Remove h5py as required dependency (#51)

gpm/io/data_integrity.py

@@ -27,7 +27,7 @@
 """This module contains functions that check the GPM files integrity."""
 import os
 
-import h5py
+import xarray as xr
 
 from gpm.io.checks import (
     check_product,
@@ -38,12 +38,23 @@
 
 
 def get_corrupted_filepaths(filepaths):
+    """Return the file paths of corrupted files."""
     l_corrupted = []
     for filepath in filepaths:
-        # Load hdf granule file
         try:
-            hdf = h5py.File(filepath, "r")  # h5py._hl.files.File
-            hdf.close()
+            # Try open the HDF file
+
+            # DataTree.close() does not work yet!
+            # dt = datatree.open_datatree(filepath, engine="netcdf4")
+            # dt.close()
+
+            # h5py it's an heavy dependency !
+            # hdf = h5py.File(filepath, "r")  # h5py._hl.files.File
+            # hdf.close()
+
+            ds = xr.open_dataset(filepath, engine="netcdf4", group="")
+            ds.close()
+
         except OSError:
             l_corrupted.append(filepath)
     return l_corrupted

pyproject.toml

@@ -38,7 +38,6 @@ keywords = ["GPM", "Precipitation", "API", "DPR", "PMW", "IMERG"]
 dependencies = [
     "xarray<=2024.2.0", # ongoing xarray datatree migration
     "xarray-datatree",  # to be deprecated as soon datatree in xarray is working
-    "h5py", # to discard as soon as possible
     "netcdf4",
     "dask",
     "distributed",
@@ -62,7 +61,7 @@ dev = ["pre-commit", "loghub",
        "pytest", "pytest-cov", "pytest-mock", "pytest-check", "pytest-sugar",
        "pytest-watcher", "deepdiff",
        "pip-tools", "bumpver", "twine", "wheel", "build", "setuptools>=61.0.0",
-       "ximage", "pyvista", "polars", "pyarrow", "pyresample",
+       "ximage", "pyvista", "polars", "pyarrow", "pyresample", "h5py",
        "sphinx", "sphinx-gallery", "sphinx-book-theme", "nbsphinx", "sphinx_mdinclude"]
 
 [project.urls]