tracker r2433 - in trunk: . src/tracker-extract

[pvanhoof svn gnome org]

Author: pvanhoof
Date: Thu Oct 30 11:40:05 2008
New Revision: 2433
URL: http://svn.gnome.org/viewvc/tracker?rev=2433&view=rev

Log:
2008-10-30  Philip Van Hoof  <philip codeminded be>

        * src/tracker-extract/tracker-albumart.c: Stripping album and artist
        strings



Modified:
   trunk/ChangeLog
   trunk/src/tracker-extract/tracker-albumart.c

Modified: trunk/src/tracker-extract/tracker-albumart.c
==============================================================================
--- trunk/src/tracker-extract/tracker-albumart.c	(original)
+++ trunk/src/tracker-extract/tracker-albumart.c	Thu Oct 30 11:40:05 2008
@@ -82,6 +82,81 @@
 
 #endif /* HAVE_STRCASESTR */
 
+
+static gchar*
+strip_characters (const gchar *original)
+{
+	const gchar *foo = "()[]<>{}_! #$^&*+=|\\/\"'?~";
+	guint osize = strlen (original);
+	gchar *retval = (gchar *) g_malloc0 (sizeof (gchar *) * osize);
+	guint i = 0, y = 0;
+
+	while (i < osize) {
+
+		/* Remove (anything) */
+
+		if (original[i] == '(') {
+			gchar *loc = strchr (original+i, ')');
+			if (loc) {
+				i = loc - original + 1;
+				continue;
+			}
+		}
+
+		/* Remove [anything] */
+
+		if (original[i] == '[') {
+			gchar *loc = strchr (original+i, ']');
+			if (loc) {
+				i = loc - original + 1;
+				continue;
+			}
+		}
+
+		/* Remove {anything} */
+
+		if (original[i] == '{') {
+			gchar *loc = strchr (original+i, '}');
+			if (loc) {
+				i = loc - original + 1;
+				continue;
+			}
+		}
+
+		/* Remove <anything> */
+
+		if (original[i] == '<') {
+			gchar *loc = strchr (original+i, '>');
+			if (loc) {
+				i = loc - original + 1;
+				continue;
+			}
+		}
+
+		/* Remove double whitespaces */
+
+		if ((y > 0) &&
+		    (original[i] == ' ' || original[i] == '\t') &&
+		    (retval[y-1] == ' ' || retval[y-1] == '\t')) {
+			i++;
+			continue;
+		}
+
+		/* Remove strange characters */
+
+		if (!strchr (foo, original[i])) {
+			retval[y] = original[i]!='\t'?original[i]:' ';
+			y++;
+		}
+
+		i++;
+	}
+
+	retval[y] = 0;
+
+	return retval;
+}
+
 static void
 perhaps_copy_to_local (const gchar *filename, const gchar *local_uri)
 {
@@ -102,8 +177,8 @@
 }
 
 static gboolean 
-heuristic_albumart (const gchar *artist,  
-		    const gchar *album, 
+heuristic_albumart (const gchar *artist_,  
+		    const gchar *album_, 
 		    const gchar *tracks_str, 
 		    const gchar *filename)
 {
@@ -116,7 +191,9 @@
 	gboolean retval;
 	gint tracks;
 	gint count;
-	
+	gchar *artist = NULL;
+	gchar *album = NULL;
+
 	file = g_file_new_for_path (filename);
 	basename = g_file_get_basename (file);
 	g_object_unref (file);
@@ -143,7 +220,12 @@
 	} else {
 		tracks = -1;
 	}
-	
+
+	if (artist_)
+		artist = strip_characters (artist_);
+	if (album_)
+		album = strip_characters (album_);
+
 	if ((tracks != -1 && tracks < count + 3 && tracks > count - 3) || 
 	    (tracks == -1 && count > 8 && count < 50)) {
 		gchar *found = NULL;
@@ -229,9 +311,11 @@
 	if (file) {
 		g_object_unref (file);
 	}
-	
+
 	g_free (target);
 	g_free (basename);
+	g_free (artist);
+	g_free (album);
 
 	return retval;
 }
@@ -286,10 +370,12 @@
 		   gchar       **path,
 		   gchar       **local)
 {
+
 	gchar *art_filename;
 	gchar *dir;
 	gchar *str;
 	gchar *down;
+	gchar *f_a = NULL, *f_b = NULL;
 
 	*path = NULL;
 
@@ -297,10 +383,20 @@
 		return;
 	}
 
-	str = g_strconcat (a ? a : "", 
+	if (a)
+		f_a = strip_characters (a);
+
+	if (b)
+		f_b = strip_characters (b);
+
+	str = g_strconcat (a ? f_a : "", 
 			   " ", 
-			   b ? b : "", 
+			   b ? f_b : "", 
 			   NULL);
+
+	g_free (f_a);
+	g_free (f_b);
+
 	down = g_utf8_strdown (str, -1);
 	g_free (str);