Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fix uniqueness of hdf5-based dask arrays #2818

Merged
merged 3 commits into from
Jun 19, 2024
Merged

Fix uniqueness of hdf5-based dask arrays #2818

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b31209e
Fix-uniqueness-of-hdf5-based-dask-arrays
mraspaud Jun 13, 2024
5062421
Make hdf5 dataset names more robust
mraspaud Jun 14, 2024
2c6af5e
Make name of h5dsets more explicit
mraspaud Jun 19, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 16 additions & 3 deletions satpy/readers/hdf5_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,19 +18,21 @@
"""Helpers for reading hdf5-based files."""

import logging
import os

import dask.array as da
import dask.config as dc
import h5py
import numpy as np
import xarray as xr
from dask.array.core import normalize_chunks
from dask.base import tokenize

from satpy.readers import open_file_or_filename
from satpy.readers.file_handlers import BaseFileHandler
from satpy.readers.utils import np2str
from satpy.utils import get_legacy_chunk_size

LOG = logging.getLogger(__name__)
CHUNK_SIZE = get_legacy_chunk_size()


class HDF5FileHandler(BaseFileHandler):
Expand Down Expand Up @@ -102,7 +104,7 @@ def __getitem__(self, key):
# these datasets are closed and inaccessible when the file is closed, need to reopen
f_obj = open_file_or_filename(self.filename)
dset = h5py.File(f_obj, "r")[key]
dset_data = da.from_array(dset, chunks=CHUNK_SIZE)
dset_data = from_h5_array(dset)
attrs = self._attrs_cache.get(key, dset.attrs)
if dset.ndim == 2:
return xr.DataArray(dset_data, dims=["y", "x"], attrs=attrs)
Expand All @@ -120,3 +122,14 @@ def get(self, item, default=None):
return self[item]
else:
return default


def from_h5_array(h5dset):
"""Create a dask array from an h5py dataset, ensuring uniqueness of the dask array name."""
chunk_size = dc.get("array.chunk-size")

chunks = normalize_chunks(chunk_size, dtype=h5dset.dtype, previous_chunks=h5dset.chunks, shape=h5dset.shape)
name = h5dset.name + "-" + tokenize(os.fspath(h5dset.file.filename), h5dset.name, chunks)

dset_data = da.from_array(h5dset, chunks=chunks, name=name)
return dset_data
10 changes: 10 additions & 0 deletions satpy/tests/reader_tests/test_hdf5_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,3 +150,13 @@ def test_all_basic(self):
assert "fake_ds" not in file_handler

assert isinstance(file_handler["ds2_f/attr/test_ref"], np.ndarray)

def test_array_name_uniqueness(self):
"""Test the dask array generated from an hdf5 dataset stay constant and unique."""
from satpy.readers.hdf5_utils import HDF5FileHandler
file_handler = HDF5FileHandler("test.h5", {}, {})

dsname = "test_group/ds1_f"

assert file_handler[dsname].data.name == file_handler[dsname].data.name
assert file_handler[dsname].data.name.startswith("/" + dsname)
Loading