_data_get_attribute_as_string (GthFileData *file_data,
 				       const char  *id)
 {
-	char    *value = NULL;
 	GObject *obj;
+	char    *value = NULL;
+	char    *utf8_value;
 
 	switch (g_file_info_get_attribute_type (file_data->info, id)) {
 	case G_FILE_ATTRIBUTE_TYPE_OBJECT:

gthumb/gth-file-properties.c

@@ -171,13 +171,17 @@ gth_file_properties_real_set_file (GthPropertyView *base,
 		}
 
 		if (value != NULL) {
+			char *utf8_value;
 			char *tmp_value;
 
-			if (g_utf8_strlen (value, -1) > MAX_ATTRIBUTE_LENGTH)
-				g_utf8_strncpy (g_utf8_offset_to_pointer (value, MAX_ATTRIBUTE_LENGTH - 1), "…", 1);
-			tmp_value = _g_utf8_replace (value, "[\r\n]", " ");
+			utf8_value = _g_utf8_from_any (value);
+			if (g_utf8_strlen (utf8_value, -1) > MAX_ATTRIBUTE_LENGTH)
+				g_utf8_strncpy (g_utf8_offset_to_pointer (utf8_value, MAX_ATTRIBUTE_LENGTH - 1), "…", 1);
+			tmp_value = _g_utf8_replace (utf8_value, "[\r\n]", " ");
 			g_free (value);
 			value = tmp_value;
+
+			g_free (utf8_value);
 		}
 		tooltip = g_markup_printf_escaped ("%s: %s", _(info->display_name), value);
 

gthumb/gth-string-list.c

@@ -148,6 +148,15 @@ gth_string_list_join (GthStringList *list,
 gboolean
 gth_string_list_equal (GthStringList  *list1,
 		       GthStringList  *list2)
+{
+	return gth_string_list_equal_custom (list1, list2, (GCompareFunc) strcmp);
+}
+
+
+gboolean
+gth_string_list_equal_custom (GthStringList  *list1,
+			      GthStringList  *list2,
+			      GCompareFunc    compare_func)
 {
 	GList *keys1;
 	GList *keys2;
@@ -165,7 +174,7 @@ gth_string_list_equal (GthStringList  *list1,
 		return FALSE;
 
 	for (scan = keys1; scan; scan = scan->next)
-		if (! g_list_find_custom (keys2, scan->data, (GCompareFunc) strcmp))
+		if (! g_list_find_custom (keys2, scan->data, compare_func))
 			return FALSE;
 
 	return TRUE;

gthumb/gth-string-list.h

@@ -58,6 +58,9 @@ char *            gth_string_list_join                (GthStringList  *list,
 						       const char     *separator);
 gboolean          gth_string_list_equal               (GthStringList  *list1,
 						       GthStringList  *list2);
+gboolean	  gth_string_list_equal_custom	      (GthStringList  *list1,
+						       GthStringList  *list2,
+						       GCompareFunc    compare_func);
 void              gth_string_list_append              (GthStringList  *list1,
 		       	       	       	       	       GthStringList  *list2);
 void              gth_string_list_concat              (GthStringList  *list1,