XrResult xrEnumerateSpatialAnnotationReferenceSourcesANDROID(
    XrInstance                                  instance,
    XrSystemId                                  systemId,
    XrSpatialCapabilityEXT                      capability,
    uint32_t                                    sourceCapacityInput,
    uint32_t*                                   sourceCountOutput,
    XrSpatialAnnotationReferenceSourceANDROID*  sources);

typedef enum XrSpatialAnnotationReferenceSourceANDROID {
    XR_SPATIAL_ANNOTATION_REFERENCE_SOURCE_CAMERA_ANDROID = 0,
    XR_SPATIAL_ANNOTATION_REFERENCE_SOURCE_MEDIA_PROJECTION_ANDROID = 1,
    XR_SPATIAL_ANNOTATION_REFERENCE_SOURCE_MAX_ENUM_ANDROID = 0x7FFFFFFF
} XrSpatialAnnotationReferenceSourceANDROID;

typedef struct XrSystemSpatialAnnotationPropertiesANDROID {
    XrStructureType    type;
    void*              next;
    uint32_t           referenceCacheTimespan;
    uint32_t           maxReferencePixelWidth;
    uint32_t           maxReferencePixelHeight;
    uint32_t           maxReferenceCount;
    uint32_t           maxAnnotationCount;
} XrSystemSpatialAnnotationPropertiesANDROID;

XR_DEFINE_HANDLE(XrSpatialReferenceCacheANDROID)

XrResult xrCreateSpatialReferenceCacheAsyncANDROID(
    XrSession                                   session,
    const XrSpatialReferenceCacheCreateInfoANDROID* createInfo,
    XrFutureEXT*                                future);

typedef struct XrSpatialReferenceCacheCreateInfoANDROID {
    XrStructureType                              type;
    const void*                                  next;
    XrSpatialCapabilityEXT                       capability;
    XrSpatialAnnotationReferenceSourceANDROID    source;
} XrSpatialReferenceCacheCreateInfoANDROID;

XrResult xrCreateSpatialReferenceCacheCompleteANDROID(
    XrSession                                   session,
    XrFutureEXT                                 future,
    XrCreateSpatialReferenceCacheCompletionANDROID* completion);

typedef struct XrCreateSpatialReferenceCacheCompletionANDROID {
    XrStructureType                   type;
    void*                             next;
    XrResult                          futureResult;
    XrSpatialReferenceCacheANDROID    referenceCache;
} XrCreateSpatialReferenceCacheCompletionANDROID;

XrResult xrDestroySpatialReferenceCacheANDROID(
    XrSpatialReferenceCacheANDROID              cacheHandle);

typedef struct XrSpatialCapabilityConfigurationAnnotationTrackingANDROID {
    XrStructureType                     type;
    const void*                         next;
    XrSpatialCapabilityEXT              capability;
    uint32_t                            enabledComponentCount;
    const XrSpatialComponentTypeEXT*    enabledComponents;
    XrSpatialReferenceCacheANDROID      cache;
    XrTime                              time;
    XrSpatialReferenceImageEXT          reference;
} XrSpatialCapabilityConfigurationAnnotationTrackingANDROID;

typedef struct XrEventDataSpatialAnnotationTrackingANDROID {
    XrStructureType        type;
    const void*            next;
    XrSpatialContextEXT    spatialContext;
    uint32_t               annotationIndex;
    XrResult               initializationResult;
} XrEventDataSpatialAnnotationTrackingANDROID;

typedef struct XrSpatialAnnotationQuadReferenceANDROID {
    XrStructureType                          type;
    const void*                              next;
    uint32_t                                 quadCount;
    const XrSpatialAnnotationQuadANDROID*    quads;
} XrSpatialAnnotationQuadReferenceANDROID;

typedef struct XrSpatialAnnotationQuadANDROID {
    XrSpatialAnnotationQuadAlignmentANDROID    alignment;
    XrVector2f                                 upperLeft;
    XrVector2f                                 upperRight;
    XrVector2f                                 lowerRight;
    XrVector2f                                 lowerLeft;
} XrSpatialAnnotationQuadANDROID;

typedef enum XrSpatialAnnotationQuadAlignmentANDROID {
    XR_SPATIAL_ANNOTATION_QUAD_ALIGNMENT_SCREEN_ANDROID = 0,
    XR_SPATIAL_ANNOTATION_QUAD_ALIGNMENT_OBJECT_ANDROID = 1,
    XR_SPATIAL_ANNOTATION_QUAD_ALIGNMENT_SURFACE_ANDROID = 2,
    XR_SPATIAL_ANNOTATION_QUAD_ALIGNMENT_MAX_ENUM_ANDROID = 0x7FFFFFFF
} XrSpatialAnnotationQuadAlignmentANDROID;

typedef struct XrSpatialAnnotationQuadDataANDROID {
    uint32_t                          annotationIndex;
    XrSpaceLocationData               origin;
    XrSpatialAnnotationQuadANDROID    quad;
} XrSpatialAnnotationQuadDataANDROID;

typedef struct XrSpatialComponentAnnotationQuadListANDROID {
    XrStructureType                        type;
    void*                                  next;
    uint32_t                               quadCount;
    XrSpatialAnnotationQuadDataANDROID*    quads;
} XrSpatialComponentAnnotationQuadListANDROID;

XrInstance instance;  // previously initialized
XrSystemId systemId;  // previously initialized
XrSession session;    // previously initialized
XrSpace localSpace;   // previously initialized, e.g. from
                      // XR_REFERENCE_SPACE_TYPE_LOCAL

PFN_xrEnumerateSpatialCapabilitiesEXT xrEnumerateSpatialCapabilitiesEXT;
PFN_xrEnumerateSpatialAnnotationReferenceSourcesANDROID xrEnumerateSpatialAnnotationReferenceSourcesANDROID;
PFN_xrEnumerateSpatialReferenceImageFormatsEXT xrEnumerateSpatialReferenceImageFormatsEXT;
PFN_xrEnumerateSpatialCapabilityComponentTypesEXT xrEnumerateSpatialCapabilityComponentTypesEXT;

// Check spatial annotation tracking capability
uint32_t capabilityCount;
CHK_XR(xrEnumerateSpatialCapabilitiesEXT(instance, systemId, 0,
                                         &capabilityCount, nullptr));
std::vector<XrSpatialCapabilityEXT> capabilities(capabilityCount);
CHK_XR(xrEnumerateSpatialCapabilitiesEXT(instance, systemId, capabilityCount,
                                         &capabilityCount,
                                         capabilities.data()));
if (std::find(capabilities.begin(), capabilities.end(),
              XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID) ==
    capabilities.end()) {
  // System does not support spatial annotation tracking.
  return;
}

// Inspect system properties for annotation tracking variables
XrSystemSpatialAnnotationPropertiesANDROID annotationSystemProperties{
    XR_TYPE_SYSTEM_SPATIAL_ANNOTATION_PROPERTIES_ANDROID};
XrSystemProperties systemProperties{.type = XR_TYPE_SYSTEM_PROPERTIES,
                                    .next = &annotationSystemProperties,
                                    .systemId = systemId};
CHK_XR(xrGetSystemProperties(instance, systemId, &systemProperties));
if (annotationSystemProperties.maxAnnotationCount == 0) {
  // System does not support any active annotations.
  return;
}

// Enumerate supported reference sources
uint32_t sourceCountOutput = 0;
XrSpatialAnnotationReferenceSourceANDROID desiredSource =
    XR_SPATIAL_ANNOTATION_REFERENCE_SOURCE_CAMERA_ANDROID;
CHK_XR(xrEnumerateSpatialAnnotationReferenceSourcesANDROID(
    instance, systemId, XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID,
    0, &sourceCountOutput, nullptr));
std::vector<XrSpatialAnnotationReferenceSourceANDROID> sources(sourceCountOutput);
CHK_XR(xrEnumerateSpatialAnnotationReferenceSourcesANDROID(
    instance, systemId, XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID,
    sourceCountOutput, &sourceCountOutput, sources.data()));
if (std::find(sources.begin(), sources.end(), desiredSource) == sources.end()) {
  // The desired source is not supported.
  return;
}

// Enumerate supported reference image formats, provided by XR_EXT_spatial_image_tracking.
uint32_t formatCountOutput = 0;
XrSpatialReferenceImageFormatEXT desiredFormat =
    XR_SPATIAL_REFERENCE_IMAGE_FORMAT_YUV_420_888_EXT; // Or XR_SPATIAL_REFERENCE_IMAGE_FORMAT_RGBA_8888_EXT for screenshot.
CHK_XR(xrEnumerateSpatialReferenceImageFormatsEXT(
    instance, systemId, XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID, 0,
    &formatCountOutput, nullptr));
std::vector<XrSpatialReferenceImageFormatEXT> formats(formatCountOutput);
CHK_XR(xrEnumerateSpatialReferenceImageFormatsEXT(
    instance, systemId, XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID,
    formats.size(), &formatCountOutput, formats.data()));
if (std::find(formats.begin(), formats.end(), desiredFormat) == formats.end()) {
  // The desired format is not supported
  return;
}

// Enumerate supported components for annotation tracking capability.
XrSpatialCapabilityComponentTypesEXT capabilityComponentTypes{
    XR_TYPE_SPATIAL_CAPABILITY_COMPONENT_TYPES_EXT};
CHK_XR(xrEnumerateSpatialCapabilityComponentTypesEXT(
    instance, systemId, XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID,
    &capabilityComponentTypes));
std::vector<XrSpatialComponentTypeEXT>
    annotationTypes(capabilityComponentTypes.componentTypeCountOutput);
capabilityComponentTypes.componentTypeCapacityInput = annotationTypes.size();
capabilityComponentTypes.componentTypes = annotationTypes.data();
CHK_XR(xrEnumerateSpatialCapabilityComponentTypesEXT(
    instance, systemId, XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID,
    &capabilityComponentTypes));

// Check supported annotation components.
XrSpatialComponentTypeEXT desiredComponent =
    XR_SPATIAL_COMPONENT_TYPE_ANNOTATION_QUAD_ANDROID;
const auto supportedComponent = [&annotationTypes](
                                    XrSpatialComponentTypeEXT component) {
  return std::find(annotationTypes.begin(), annotationTypes.end(), component) !=
         annotationTypes.end();
};
if (!supportedComponent(desiredComponent)) {
  // The desired annotation is not supported.
  return;
}

XrInstance instance;  // previously initialized
XrSystemId systemId;  // previously initialized
XrSession session;    // previously initialized
XrSpace localSpace;   // previously initialized, e.g. from
                      // XR_REFERENCE_SPACE_TYPE_LOCAL

PFN_xrCreateSpatialReferenceCacheAsyncANDROID xrCreateSpatialReferenceCacheAsyncANDROID;
PFN_xrCreateSpatialReferenceCacheCompleteANDROID xrCreateSpatialReferenceCacheCompleteANDROID;
PFN_xrDestroySpatialReferenceCacheANDROID xrDestroySpatialReferenceCacheANDROID;
PFN_xrCreateSpatialContextAsyncEXT xrCreateSpatialContextAsyncEXT;
PFN_xrCreateSpatialContextCompleteEXT xrCreateSpatialContextCompleteEXT;
void (*waitUntilReady)(XrFutureEXT);

XrSpatialReferenceImageFormatEXT desiredFormat;
XrSpatialAnnotationReferenceSourceANDROID desiredSource;
XrSpatialComponentTypeEXT desiredComponent;

// Create and start reference cache to prepare for capturing reference images.
XrSpatialReferenceCacheANDROID referenceCache;
XrSpatialReferenceCacheCreateInfoANDROID cacheCreateInfo{
  .type = XR_TYPE_SPATIAL_REFERENCE_CACHE_CREATE_INFO_ANDROID,
  .next = nullptr,
  .capability = XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID,
  .source = desiredSource,
};
XrFutureEXT cacheFuture = XR_NULL_FUTURE_EXT;
CHK_XR(xrCreateSpatialReferenceCacheAsyncANDROID(session, &cacheCreateInfo, &cacheFuture));

waitUntilReady(cacheFuture);

XrCreateSpatialReferenceCacheCompletionANDROID cacheCompletion{
    XR_TYPE_CREATE_SPATIAL_REFERENCE_CACHE_COMPLETION_ANDROID};
CHK_XR(xrCreateSpatialReferenceCacheCompleteANDROID(session, cacheFuture,
                                                     &cacheCompletion));
if (cacheCompletion.futureResult != XR_SUCCESS) {
  // Error, reference cache creation failed
  return;
}

referenceCache = cacheCompletion.referenceCache;

// Create the reference image with quad annotations

//  YUV 4:2:0 Planar:
// ----------------------
// |     Y      |Cb |Cr |
// ----------------------
XrSpatialReferenceImagePlaneEXT cameraImagePlanes[3];
// for each plane (0=Y, 1=U(cb), 2=V(Cr)) return by Image#getPlanes().
for (int i = 0; i < 3; ++i) {
  cameraImagePlanes[i].buffer; // read from Plane#getBuffer().
  cameraImagePlanes[i].bufferSize; // set each buffer size. Y = width * height (bytes); U = Y / 4 (bytes); V = Y / 4 (bytes).
  cameraImagePlanes[i].rowStride; // read from Plane#getRowStride().
  cameraImagePlanes[i].pixelStride; // read from Plane#getPixelStride().
}
XrSpatialReferenceImageEXT referenceImage{
      XR_TYPE_SPATIAL_REFERENCE_IMAGE_EXT};
referenceImage.width = 640;
referenceImage.height = 480;
referenceImage.format = desiredFormat;
referenceImage.planeCount = 3;
referenceImage.planes = cameraImagePlanes;

XrSpatialAnnotationQuadANDROID
    quad; // quad coordinates within the reference image.
quad.alignment = XR_SPATIAL_ANNOTATION_QUAD_ALIGNMENT_OBJECT_ANDROID;
// Set unnormalized pixel coordinates of 4 corners.
quad.upperLeft = {0.0, 0.0};
quad.upperRight = {0.0, 100.0};
quad.lowerRight = {100.0, 100.0};
quad.lowerLeft = {0.0, 100.0};

XrSpatialAnnotationQuadReferenceANDROID quadReference{
    XR_TYPE_SPATIAL_ANNOTATION_QUAD_REFERENCE_ANDROID};
quadReference.quads = &quad;
quadReference.quadCount = 1;
referenceImage.next = &quadReference;

// Create the spatial context with annotation configuration.
XrSpatialContextEXT spatialContext{};
{
  const std::array<XrSpatialComponentTypeEXT, 1> enabledComponents = {
      desiredComponent,
  };

  // Configure annotation tracking
  XrSpatialCapabilityConfigurationAnnotationTrackingANDROID annotationConfig{
      XR_TYPE_SPATIAL_CAPABILITY_CONFIGURATION_ANNOTATION_TRACKING_ANDROID};
  annotationConfig.capability =
      XR_SPATIAL_CAPABILITY_ANNOTATION_TRACKING_ANDROID;
  annotationConfig.enabledComponentCount = enabledComponents.size();
  annotationConfig.enabledComponents = enabledComponents.data();
  annotationConfig.cache = referenceCache;
  annotationConfig.time; // The time when this reference image is captured.
  annotationConfig.reference = referenceImage;

  std::vector<XrSpatialCapabilityConfigurationBaseHeaderEXT *>
      capabilityConfigs;
  capabilityConfigs.push_back(
      reinterpret_cast<XrSpatialCapabilityConfigurationBaseHeaderEXT *>(
          &annotationConfig));

  XrSpatialContextCreateInfoEXT spatialContextCreateInfo{
      XR_TYPE_SPATIAL_CONTEXT_CREATE_INFO_EXT};
  spatialContextCreateInfo.capabilityConfigCount = capabilityConfigs.size();
  spatialContextCreateInfo.capabilityConfigs = capabilityConfigs.data();

  XrFutureEXT createContextFuture = XR_NULL_FUTURE_EXT;
  CHK_XR(xrCreateSpatialContextAsyncEXT(session, &spatialContextCreateInfo,
                                        &createContextFuture));

  waitUntilReady(createContextFuture);

  XrCreateSpatialContextCompletionEXT completion{
      XR_TYPE_CREATE_SPATIAL_CONTEXT_COMPLETION_EXT};
  CHK_XR(xrCreateSpatialContextCompleteEXT(session, createContextFuture,
                                           &completion));
  if (completion.futureResult != XR_SUCCESS) {
    // Error, spatial context creation failed
    return;
  }

  spatialContext = completion.spatialContext;
  CHK_XR(xrDestroySpatialReferenceCacheANDROID(referenceCache));
}

XrInstance instance;  // previously initialized
XrSession session;    // previously initialized
XrSpace localSpace;   // previously initialized, e.g. from
                      // XR_REFERENCE_SPACE_TYPE_LOCAL
XrSpatialContextEXT spatialContext; // previously created

PFN_xrCreateSpatialDiscoverySnapshotAsyncEXT xrCreateSpatialDiscoverySnapshotAsyncEXT;
PFN_xrCreateSpatialDiscoverySnapshotCompleteEXT xrCreateSpatialDiscoverySnapshotCompleteEXT;
PFN_xrQuerySpatialComponentDataEXT xrQuerySpatialComponentDataEXT;
PFN_xrDestroySpatialSnapshotEXT xrDestroySpatialSnapshotEXT;
PFN_xrPollEvent xrPollEvent;
PFN_xrDestroySpatialContextEXT xrDestroySpatialContextEXT;
void (*waitUntilReady)(XrFutureEXT);

XrSpatialComponentTypeEXT desiredComponent; // e.g. XR_SPATIAL_COMPONENT_TYPE_ANNOTATION_QUAD_ANDROID

auto discoverAnnotations = [&](XrSpatialContextEXT spatialContext, XrTime time,
                               XrSpace baseSpace) {
  std::array<XrSpatialComponentTypeEXT, 1> snapshotComponents{desiredComponent};
  XrSpatialDiscoverySnapshotCreateInfoEXT snapshotCreateInfo{
      XR_TYPE_SPATIAL_DISCOVERY_SNAPSHOT_CREATE_INFO_EXT};
  snapshotCreateInfo.componentTypeCount = snapshotComponents.size();
  snapshotCreateInfo.componentTypes = snapshotComponents.data();
  XrFutureEXT discoveryFuture = XR_NULL_FUTURE_EXT;
  CHK_XR(xrCreateSpatialDiscoverySnapshotAsyncEXT(
      spatialContext, &snapshotCreateInfo, &discoveryFuture));

  waitUntilReady(discoveryFuture);

  XrCreateSpatialDiscoverySnapshotCompletionInfoEXT completionInfo{
      XR_TYPE_CREATE_SPATIAL_DISCOVERY_SNAPSHOT_COMPLETION_INFO_EXT};
  completionInfo.baseSpace = baseSpace;
  completionInfo.time = time;
  completionInfo.future = discoveryFuture;

  XrCreateSpatialDiscoverySnapshotCompletionEXT completion{
      XR_TYPE_CREATE_SPATIAL_DISCOVERY_SNAPSHOT_COMPLETION_EXT};
  CHK_XR(xrCreateSpatialDiscoverySnapshotCompleteEXT(
      spatialContext, &completionInfo, &completion));
  if (completion.futureResult == XR_SUCCESS) {
    // Query for desired annotation data, e.g. quad.
    XrSpatialComponentTypeEXT componentsToQuery[] = {
        XR_SPATIAL_COMPONENT_TYPE_ANNOTATION_QUAD_ANDROID};
    XrSpatialComponentDataQueryConditionEXT queryCond{
        XR_TYPE_SPATIAL_COMPONENT_DATA_QUERY_CONDITION_EXT};
    queryCond.componentTypeCount = 1;
    queryCond.componentTypes = componentsToQuery;
    XrSpatialComponentDataQueryResultEXT queryResult{
        XR_TYPE_SPATIAL_COMPONENT_DATA_QUERY_RESULT_EXT};
    CHK_XR(xrQuerySpatialComponentDataEXT(completion.snapshot, &queryCond,
                                          &queryResult));

    std::vector<XrSpatialEntityIdEXT> entityIds(
        queryResult.entityIdCountOutput);
    std::vector<XrSpatialEntityTrackingStateEXT> entityStates(
        queryResult.entityStateCountOutput);
    queryResult.entityIdCapacityInput = entityIds.size();
    queryResult.entityIds = entityIds.data();
    queryResult.entityStateCapacityInput = entityStates.size();
    queryResult.entityStates = entityStates.data();

    std::vector<XrSpatialAnnotationQuadDataANDROID> quads(
        queryResult.entityIdCountOutput);
    XrSpatialComponentAnnotationQuadListANDROID quadList{
        XR_TYPE_SPATIAL_COMPONENT_ANNOTATION_QUAD_LIST_ANDROID};
    quadList.quadCount = quads.size();
    quadList.quads = quads.data();
    queryResult.next = &quadList;

    CHK_XR(xrQuerySpatialComponentDataEXT(completion.snapshot, &queryCond,
                                          &queryResult));

    for (int32_t i = 0; i < queryResult.entityIdCountOutput; ++i) {
      if (entityStates[i] == XR_SPATIAL_ENTITY_TRACKING_STATE_TRACKING_EXT) {
        const XrSpatialAnnotationQuadDataANDROID &currentQuad = quads[i];
        // Rendering quad in the view.
      }
    }

    CHK_XR(xrDestroySpatialSnapshotEXT(completion.snapshot));
  }
};

while (1) {
  // For every frame in frame loop

  XrSpace space;           // Application's play space.
  XrFrameState frameState; // Previously returned from xrWaitFrame
  const XrTime time = frameState.predictedDisplayTime;

  // Poll for the XR_TYPE_EVENT_DATA_SPATIAL_DISCOVERY_RECOMMENDED_EXT event
  XrEventDataBuffer event = {XR_TYPE_EVENT_DATA_BUFFER};
  XrResult result = xrPollEvent(instance, &event);
  if (result == XR_SUCCESS) {
    switch (event.type) {
    case XR_TYPE_EVENT_DATA_SPATIAL_ANNOTATION_TRACKING_ANDROID: {
      const XrEventDataSpatialAnnotationTrackingANDROID &eventdata =
          *reinterpret_cast<XrEventDataSpatialAnnotationTrackingANDROID *>(
              &event);
      if (eventdata.initializationResult != XR_SUCCESS) {
        // handle initialization failure.
        // e.g. CHK_XR(xrDestroySpatialContextEXT(spatialContext));
      }
      break;
    }
    case XR_TYPE_EVENT_DATA_SPATIAL_DISCOVERY_RECOMMENDED_EXT: {
      const XrEventDataSpatialDiscoveryRecommendedEXT &eventdata =
          *reinterpret_cast<XrEventDataSpatialDiscoveryRecommendedEXT *>(
              &event);
      // Discover spatial entities for the context that we received the
      // "discovery recommended" event for.
      //discoverSpatialEntities(eventdata.spatialContext, time, space);
      break;
    }
    }
  }

  // Finish frame loop
}

CHK_XR(xrDestroySpatialContextEXT(spatialContext));