typefindfunctions: UTF-8 MSS Manifest detection support

Check if the first bytes of data contain an UTF-8 BOM.

https://bugzilla.gnome.org/show_bug.cgi?id=750802
This commit is contained in:
Philippe Normand 2015-06-11 18:14:47 +02:00 committed by Sebastian Dröge
parent 9c47e7d5e6
commit d182e66bce

View File

@ -3580,6 +3580,7 @@ mss_manifest_type_find (GstTypeFind * tf, gpointer unused)
glong n_read = 0, size = 0; glong n_read = 0, size = 0;
guint length; guint length;
gchar *utf8; gchar *utf8;
gboolean utf8_bom_detected = FALSE;
if (xml_check_first_element (tf, "SmoothStreamingMedia", 20, TRUE)) { if (xml_check_first_element (tf, "SmoothStreamingMedia", 20, TRUE)) {
gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS); gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS);
@ -3595,7 +3596,9 @@ mss_manifest_type_find (GstTypeFind * tf, gpointer unused)
return; return;
/* look for a possible BOM */ /* look for a possible BOM */
if (data[0] == 0xFF && data[1] == 0xFE) if (data[0] == 0xEF && data[1] == 0xBB)
utf8_bom_detected = TRUE;
else if (data[0] == 0xFF && data[1] == 0xFE)
data_endianness = G_LITTLE_ENDIAN; data_endianness = G_LITTLE_ENDIAN;
else if (data[0] == 0xFE && data[1] == 0xFF) else if (data[0] == 0xFE && data[1] == 0xFF)
data_endianness = G_BIG_ENDIAN; data_endianness = G_BIG_ENDIAN;
@ -3619,20 +3622,29 @@ mss_manifest_type_find (GstTypeFind * tf, gpointer unused)
data += 2; data += 2;
length -= 2; length -= 2;
length = GST_ROUND_DOWN_2 (length); if (utf8_bom_detected) {
/* skip last byte of the BOM */
data++;
length--;
/* convert to native endian UTF-16 */ if (xml_check_first_element_from_data (data, length,
mss_manifest_load_utf16 (utf16_ne, data, length, data_endianness);
/* and now convert to UTF-8 */
utf8 = g_utf16_to_utf8 (utf16_ne, length / 2, &n_read, &size, NULL);
if (utf8 != NULL && n_read > 0) {
if (xml_check_first_element_from_data ((const guint8 *) utf8, size,
"SmoothStreamingMedia", 20, TRUE)) "SmoothStreamingMedia", 20, TRUE))
gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS); gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS);
} } else {
length = GST_ROUND_DOWN_2 (length);
g_free (utf8); /* convert to native endian UTF-16 */
mss_manifest_load_utf16 (utf16_ne, data, length, data_endianness);
/* and now convert to UTF-8 */
utf8 = g_utf16_to_utf8 (utf16_ne, length / 2, &n_read, &size, NULL);
if (utf8 != NULL && n_read > 0) {
if (xml_check_first_element_from_data ((const guint8 *) utf8, size,
"SmoothStreamingMedia", 20, TRUE))
gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS);
}
g_free (utf8);
}
} }
/*** image/jpeg ***/ /*** image/jpeg ***/