Parallel spatial extraction processor for image-modality containers. More...
#include <SpatialRegionProcessor.hpp>
Inheritance diagram for MayaFlux::Kakshya::SpatialRegionProcessor:
Collaboration diagram for MayaFlux::Kakshya::SpatialRegionProcessor:
Public Member Functions | |
| SpatialRegionProcessor ()=default | |
| ~SpatialRegionProcessor () override=default | |
| void | on_attach (const std::shared_ptr< SignalSourceContainer > &container) override |
| Attach to a spatial container. | |
| void | on_detach (const std::shared_ptr< SignalSourceContainer > &container) override |
| Detach; delegates to RegionProcessorBase::on_detach. | |
| void | process (const std::shared_ptr< SignalSourceContainer > &container) override |
| Extract all active regions from processed_data[0] in parallel. | |
| void | refresh () |
| Re-sync m_organized_regions from the container's current group map. | |
| Public Member Functions inherited from MayaFlux::Kakshya::RegionProcessorBase | |
| virtual | ~RegionProcessorBase ()=default |
| bool | is_processing () const override |
| Query if the processor is currently performing processing. | |
| void | set_cache_limit (size_t max_cached_elements) |
| Set the maximum cache size for regions (in elements). | |
| void | set_auto_caching (bool enabled) |
| Enable or disable automatic region caching. | |
| const std::vector< uint64_t > & | get_current_position () const |
| Get the current processing position (N-dimensional coordinates). | |
| void | set_current_position (const std::vector< uint64_t > &position) |
| Set the current processing position (N-dimensional coordinates). | |
| Public Member Functions inherited from MayaFlux::Kakshya::DataProcessor | |
| virtual | ~DataProcessor ()=default |
| Virtual destructor for proper cleanup. | |
Protected Member Functions | |
| void | organize_container_data (const std::shared_ptr< SignalSourceContainer > &container) override |
| Build m_organized_regions from get_all_region_groups(). | |
| Protected Member Functions inherited from MayaFlux::Kakshya::RegionProcessorBase | |
| virtual void | cache_region_if_needed (const RegionSegment &segment, const std::shared_ptr< SignalSourceContainer > &container) |
| Cache a region's data if beneficial and not already cached. | |
| virtual bool | advance_position (std::vector< uint64_t > &position, uint64_t steps=1, const OrganizedRegion *region=nullptr) |
| Advance position according to memory layout and looping. | |
| virtual void | ensure_output_dimensioning (std::vector< DataVariant > &output_data, const std::vector< uint64_t > &required_shape) |
| Ensure output data is properly dimensioned for region extraction. | |
Additional Inherited Members | |
| Protected Attributes inherited from MayaFlux::Kakshya::RegionProcessorBase | |
| std::weak_ptr< SignalSourceContainer > | m_container_weak |
| std::atomic< bool > | m_is_processing { false } |
| std::vector< OrganizedRegion > | m_organized_regions |
| size_t | m_current_region_index = 0 |
| std::vector< uint64_t > | m_current_position |
| std::unique_ptr< RegionCacheManager > | m_cache_manager |
| size_t | m_max_cache_size { static_cast<size_t>(1024 * 1024) } |
| bool | m_auto_caching = true |
| ContainerDataStructure | m_structure |
Detailed Description
Parallel spatial extraction processor for image-modality containers.
Operates on any SignalSourceContainer whose structure carries at least one spatial dimension (SPATIAL_Y, SPATIAL_X, or SPATIAL_Z). Unlike RegionOrganizationProcessor — which advances a sequential temporal cursor through audio regions — this processor extracts all active regions on every process() call, reflecting the parallel readback semantics of spatial data.
Region source: Regions are read from the container's RegionGroup map (get_all_region_groups()). Each Region inside each group becomes one OrganizedRegion in m_organized_regions, inheriting both group-level and region-level attributes. organize_container_data() re-syncs this list from the live group map, so callers can mutate groups between frames by calling refresh().
Processing contract (process()):
- Expects processed_data[0] to hold the full-surface readback as a std::vector<uint8_t> placed there by the default processor (e.g. WindowAccessProcessor) before the chain runs.
- Replaces processed_data with one DataVariant per active region, in group-iteration order, then region-insertion order within each group.
- Each DataVariant carries the attributes from its OrganizedRegion so downstream consumers can identify the source ("group_name", "region_index").
- If processed_data[0] is absent or empty, state is set to IDLE and the method returns without modifying processed_data.
Caching: Auto-caching is disabled by default. Spatial data from a live surface changes every frame, so caching individual regions yields no benefit under normal operation. It can be re-enabled via set_auto_caching(true) for static or infrequently-updated surfaces (e.g. a paused framebuffer).
Container neutrality: No WindowContainer-specific code. Any container that satisfies the NDDimensionalContainer interface and populates processed_data[0] with a flat spatial buffer is compatible.
Definition at line 48 of file SpatialRegionProcessor.hpp.
The documentation for this class was generated from the following files:
