FLV文件内容解析(一)
作者： 孙悟空 http://bbs.chinaffmpeg.com参考flv的spec文档，写了个解析onMetaData，代码没有整理
[*]
/*
* Copyright (c) 2013 Steven Liu
*
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <fcntl.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <errno.h>
#define TAG_FLV_F 'F'
#define TAG_FLV_L 'L'
#define TAG_FLV_V 'V'
#define TAG_TYPE_SCRIPTDATA 0x12
int script_type_parse(unsigned char *data);
union av_intfloat64 {
uint64_t i;
double f;
};
enum script_data_type {
Number = 0,
Boolean,
String,
Object,
MovieClip,
Null,
Undefined,
Reference,
EcmaArray,
ObjectEndMarker,
StringArray,
Date,
LongString,
};
typedef enum script_data_type script_type_t;
struct script_string {
char len;
};
typedef struct script_string String_t;
struct flv_header {
unsigned char tag_f;
unsigned char tag_l;
unsigned char tag_v;
unsigned char version;
unsigned char audio_and_video;
unsigned char DataOffset;
};
typedef struct flv_header header_t;
struct flv_tag {
unsigned char tagtype;
unsigned char size;
unsigned char timestamp;
unsigned char extent_timestamp;
unsigned char stream_id;
};
typedef struct flv_tag tag_t;
struct flv_body {
tag_t tag;
// unsigned char PreviousTagSize;
};
typedef struct flv_body body_t;
struct object_name {
unsigned short length;
unsigned char name;
};
typedef struct object_name object_name_t;
struct script_object {
object_name_t name;
};
typedef struct script_object_name script_object_t;
/**
* Reinterpret a 64-bit integer as a double.
*/
static double int2double(uint64_t i)
{
union av_intfloat64 v;
v.i = i;
return v.f;
}
/*
* get_header_audio check if the flv have audio stream
* info, a byte with video and audio bit flag
*
* return 0 or 1
*/
int get_header_audio(char info)
{
return (info & 0x4) >> 2;
}
/*
* get_header_video check if the flv have video stream
* info, a byte with video and audio bit flag
*
* return 0 or 1
*/
int get_header_video(char info)
{
return info & 0x1;
}
/*
* flv_header_parse just parse the FLV VERSION tag
* fd, file describe handle
* return flv version
*/
int flv_header_parse(int fd)
{
int ret = 0;
header_t header;
memset(&header, 0, sizeof(header));
ret = read(fd, &header, sizeof(header));
if (ret <= 0) {
fprintf(stderr, "read header error, %d\n", -EINVAL);
return -EINVAL;
}
fprintf(stdout, "flv = [%c %c %c] version = [%x]\n"
"TypeFlagsAudio = [%x], TypeFlagsVideo = [%x], DataOffset = [%x %x %x %x]\n",
header.tag_f, header.tag_l, header.tag_v, header.version,
get_header_audio(header.audio_and_video), get_header_video(header.audio_and_video),
header.DataOffset, header.DataOffset, header.DataOffset, header.DataOffset);
return (int)header.version;
}
/*
* flv_do_tag just parse all the flv tag
* fd, file describe handle
* size, the size of the flv tag body
*
* return tag type
*/
int flv_do_tag(int fd, unsigned long *size)
{
int ret = 0;
char *p = NULL;
char data_size = {0};
body_t body;
ret = read(fd, &body, sizeof(body));
if (ret <= 0) {
fprintf(stderr, "Can not ret the fd errno [%d]\n", -EINVAL);
return -EINVAL;
}
fprintf(stdout, "size = %lu tag->type = [%x]"
"tag->size = [%0.2x %0.2x %0.2x], tag->timestamp = [%0.2x %0.2x %0.2x], "
"tag->extent_timestamp = [%0.2x], tag->streamid = [%0.2x %0.2x %0.2x]\n",
sizeof(body), body.tag.tagtype,
body.tag.size, body.tag.size,body.tag.size, body.tag.timestamp, body.tag.timestamp,
body.tag.timestamp, body.tag.extent_timestamp, body.tag.stream_id,
body.tag.stream_id, body.tag.stream_id);
memcpy(data_size, body.tag.size, 3);
p = data_size;
snprintf(data_size, sizeof(data_size), "0x%x%x%x", body.tag.size, body.tag.size, body.tag.size);
*size = strtoul(p, NULL, 16);
fprintf(stdout, "size = [%lu] \n", *size);
return body.tag.tagtype;
}
/*
* get_string_len get the key-value value string length
* data, the data will be parsed
*
* return the String length by int
*/
int get_string_len(unsigned char *data)
{
unsigned char *p = data;
unsigned char len_char;
int len_int = 0;
memset(len_char, 0, sizeof(len_char));
snprintf((char *)len_char, sizeof(len_char), "0x%x%x", data, data);
p = len_char;
len_int = strtoul((char *)p, NULL, 16);
return len_int;
}
/*
* get_bool_value get bool type value (key-value value)
* data, the data will be parsed
*
* return bool value 0 or 1
*/
int get_bool_value(unsigned char *data)
{
return *data;
}
/*
* get_key_len the key-value key string length unit: four bytes
* data, the data will be parsed
*
* return the key length by int
*/
int get_key_len(unsigned char *data)
{
char len_char;
char *p = len_char;
memset(len_char, 0, sizeof(len_char));
snprintf(len_char, sizeof(len_char), "0x%x%x%x%x", data, data, data, data);
return strtoul(p, NULL, 16);
}
int process_ecma_array_end(unsigned char *data)
{
return 0;
}
/*
* process_ecma_array The onMetaData sub type named ECMA Array
* data, the data will be parsed
*
* return the offset of the data
*/
int process_ecma_array(unsigned char *data)
{
int ecma_array_len = 0;
int keyname_len = 0;
unsigned char keyname;
unsigned char *p = data;
int i = 0;
ecma_array_len = get_key_len(p);
p += 4;
for (i = 0; i < ecma_array_len; i++) {
keyname_len = get_string_len(p);
p += 2;
memset(keyname, 0, sizeof(keyname));
strncpy((char *)keyname, (const char *)p, keyname_len);
fprintf(stdout, "keyname = [%s]\n", keyname);
p += keyname_len;
p += script_type_parse(p);
}
if (*p == 0 && *(p + 1) == 0 && *(p + 2) == 9) {
p += 3;
}
return p - data;
}
/*
* get_double get the double number by long long type
* data, the data will be parsed
*
* return the parse out number of long long
*/
unsigned long long get_double(unsigned char *data)
{
unsigned char double_number;
unsigned char *p = double_number;
memset(double_number, 0, sizeof(double_number));
snprintf((char *)double_number, sizeof(double_number), "0x%0.2x%0.2x%0.2x%0.2x%0.2x%0.2x%0.2x%0.2x",
*data, *(data + 1), *(data + 2), *(data + 3), *(data + 4), *(data + 5), *(data + 6), *(data + 7));
return strtoull((char *)p, NULL, 16);
}
/*
* script_type_parse parse the script object type and parse the data
* data, the data will be parsed
*
* return the data offset
*/
int script_type_parse(unsigned char *data)
{
int ret = 0;
unsigned char *p = data;
unsigned char string_output;
unsigned long long number = 0;
double double_number = 0.0;
switch (*p) {
case Number:
p++;
number = get_double(p);
double_number = int2double(number);
fprintf(stdout, "number = [%.2f]\n", double_number);
p += 8;
break;
case Boolean:
p++;
ret = get_bool_value(p);
p++;
break;
case String:
p++;
memset(string_output, 0, sizeof(string_output));
ret = get_string_len(p);
p += 2;
strncpy((char *)string_output, (const char *)p, ret);
fprintf(stdout, "String = [%s]\n", string_output);
p += ret;
break;
case Object:
p++;
break;
case MovieClip:
p++;
break;
case Null:
p++;
break;
case Undefined:
p++;
break;
case Reference:
p++;
break;
case EcmaArray:
p++;
ret = process_ecma_array(p);
p += ret;
break;
case ObjectEndMarker:
p++;
break;
case StringArray:
p++;
break;
case Date:
p++;
break;
case LongString:
p++;
break;
default:
break;
}
return p - data;
}
/*
* do_tag_onMetaData parse the onMetaData tags
* fd, file describe handle
* size, will read size of the data, unit: byte
*
* return offset of the data
*/
int do_tag_onMetaData(int fd, int size)
{
int ret = 0;
unsigned char *p = NULL;
unsigned char *onMetaData = malloc(size);
fprintf(stdout, "size = [%d]\n", size);
memset(onMetaData, 0, size);
ret = read(fd, onMetaData, size);
fprintf(stdout, "ret = [%d]\n", ret);
int fuck = open("./xxx.dat", O_RDWR);
write(fuck, onMetaData, ret);
close(fuck);
p = onMetaData;
while (p) {
if (p - onMetaData >= size) {
break;
}
ret = script_type_parse(p);
fprintf(stdout, "in while 1 offset = [%d]\n", p - onMetaData);
p += ret;
}
return p - onMetaData;
}
/*
* dump_flv_info just print the flv onMetaData info
* filename, the flv filename
*
* return always right
*/
int dump_flv_info(char *filename)
{
int fd = 0;
int tag_type = -1;
int ret = 0;
unsigned long size = 0;
unsigned char PreviousTagSize;
fd = open(filename, O_RDONLY);
if (fd < 0) {
fprintf(stdout, "Can not open the file [%s]\n", filename);
return -ENOENT;
}
fprintf(stdout, "filename = [%s]\n", filename);
flv_header_parse(fd);
ret = read(fd, PreviousTagSize, sizeof(PreviousTagSize));
tag_type = flv_do_tag(fd, &size);
switch(tag_type) {
case TAG_TYPE_SCRIPTDATA:
fprintf(stdout, "data_size = [%lu]\n", size);
do_tag_onMetaData(fd, size);
break;
}
return 0;
}
void usage(char **argv)
{
fprintf(stderr, "Usage:\n"
"\t%s flv_file_name\n", argv);
return ;
}
int main(int argc, char *argv[])
{
char *filename = NULL;
if (argc < 2) {
usage(argv);
return -EINVAL;
}
filename = argv;
dump_flv_info(filename);
return 0;
}
[*]
Makefile为
all:
gcc -O2 -Wall -g dump_flv_info.c -o flv_info
clean:
rm -rf flv_info
[*]
执行结果如下：
2.flv的内容：
主要参考文档为：video_file_format_spec_v10_1_20131118111845.pdf
根据参考文档可以看出来解析步骤如下首先解析文件头
根据上面的内容可以看出来，FLV和Version占用了4个字节，而Version总是1,其中的DataOffset一般来在flv的version是1的时候总是9，如图
464c 5601为四个字节，就是FLV 和Version部分，接下来的05为一个字节，这个字节为TypeFlagsReserved, TypeFlagsAudio,TyopeFlagsReserved, TypeFlagsVideo的内容，可以将05转换为2进制，就是00000101, 也就是video是1，audio也是1，接下爱就是DataOffset，DataOffset一次读取32位，就是4个字节，内容为00 0000 09，也就是上面参考中的内容，The DataOffset field usually has a value of 9；接下来就开始解析FLV的File Body:部分，这部分排列方式位PreviousTagSizeN（N >= 0）,如图占用了4个字节，全部为0.在前面的spec里面可以看到，接下来的就是FLVTAG，这个FLVTAG为另外一个结构，结构如下：
通过结构可以看出，Reserved,Filter,TagType占用了一个字节，也就是前面的12，这里说的12其实不是十进制的12，而是16进制的0x12，将其转成十进制就是18，可以转成2进制进行对比，最终TagType来判断这个FLVTAG是音频，还是视频，还是脚本数据(ScriptData)，如果TagType是8就是audio，如果TagType是9就是video，如果TagType是18，就是ScriptData；接下来继续读取24个字节，来判断将要读取得FLVTAG对应的数据字节数，例如onMetaData这部分是ScriptData，那么读取ScriptData的数据的时候，就要读取DataSize这么大，如图：
可以看到，需要读取0x000125这么多数据，转换成十进制的话，就是293个字节的数据，这个数据我们暂且记下，后面会用到，千万要记住！！！接下来就是读取TimeStamp（3个字节）和Extent Timestamp（1个字节），两个字段加起来，占用了4个字节，全部为0，具体参考上面的spec可以看到为什么，过了四个字节以后，进入到StreamID部分，StreamID占用了3个字节，总是0，Spec中已经可以看到，接下来就是0200 0a6f ....部分，这部分在前面的spec中可以看到说明，由于前面我们看到的是TagType是18，那么这个地方就做ScriptData相关的处理办法，直接按照ScriptData来处理：
从上面spec中可以看到，ScriptData里面使用了ScriptTagBody，而ScriptTagBody的定义为一个Type和ScriptDataValue，其中的DataValue是根据Type来决定的，我们读数据读到offset为0x0000010 + 8位置，这时下一个字节为02,也就是说，Type为SCRIPTDATASTRING，这个时候会处理SCRIPTDATASTRING,
SCRIPTDATASTRING份两部分，一个是2两个字节的StringLength，一个是StringData，这个StringData部分占用的字节多少由StringLength部分决定，那么就继续上面的位置读取2字节，为00 0a,然后接下来读取00 0a个字节，作为String内容，也就是看到的onMetaData这个字符串，这样，onMetaData这个字符串读取完毕，也就是前面提到的ScriptTagBody的Name部分读取完毕。接下来内容进入到了高潮：读取ECMA ARRAY部分，同样使用前面的办法读取对应的内容，如图：
其中08为ECMAARRAY的类型，接下来的0000 000d为ECMAARRAY中的结构的ECMAArrayLength，一次读取4个字节，得到的结果为13(0x0d)。
也就是说，ECMAARRAY由13个对应的项，可以建立一个递归或者循环，去读取这13个Variables，接下来开始解析第一个Variable，也就是SCRIPTDATAOBJECTPROPERTY。
接下来就是解析PropertyName，也是SCRIPTDATASTRING类型，对应解析办法前面有提到过，一次读两个字节，然后获得PropertyName的字符串，然后获得对应的PropertyData数据，要注意的是PropertyData里面的数据类型，读到的是00的话，就是Number方法，这个Number存的值就是个DOUBLE,通过前面的spec参考可以看到对应的信息，下图是对应的字节：
接下来可以按照前面的方法继续循环或者递归，这样，就将onMetaData信息全部读取出来了，读取最后的时候不要忘记有三个字节，0000 09,为SCRIPTDATAOBJECTEND，这样，FLV的onMetaData部分解析完毕
以前是看解析器进行解析，这个直接码字，，哈哈 小胖子 发表于 2013-12-19 09:02
以前是看解析器进行解析，这个直接码字，，哈哈
献丑了，就是没事的时候看看spec，然后写一个分享一下大概的思路，代码很屎，希望能看明白
页:
[1]