# Camera
## 概述
### 功能简介
OpenHarmony相机驱动框架模型对上实现相机HDI（Hardware Device Interface）接口，对下实现相机Pipeline模型，管理相机各个硬件设备。
该驱动框架模型内部分为三层，依次为HDI实现层、框架层和设备适配层。各层基本概念如下：
+ HDI实现层：实现OHOS（OpenHarmony Operation System）相机标准南向接口。
+ 框架层：对接HDI实现层的控制、流的转发，实现数据通路的搭建，管理相机各个硬件设备等功能。
+ 设备适配层：屏蔽底层芯片和OS（Operation System）差异，支持多平台适配。
### 运作机制
Camera模块主要包含服务、设备的初始化，数据通路的搭建，流的配置、创建、下发、捕获等，具体运作机制参考以下图文解析：
**图 1** 基于HDF驱动框架的Camera驱动模型
1. 系统启动时创建camera_host进程。进程创建后，首先枚举底层设备，创建（也可以通过配置表创建）管理设备树的DeviceManager类及其内部各个底层设备的对象，创建对应的CameraHost类实例并且将其注册到UHDF（用户态HDF驱动框架）服务中，方便相机服务层通过UHDF服务获取底层CameraDeviceHost的服务，从而操作硬件设备。
2. Service通过CameraDeviceHost服务获取CameraHost实例，CameraHost可以获取底层的Camera能力，开启闪光灯、调用Open接口打开Camera创建连接、创建DeviceManager（负责底层硬件模块上电）、创建CameraDevice（向上提供设备控制接口）。创建CameraDevice时会实例化PipelineCore的各个子模块，其中StreamPipelineCore负责创建Pipeline，MetaQueueManager负责上报metaData。
3. Service通过CameraDevice模块配置流、创建Stream类。StreamPipelineStrategy模块通过上层下发的模式和查询配置表创建对应流的Node连接方式，StreamPipelineBuilder模块创建Node实例并且连接返回该Pipeline给StreamPipelineDispatcher。StreamPipelineDispatcher提供统一的Pipeline调用管理。
4. Service通过Stream控制整个流的操作，AttachBufferQueue接口将从显示模块申请的BufferQueue下发到底层，由CameraDeviceDriverModel自行管理buffer，当Capture接口下发命令后，底层开始向上传递buffer。Pipeline的IspNode依次从BufferQueue获取指定数量buffer，然后下发到底层ISP（Image Signal Processor，图像信号处理器）硬件，ISP填充完之后将buffer传递给CameraDeviceDriverModel，CameraDeviceDriverModel通过循环线程将buffer填充到已经创建好的Pipeline中，各个Node处理后通过回调传递给上层，同时buffer返回BufferQueue等待下一次下发。
5. Service通过Capture接口下发拍照命令。ChangeToOfflineStream接口查询拍照buffer位置，如果ISP已经出图，并且图像数据已经送到IPP node，可以将普通拍照流转换为离线流，否则直接走关闭流程。ChangeToOfflineStream接口通过传递StreamInfo使离线流获取到普通流的流信息，并且通过配置表确认离线流的具体Node连接方式，创建离线流的Node连接（如果已创建则通过CloseCamera释放非离线流所需的Node），等待buffer从底层Pipeline回传到上层再释放持有的Pipeline相关资源。
6. Service通过CameraDevice的UpdateSettings接口向下发送CaptureSetting参数，CameraDeviceDriverModel通过StreamPipelineDispatcher模块向各个Node转发，StartStreamingCapture和Capture接口携带的CaptureSetting通过StreamPipelineDispatcher模块向该流所属的Node转发。
7. Service通过EnableResult和DisableResult接口控制底层metaData的上报。如果需要底层metaData上报，pipeline会创建CameraDeviceDriverModel内部的一个Bufferqueue用来收集和传递metaData，根据StreamPipelineStrategy模块查询配置表并通过StreamPipelineBuilder创建和连接Node，MetaQueueManager下发buffer至底层，底层相关Node填充数据，MetaQueueManager模块再调用上层回调传递给上层。
8. Service调用CameraDevice的Close接口，CameraDevice调用对应的DeviceManager模块对各个硬件下电；如果此时在Ipp的SubPipeline中存在OfflineStream，则需要保留OfflineStream，直到执行完毕。
9. 动态帧率控制。在StreamOperator中起一个CollectBuffer线程，CollectBuffer线程从每一路stream的BufferQueue中获取buffer，如果某一路流的帧率需要控制（为sensor出帧帧率的1/n），可以根据需求控制每一帧的buffer打包，并决定是否collect此路流的buffer（比如sensor出帧帧率为120fps，预览流的帧率为30fps，CollectBuffer线程collect预览流的buffer时，每隔4fps collect一次）。
## 开发指导
### 场景介绍
Camera模块主要针对相机预览、拍照、视频流等场景，对这些场景下的相机操作进行封装，使开发者更易操作相机硬件，提高开发效率。
### 接口说明
注：以下接口列举的为IDL接口描述生成的对应C++语言函数接口，接口声明见idl文件（/drivers/interface/camera/v1_0/）。
在HDI使用中下发的配置参数不能超出GetCameraAbility上报的能力范围。即使通过UpdateSettings、CommitStreams、Capture等接口可以下发超出该范围的配置参数，且接口调用不会返回失败，但设置后的行为是不确定的。
- icamera_device.h
| 功能描述 | 接口名称 |
| ---------------------------- | ------------------------------------------------------------ |
| 获取流控制器 | int32_t GetStreamOperator(const sptr& callbackObj,
sptr& streamOperator) |
| 更新设备控制参数 | int32_t UpdateSettings(const std::vector& settings) |
| 设置Result回调模式和回调函数 | int32_t SetResultMode(ResultCallbackMode mode) |
| 获取使能的ResultMeta | int32_t GetEnabledResults(std::vector& results) |
| 使能具体的ResultMeta | int32_t EnableResult(const std::vector& results) |
| 禁止具体的ResultMeta | int32_t DisableResult(const std::vector& results) |
| 关闭Camera设备 | int32_t Close() |
- icamera_device_callback.h
| 功能描述 | 接口名称 |
| ---------------------------------------------------------- | ------------------------------------------------------------ |
| 设备发生错误时调用，由调用者实现，用于返回错误信息给调用者 | int32_t OnError(ErrorType type, int32_t errorCode) |
| 上报camera设备相关的metadata的回调 | int32_t OnResult(uint64_t timestamp, const std::vector& result) |
- icamera_host.h
| 功能描述 | 接口名称 |
| ------------------------------ | ------------------------------------------------------------ |
| 设置ICameraHost回调接口 | int32_t SetCallback(const sptr& callbackObj) |
| 获取当前可用的Camera设备ID列表 | int32_t GetCameraIds(std::vector& cameraIds) |
| 获取Camera设备能力集合 | int32_t GetCameraAbility(const std::string& cameraId, std::vector& cameraAbility) |
| 打开Camera设备 | int32_t OpenCamera(const std::string& cameraId, const sptr& callbackObj,
sptr& device) |
| 打开或关闭闪光灯 | int32_t SetFlashlight(const std::string& cameraId, bool isEnable) |
- icamera_host_callback.h
| 功能描述 | 接口名称 |
| ---------------------- | ------------------------------------------------------------ |
| Camera设备状态变化上报 | int32_t OnCameraStatus(const std::string& cameraId, CameraStatus status) |
| 闪光灯状态变化回调 | int32_t OnFlashlightStatus(const std::string& cameraId, FlashlightStatus status) |
| Camera事件回调 | int32_t OnCameraEvent(const std::string& cameraId, CameraEvent event) |
- ioffline_stream_operator.h
| 功能描述 | 接口名称 |
| -------------- | ------------------------------------------------------------ |
| 取消捕获请求 | int32_t CancelCapture(int32_t captureId) |
| 释放流 | int32_t ReleaseStreams(const std::vector& streamIds) |
| 释放所有离线流 | int32_t Release() |
- istream_operator.h
| 功能描述 | 接口名称 |
| -------------------------------- | ------------------------------------------------------------ |
| 查询是否支持添加参数对应的流 | int32_t IsStreamsSupported(
OperationMode mode,
const std::vector& modeSetting,
const std::vector& infos,
StreamSupportType& type) |
| 创建流 | int32_t CreateStreams(const std::vector& streamInfos) |
| 释放流 | int32_t ReleaseStreams(const std::vector& streamIds) |
| 配置流 | int32_t CommitStreams(OperationMode mode, const std::vector& modeSetting) |
| 获取流的属性 | int32_t GetStreamAttributes(std::vector& attributes) |
| 绑定生产者句柄和指定流 | int32_t AttachBufferQueue(int32_t streamId, const sptr& bufferProducer) |
| 解除生产者句柄和指定流的绑定关系 | int32_t DetachBufferQueue(int32_t streamId) |
| 捕获图像 | int32_t Capture(int32_t captureId, const CaptureInfo& info, bool isStreaming) |
| 取消捕获 | int32_t CancelCapture(int32_t captureId) |
| 将指定流转换成离线流 | int32_t ChangeToOfflineStream(const std::vector& streamIds,
const sptr& callbackObj,
sptr& offlineOperator) |
- istream_operator_callback.h
| 功能描述 | 接口名称 |
| ---------------------------------------- | ------------------------------------------------------------ |
| 捕获开始回调，在捕获开始时调用 | int32_t OnCaptureStarted(int32_t captureId, const std::vector& streamIds) |
| 捕获结束回调，在捕获结束时调用 | int32_t OnCaptureEnded(int32_t captureId, const std::vector& infos) |
| 捕获错误回调，在捕获过程中发生错误时调用 | int32_t OnCaptureError(int32_t captureId, const std::vector& infos) |
| 帧捕获回调 | int32_t OnFrameShutter(int32_t captureId, const std::vector& streamIds, uint64_t timestamp) |
### 开发步骤
Camera驱动的开发过程主要包含以下步骤：
1. 注册CameraHost
定义Camera的HdfDriverEntry结构体，该结构体中定义了CameraHost初始化的方法（代码目录drivers/peripheral/camera/interfaces/hdi_ipc/camera_host_driver.cpp）。
```c++
struct HdfDriverEntry g_cameraHostDriverEntry = {
.moduleVersion = 1,
.moduleName = "camera_service",
.Bind = HdfCameraHostDriverBind,
.Init = HdfCameraHostDriverInit,
.Release = HdfCameraHostDriverRelease,
};
HDF_INIT(g_cameraHostDriverEntry); // 将Camera的HdfDriverEntry结构体注册到HDF上
```
2. 初始化Host服务
步骤1中提到的HdfCameraHostDriverBind接口提供了CameraServiceDispatch和CameraHostStubInstance的注册。CameraServiceDispatch接口是远端调用CameraHost的方法，如OpenCamera()，SetFlashlight()等，CameraHostStubInstance接口是Camera设备的初始化，在开机时被调用。
```c++
static int HdfCameraHostDriverBind(struct HdfDeviceObject *deviceObject)
{
HDF_LOGI("HdfCameraHostDriverBind enter");
auto *hdfCameraHostHost = new (std::nothrow) HdfCameraHostHost;
if (hdfCameraHostHost == nullptr) {
HDF_LOGE("%{public}s: failed to create HdfCameraHostHost object", __func__);
return HDF_FAILURE;
}
hdfCameraHostHost->ioService.Dispatch = CameraHostDriverDispatch; // 提供远端CameraHost调用方法
hdfCameraHostHost->ioService.Open = NULL;
hdfCameraHostHost->ioService.Release = NULL;
auto serviceImpl = ICameraHost::Get(true);
if (serviceImpl == nullptr) {
HDF_LOGE("%{public}s: failed to get of implement service", __func__);
delete hdfCameraHostHost;
return HDF_FAILURE;
}
hdfCameraHostHost->stub = OHOS::HDI::ObjectCollector::GetInstance().GetOrNewObject(serviceImpl,
ICameraHost::GetDescriptor()); // 初始化Camera设备
if (hdfCameraHostHost->stub == nullptr) {
HDF_LOGE("%{public}s: failed to get stub object", __func__);
delete hdfCameraHostHost;
return HDF_FAILURE;
}
deviceObject->service = &hdfCameraHostHost->ioService;
return HDF_SUCCESS;
}
```
下面的函数是远端CameraHost调用的方法：
```c++
int32_t CameraHostStub::CameraHostServiceStubOnRemoteRequest(int cmdId, MessageParcel &data,
MessageParcel &reply, MessageOption &option)
{
switch(cmdId) {
case CMD_CAMERA_HOST_SET_CALLBACK: {
return CameraHostStubSetCallback(data, reply, option);
}
case CMD_CAMERA_HOST_GET_CAMERAID: {
return CameraHostStubGetCameraIds(data, reply, option);
}
case CMD_CAMERA_HOST_GET_CAMERA_ABILITY: {
return CameraHostStubGetCameraAbility(data, reply, option);
}
case CMD_CAMERA_HOST_OPEN_CAMERA: {
return CameraHostStubOpenCamera(data, reply, option);
}
case CMD_CAMERA_HOST_SET_FLASH_LIGHT: {
return CameraHostStubSetFlashlight(data, reply, option);
}
default: {
HDF_LOGE("%s: not support cmd %d", __func__, cmdId);
return HDF_ERR_INVALID_PARAM;
}
}
return HDF_SUCCESS;
}
```
CameraHostStubInstance()接口最终调用CameraHostImpl::Init()方法，该方法会获取物理Camera，并对DeviceManager和PipelineCore进行初始化。
3. 获取Host服务
调用Get()接口从远端CameraService中获取CameraHost对象。get()方法如下：
```c++
sptr ICameraHost::Get(const char *serviceName)
{
do {
using namespace OHOS::HDI::ServiceManager::V1_0;
auto servMgr = IServiceManager::Get();
if (servMgr == nullptr) {
HDF_LOGE("%s: IServiceManager failed!", __func__);
break;
}
auto remote = servMgr->GetService(serviceName); // 根据serviceName名称获取CameraHost
if (remote != nullptr) {
sptr hostSptr = iface_cast(remote); // 将CameraHostProxy对象返回给调用者，该对象中包含OpenCamera()等方法。
return hostSptr;
}
HDF_LOGE("%s: GetService failed! serviceName = %s", __func__, serviceName);
} while(false);
HDF_LOGE("%s: get %s failed!", __func__, serviceName);
return nullptr;
}
```
4. 打开设备
CameraHostProxy对象中有五个方法，分别是SetCallback、GetCameraIds、GetCameraAbility、OpenCamera和SetFlashlight。下面着重描述OpenCamera接口。
CameraHostProxy的OpenCamera()接口通过CMD_CAMERA_HOST_OPEN_CAMERA调用远端CameraHostStubOpenCamera()接口并获取ICameraDevice对象。
```c++
int32_t CameraHostProxy::OpenCamera(const std::string& cameraId, const sptr& callbackObj,
sptr& device)
{
MessageParcel cameraHostData;
MessageParcel cameraHostReply;
MessageOption cameraHostOption(MessageOption::TF_SYNC);
if (!cameraHostData.WriteInterfaceToken(ICameraHost::GetDescriptor())) {
HDF_LOGE("%{public}s: failed to write interface descriptor!", __func__);
return HDF_ERR_INVALID_PARAM;
}
if (!cameraHostData.WriteCString(cameraId.c_str())) {
HDF_LOGE("%{public}s: write cameraId failed!", __func__);
return HDF_ERR_INVALID_PARAM;
}
if (!cameraHostData.WriteRemoteObject(OHOS::HDI::ObjectCollector::GetInstance().GetOrNewObject(callbackObj,
ICameraDeviceCallback::GetDescriptor()))) {
HDF_LOGE("%{public}s: write callbackObj failed!", __func__);
return HDF_ERR_INVALID_PARAM;
}
int32_t cameraHostRet = Remote()->SendRequest(CMD_CAMERA_HOST_OPEN_CAMERA, cameraHostData, cameraHostReply, cameraHostOption);
if (cameraHostRet != HDF_SUCCESS) {
HDF_LOGE("%{public}s failed, error code is %{public}d", __func__, cameraHostRet);
return cameraHostRet;
}
device = hdi_facecast(cameraHostReply.ReadRemoteObject());
return cameraHostRet;
}
```
Remote()->SendRequest调用上文提到的CameraHostServiceStubOnRemoteRequest()，根据cmdId进入CameraHostStubOpenCamera()接口，最终调用CameraHostImpl::OpenCamera()，该接口获取了CameraDevice并对硬件进行上电等操作。
```c++
int32_t CameraHostImpl::OpenCamera(const std::string& cameraId, const sptr& callbackObj,
sptr& device)
{
CAMERA_LOGD("OpenCamera entry");
DFX_LOCAL_HITRACE_BEGIN;
if (CameraIdInvalid(cameraId) != RC_OK || callbackObj == nullptr) {
CAMERA_LOGW("open camera id is empty or callback is null.");
return INVALID_ARGUMENT;
}
auto itr = cameraDeviceMap_.find(cameraId);
if (itr == cameraDeviceMap_.end()) {
CAMERA_LOGE("camera device not found.");
return INSUFFICIENT_RESOURCES;
}
CAMERA_LOGD("OpenCamera cameraId find success.");
std::shared_ptr cameraDevice = itr->second;
if (cameraDevice == nullptr) {
CAMERA_LOGE("camera device is null.");
return INSUFFICIENT_RESOURCES;
}
CamRetCode ret = cameraDevice->SetCallback(callbackObj);
CHECK_IF_NOT_EQUAL_RETURN_VALUE(ret, HDI::Camera::V1_0::NO_ERROR, ret);
CameraHostConfig *config = CameraHostConfig::GetInstance();
CHECK_IF_PTR_NULL_RETURN_VALUE(config, INVALID_ARGUMENT);
std::vector phyCameraIds;
RetCode rc = config->GetPhysicCameraIds(cameraId, phyCameraIds);
if (rc != RC_OK) {
CAMERA_LOGE("get physic cameraId failed.");
return DEVICE_ERROR;
}
if (CameraPowerUp(cameraId, phyCameraIds) != RC_OK) { // 对Camera硬件上电
CAMERA_LOGE("camera powerup failed.");
CameraPowerDown(phyCameraIds);
return DEVICE_ERROR;
}
auto sptrDevice = deviceBackup_.find(cameraId);
if (sptrDevice == deviceBackup_.end()) {
#ifdef CAMERA_BUILT_ON_OHOS_LITE
deviceBackup_[cameraId] = cameraDevice;
#else
deviceBackup_[cameraId] = cameraDevice.get();
#endif
}
device = deviceBackup_[cameraId];
cameraDevice->SetStatus(true);
CAMERA_LOGD("open camera success.");
DFX_LOCAL_HITRACE_END;
return HDI::Camera::V1_0::NO_ERROR;
}
```
5. 获取流
CameraDeviceImpl定义了GetStreamOperator、UpdateSettings、SetResultMode和GetEnabledResult等方法，获取流操作方法如下：
```c++
int32_t CameraDeviceImpl::GetStreamOperator(const sptr& callbackObj,
sptr& streamOperator)
{
HDI_DEVICE_PLACE_A_WATCHDOG;
DFX_LOCAL_HITRACE_BEGIN;
if (callbackObj == nullptr) {
CAMERA_LOGW("input callback is null.");
return INVALID_ARGUMENT;
}
spCameraDeciceCallback_ = callbackObj;
if (spStreamOperator_ == nullptr) {
#ifdef CAMERA_BUILT_ON_OHOS_LITE
// 这里创建一个spStreamOperator_ 对象传递给调用者，以便对stream进行各种操作
spStreamOperator_ = std::make_shared(spCameraDeciceCallback_, shared_from_this());
#else
spStreamOperator_ = new(std::nothrow) StreamOperator(spCameraDeciceCallback_, shared_from_this());
#endif
if (spStreamOperator_ == nullptr) {
CAMERA_LOGW("create stream operator failed.");
return DEVICE_ERROR;
}
spStreamOperator_->Init();
ismOperator_ = spStreamOperator_;
}
streamOperator = ismOperator_;
#ifndef CAMERA_BUILT_ON_OHOS_LITE
CAMERA_LOGI("CameraDeviceImpl %{public}s: line: %{public}d", __FUNCTION__, __LINE__);
pipelineCore_->GetStreamPipelineCore()->SetCallback(
[this](const std::shared_ptr &metadata) {
OnMetadataChanged(metadata);
});
#endif
DFX_LOCAL_HITRACE_END;
return HDI::Camera::V1_0::NO_ERROR;
}
```
6. 创建流
调用CreateStreams创建流前需要填充StreamInfo结构体，具体内容如下：
```c++
using StreamInfo = struct _StreamInfo {
int streamId_;
int width_; // 数据流宽
int height_; // 数据流高
int format_; // 数据流格式，如PIXEL_FMT_YCRCB_420_SP
int dataSpace_;
StreamIntent intent_; // StreamIntent 如PREVIEW
bool tunneledMode_;
BufferProducerSequenceable bufferQueue_; // 数据流bufferQueue可用streamCustomer->CreateProducer()接口创建
int minFrameDuration_;
EncodeType encodeType_;
};
```
CreateStreams()接口是StreamOperator（StreamOperatorImpl类是StreamOperator的基类）类中的方法，该接口的主要作用是创建一个StreamBase对象，通过StreamBase的Init方法初始化CreateBufferPool等操作。
```c++
int32_t StreamOperator::CreateStreams(const std::vector& streamInfos)
{
PLACE_A_NOKILL_WATCHDOG(requestTimeoutCB_);
DFX_LOCAL_HITRACE_BEGIN;
for (const auto& it : streamInfos) {
CHECK_IF_NOT_EQUAL_RETURN_VALUE(CheckStreamInfo(it), true, INVALID_ARGUMENT);
CAMERA_LOGI("streamId:%{public}d and format:%{public}d and width:%{public}d and height:%{public}d",
it.streamId_, it.format_, it.width_, it.height_);
if (streamMap_.count(it.streamId_) > 0) {
CAMERA_LOGE("stream [id = %{public}d] has already been created.", it.streamId_);
return INVALID_ARGUMENT;
}
std::shared_ptr stream = StreamFactory::Instance().CreateShared( // 创建Stream实例
IStream::g_availableStreamType[it.intent_], it.streamId_, it.intent_, pipelineCore_, messenger_);
if (stream == nullptr) {
CAMERA_LOGE("create stream [id = %{public}d] failed.", it.streamId_);
return INSUFFICIENT_RESOURCES;
}
StreamConfiguration scg;
StreamInfoToStreamConfiguration(scg, it);
RetCode rc = stream->ConfigStream(scg);
if (rc != RC_OK) {
CAMERA_LOGE("configure stream %{public}d failed", it.streamId_);
return INVALID_ARGUMENT;
}
if (!scg.tunnelMode && (it.bufferQueue_)->producer_ != nullptr) {
CAMERA_LOGE("stream [id:%{public}d] is not tunnel mode, can't bind a buffer producer", it.streamId_);
return INVALID_ARGUMENT;
}
if ((it.bufferQueue_)->producer_ != nullptr) {
auto tunnel = std::make_shared();
CHECK_IF_PTR_NULL_RETURN_VALUE(tunnel, INSUFFICIENT_RESOURCES);
rc = tunnel->AttachBufferQueue((it.bufferQueue_)->producer_);
CHECK_IF_NOT_EQUAL_RETURN_VALUE(rc, RC_OK, INVALID_ARGUMENT);
if (stream->AttachStreamTunnel(tunnel) != RC_OK) {
CAMERA_LOGE("attach buffer queue to stream [id = %{public}d] failed", it.streamId_);
return INVALID_ARGUMENT;
}
}
{
std::lock_guard l(streamLock_);
streamMap_[stream->GetStreamId()] = stream;
}
CAMERA_LOGI("create stream success [id:%{public}d] [type:%{public}s]", stream->GetStreamId(),
IStream::g_availableStreamType[it.intent_].c_str());
}
DFX_LOCAL_HITRACE_END;
return HDI::Camera::V1_0::NO_ERROR;
}
```
7. 配置流
CommitStreams()是配置流的接口，必须在创建流之后调用，其主要作用是初始化Pipeline和创建Pipeline。
```c++
int32_t StreamOperator::CommitStreams(OperationMode mode, const std::vector& modeSetting)
{
CAMERA_LOGV("enter");
CHECK_IF_PTR_NULL_RETURN_VALUE(streamPipeline_, DEVICE_ERROR);
PLACE_A_NOKILL_WATCHDOG(requestTimeoutCB_);
if (modeSetting.empty()) {
CAMERA_LOGE("input vector is empty");
return INVALID_ARGUMENT;
}
DFX_LOCAL_HITRACE_BEGIN;
std::vector configs = {};
{
std::lock_guard l(streamLock_);
std::transform(streamMap_.begin(), streamMap_.end(), std::back_inserter(configs),
[](auto &iter) { return iter.second->GetStreamAttribute(); });
}
std::shared_ptr setting;
MetadataUtils::ConvertVecToMetadata(modeSetting, setting);
DynamicStreamSwitchMode method = streamPipeline_->CheckStreamsSupported(mode, setting, configs);
if (method == DYNAMIC_STREAM_SWITCH_NOT_SUPPORT) {
return INVALID_ARGUMENT;
}
if (method == DYNAMIC_STREAM_SWITCH_NEED_INNER_RESTART) {
std::lock_guard l(streamLock_);
for (auto it : streamMap_) {
it.second->StopStream();
}
}
{
std::lock_guard l(streamLock_);
for (auto it : streamMap_) {
if (it.second->CommitStream() != RC_OK) {
CAMERA_LOGE("commit stream [id = %{public}d] failed.", it.first);
return DEVICE_ERROR;
}
}
}
RetCode rc = streamPipeline_->PreConfig(setting); // 设备流配置
if (rc != RC_OK) {
CAMERA_LOGE("prepare mode settings failed");
return DEVICE_ERROR;
}
rc = streamPipeline_->CreatePipeline(mode); // 创建一个pipeline
if (rc != RC_OK) {
CAMERA_LOGE("create pipeline failed.");
return INVALID_ARGUMENT;
}
DFX_LOCAL_HITRACE_END;
return HDI::Camera::V1_0::NO_ERROR;
}
```
8. 捕获图像
在调用Capture()接口前需要先填充CaptureInfo结构体，具体内容如下：
```c++
using CaptureInfo = struct _CaptureInfo {
int[] streamIds_; // 需要Capture的streamIds
unsigned char[] captureSetting_; // 这里填充camera ability 可通过CameraHost 的GetCameraAbility()接口获取
bool enableShutterCallback_;
};
```
StreamOperator中的Capture方法主要是捕获数据流：
```c++
int32_t StreamOperator::Capture(int32_t captureId, const CaptureInfo& info, bool isStreaming)
{
CHECK_IF_EQUAL_RETURN_VALUE(captureId < 0, true, INVALID_ARGUMENT);
PLACE_A_NOKILL_WATCHDOG(requestTimeoutCB_);
DFX_LOCAL_HITRACE_BEGIN;
for (auto id : info.streamIds_) {
std::lock_guard l(streamLock_);
auto it = streamMap_.find(id);
if (it == streamMap_.end()) {
return INVALID_ARGUMENT;
}
}
{
std::lock_guard l(requestLock_);
auto itr = requestMap_.find(captureId);
if (itr != requestMap_.end()) {
return INVALID_ARGUMENT;
}
}
std::shared_ptr captureSetting;
MetadataUtils::ConvertVecToMetadata(info.captureSetting_, captureSetting);
CaptureSetting setting = captureSetting;
auto request =
std::make_shared(captureId, info.streamIds_.size(), setting,
info.enableShutterCallback_, isStreaming);
for (auto id : info.streamIds_) {
RetCode rc = streamMap_[id]->AddRequest(request);
if (rc != RC_OK) {
return DEVICE_ERROR;
}
}
{
std::lock_guard l(requestLock_);
requestMap_[captureId] = request;
}
return HDI::Camera::V1_0::NO_ERROR;
}
```
9. 取消捕获和释放离线流
StreamOperator类中的CancelCapture()接口的主要作用是根据captureId取消数据流的捕获。
```c++
int32_t StreamOperator::CancelCapture(int32_t captureId)
{
CHECK_IF_EQUAL_RETURN_VALUE(captureId < 0, true, INVALID_ARGUMENT);
PLACE_A_NOKILL_WATCHDOG(requestTimeoutCB_);
DFX_LOCAL_HITRACE_BEGIN;
std::lock_guard l(requestLock_);
auto itr = requestMap_.find(captureId); // 根据captureId 在Map中查找对应的CameraCapture对象
if (itr == requestMap_.end()) {
CAMERA_LOGE("can't cancel capture [id = %{public}d], this capture doesn't exist", captureId);
return INVALID_ARGUMENT;
}
RetCode rc = itr->second->Cancel(); // 调用CameraCapture中Cancel方法结束数据捕获
if (rc != RC_OK) {
return DEVICE_ERROR;
}
requestMap_.erase(itr); // 擦除该CameraCapture对象
DFX_LOCAL_HITRACE_END;
return HDI::Camera::V1_0::NO_ERROR;
}
```
StreamOperator类中的ReleaseStreams接口的主要作用是释放之前通过CreateStream()和CommitStreams()接口创建的流，并销毁Pipeline。
```c++
int32_t StreamOperator::ReleaseStreams(const std::vector& streamIds)
{
PLACE_A_NOKILL_WATCHDOG(requestTimeoutCB_);
DFX_LOCAL_HITRACE_BEGIN;
for (auto id : streamIds) {
std::lock_guard l(streamLock_);
auto it = streamMap_.find(id);
if (it == streamMap_.end()) {
continue;
}
if (it->second->IsRunning()) {
it->second->StopStream();
}
it->second->DumpStatsInfo();
streamMap_.erase(it);
}
for (auto id : streamIds) {
CHECK_IF_EQUAL_RETURN_VALUE(id < 0, true, INVALID_ARGUMENT);
}
DFX_LOCAL_HITRACE_END;
return HDI::Camera::V1_0::NO_ERROR;
}
```
10. 关闭Camera设备
调用CameraDeviceImpl中的Close()来关闭CameraDevice，该接口调用deviceManager中的PowerDown()来给设备下电。
### 开发实例
在/drivers/peripheral/camera/hal/init目录下有一个关于Camera的demo，开机后会在/vendor/bin下生成可执行文件ohos_camera_demo，该demo可以完成Camera的预览，拍照等基础功能。下面我们就以此demo为例讲述怎样用HDI接口去编写预览PreviewOn()和拍照CaptureON()的用例，可参考[ohos_camera_demo](https://gitee.com/openharmony/drivers_peripheral/tree/master/camera/hal/init)。
1. 在main函数中构造一个CameraDemo 对象，该对象中有对Camera初始化、启停流、释放等控制的方法。下面mainDemo->InitSensors()函数为初始化CameraHost，mainDemo->InitCameraDevice()函数为初始化CameraDevice。
```c++
int main(int argc, char** argv)
{
RetCode rc = RC_OK;
auto mainDemo = std::make_shared();
rc = mainDemo->InitSensors(); // 初始化CameraHost
if (rc == RC_ERROR) {
CAMERA_LOGE("main test: mainDemo->InitSensors() error\n");
return -1;
}
rc = mainDemo->InitCameraDevice(); // 初始化CameraDevice
if (rc == RC_ERROR) {
CAMERA_LOGE("main test: mainDemo->InitCameraDevice() error\n");
return -1;
}
rc = PreviewOn(0, mainDemo); // 配流和启流
if (rc != RC_OK) {
CAMERA_LOGE("main test: PreviewOn() error demo exit");
return -1;
}
ManuList(mainDemo, argc, argv); // 打印菜单到控制台
return RC_OK;
}
```
初始化CameraHost函数实现如下，这里调用了HDI接口ICameraHost::Get()去获取demoCameraHost，并对其设置回调函数。
```c++
RetCode OhosCameraDemo::InitSensors()
{
int rc = 0;
CAMERA_LOGD("demo test: InitSensors enter");
if (demoCameraHost_ != nullptr) {
return RC_OK;
}
#ifdef CAMERA_BUILT_ON_OHOS_LITE
demoCameraHost_ = OHOS::Camera::CameraHost::CreateCameraHost();
#else
constexpr const char *DEMO_SERVICE_NAME = "camera_service";
demoCameraHost_ = ICameraHost::Get(DEMO_SERVICE_NAME, false);
#endif
if (demoCameraHost_ == nullptr) {
CAMERA_LOGE("demo test: ICameraHost::Get error");
return RC_ERROR;
}
#ifdef CAMERA_BUILT_ON_OHOS_LITE
hostCallback_ = std::make_shared();
#else
hostCallback_ = new DemoCameraHostCallback();
#endif
rc = demoCameraHost_->SetCallback(hostCallback_);
if (rc != HDI::Camera::V1_0::NO_ERROR) {
CAMERA_LOGE("demo test: demoCameraHost_->SetCallback(hostCallback_) error");
return RC_ERROR;
}
CAMERA_LOGD("demo test: InitSensors exit");
return RC_OK;
}
```
初始化CameraDevice函数实现如下，这里调用了GetCameraIds(cameraIds_)，GetCameraAbility(cameraId, ability_)，OpenCamera(cameraIds_.front(), callback, demoCameraDevice_)等接口实现了demoCameraHost的获取。
```c++
RetCode OhosCameraDemo::InitCameraDevice()
{
int rc = 0;
CAMERA_LOGD("demo test: InitCameraDevice enter");
if (demoCameraHost_ == nullptr) {
CAMERA_LOGE("demo test: InitCameraDevice demoCameraHost_ == nullptr");
return RC_ERROR;
}
(void)demoCameraHost_->GetCameraIds(cameraIds_);
if (cameraIds_.empty()) {
return RC_ERROR;
}
const std::string cameraId = cameraIds_.front();
demoCameraHost_->GetCameraAbility(cameraId, cameraAbility_);
MetadataUtils::ConvertVecToMetadata(cameraAbility_, ability_);
GetFaceDetectMode(ability_);
GetFocalLength(ability_);
GetAvailableFocusModes(ability_);
GetAvailableExposureModes(ability_);
GetExposureCompensationRange(ability_);
GetExposureCompensationSteps(ability_);
GetAvailableMeterModes(ability_);
GetAvailableFlashModes(ability_);
GetMirrorSupported(ability_);
GetStreamBasicConfigurations(ability_);
GetFpsRange(ability_);
GetCameraPosition(ability_);
GetCameraType(ability_);
GetCameraConnectionType(ability_);
GetFaceDetectMaxNum(ability_);
#ifdef CAMERA_BUILT_ON_OHOS_LITE
std::shared_ptr callback = std::make_shared();
#else
sptr callback = new DemoCameraDeviceCallback();
#endif
rc = demoCameraHost_->OpenCamera(cameraIds_.front(), callback, demoCameraDevice_);
if (rc != HDI::Camera::V1_0::NO_ERROR || demoCameraDevice_ == nullptr) {
CAMERA_LOGE("demo test: InitCameraDevice OpenCamera failed");
return RC_ERROR;
}
CAMERA_LOGD("demo test: InitCameraDevice exit");
return RC_OK;
}
```
2. PreviewOn()接口包含配置流、开启预览流和启动Capture动作。该接口执行完成后Camera预览通路已经开始运转并开启了两路流，一路流是preview，另外一路流是capture或者video，两路流中仅对preview流进行capture动作。
```c++
static RetCode PreviewOn(int mode, const std::shared_ptr& mainDemo)
{
RetCode rc = RC_OK;
CAMERA_LOGD("main test: PreviewOn enter");
rc = mainDemo->StartPreviewStream(); // 配置preview流
if (rc != RC_OK) {
CAMERA_LOGE("main test: PreviewOn StartPreviewStream error");
return RC_ERROR;
}
if (mode == 0) {
rc = mainDemo->StartCaptureStream(); // 配置capture流
if (rc != RC_OK) {
CAMERA_LOGE("main test: PreviewOn StartCaptureStream error");
return RC_ERROR;
}
} else {
rc = mainDemo->StartVideoStream(); // 配置video流
if (rc != RC_OK) {
CAMERA_LOGE("main test: PreviewOn StartVideoStream error");
return RC_ERROR;
}
}
rc = mainDemo->CaptureON(STREAM_ID_PREVIEW, CAPTURE_ID_PREVIEW, CAPTURE_PREVIEW);
if (rc != RC_OK) {
CAMERA_LOGE("main test: PreviewOn mainDemo->CaptureON() preview error");
return RC_ERROR;
}
CAMERA_LOGD("main test: PreviewOn exit");
return RC_OK;
}
```
StartCaptureStream()、StartVideoStream()和StartPreviewStream()接口都会调用CreateStream()接口，只是传入的参数不同。
CreateStream()方法调用HDI接口去配置和创建流，首先调用HDI接口去获取StreamOperation对象，然后创建一个StreamInfo。调用CreateStreams()和CommitStreams()实际创建流并配置流。
```c++
RetCode OhosCameraDemo::CreateStream(const int streamId, std::shared_ptr &streamCustomer,
StreamIntent intent)
{
int rc = 0;
CAMERA_LOGD("demo test: CreateStream enter");
GetStreamOpt(); // 获取StreamOperator对象
if (streamOperator_ == nullptr) {
CAMERA_LOGE("demo test: CreateStream GetStreamOpt() is nullptr\n");
return RC_ERROR;
}
StreamInfo streamInfo = {0};
SetStreamInfo(streamInfo, streamCustomer, streamId, intent); // 填充StreamInfo流
if (streamInfo.bufferQueue_->producer_ == nullptr) {
CAMERA_LOGE("demo test: CreateStream CreateProducer(); is nullptr\n");
return RC_ERROR;
}
std::vector streamInfos;
streamInfos.push_back(streamInfo);
rc = streamOperator_->CreateStreams(streamInfos); // 创建流
if (rc != HDI::Camera::V1_0::NO_ERROR) {
CAMERA_LOGE("demo test: CreateStream CreateStreams error\n");
return RC_ERROR;
}
rc = streamOperator_->CommitStreams(NORMAL, cameraAbility_);
if (rc != HDI::Camera::V1_0::NO_ERROR) {
CAMERA_LOGE("demo test: CreateStream CommitStreams error\n");
std::vector streamIds;
streamIds.push_back(streamId);
streamOperator_->ReleaseStreams(streamIds);
return RC_ERROR;
}
CAMERA_LOGD("demo test: CreateStream exit");
return RC_OK;
}
```
CaptureON()接口调用streamOperator的Capture()方法获取Camera数据并轮转buffer，拉起一个线程接收相应类型的数据。
```c++
RetCode OhosCameraDemo::CaptureON(const int streamId,
const int captureId, CaptureMode mode)
{
CAMERA_LOGI("demo test: CaptureON enter streamId == %{public}d and captureId == %{public}d and mode == %{public}d",
streamId, captureId, mode);
std::lock_guard l(metaDatalock_);
if (mode == CAPTURE_SNAPSHOT) {
constexpr double latitude = 27.987500; // dummy data: Qomolangma latitde
constexpr double longitude = 86.927500; // dummy data: Qomolangma longituude
constexpr double altitude = 8848.86; // dummy data: Qomolangma altitude
constexpr size_t entryCapacity = 100;
constexpr size_t dataCapacity = 2000;
captureSetting_ = std::make_shared(entryCapacity, dataCapacity);
captureQuality_ = OHOS_CAMERA_JPEG_LEVEL_HIGH;
captureOrientation_ = OHOS_CAMERA_JPEG_ROTATION_270;
mirrorSwitch_ = OHOS_CAMERA_MIRROR_ON;
gps_.push_back(latitude);
gps_.push_back(longitude);
gps_.push_back(altitude);
captureSetting_->addEntry(OHOS_JPEG_QUALITY, static_cast(&captureQuality_),
sizeof(captureQuality_));
captureSetting_->addEntry(OHOS_JPEG_ORIENTATION, static_cast(&captureOrientation_),
sizeof(captureOrientation_));
captureSetting_->addEntry(OHOS_CONTROL_CAPTURE_MIRROR, static_cast(&mirrorSwitch_),
sizeof(mirrorSwitch_));
captureSetting_->addEntry(OHOS_JPEG_GPS_COORDINATES, gps_.data(), gps_.size());
}
std::vector setting;
MetadataUtils::ConvertMetadataToVec(captureSetting_, setting);
captureInfo_.streamIds_ = {streamId};
if (mode == CAPTURE_SNAPSHOT) {
captureInfo_.captureSetting_ = setting;
} else {
captureInfo_.captureSetting_ = cameraAbility_;
}
captureInfo_.enableShutterCallback_ = false;
int rc = streamOperator_->Capture(captureId, captureInfo_, true); // 实际capture开始，buffer轮转开始
if (rc != HDI::Camera::V1_0::NO_ERROR) {
CAMERA_LOGE("demo test: CaptureStart Capture error\n");
streamOperator_->ReleaseStreams(captureInfo_.streamIds_);
return RC_ERROR;
}
if (mode == CAPTURE_PREVIEW) {
streamCustomerPreview_->ReceiveFrameOn(nullptr); // 创建预览线程接收传递上来的buffer
} else if (mode == CAPTURE_SNAPSHOT) {
streamCustomerCapture_->ReceiveFrameOn([this](void* addr, const uint32_t size) { // 创建capture线程通过StoreImage回调接收传递上来的buffer
StoreImage(addr, size);
});
} else if (mode == CAPTURE_VIDEO) {
OpenVideoFile();
streamCustomerVideo_->ReceiveFrameOn([this](void* addr, const uint32_t size) { // 创建video线程通过StoreImage回调接收传递上来的buffer
StoreVideo(addr, size);
});
}
CAMERA_LOGD("demo test: CaptureON exit");
return RC_OK;
}
```
3. ManuList()函数从控制台通过fgets()接口获取字符，不同字符所对应demo支持的功能不同，并打印出该demo所支持功能的菜单。
```c++
static void ManuList(const std::shared_ptr& mainDemo,
const int argc, char** argv)
{
int idx, c;
bool isAwb = true;
const char *shortOptions = "h:cwvaeqof:";
c = getopt_long(argc, argv, shortOptions, LONG_OPTIONS, &idx);
while (1) {
switch (c) {
case 'h':
c = PutMenuAndGetChr(); // 打印菜单
break;
case 'f':
FlashLightTest(mainDemo); // 手电筒功能测试
c = PutMenuAndGetChr();
break;
case 'o':
OfflineTest(mainDemo); // Offline功能测试
c = PutMenuAndGetChr();
break;
case 'c':
CaptureTest(mainDemo); // Capture功能测试
c = PutMenuAndGetChr();
break;
case 'w': // AWB功能测试
if (isAwb) {
mainDemo->SetAwbMode(OHOS_CAMERA_AWB_MODE_INCANDESCENT);
} else {
mainDemo->SetAwbMode(OHOS_CAMERA_AWB_MODE_OFF);
}
isAwb = !isAwb;
c = PutMenuAndGetChr();
break;
case 'a': // AE功能测试
mainDemo->SetAeExpo();
c = PutMenuAndGetChr();
break;
case 'e': // Metadata测试
mainDemo->SetMetadata();
c = PutMenuAndGetChr();
break;
case 'v': // VIDEO功能测试
VideoTest(mainDemo);
c = PutMenuAndGetChr();
break;
case 'q': // 退出demo
PreviewOff(mainDemo);
mainDemo->QuitDemo();
return;
default:
CAMERA_LOGE("main test: command error please retry input command");
c = PutMenuAndGetChr();
break;
}
}
}
```
PutMenuAndGetChr()接口打印了demo程序的菜单，并调用fgets()等待从控制台输入命令，内容如下：
```c++
static int PutMenuAndGetChr(void)
{
constexpr uint32_t inputCount = 50;
int c = 0;
char strs[inputCount];
Usage(stdout);
CAMERA_LOGD("pls input command(input -q exit this app)\n");
fgets(strs, inputCount, stdin);
for (int i = 0; i < inputCount; i++) {
if (strs[i] != '-') {
c = strs[i];
break;
}
}
return c;
}
```
控制台输出菜单详情如下：
```c++
"Options:\n"
"-h | --help Print this message\n"
"-o | --offline stream offline test\n"
"-c | --capture capture one picture\n"
"-w | --set WB Set white balance Cloudy\n"
"-v | --video capture Video of 10s\n"
"-a | --Set AE Set Auto exposure\n"
"-e | --Set Metadeta Set Metadata\n"
"-f | --Set Flashlight Set flashlight ON 5s OFF\n"
"-q | --quit stop preview and quit this app\n");
```
4. 编译用例
在drivers/peripheral/camera/hal/BUILD.gn文件中的deps中添加“init:ohos_camera_demo”，示例代码如下：
```
deps = [
"buffer_manager:camera_buffer_manager",
"device_manager:camera_device_manager",
"hdi_impl:camera_host_service_1.0",
"pipeline_core:camera_pipeline_core",
"utils:camera_utils",
"init:ohos_camera_demo",
]
```
以RK3568为例：
1. 执行全量编译命令./build.sh --product-name rk3568 --ccache，生成可执行二进制文件ohos_camera_demo，路径为：out/rk3568/packages/phone/vendor/bin/。
2. 将可执行文件ohos_camera_demo导入开发板，修改权限直接运行即可。