Source code for photon_mosaic.core.loading

"""General-purpose load function for photon-mosaic objects.

Inspired by ``spikeinterface.core.loading.load``, this module provides a
single ``load`` entry-point that auto-detects the object type and format.

Loadable objects
----------------
* ``RoiAnalyzer``  — from binary folder or zarr
* ``BaseImaging``  — delegates to ``spikeinterface.core.loading.load``
* ``BaseRois``     — from binary folder (``BinaryFolderRois``)
                     or zarr (``ZarrRois``)
"""

from pathlib import Path

from spikeinterface.core.core_tools import is_path_remote

_error_msg = (
    "{file_path} is not a recognised photon-mosaic folder. "
    "It should be the result of save() or save_as() on a "
    "RoiAnalyzer, BaseRois, or BaseImaging object."
)


[docs] def load(file_or_folder, **kwargs): """Load a photon-mosaic (or spikeinterface) object from disk. Parameters ---------- file_or_folder : str or Path Path to a folder or file. **kwargs Forwarded to the specific loader. Recognised keys include: * ``load_extensions`` (bool) — for ``RoiAnalyzer`` * ``format`` (str) — for ``RoiAnalyzer`` * ``backend_options`` (dict) — for ``RoiAnalyzer`` / zarr * ``storage_options`` (dict) — for zarr ROIs * ``base_folder`` — for SI extractor json/pkl files Returns ------- object A ``RoiAnalyzer``, ``BinaryFolderRois``, ``ZarrRois``, or a spikeinterface ``BaseExtractor`` (Recording/Sorting). """ # --- remote zarr store (e.g. s3://, gcs://) -------------------------------- if is_path_remote(str(file_or_folder)): return _load_zarr(file_or_folder, **kwargs) path = Path(file_or_folder) # --- zarr store --------------------------------------------------------- if str(path).endswith(".zarr") or (path.is_dir() and (path / ".zgroup").exists()): return _load_zarr(path, **kwargs) # --- local folder ------------------------------------------------------- if path.is_dir(): return _load_folder(path, **kwargs) # --- single file (json / pkl) — fall back to spikeinterface ------------- if path.is_file(): from spikeinterface.core.loading import load as si_load return si_load(file_or_folder, **kwargs) raise ValueError(_error_msg.format(file_path=file_or_folder))
# --------------------------------------------------------------------------- # Internal helpers # --------------------------------------------------------------------------- def _load_folder(folder, **kwargs): """Dispatch a local folder to the right loader.""" folder = Path(folder) # RoiAnalyzer — identified by photon_mosaic_info.json if (folder / "photon_mosaic_info.json").is_file(): from .roianalyzer import load_roi_analyzer return load_roi_analyzer(folder, **kwargs) # BinaryFolderRois — identified by binary.json + roi_image_masks.npy # sparse masks are saved as .npz, dense ones as .npy has_masks = (folder / "roi_image_masks.npy").is_file() or (folder / "roi_image_masks.npz").is_file() if has_masks and (folder / "metadata.json").is_file(): from .binaryrois import BinaryFolderRois return BinaryFolderRois(folder_path=folder) # BinaryFolderImaging — identified by binary.json with BinaryImaging class if (folder / "binary.json").is_file(): from .binaryimaging import BinaryFolderImaging return BinaryFolderImaging(folder_path=folder) # Use spikeinterface load to load from dict/json/pkl from spikeinterface.core.loading import load as si_load return si_load(str(folder), **kwargs) def _load_zarr(path, **kwargs): """Dispatch a zarr store to the right loader.""" from spikeinterface.core.zarrextractors import super_zarr_open storage_options = kwargs.get("storage_options", kwargs.get("backend_options", {}).get("storage_options", {})) zarr_root = super_zarr_open(str(path), mode="r", storage_options=storage_options or {}) # RoiAnalyzer — identified by photon_mosaic_info info = zarr_root.attrs.get("photon_mosaic_info") if info is not None and info.get("object") == "RoiAnalyzer": from .roianalyzer import load_roi_analyzer return load_roi_analyzer(path, **kwargs) # Standalone zarr rois — has a "rois" group with roi_image_masks if "rois" in zarr_root and "roi_image_masks" in zarr_root["rois"]: from .zarrrois import ZarrRois return ZarrRois(path, zarr_group_name="rois", storage_options=storage_options) # Fall back to spikeinterface from spikeinterface.core.loading import load as si_load return si_load(str(path), **kwargs)