webm EBML文件頭分析
EBML element
分析webm格式，主要是對ebml element的理解，EBML是類似於XML那樣的層次化結構，每一個元素都有它的ID、值，在二進制存儲中每個元素的排列是ID，長度，值
Element IDs (also called EBML IDs) are outlined as follows, beginning with the ID itself, followed by the Data Size, and then the non-interpreted Binary itself:
-
- Element ID coded with an UTF-8 like system :
- bits, big-endian
- 1xxx xxxx - Class A IDs (2^7 -1 possible values) (base 0x8X)
- 01xx xxxx xxxx xxxx - Class B IDs (2^14-1 possible values) (base 0x4X 0xXX)
- 001x xxxx xxxx xxxx xxxx xxxx - Class C IDs (2^21-1 possible values) (base 0x2X 0xXX 0xXX)
- 0001 xxxx xxxx xxxx xxxx xxxx xxxx xxxx - Class D IDs (2^28-1 possible values) (base 0x1X 0xXX 0xXX 0xXX)
Data size, in octets, is also coded with an UTF-8 like system :
去掉001這樣的前綴，xxx就是實際需要的element id, element data size的值。
Data
element id, element data size 都是以001這樣開頭，element data直接跟在element data size，沒有了前面的001這樣的前綴。
android 中的 EBML header的ID( external/libvpx/libmkv/EbmlIDs.h)：
EBML = 0x1A45DFA3,
EBMLVersion = 0x4286,
EBMLReadVersion = 0x42F7,
EBMLMaxIDLength = 0x42F2,
EBMLMaxSizeLength = 0x42F3,
DocType = 0x4282,
DocTypeVersion = 0x4287,
DocTypeReadVersion = 0x4285,
//segment
Segment = 0x18538067,
Segment = 0x18538067,
所以判斷一個文件是否webm文件需要滿足兩個主要條件：
- 是否有EBML文件頭0x1A45DFA3
- doctype是不是webm
分析webm文件頭
這是通過ghex程序拷貝的一個文件的ebml文件頭信息（ghex打開文件後，可以通過save as菜單把hex保存爲html）：
1a 45 df a3 01 00 00 00 00 00 00 1f 42 86 81 01 42 f7
|
81 01 42 f2 81 04 42 f3 81 08 42 82 84 77 65 62 6d 42 |
87 81 02 42 85 81 02 18 53 80 67 01 00 00 00 00 18 ab
|
這個文件的EBML header可以這樣理解：
Element ID：1a 45 df a3
Element data size : 01 [0000 0001, 8個字節]
Element data: 00 00 00 00 00 00 1f [十進制是31，表示了後面所有Element總長度（字節）,所以對於EBML header 的level 0，data的內容就是header中sub element的總字節數]
以42 82爲例分析doctype：
Element ID：42 82
Element data size : 84 [84二進制就是1000 0100，去掉1，後面就是000 0100，十進制是4，表示後面的數據佔四個字節]
Element data: 77 65 62 6d [對應的ascii字符就是w e b m]
gstreamer中gsttypefindfunctions.c 中 EBML 文件頭解析的部分代碼如下：
- /* EBML typefind helper */
- static gboolean
- ebml_check_header (GstTypeFind * tf, const gchar * doctype, int doctype_len)
- {
- /* 4 bytes for EBML ID, 1 byte for header length identifier */
- guint8 *data = gst_type_find_peek (tf, 0, 4 + 1);
- gint len_mask = 0x80, size = 1, n = 1, total;
-
- if (!data)
- return FALSE;
-
- /* ebml header? */
- if (data[0] != 0x1A || data[1] != 0x45 || data[2] != 0xDF || data[3] != 0xA3)
- return FALSE;
-
- /* length of header */
- total = data[4];
- /*
- * len_mask binary: 1000 0000, while循環 total & len_mask 就可計算出前面0的個數,
- * 碰到1結束循環，size的值剛好就是ebml head element的字節數。
- */
- while (size <= 8 && !(total & len_mask)) {
- size++;
- len_mask >>= 1;
- }
- if (size > 8) /* 得出ebml header(level 0) data 的字節數 */
- return FALSE;
-
- total &= (len_mask - 1);
- while (n < size)
- total = (total << 8) | data[4 + n++];
-
- /* get new data for full header, 4 bytes for EBML ID,
- * EBML length tag and the actual header */
- data = gst_type_find_peek (tf, 0, 4 + size + total);
- if (!data)
- return FALSE;
-
- /* only check doctype if asked to do so */
- if (doctype == NULL || doctype_len == 0)
- return TRUE;
-
- /* the header must contain the doctype. For now, we don't parse the
- * whole header but simply check for the availability of that array
- * of characters inside the header. Not fully fool-proof, but good
- * enough. */
- for (n = 4 + size; n <= 4 + size + total - doctype_len; n++)
- if (!memcmp (&data[n], doctype, doctype_len))
- return TRUE;
-
- return FALSE;
- }
調用ebml_check_header的時候指定參數doctype爲"matroska", "webm"即可。
- static void
- matroska_type_find (GstTypeFind * tf, gpointer ununsed)
- {
- if (ebml_check_header (tf, "matroska", 8))
- gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MATROSKA_CAPS);
- else if (ebml_check_header (tf, NULL, 0))
- gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, MATROSKA_CAPS);
- }
-
參考:
多媒體封裝格式詳解---MKV【1】【2】【3】
http://blog.csdn.net/tx3344/article/details/8162656http://blog.csdn.net/tx3344/article/details/8176288
http://blog.csdn.net/tx3344/article/details/8203260
MKV的EBML格式
http://tigersoldier.is-programmer.com/2008/6/30/ebml-in-mkv.4052.html
MKV文件格式
http://blog.chinaunix.net/uid-12845622-id-311943.html
Matroska文件解析之SimpleBlock
http://www.cnblogs.com/tangdoudou/archive/2012/05/14/2499063.html
工具MKVtoolnix:
http://www.cinker.com/2009/01/13/mkv-movie-split-merge-mkvtoolnix/
mkv ebml官方文檔：
http://www.matroska.org/technical/specs/index.html