onvif成功实现ONVIF协议RTSP-Video-Stream与OnvifDeviceManager的视频对接

原文地址：http://blog.csdn.net/ghostyu/article/details/8208428

有了前几篇的基础，现在可以正式开始onvif的实现工作，其中一项非常重要的部分就是 视频流的对接，即能够在符合onvif标准的监控客户端软件里接收到设备端NVT发来的RTSP视频流。这里，我所用的客户端软件是Onvif Device Manager v2.2。【来自http://blog.csdn.net/ghostyu】

ONVIF Profile S Specification文档描述了Device或者说DVT和Client可以使用的一种Profile，Profile这个词在计算机领域非常常见，我们可以理解成一种方案、配置、框架等。

文档里描述了如果实现VideoStream，device和client应该具备的条件，当然如果实现文档的所有条件，就可以说该设备符合Profile S

如果单纯实现VideoStream，只需完成下列命令。

1. 1、GetProfiles  
2. 2、GetStreamUri   
3. 填充rtsp路径，例如：rtsp://192.168.1.201/petrov.m4e  
4. 3、Media Streaming using RTSP  
5. 这里使用开源的live555，完成rtsp功能  
6. 4、GetVideoEncoderConfiguration  
7. 5、GetVideoEncoderConfigurationOptions  
8. 6、GetCapabilities  
9. NVC为了获取DVT所支持的功能的命令  

参考文档：

1. 1、ONVIF Profile S Specification  
2. 描述ProfileS是什么样的一个东西，如何实现  
3. 2、Reference_of_ONVIF_Development_v1.01.02  
4. Onvif DVT设计参考，指明了一条道路，但没有具体内容  
5. 3、ONVIF-Media-Service-Spec-v220  
6. Onvif Media的说明介绍  
7. 4、
8. onvif几乎全部命令的详细说明，非常重要。该文档告诉我们结构体成员的意义和如何填充。Onvif开发其实就是各种结构体的填充。  

一、产生onvif源码框架

1、从wsdl生成C头文件

wsdl2h -o onvif.h -c -s -t .\typemap.dat http://www.onvif.org/onvif/ver10/device/wsdl/devicemgmt.wsdl http://www.onvif.org/onvif/ver10/event/wsdl/event.wsdl http://www.onvif.org/onvif/ver10/display.wsdl http://www.onvif.org/onvif/ver10/deviceio.wsdl http://www.onvif.org/onvif/ver20/imaging/wsdl/imaging.wsdl http://www.onvif.org/onvif/ver10/media/wsdl/media.wsdl http://www.onvif.org/onvif/ver20/ptz/wsdl/ptz.wsdl  http://www.onvif.org/onvif/ver10/receiver.wsdl http://www.onvif.org/onvif/ver10/recording.wsdl  http://www.onvif.org/onvif/ver10/search.wsdl http://www.onvif.org/onvif/ver10/network/wsdl/remotediscovery.wsdl http://www.onvif.org/onvif/ver10/replay.wsdl http://www.onvif.org/onvif/ver20/analytics/wsdl/analytics.wsdl http://www.onvif.org/onvif/ver10/analyticsdevice.wsdl  http://www.onvif.org/onvif/ver10/schema/onvif.xsd  http://www.onvif.org/ver10/actionengine.wsdl

跟前一篇discovery唯一不同的是，这里多了很多wsdl文件，这次创建完整的onvif代码框架

2、从头文件生成源码框架

soapcpp2 -c onvif.h -x -I /root/onvif/gsoap-2.8/gsoap/import -I /root/onvif/gsoap-2.8/gsoap/

产生的C文件比较庞大，最大的有十几兆，大部分的内容没有复用导致。

二、创建soap运行环境

int main(int argc, char **argv)
{
    int m, s;
    struct soap add_soap;
    int server_udp;

    server_udp = create_server_socket_udp();
    //bind_server_udp1(server_udp);
    pthread_t thrHello;
    pthread_t thrProbe;
    //pthread_create(&thrHello,NULL,main_Hello,server_udp);
    //sleep(2);
    pthread_create(&thrProbe,NULL,main_Probe,server_udp);

    soap_init(&add_soap);
    soap_set_namespaces(&add_soap, namespaces);


    if (argc < 0) {
        printf("usage: %s  \n", argv[0]);
        exit(1);
    } else {
        m = soap_bind(&add_soap, NULL, 80, 100);
        if (m < 0) {
            soap_print_fault(&add_soap, stderr);
            exit(-1);
        }
        fprintf(stderr, "Socket connection successful: master socket = %d\n", m);
        for (;;) {
            s = soap_accept(&add_soap);
            if (s < 0) {
                soap_print_fault(&add_soap, stderr);
                exit(-1);
            }
            fprintf(stderr, "Socket connection successful: slave socket = %d\n", s);
            soap_serve(&add_soap);
            soap_end(&add_soap);
        }
    }
    return 0;
}

注 意，这里绑定了80端口，onvif使用的是http请求，然后附带xml，其实正常的是将onvif集成到web服务器中，普通的http请求有web 服务器处理，onvif的http请求则有soap处理。我们这里的做法也可行，只不过onvif的访问web服务器的功能是无法使用的。

三、RTSP视频对接

1、实现GetCapabilities命令

客户端发送GetCapabilities命令来得到设备端的能力，然后依据GetCapabilities返回的结果再来进行下一步操作

在__tds__GetCapabilities函数中我们只需要填充Media部分和一些必要的即可

//想要对接RTSP视频，必须设置Media
tds__GetCapabilitiesResponse->Capabilities->Media = (struct tt__MediaCapabilities*)soap_malloc(soap, sizeof(struct tt__MediaCapabilities));
tds__GetCapabilitiesResponse->Capabilities->Media->XAddr = (char *) soap_malloc(soap, sizeof(char) * LARGE_INFO_LENGTH);
strcpy(tds__GetCapabilitiesResponse->Capabilities->Media->XAddr, _IPv4Address);
tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities = (struct tt__RealTimeStreamingCapabilities*)soap_malloc(soap, sizeof(struct tt__RealTimeStreamingCapabilities));
tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->RTPMulticast = (int *)soap_malloc(soap, sizeof(int));
*tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->RTPMulticast = _false;
tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->RTP_USCORETCP = (int *)soap_malloc(soap, sizeof(int));
*tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->RTP_USCORETCP = _true;
tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->RTP_USCORERTSP_USCORETCP = (int *)soap_malloc(soap, sizeof(int));
*tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->RTP_USCORERTSP_USCORETCP = _true;
tds__GetCapabilitiesResponse->Capabilities->Media->StreamingCapabilities->Extension = NULL;
tds__GetCapabilitiesResponse->Capabilities->Media->Extension = NULL;
tds__GetCapabilitiesResponse->Capabilities->Media->__size = 0;
tds__GetCapabilitiesResponse->Capabilities->Media->__any = 0;

另外必要填充的还有

//下面的重要，这里只实现视频流，需要设置VideoSources
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->VideoSources = TRUE;
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->VideoOutputs = FALSE;
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->AudioSources = FALSE;
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->AudioOutputs = FALSE;
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->RelayOutputs = FALSE;
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->__size = 0;
tds__GetCapabilitiesResponse->Capabilities->Extension->DeviceIO->__any = NULL;

tds__GetCapabilitiesResponse->Capabilities->Extension->Display = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->Recording = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->Search = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->Replay = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->Receiver = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->AnalyticsDevice = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->Extensions = NULL;
tds__GetCapabilitiesResponse->Capabilities->Extension->__size = 0;
tds__GetCapabilitiesResponse->Capabilities->Extension->__any = NULL;

2、实现GetServices命令

int  __tds__GetServices(struct soap* soap, struct _tds__GetServices *tds__GetServices, struct _tds__GetServicesResponse *tds__GetServicesResponse)
{
    DBG("__tds__GetServices\n");
    /*该函数很必要*/
    char _IPAddr[INFO_LENGTH];
    int i = 0;
    sprintf(_IPAddr, "http://%03d.%03d.%03d.%03d/onvif/services", 192, 168, 1, 233);
    tds__GetServicesResponse->__sizeService = 1;

    tds__GetServicesResponse->Service = (struct tds__Service *)soap_malloc(soap, sizeof(struct tds__Service));
    tds__GetServicesResponse->Service[0].XAddr = (char *)soap_malloc(soap, sizeof(char) * INFO_LENGTH);
    tds__GetServicesResponse->Service[0].Namespace = (char *)soap_malloc(soap, sizeof(char) * INFO_LENGTH);
    strcpy(tds__GetServicesResponse->Service[0].Namespace, "");
    strcpy(tds__GetServicesResponse[0].Service->XAddr, _IPAddr);
    tds__GetServicesResponse->Service[0].Capabilities = NULL;
    tds__GetServicesResponse->Service[0].Version = (struct tt__OnvifVersion *)soap_malloc(soap, sizeof(struct tt__OnvifVersion));
    tds__GetServicesResponse->Service[0].Version->Major = 0;
    tds__GetServicesResponse->Service[0].Version->Minor = 3;
    tds__GetServicesResponse->Service[0].__any = (char **)soap_malloc(soap, sizeof(char *));
    tds__GetServicesResponse->Service[0].__any[0] = (char *)soap_malloc(soap, sizeof(char) * INFO_LENGTH);
    strcpy(tds__GetServicesResponse->Service[0].__any[0],"why1");
    tds__GetServicesResponse->Service[0].__any[1] = (char *)soap_malloc(soap,sizeof(char) * INFO_LENGTH);
    strcpy(tds__GetServicesResponse->Service[0].__any[1],"why2");
    tds__GetServicesResponse->Service[0].__size = NULL;
    tds__GetServicesResponse->Service[0].__anyAttribute = NULL;
    return SOAP_OK;
}

3、实现GetVideoSources命令

int  __tmd__GetVideoSources(struct soap* soap, struct _trt__GetVideoSources *trt__GetVideoSources, struct _trt__GetVideoSourcesResponse *trt__GetVideoSourcesResponse)
{
    DBG("__tmd__GetVideoSources\n");

    int size1;
    size1 = 1;
    trt__GetVideoSourcesResponse->__sizeVideoSources = size1;
    trt__GetVideoSourcesResponse->VideoSources = (struct tt__VideoSource *)soap_malloc(soap, sizeof(struct tt__VideoSource) * size1);
    trt__GetVideoSourcesResponse->VideoSources[0].Framerate = 30;
    trt__GetVideoSourcesResponse->VideoSources[0].Resolution = (struct tt__VideoResolution *)soap_malloc(soap, sizeof(struct tt__VideoResolution));
    trt__GetVideoSourcesResponse->VideoSources[0].Resolution->Height = 720;
    trt__GetVideoSourcesResponse->VideoSources[0].Resolution->Width = 1280;
    trt__GetVideoSourcesResponse->VideoSources[0].token = (char *)soap_malloc(soap, sizeof(char)*INFO_LENGTH);
    strcpy(trt__GetVideoSourcesResponse->VideoSources[0].token,"GhostyuSource_token"); //注意这里需要和GetProfile中的sourcetoken相同

    trt__GetVideoSourcesResponse->VideoSources[0].Imaging =(struct tt__ImagingSettings*)soap_malloc(soap, sizeof(struct tt__ImagingSettings));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Brightness = (float*)soap_malloc(soap,sizeof(float));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Brightness[0] = 128;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->ColorSaturation = (float*)soap_malloc(soap,sizeof(float));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->ColorSaturation[0] = 128;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Contrast = (float*)soap_malloc(soap,sizeof(float));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Contrast[0] = 128;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->IrCutFilter = (int *)soap_malloc(soap,sizeof(int));
    *trt__GetVideoSourcesResponse->VideoSources[0].Imaging->IrCutFilter = 0;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Sharpness = (float*)soap_malloc(soap,sizeof(float));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Sharpness[0] = 128;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->BacklightCompensation = (struct tt__BacklightCompensation*)soap_malloc(soap, sizeof(struct tt__BacklightCompensation));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->BacklightCompensation->Mode = 0;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->BacklightCompensation->Level = 20;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Exposure = NULL;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Focus = NULL;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WideDynamicRange = (struct tt__WideDynamicRange*)soap_malloc(soap, sizeof(struct tt__WideDynamicRange));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WideDynamicRange->Mode = 0;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WideDynamicRange->Level = 20;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WhiteBalance = (struct tt__WhiteBalance*)soap_malloc(soap, sizeof(struct tt__WhiteBalance));
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WhiteBalance->Mode = 0;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WhiteBalance->CrGain = 0;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->WhiteBalance->CbGain = 0;
    trt__GetVideoSourcesResponse->VideoSources[0].Imaging->Extension = NULL;
    trt__GetVideoSourcesResponse->VideoSources[0].Extension = NULL;
    return SOAP_OK;
}

__tmd__GetVideoSources最重要的是token的填充，必须要和下面profile中的sourcetoken相同，需要匹配到这个视频源

4、实现GetProfiles命令

size = 1;
trt__GetProfilesResponse->Profiles =(struct tt__Profile *)soap_malloc(soap, sizeof(struct tt__Profile) * size);
trt__GetProfilesResponse->__sizeProfiles = size;

i=0;
trt__GetProfilesResponse->Profiles[i].Name = (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
strcpy(trt__GetProfilesResponse->Profiles[i].Name,"my_profile");
trt__GetProfilesResponse->Profiles[i].token= (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
strcpy(trt__GetProfilesResponse->Profiles[i].token,"token_profile");
trt__GetProfilesResponse->Profiles[i].fixed = _false;
trt__GetProfilesResponse->Profiles[i].__anyAttribute = NULL;

除了上面的基本信息，还需要填充两大项：VideoSourceConfiguration和VideoEncoderConfiguration，一个用于描述视频源的信息，另外一个描述视频的编码信息

先给VideoSourceConfiguration分配空间

trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration = (struct tt__VideoSourceConfiguration *)soap_malloc(soap,sizeof(struct tt__VideoSourceConfiguration ));
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Name = (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->token = (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->SourceToken = (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Bounds = (struct tt__IntRectangle *)soap_malloc(soap,sizeof(struct tt__IntRectangle));

然后在填充它

/*注意SourceToken*/
strcpy(trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Name,"VS_Name");
strcpy(trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->token,"VS_Token");
strcpy(trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->SourceToken,"GhostyuSource_token"); /*必须与__tmd__GetVideoSources中的token相同*/
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->UseCount = 1;
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Bounds->x = 1;
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Bounds->y = 1;
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Bounds->height = 720;
trt__GetProfilesResponse->Profiles[i].VideoSourceConfiguration->Bounds->width = 1280;

如果是指针必须先用soap_malloc分配内存，然后才能赋值

下面是VideoEncoderConfiguration

trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration = (struct tt__VideoEncoderConfiguration *)soap_malloc(soap,sizeof(struct tt__VideoEncoderConfiguration));
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Name = (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->token= (char *)soap_malloc(soap,sizeof(char)*MAX_PROF_TOKEN);
strcpy(trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Name,"VE_Name1");
strcpy(trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->token,"VE_token1");
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->UseCount = 1;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Quality = 10;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Encoding = 1;//JPEG = 0, MPEG4 = 1, H264 = 2;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Resolution = (struct tt__VideoResolution *)soap_malloc(soap, sizeof(struct tt__VideoResolution));
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Resolution->Height = 720;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->Resolution->Width = 1280;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->RateControl = (struct tt__VideoRateControl *)soap_malloc(soap, sizeof(struct tt__VideoRateControl));
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->RateControl->FrameRateLimit = 30;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->RateControl->EncodingInterval = 1;
trt__GetProfilesResponse->Profiles[i].VideoEncoderConfiguration->RateControl->BitrateLimit = 500;

5、GetVideoSourceConfiguration和GetVideoEncoderConfiguration

int  __trt__GetVideoSourceConfiguration(struct soap* soap, struct _trt__GetVideoSourceConfiguration *trt__GetVideoSourceConfiguration, struct _trt__GetVideoSourceConfigurationResponse *trt__GetVideoSourceConfigurationResponse)
{
    DBG("__trt__GetVideoSourceConfiguration\n");
    //该函数必要，live video需要
    return SOAP_OK;
}

int  __trt__GetVideoEncoderConfiguration(struct soap* soap, struct _trt__GetVideoEncoderConfiguration *trt__GetVideoEncoderConfiguration, struct _trt__GetVideoEncoderConfigurationResponse *trt__GetVideoEncoderConfigurationResponse)
{
    DBG("__trt__GetVideoEncoderConfiguration\n");
    return SOAP_OK;
}

6、GetVideoEncoderConfigurationOptions

int  __trt__GetVideoEncoderConfigurationOptions(struct soap* soap, struct _trt__GetVideoEncoderConfigurationOptions *trt__GetVideoEncoderConfigurationOptions, struct _trt__GetVideoEncoderConfigurationOptionsResponse *trt__GetVideoEncoderConfigurationOptionsResponse)
{
    DBG("__trt__GetVideoEncoderConfigurationOptions\n");
    //该函数必要，video streaming需要
    return SOAP_OK;
}

以上5、6不分的代码直接返回SOAP_OK即可，正常来说是应该填充的，这里不影响RTSP Video Stream，暂时就不去动它

四、运行live555MediaServer服务器

live555官网有很多测试文件，我这里用的是MPEG4的测试文件路劲为rtsp://192.168.1.201/petrov.m4e

五、启动Onvif Device Manager测试

有一个问题，OnvifDeviceManager的并不能自动发现设备（OnvifTestTool可以），还好它提供了手动添加功能

单击add，添加如下内容：http://192.168.1.233/onvif/device_service

注意，我在程序中固定了两个IP：linux192.168.1.233，windows：192.168.1.201，这里需看情况修改

测试截图：

1、Live video

2、Video streaming

3、Profiles

最后是运行的live555 rtsp服务器

终端打印的DEBUG信息

源代码下载地址：

【来自http://blog.csdn.net/ghostyu，ghostyu#163.com】