Re: [Rhythmbox-devel] Patch for [Bug 22483] Rhythmbox has case-sensitivity

["Surya Kiran (Virus)" <suryaoruganti gmail com>]

Hi,
Sorry for the multiple posts but I've attached a better version of the patch :)
It doesn't ignore special characters and also the memory leak has been fixed.
Cheers,

On Sun, Mar 22, 2009 at 10:29 PM, Surya Kiran (Virus) <suryaoruganti gmail com> wrote:

On Sun, Mar 22, 2009 at 10:12 PM, Jud Craft <craftjml gmail com> wrote:

2009/3/19 Surya Kiran (Virus):

> Expected behaviour: No case-sesntivity. All 3 artists must be grouped into
> one,
> as *never* is an artist differentiated by case.

Are you sure no one has ever tried to remix an existing author's name
with different case for "originality?"

I mean, I think you're almost right here...but I'm not completely
sure.  If some artist did re-case an existing name and call it
"creativity" I wouldn't be surprised.  But even if so, it may still be
so rare that this patch wouldn't be a problem at all.

Hi Jud,
This patch is in no way complete. It is just a workaround that will be useful in a majority of the cases. I wrote this one just for my convenience and thought that it might be useful to some other people too!
As this bug discussion shows ( http://bugzilla.gnome.org/show_bug.cgi?id=144283 ) , there are a lot of flaws and this cannot be accepted into the rhythmbox source, but it was worth a try. Also, I need to add a little free statement at the end to clear up the memory.
Cheers,
--
Surya Kiran Oruganti

-- 
Surya Kiran Oruganti
Index: rhythmdb/rb-refstring.c
===================================================================
--- rhythmdb/rb-refstring.c	(revision 6240)
+++ rhythmdb/rb-refstring.c	(working copy)
@@ -145,6 +145,18 @@
 }
 
 const char *
+rb_refstring_get_titlecase (const RBRefString *val)
+{
+	const char *val_lower;
+	if (val) {
+		val_lower = rb_search_fold_titlecase (rb_refstring_get (val));
+		return val_lower;
+		}
+	else
+		return NULL;
+}
+
+const char *
 rb_refstring_get (const RBRefString *val)
 {
 	return val ? val->value : NULL;
Index: rhythmdb/rb-refstring.h
===================================================================
--- rhythmdb/rb-refstring.h	(revision 6240)
+++ rhythmdb/rb-refstring.h	(working copy)
@@ -46,9 +46,12 @@
 void		rb_refstring_unref (RBRefString *val);
 
 const char *	rb_refstring_get (const RBRefString *val);
+const char *	rb_refstring_get_titlecase (const RBRefString *val);
 const char *	rb_refstring_get_folded (RBRefString *val);
 const char *	rb_refstring_get_sort_key (RBRefString *val);
+gchar* rb_search_fold_titlecase (const char *original);
 
+
 guint rb_refstring_hash (gconstpointer p);
 gboolean rb_refstring_equal (gconstpointer ap, gconstpointer bp);
 
Index: rhythmdb/rhythmdb.c
===================================================================
--- rhythmdb/rhythmdb.c	(revision 6240)
+++ rhythmdb/rhythmdb.c	(working copy)
@@ -1489,6 +1489,7 @@
  *
  * Returns: the newly allocated #RhythmDBEntry
  **/
+
 RhythmDBEntry *
 rhythmdb_entry_allocate (RhythmDB *db,
 			 RhythmDBEntryType type)
@@ -4907,13 +4908,13 @@
 
 	switch (propid) {
 	case RHYTHMDB_PROP_TITLE:
-		return rb_refstring_get (entry->title);
+		return rb_refstring_get_titlecase (entry->title);
 	case RHYTHMDB_PROP_ALBUM:
-		return rb_refstring_get (entry->album);
+		return rb_refstring_get_titlecase (entry->album);
 	case RHYTHMDB_PROP_ARTIST:
-		return rb_refstring_get (entry->artist);
+		return rb_refstring_get_titlecase (entry->artist);
 	case RHYTHMDB_PROP_GENRE:
-		return rb_refstring_get (entry->genre);
+		return rb_refstring_get_titlecase (entry->genre);
 	case RHYTHMDB_PROP_MUSICBRAINZ_TRACKID:
 		return rb_refstring_get (entry->musicbrainz_trackid);
 	case RHYTHMDB_PROP_MUSICBRAINZ_ARTISTID:
Index: lib/rb-util.c
===================================================================
--- lib/rb-util.c	(revision 6240)
+++ lib/rb-util.c	(working copy)
@@ -497,6 +497,78 @@
 }
 
 gchar*
+rb_search_fold_titlecase (const char *original)
+{
+	GString *string;
+	gchar *normalized;
+	gunichar *unicode, *cur, prev = 0x0020;
+	
+	g_return_val_if_fail (original != NULL, NULL);
+
+	/* old behaviour is equivalent to: return g_utf8_casefold (original, -1); */
+	
+	string = g_string_new (NULL);
+	normalized = g_utf8_normalize(original, -1, G_NORMALIZE_DEFAULT);
+	unicode = g_utf8_to_ucs4_fast (normalized, -1, NULL);
+	cur = unicode;
+	
+	for (cur = unicode; *cur != 0; cur++) {
+		if (*cur == 0x005f)		//underscore
+			*cur = 0x0020;
+		switch (g_unichar_type (*cur)) {
+		case G_UNICODE_COMBINING_MARK:
+		case G_UNICODE_ENCLOSING_MARK:
+		case G_UNICODE_NON_SPACING_MARK:
+		case G_UNICODE_CONNECT_PUNCTUATION:
+		case G_UNICODE_DASH_PUNCTUATION:
+		case G_UNICODE_CLOSE_PUNCTUATION:
+		case G_UNICODE_FINAL_PUNCTUATION:
+		case G_UNICODE_INITIAL_PUNCTUATION:
+		case G_UNICODE_OTHER_PUNCTUATION:
+		case G_UNICODE_OPEN_PUNCTUATION:
+
+		case G_UNICODE_LOWERCASE_LETTER:
+		case G_UNICODE_MODIFIER_LETTER:
+		case G_UNICODE_OTHER_LETTER:
+		case G_UNICODE_TITLECASE_LETTER:
+		case G_UNICODE_UPPERCASE_LETTER:
+			/* convert to title case */
+			if (prev == 0x0020)
+				*cur = g_unichar_toupper (*cur);
+			else
+				*cur = g_unichar_tolower (*cur);
+			/* ... and fall through */\
+		case G_UNICODE_DECIMAL_NUMBER:
+		case G_UNICODE_LETTER_NUMBER:
+		case G_UNICODE_OTHER_NUMBER:
+		/* should be keep symbols? */
+		case G_UNICODE_CURRENCY_SYMBOL:
+		case G_UNICODE_MODIFIER_SYMBOL:
+		case G_UNICODE_MATH_SYMBOL:
+		case G_UNICODE_OTHER_SYMBOL:
+			g_string_append_unichar (string, *cur);
+			break;
+
+		case G_UNICODE_UNASSIGNED:
+			rb_debug ("unassigned unicode character type found");
+			/* fall through */
+
+		default:
+			/* leave these in */
+			g_string_append_unichar (string, *cur);
+		}
+		prev = *cur;
+	}
+	
+	g_free (unicode);
+	g_free (normalized);
+			
+	return g_string_free (string, FALSE);
+}
+
+
+
+gchar*
 rb_search_fold (const char *original)
 {
 	GString *string;
Index: lib/rb-util.h
===================================================================
--- lib/rb-util.h	(revision 6240)
+++ lib/rb-util.h	(working copy)
@@ -60,6 +60,7 @@
 gboolean rb_is_main_thread (void);
 
 gchar* rb_search_fold (const char *original);
+gchar* rb_search_fold_titlecase (const char *original);
 gchar** rb_string_split_words (const gchar *string);
 
 gboolean rb_string_list_equal (GList *a, GList *b);