ocf-data-sampler 0.2.38__tar.gz → 0.3.1__tar.gz

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocf-data-sampler
-Version: 0.2.38
+Version: 0.3.1
 Author: James Fulton, Peter Dudfield
 Author-email: Open Climate Fix team <info@openclimatefix.org>
 License: MIT License
@@ -44,6 +44,7 @@ Requires-Dist: pyproj
 Requires-Dist: pyaml_env
 Requires-Dist: pyresample
 Requires-Dist: h5netcdf
+Requires-Dist: xarray-tensorstore==0.1.5
 
 # ocf-data-sampler
  

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/config/model.py

@@ -90,11 +90,10 @@ class DropoutMixin(Base):
         "negative or zero.",
     )
 
-    dropout_fraction: float = Field(
+    dropout_fraction: float|list[float] = Field(
         default=0,
-        description="Chance of dropout being applied to each sample",
-        ge=0,
-        le=1,
+        description="Either a float(Chance of dropout being applied to each sample) or a list of "
+        "floats (probability that dropout of the corresponding timedelta is applied)",
     )
 
     @field_validator("dropout_timedeltas_minutes")
@@ -105,6 +104,36 @@ class DropoutMixin(Base):
                 raise ValueError("Dropout timedeltas must be negative")
         return v
 
+
+    @field_validator("dropout_fraction")
+    def dropout_fractions(cls, dropout_frac: float|list[float]) -> float|list[float]:
+        """Validate 'dropout_frac'."""
+        from math import isclose
+        if isinstance(dropout_frac, float):
+            if not (dropout_frac <= 1):
+                raise ValueError("Input should be less than or equal to 1")
+            elif not (dropout_frac >= 0):
+                raise ValueError("Input should be greater than or equal to 0")
+
+        elif isinstance(dropout_frac, list):
+            if not dropout_frac:
+                raise ValueError("List cannot be empty")
+
+            if not all(isinstance(i, float) for i in dropout_frac):
+                raise ValueError("All elements in the list must be floats")
+
+            if not all(0 <= i <= 1 for i in dropout_frac):
+                raise ValueError("Each float in the list must be between 0 and 1")
+
+            if not isclose(sum(dropout_frac), 1.0, rel_tol=1e-9):
+                raise ValueError("Sum of all floats in the list must be 1.0")
+
+
+        else:
+            raise TypeError("Must be either a float or a list of floats")
+        return dropout_frac
+
+
     @model_validator(mode="after")
     def dropout_instructions_consistent(self) -> "DropoutMixin":
         """Validator for dropout instructions."""

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/load_dataset.py

@@ -25,7 +25,7 @@ def get_dataset_dict(
             zarr_path=input_config.gsp.zarr_path,
             boundaries_version=input_config.gsp.boundaries_version,
             public=input_config.gsp.public,
-        ).compute()
+        )
 
         if gsp_ids is None:
             # Remove national (gsp_id=0)

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/nwp/providers/cloudcasting.py

@@ -28,7 +28,7 @@ def open_cloudcasting(zarr_path: str | list[str]) -> xr.DataArray:
             [3] https://github.com/openclimatefix/sat_pred
     """
     # Open the data
-    ds = open_zarr_paths(zarr_path)
+    ds = open_zarr_paths(zarr_path, backend="tensorstore")
 
     # Rename
     ds = ds.rename(

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/nwp/providers/ecmwf.py

@@ -19,7 +19,7 @@ def open_ifs(zarr_path: str | list[str]) -> xr.DataArray:
     Returns:
         Xarray DataArray of the NWP data
     """
-    ds = open_zarr_paths(zarr_path)
+    ds = open_zarr_paths(zarr_path, backend="tensorstore")
 
     # LEGACY SUPPORT - rename variable to channel if it exists
     ds = ds.rename({"init_time": "init_time_utc", "variable": "channel"})

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/nwp/providers/gfs.py

@@ -23,7 +23,12 @@ def open_gfs(zarr_path: str | list[str], public: bool = False) -> xr.DataArray:
     _log.info("Loading NWP GFS data")
 
     # Open data
-    gfs: xr.Dataset = open_zarr_paths(zarr_path, time_dim="init_time_utc", public=public)
+    gfs: xr.Dataset = open_zarr_paths(
+        zarr_path,
+        time_dim="init_time_utc",
+        public=public,
+        backend="dask",
+    )
     nwp: xr.DataArray = gfs.to_array(dim="channel")
 
     del gfs

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/nwp/providers/icon.py

@@ -20,7 +20,7 @@ def open_icon_eu(zarr_path: str | list[str]) -> xr.DataArray:
         Xarray DataArray of the NWP data
     """
     # Open and check initially
-    ds = open_zarr_paths(zarr_path, time_dim="init_time_utc")
+    ds = open_zarr_paths(zarr_path, time_dim="init_time_utc", backend="dask")
 
     if "icon_eu_data" in ds.data_vars:
         nwp = ds["icon_eu_data"]

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/nwp/providers/ukv.py

@@ -19,7 +19,7 @@ def open_ukv(zarr_path: str | list[str]) -> xr.DataArray:
     Returns:
         Xarray DataArray of the NWP data
     """
-    ds = open_zarr_paths(zarr_path)
+    ds = open_zarr_paths(zarr_path, backend="tensorstore")
 
     ds = ds.rename(
         {

ocf_data_sampler-0.3.1/ocf_data_sampler/load/nwp/providers/utils.py

@@ -0,0 +1,83 @@
+"""Utility functions for the NWP data processing."""
+
+from glob import glob
+
+import xarray as xr
+from xarray_tensorstore import open_zarr
+
+from ocf_data_sampler.load.open_tensorstore_zarrs import open_zarrs
+
+
+def open_zarr_paths(
+    zarr_path: str | list[str],
+    time_dim: str = "init_time",
+    public: bool = False,
+    backend: str = "dask",
+) -> xr.Dataset:
+    """Opens the NWP data.
+
+    Args:
+        zarr_path: Path to the zarr(s) to open
+        time_dim: Name of the time dimension
+        public: Whether the data is public or private. Only available for the dask backend.
+        backend: The xarray backend to use.
+
+    Returns:
+        The opened Xarray Dataset
+    """
+    if backend not in ["dask", "tensorstore"]:
+        raise ValueError(
+            f"Unsupported backend: {backend}. Supported backends are 'dask' and 'tensorstore'.",
+        )
+
+    if public and backend == "tensorstore":
+        raise ValueError("Public data is only supported with the 'dask' backend.")
+
+    if backend == "tensorstore":
+        ds = _tensostore_open_zarr_paths(zarr_path, time_dim)
+
+    elif backend == "dask":
+        ds = _dask_open_zarr_paths(zarr_path, time_dim, public)
+
+    return ds
+
+
+def _dask_open_zarr_paths(zarr_path: str | list[str], time_dim: str, public: bool) -> xr.Dataset:
+    general_kwargs = {
+        "engine": "zarr",
+        "chunks": "auto",
+        "decode_timedelta": True,
+    }
+
+    if public:
+        # note this only works for s3 zarr paths at the moment
+        general_kwargs["storage_options"] = {"anon": True}
+
+    if isinstance(zarr_path, list | tuple) or "*" in str(zarr_path):  # Multi-file dataset
+        ds = xr.open_mfdataset(
+            zarr_path,
+            concat_dim=time_dim,
+            combine="nested",
+            **general_kwargs,
+        ).sortby(time_dim)
+    else:
+        ds = xr.open_dataset(
+            zarr_path,
+            consolidated=True,
+            mode="r",
+            **general_kwargs,
+        )
+    return ds
+
+
+def _tensostore_open_zarr_paths(zarr_path: str | list[str], time_dim: str) -> xr.Dataset:
+
+    if "*" in str(zarr_path):
+        zarr_path = sorted(glob(zarr_path))
+
+    if isinstance(zarr_path, list | tuple):
+        ds = open_zarrs(zarr_path, concat_dim=time_dim).sortby(time_dim)
+    else:
+        ds = open_zarr(zarr_path)
+    return ds
+

ocf_data_sampler-0.3.1/ocf_data_sampler/load/open_tensorstore_zarrs.py

@@ -0,0 +1,92 @@
+"""Open multiple zarrs with TensorStore.
+
+This extendds the functionality of xarray_tensorstore to open multiple zarr stores
+"""
+
+import os
+
+import tensorstore as ts
+import xarray as xr
+from xarray_tensorstore import (
+    _raise_if_mask_and_scale_used_for_data_vars,
+    _TensorStoreAdapter,
+    _zarr_spec_from_path,
+)
+
+
+def tensorstore_open_multi_zarrs(
+    paths: list[str],
+    data_vars: list[str],
+    concat_axes: list[int],
+    context: ts.Context,
+    write: bool,
+) -> dict[str, ts.TensorStore]:
+    """Open multiple zarrs with TensorStore.
+
+    Args:
+        paths: List of paths to zarr stores.
+        data_vars: List of data variable names to open.
+        concat_axes: List of axes along which to concatenate the data variables.
+        context: TensorStore context.
+        write: Whether to open the stores for writing.
+    """
+    arrays_list = []
+    for path in paths:
+        specs = {k: _zarr_spec_from_path(os.path.join(path, k)) for k in data_vars}
+        array_futures = {
+          k: ts.open(spec, read=True, write=write, context=context)
+          for k, spec in specs.items()
+        }
+        arrays_list.append({k: v.result() for k, v in array_futures.items()})
+
+    arrays = {}
+    for k, axis in zip(data_vars, concat_axes, strict=False):
+        datasets = [d[k] for d in arrays_list]
+        arrays[k] = ts.concat(datasets, axis=axis)
+
+    return arrays
+
+
+def open_zarrs(
+    paths: list[str],
+    concat_dim: str,
+    *,
+    context: ts.Context | None = None,
+    mask_and_scale: bool = True,
+    write: bool = False,
+) -> xr.Dataset:
+    """Open multiple zarrs with TensorStore.
+
+    Args:
+        paths: List of paths to zarr stores.
+        concat_dim: Dimension along which to concatenate the data variables.
+        context: TensorStore context.
+        mask_and_scale: Whether to mask and scale the data.
+        write: Whether to open the stores for writing.
+    """
+    if context is None:
+        context = ts.Context()
+
+    ds = xr.open_mfdataset(
+        paths,
+        concat_dim=concat_dim,
+        combine="nested",
+        mask_and_scale=mask_and_scale,
+        decode_timedelta=True,
+    )
+
+    if mask_and_scale:
+        # Data variables get replaced below with _TensorStoreAdapter arrays, which
+        # don't get masked or scaled. Raising an error avoids surprising users with
+        # incorrect data values.
+        _raise_if_mask_and_scale_used_for_data_vars(ds)
+
+    data_vars = list(ds.data_vars)
+
+    concat_axes = [ds[v].dims.index(concat_dim) for v in data_vars]
+
+    arrays = tensorstore_open_multi_zarrs(paths, data_vars, concat_axes, context, write)
+
+    new_data = {k: _TensorStoreAdapter(v) for k, v in arrays.items()}
+
+    return ds.copy(data=new_data)

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/satellite.py

@@ -1,6 +1,7 @@
 """Satellite loader."""
 import numpy as np
 import xarray as xr
+from xarray_tensorstore import open_zarr
 
 from ocf_data_sampler.load.utils import (
     check_time_unique_increasing,
@@ -8,39 +9,7 @@ from ocf_data_sampler.load.utils import (
     make_spatial_coords_increasing,
 )
 
-
-def get_single_sat_data(zarr_path: str) -> xr.Dataset:
-    """Helper function to open a zarr from either a local or GCP path.
-
-    Args:
-        zarr_path: path to a zarr file. Wildcards (*) are supported only for local paths
-                   GCS paths (gs://) do not support wildcards
-
-    Returns:
-        An xarray Dataset containing satellite data
-
-    Raises:
-        ValueError: If a wildcard (*) is used in a GCS (gs://) path
-    """
-    # Raise an error if a wildcard is used in a GCP path
-    if "gs://" in str(zarr_path) and "*" in str(zarr_path):
-        raise ValueError("Wildcard (*) paths are not supported for GCP (gs://) URLs")
-
-    # Handle multi-file dataset for local paths
-    if "*" in str(zarr_path):
-        ds = xr.open_mfdataset(
-            zarr_path,
-            engine="zarr",
-            concat_dim="time",
-            combine="nested",
-            chunks="auto",
-            join="override",
-        )
-        check_time_unique_increasing(ds.time)
-    else:
-        ds = xr.open_dataset(zarr_path, engine="zarr", chunks="auto")
-
-    return ds
+from .open_tensorstore_zarrs import open_zarrs
 
 
 def open_sat_data(zarr_path: str | list[str]) -> xr.DataArray:
@@ -52,14 +21,11 @@ def open_sat_data(zarr_path: str | list[str]) -> xr.DataArray:
     """
     # Open the data
     if isinstance(zarr_path, list | tuple):
-        ds = xr.combine_nested(
-            [get_single_sat_data(path) for path in zarr_path],
-            concat_dim="time",
-            combine_attrs="override",
-            join="override",
-        )
+        ds = open_zarrs(zarr_path, concat_dim="time")
     else:
-        ds = get_single_sat_data(zarr_path)
+        ds = open_zarr(zarr_path)
+
+    check_time_unique_increasing(ds.time)
 
     ds = ds.rename(
         {

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/load/utils.py

@@ -47,7 +47,7 @@ def get_xr_data_array_from_xr_dataset(ds: xr.Dataset) -> xr.DataArray:
     Args:
         ds: xr.Dataset to extract xr.DataArray from
     """
-    datavars = list(ds.var())
+    datavars = list(ds.data_vars)
     if len(datavars) != 1:
         raise ValueError("Cannot open as xr.DataArray: dataset contains multiple variables")
     return ds[datavars[0]]

ocf_data_sampler-0.3.1/ocf_data_sampler/select/dropout.py

@@ -0,0 +1,61 @@
+"""Functions for simulating dropout in time series data.
+
+This is used for the following types of data: GSP, Satellite and Site
+This is not used for NWP
+"""
+
+import numpy as np
+import pandas as pd
+import xarray as xr
+
+
+def apply_sampled_dropout_time(
+    t0: pd.Timestamp,
+    dropout_timedeltas: list[pd.Timedelta],
+    dropout_frac: float|list[float],
+    da: xr.DataArray,
+) -> xr.DataArray:
+    """Randomly pick a dropout time from a list of timedeltas and apply dropout time to the data.
+
+    Args:
+        t0: The forecast init-time
+        dropout_timedeltas: List of timedeltas relative to t0 to pick from
+        dropout_frac: Either a probability that dropout will be applied.
+            This should be between 0 and 1 inclusive.
+            Or a list of probabilities for each of the corresponding timedeltas
+        da: Xarray DataArray with 'time_utc' coordinate
+    """
+    if  isinstance(dropout_frac, list):
+        # checking if len match
+        if len(dropout_frac) != len(dropout_timedeltas):
+            raise ValueError("Lengths of dropout_frac and dropout_timedeltas should match")
+
+
+
+
+        dropout_time = t0 + np.random.choice(dropout_timedeltas,p=dropout_frac)
+
+        return da.where(da.time_utc <= dropout_time)
+
+
+
+    # old logic
+    else:
+        # sample dropout time
+        if dropout_frac > 0 and len(dropout_timedeltas) == 0:
+            raise ValueError("To apply dropout, dropout_timedeltas must be provided")
+
+
+        if not (0 <= dropout_frac <= 1):
+            raise ValueError("dropout_frac must be between 0 and 1 inclusive")
+
+        if (len(dropout_timedeltas) == 0) or (np.random.uniform() >= dropout_frac):
+            dropout_time = None
+        else:
+            dropout_time = t0 + np.random.choice(dropout_timedeltas)
+
+        # apply dropout time
+        if dropout_time is None:
+            return da
+        # This replaces the times after the dropout with NaNs
+        return da.where(da.time_utc <= dropout_time)

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler/torch_datasets/datasets/pvnet_uk.py

@@ -270,6 +270,8 @@ class PVNetUKRegionalDataset(AbstractPVNetUKDataset):
     def __getitem__(self, idx: int) -> NumpySample:
         # Get the coordinates of the sample
 
+        idx = int(idx)
+
         if idx >= len(self):
             raise ValueError(f"Index {idx} out of range for dataset of length {len(self)}")
 

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocf-data-sampler
-Version: 0.2.38
+Version: 0.3.1
 Author: James Fulton, Peter Dudfield
 Author-email: Open Climate Fix team <info@openclimatefix.org>
 License: MIT License
@@ -44,6 +44,7 @@ Requires-Dist: pyproj
 Requires-Dist: pyaml_env
 Requires-Dist: pyresample
 Requires-Dist: h5netcdf
+Requires-Dist: xarray-tensorstore==0.1.5
 
 # ocf-data-sampler
  

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler.egg-info/SOURCES.txt

@@ -17,6 +17,7 @@ ocf_data_sampler/data/uk_gsp_locations_20250109.csv
 ocf_data_sampler/load/__init__.py
 ocf_data_sampler/load/gsp.py
 ocf_data_sampler/load/load_dataset.py
+ocf_data_sampler/load/open_tensorstore_zarrs.py
 ocf_data_sampler/load/satellite.py
 ocf_data_sampler/load/site.py
 ocf_data_sampler/load/utils.py

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/ocf_data_sampler.egg-info/requires.txt

@@ -12,3 +12,4 @@ pyproj
 pyaml_env
 pyresample
 h5netcdf
+xarray-tensorstore==0.1.5

{ocf_data_sampler-0.2.38 → ocf_data_sampler-0.3.1}/pyproject.toml

@@ -35,6 +35,7 @@ dependencies = [
     "pyaml_env",
     "pyresample",
     "h5netcdf",
+    "xarray-tensorstore==0.1.5",
 ]
 
 [dependency-groups]

ocf_data_sampler-0.2.38/ocf_data_sampler/load/nwp/providers/utils.py

@@ -1,43 +0,0 @@
-"""Utility functions for the NWP data processing."""
-
-import xarray as xr
-
-
-def open_zarr_paths(
-    zarr_path: str | list[str], time_dim: str = "init_time", public: bool = False,
-) -> xr.Dataset:
-    """Opens the NWP data.
-
-    Args:
-        zarr_path: Path to the zarr(s) to open
-        time_dim: Name of the time dimension
-        public: Whether the data is public or private
-
-    Returns:
-        The opened Xarray Dataset
-    """
-    general_kwargs = {
-        "engine": "zarr",
-        "chunks": "auto",
-        "decode_timedelta": True,
-    }
-
-    if public:
-        # note this only works for s3 zarr paths at the moment
-        general_kwargs["storage_options"] = {"anon": True}
-
-    if type(zarr_path) in [list, tuple] or "*" in str(zarr_path):  # Multi-file dataset
-        ds = xr.open_mfdataset(
-            zarr_path,
-            concat_dim=time_dim,
-            combine="nested",
-            **general_kwargs,
-        ).sortby(time_dim)
-    else:
-        ds = xr.open_dataset(
-            zarr_path,
-            consolidated=True,
-            mode="r",
-            **general_kwargs,
-        )
-    return ds

ocf_data_sampler-0.2.38/ocf_data_sampler/select/dropout.py

@@ -1,47 +0,0 @@
-"""Functions for simulating dropout in time series data.
-
-This is used for the following types of data: GSP, Satellite and Site
-This is not used for NWP
-"""
-
-import numpy as np
-import pandas as pd
-import xarray as xr
-
-
-def apply_sampled_dropout_time(
-    t0: pd.Timestamp,
-    dropout_timedeltas: list[pd.Timedelta],
-    dropout_frac: float,
-    da: xr.DataArray,
-) -> xr.DataArray:
-    """Randomly pick a dropout time from a list of timedeltas and apply dropout time to the data.
-
-    Args:
-        t0: The forecast init-time
-        dropout_timedeltas: List of timedeltas relative to t0 to pick from
-        dropout_frac: Probability that dropout will be applied.
-            This should be between 0 and 1 inclusive
-        da: Xarray DataArray with 'time_utc' coordinate
-    """
-    # sample dropout time
-    if dropout_frac > 0 and len(dropout_timedeltas) == 0:
-        raise ValueError("To apply dropout, dropout_timedeltas must be provided")
-
-    for t in dropout_timedeltas:
-        if t > pd.Timedelta("0min"):
-            raise ValueError("Dropout timedeltas must be negative")
-
-    if not (0 <= dropout_frac <= 1):
-        raise ValueError("dropout_frac must be between 0 and 1 inclusive")
-
-    if (len(dropout_timedeltas) == 0) or (np.random.uniform() >= dropout_frac):
-        dropout_time = None
-    else:
-        dropout_time = t0 + np.random.choice(dropout_timedeltas)
-
-    # apply dropout time
-    if dropout_time is None:
-        return da
-    # This replaces the times after the dropout with NaNs
-    return da.where(da.time_utc <= dropout_time)