Every object (Groups or Dataset) in an HDF5 file has a name
Names are arranged in a POSIX-style hierarchy with /-separators
f is the entry, as well as a root group.
f = h5py.File('mytestfile.hdf5', 'r')
like folders, can have contents (i.e. datasets), and subfolders (i.e. sub groups)
works like a python dict, .keys(), .values(), .items(), iter() and get()
create group
>>> grp = f.create_group("bar")
>>> grp.name
'/bar'
>>> subgrp = grp.create_group("baz")
>>> subgrp.name
'/bar/baz'
f.create_group(“/some/long/path”)
iterating over a group only yields its directly-attached members
iter over whole file using .visit() and .visititems() methods of group.
>>> def printname(name):
... print name
>>> f.visit(printname)
mydataset
subgroup
subgroup/another_dataset
subgroup2
subgroup2/dataset_three
similar to numpy array
each dataset has basic attribute like numpy array
indexing like numpy array
# get numpy array from dataset
arr = dset[:]
# 可以赋值
dset[…]=np.arrange(100)
# using existing array
>>> arr = np.arange(100)
>>> dset = f.create_dataset("init", data=arr)
Attributes, also works like dict dset.attrs['temperature'] = 99.5
Attributs有如下特点
可以由任何numpy array或者scalar创建
每个attribute应该小 <64k
不可以slice，读的时候必须整个读