From c4335cf66395d2b272cbbf22eaa7703621f64ff9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tim-Philipp=20M=C3=BCller?= <tim@centricular.net>
Date: Sun, 11 Nov 2012 20:42:42 +0000
Subject: [PATCH] cdio: use gst_tag_freeform_string_to_utf8() for CD-TEXT
 strings

They're probably not going to be UTF-8, but ISO-8859-1 or somesuch
if they're not plain ASCII.
---
 ext/cdio/Makefile.am |  2 +-
 ext/cdio/gstcdio.c   | 24 +++++++++++++++++-------
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/ext/cdio/Makefile.am b/ext/cdio/Makefile.am
index 8ad13c3f12..e4865d1892 100644
--- a/ext/cdio/Makefile.am
+++ b/ext/cdio/Makefile.am
@@ -11,7 +11,7 @@ libgstcdio_la_CFLAGS = \
 	$(CDIO_CFLAGS)
 
 libgstcdio_la_LIBADD = \
-	$(GST_PLUGINS_BASE_LIBS) -lgstaudio-$(GST_API_VERSION) \
+	$(GST_PLUGINS_BASE_LIBS) -lgstaudio-$(GST_API_VERSION) -lgsttag-$(GST_API_VERSION) \
 	$(GST_BASE_LIBS) \
 	$(CDIO_LIBS)
 
diff --git a/ext/cdio/gstcdio.c b/ext/cdio/gstcdio.c
index 4964fd8a69..e89e1055a9 100644
--- a/ext/cdio/gstcdio.c
+++ b/ext/cdio/gstcdio.c
@@ -23,6 +23,7 @@
 #endif
 
 #include <gst/gst.h>
+#include <gst/tag/tag.h>
 
 static gboolean plugin_init (GstPlugin * plugin);
 
@@ -31,7 +32,7 @@ GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
     GST_VERSION_MINOR,
     cdio,
     "Read audio from audio CDs",
-    plugin_init, VERSION, "GPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)
+    plugin_init, VERSION, "GPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN);
 
 #include "gstcdio.h"
 #include "gstcdiocddasrc.h"
@@ -44,7 +45,9 @@ void
 gst_cdio_add_cdtext_field (GstObject * src, cdtext_t * cdtext, track_t track,
     cdtext_field_t field, const gchar * gst_tag, GstTagList ** p_tags)
 {
+  const gchar *vars[] = { "GST_CDTEXT_TAG_ENCODING", "GST_TAG_ENCODING", NULL };
   const gchar *txt;
+  gchar *txt_utf8;
 
 #if LIBCDIO_VERSION_NUM > 83
   txt = cdtext_get_const (cdtext, field, track);
@@ -56,19 +59,26 @@ gst_cdio_add_cdtext_field (GstObject * src, cdtext_t * cdtext, track_t track,
     return;
   }
 
-  /* FIXME: beautify strings (they might be all uppercase for example)? */
-  /* FIXME: what encoding are these strings in? Let's hope ASCII or UTF-8 */
-  if (!g_utf8_validate (txt, -1, NULL)) {
-    GST_WARNING_OBJECT (src, "CD-TEXT string is not UTF-8! (%s)", gst_tag);
+  /* The character encoding is not specified, and there is no provision
+   * for indicating in the CD-Text data which encoding is in use.. */
+  txt_utf8 = gst_tag_freeform_string_to_utf8 (txt, -1, vars);
+
+  if (txt_utf8 == NULL) {
+    GST_WARNING_OBJECT (src, "CD-TEXT %s could not be converted to UTF-8, "
+        "try setting the GST_CDTEXT_TAG_ENCODING or GST_TAG_ENCODING "
+        "environment variable", gst_tag);
     return;
   }
 
+  /* FIXME: beautify strings (they might be all uppercase for example)? */
+
   if (*p_tags == NULL)
     *p_tags = gst_tag_list_new_empty ();
 
-  gst_tag_list_add (*p_tags, GST_TAG_MERGE_REPLACE, gst_tag, txt, NULL);
+  gst_tag_list_add (*p_tags, GST_TAG_MERGE_REPLACE, gst_tag, txt_utf8, NULL);
 
-  GST_DEBUG_OBJECT (src, "CD-TEXT: %s = %s", gst_tag, txt);
+  GST_DEBUG_OBJECT (src, "CD-TEXT: %s = %s", gst_tag, txt_utf8);
+  g_free (txt_utf8);
 }
 
 GstTagList *