How to inspect .h5 file in Python

Question

How do I see what variables, datasets, etc. a given .h5 file has in Python?

I can read the file by running this

import h5py
f = h5py.File(filename, 'r')

How can I now see which variables my .h5 file have?

Running f.keys() outputs the non-informative

KeysView(<HDF5 file filename (mode r)>)

In Matlab I simply call h5disp(filename) but would like to know how to do it in Python

In Python3, a `dictionary.keys()` gives a 'KeysView` which has to be expanded with `list` (or can be used directly in iteration). — hpaulj, Apr 12 '17 at 15:25

Imanol Luengo · Answer 1 · 2017-04-12T17:12:02.427

Maybe overkill, but I had this and might be useful for someone:

from __future__ import print_function

def scan_hdf5(path, recursive=True, tab_step=2):
    def scan_node(g, tabs=0):
        print(' ' * tabs, g.name)
        for k, v in g.items():
            if isinstance(v, h5.Dataset):
                print(' ' * tabs + ' ' * tab_step + ' -', v.name)
            elif isinstance(v, h5.Group) and recursive:
                scan_node(v, tabs=tabs + tab_step)
    with h5.File(path, 'r') as f:
        scan_node(f)

And simple input:

>>> scan_hdf5('/tmp/dummy.h5')
/
   - /d1
   /g1
     - /g1/d2
     - /g1/d3
   /g2
     - /g2/d4
     /g2/g3
       - /g2/g3/d5

Or an alternative version that returns the elements in something more usable:

def scan_hdf52(path, recursive=True, tab_step=2):
    def scan_node(g, tabs=0):
        elems = []
        for k, v in g.items():
            if isinstance(v, h5.Dataset):
                elems.append(v.name)
            elif isinstance(v, h5.Group) and recursive:
                elems.append((v.name, scan_node(v, tabs=tabs + tab_step)))
        return elems
    with h5.File(path, 'r') as f:
        return scan_node(f)

with returns:

>>> scan_hdf5_2('/tmp/dummy.h5')
[u'/d1',
 (u'/g1', [u'/g1/d2', u'/g1/d3']),
 (u'/g2', [u'/g2/d4', (u'/g2/g3', [u'/g2/g3/d5'])])]

This is amazing, thank you for posting it, saved me a ton of time here! — Nek0pi, Jun 07 '23 at 21:05

score 7 · Accepted Answer · answered Apr 12 '17 at 13:55

7

Did you try?

print(list(f.keys()))

That should give you all the group inside your hdf5 file. You can do the same for the datasets if f is a group.

answered Apr 12 '17 at 13:55

Astrom

767
5
20

1

This is needed since in Python3 f.keys() returns and iterator that should be expanded into a list explicitly. – Dima Chubarov Apr 12 '17 at 14:06
It outputs '['#refs#', 'var1', 'var2', 'var3', 'var4']' what does #refs# mean? – gota Apr 12 '17 at 14:11
1

I would suggest to look inside '#ref#' to see what is inside. As far as we can judge it is a group inside your hdf5 file but without giving more information (that you may not have) about how it has been created it is hard to tell. – Astrom Apr 12 '17 at 14:15
This also works if one likes to get list of keys for debugging: `keys = [key for key in f.keys()] ` – Good Will Feb 13 '19 at 00:33

A. Ahmed · Answer 3 · 2020-05-16T01:56:35.927

I came across this question while trying to figure out a way to display every single node in h5 file to be able to extract only wanted nodes with corresponding datasets.

I think this is very simple yet easy to understand (for me) snippet:

h5 = h5py.File(filename, 'r')
def hierarchy(d):
    for item in d:
        if ' 0 member' in str(d[item]):
            print(d[item].name, ['empty group'])
        if isinstance(d[item], h5py.Group):
            hierarchy(d[item])
        else: #Dataset
            print(d[item].name, ['dataset'])
hierarchy(h5)

Hence I will be using this in GUI application, I am going to enable selecting only 'dataset' flagged items.

score 0 · Answer 4 · edited Jul 25 '23 at 19:19

This prints out data structure in a hierarchy both groups and datasets contained within the h5 file.

import h5py

def printGroup(group):
    for i in list(group. keys()):
        try:
            if list(group[i].keys()):
                print(f"{group.name}/{i}/")
                printGroup(group[i])
        except:
            print(group[i].name, group[i].dtype, group[i].shape)

print("\n\nData File Structure")
f = h5py.File('filePath.h5', 'r')
printGroup(f)

How to inspect .h5 file in Python

4 Answers4

Linked