Closed jreadey closed 8 years ago
I think something must be wrong in your environment. I have no issue in reading szip using python. See the test script below and result on joe-test-issue25 instance at OSDC that runs issue25 image.
ubuntu@joe-test-issue25:~/data$ python test_compression.py
-21.3293
ubuntu@joe-test-issue25:~/data$ more test_compression.py
import h5py
import numpy
# file_path = 'GSSTF_NCEP.3.1987.07.01.blosc.he5'
file_path = 'GSSTF_NCEP.3.1987.07.01.szip.he5'
h5path = '/HDFEOS/GRIDS/NCEP/Data Fields/Tair_2m'
with h5py.File(file_path, 'r') as f:
dset = f[h5path]
# mask fill value
if '_FillValue' in dset.attrs:
arr = dset[...]
fill = dset.attrs['_FillValue'][0]
v = arr[arr != fill]
else:
v = dset[...]
# file name GSSTF_NCEP.3.YYYY.MM.DD.he5
# Python 3 style.
print(numpy.min(v))
@hyoklee - Would you try out using snapshot13? I'm not clear on what is different about the env setup there compared to the issue25 image.
@jreadey snapshot13 uses anaconda which I don't use. I built Python, hdf5, h5py from scratch for issue25 image.
Last login: Tue Nov 24 03:11:16 2015 from 172.17.192.2
discarding /home/ubuntu/anaconda/bin from PATH
prepending /home/ubuntu/anaconda/envs/py34/bin to PATH
h5py can read the other compressed formats (blosc, mafic, zlib), but not szip. Reading the data raised the error: OSError: Can't read data (Required filter (name unavailable) is not registered)
Strangely, h5dump works fine and the dataset values get printed normally.