h5netcdf

A Python interface for the netCDF4 file format that reads and writes local or remote HDF5 files directly via h5py or h5pyd, without relying on the Unidata netCDF library.
https://github.com/h5netcdf/h5netcdf

Category: Climate Change
Sub Category: Climate Data Standards

Keywords

h5py hdf5 netcdf python

Keywords from Contributors

pydata closember measur flexible transforms observation netcdf4 fishing qt alignment

Last synced: about 9 hours ago
JSON representation

Repository metadata

Pythonic interface to netCDF4 via h5py

• Host: GitHub
• URL: https://github.com/h5netcdf/h5netcdf
• Owner: h5netcdf
• License: bsd-3-clause
• Created: 2015-04-07T18:44:42.000Z (about 10 years ago)
• Default Branch: main
• Last Pushed: 2025-03-07T14:51:02.000Z (2 months ago)
• Last Synced: 2025-05-16T23:34:03.966Z (4 days ago)
• Topics: h5py, hdf5, netcdf, python
• Language: Python
• Homepage: https://h5netcdf.org
• Size: 6.99 MB
• Stars: 198
• Watchers: 5
• Forks: 36
• Open Issues: 18
• Releases: 18
• Metadata Files:
  • Readme: README.rst
  • Changelog: CHANGELOG.rst
  • License: LICENSE
  • Authors: AUTHORS.txt

README.rst

          h5netcdf
========

.. image:: https://github.com/h5netcdf/h5netcdf/workflows/CI/badge.svg
    :target: https://github.com/h5netcdf/h5netcdf/actions
.. image:: https://badge.fury.io/py/h5netcdf.svg
    :target: https://pypi.org/project/h5netcdf/
.. image:: https://github.com/h5netcdf/h5netcdf/actions/workflows/pages/pages-build-deployment/badge.svg?branch=gh-pages
    :target: https://h5netcdf.github.io/h5netcdf/

A Python interface for the `netCDF4`_ file-format that reads and writes local or
remote HDF5 files directly via `h5py`_ or `h5pyd`_, without relying on the Unidata
netCDF library.

.. _netCDF4: https://docs.unidata.ucar.edu/netcdf-c/current/file_format_specifications.html#netcdf_4_spec
.. _h5py: https://www.h5py.org/
.. _h5pyd: https://github.com/HDFGroup/h5pyd


.. why-h5netcdf

Why h5netcdf?
-------------

- It has one less binary dependency (netCDF C). If you already have h5py
  installed, reading netCDF4 with h5netcdf may be much easier than installing
  netCDF4-Python.
- We've seen occasional reports of better performance with h5py than
  netCDF4-python, though in many cases performance is identical. For
  `one workflow`_, h5netcdf was reported to be almost **4x faster** than
  `netCDF4-python`_.
- Anecdotally, HDF5 users seem to be unexcited about switching to netCDF --
  hopefully this will convince them that netCDF4 is actually quite sane!
- Finally, side-stepping the netCDF C library (and Cython bindings to it)
  gives us an easier way to identify the source of performance issues and
  bugs in the netCDF libraries/specification.

.. _one workflow: https://github.com/Unidata/netcdf4-python/issues/390#issuecomment-93864839
.. _xarray: https://github.com/pydata/xarray/

Install
-------

Ensure you have a recent version of h5py installed (I recommend using `conda`_ or
the community effort `conda-forge`_).
At least version 3.0 is required. Then::

    $ pip install h5netcdf

Or if you are already using conda::

    $ conda install h5netcdf

Note:

From version 1.2. h5netcdf tries to align with a `nep29`_-like support policy with regard
to it's upstream dependencies.

.. _conda: https://conda.io/
.. _conda-forge: https://conda-forge.org/
.. _nep29: https://numpy.org/neps/nep-0029-deprecation_policy.html

Usage
-----

h5netcdf has two APIs, a new API and a legacy API. Both interfaces currently
reproduce most of the features of the netCDF interface, with the notable
exception of support for operations that rename or delete existing objects.
We simply haven't gotten around to implementing this yet. Patches
would be very welcome.

New API
~~~~~~~

The new API supports direct hierarchical access of variables and groups. Its
design is an adaptation of h5py to the netCDF data model. For example:

.. code-block:: python

    import h5netcdf
    import numpy as np

    with h5netcdf.File("mydata.nc", "w") as f:
        # set dimensions with a dictionary
        f.dimensions = {"x": 5}
        # and update them with a dict-like interface
        # f.dimensions['x'] = 5
        # f.dimensions.update({'x': 5})

        v = f.create_variable("hello", ("x",), float)
        v[:] = np.ones(5)

        # you don't need to create groups first
        # you also don't need to create dimensions first if you supply data
        # with the new variable
        v = f.create_variable("/grouped/data", ("y",), data=np.arange(10))

        # access and modify attributes with a dict-like interface
        v.attrs["foo"] = "bar"

        # you can access variables and groups directly using a hierarchical
        # keys like h5py
        print(f["/grouped/data"])

        # add an unlimited dimension
        f.dimensions["z"] = None
        # explicitly resize a dimension and all variables using it
        f.resize_dimension("z", 3)

Notes:

- Automatic resizing of unlimited dimensions with array indexing is not available.
- Dimensions need to be manually resized with ``Group.resize_dimension(dimension, size)``.
- Arrays are returned padded with ``fillvalue`` (taken from underlying hdf5 dataset) up to
  current size of variable's dimensions. The behaviour is equivalent to netCDF4-python's
  ``Dataset.set_auto_mask(False)``.

Legacy API
~~~~~~~~~~

The legacy API is designed for compatibility with `netCDF4-python`_. To use it, import
``h5netcdf.legacyapi``:

.. _netCDF4-python: https://github.com/Unidata/netcdf4-python

.. code-block:: python

    import h5netcdf.legacyapi as netCDF4

    # everything here would also work with this instead:
    # import netCDF4
    import numpy as np

    with netCDF4.Dataset("mydata.nc", "w") as ds:
        ds.createDimension("x", 5)
        v = ds.createVariable("hello", float, ("x",))
        v[:] = np.ones(5)

        g = ds.createGroup("grouped")
        g.createDimension("y", 10)
        g.createVariable("data", "i8", ("y",))
        v = g["data"]
        v[:] = np.arange(10)
        v.foo = "bar"
        print(ds.groups["grouped"].variables["data"])

The legacy API is designed to be easy to try-out for netCDF4-python users, but it is not an
exact match. Here is an incomplete list of functionality we don't include:

- Utility functions ``chartostring``, ``num2date``, etc., that are not directly necessary
  for writing netCDF files.
- h5netcdf variables do not support automatic masking or scaling (e.g., of values matching
  the ``_FillValue`` attribute). We prefer to leave this functionality to client libraries
  (e.g., `xarray`_), which can implement their exact desired scaling behavior. Nevertheless
  arrays are returned padded with ``fillvalue`` (taken from underlying hdf5 dataset) up to
  current size of variable's dimensions. The behaviour is equivalent to netCDF4-python's
  ``Dataset.set_auto_mask(False)``.

.. _invalid netcdf:

Invalid netCDF files
~~~~~~~~~~~~~~~~~~~~

h5py implements some features that do not (yet) result in valid netCDF files:

- Data types:
    - Booleans
    - Reference types
- Arbitrary filters:
    - Scale-offset filters

By default [#]_, h5netcdf will not allow writing files using any of these features,
as files with such features are not readable by other netCDF tools.

However, these are still valid HDF5 files. If you don't care about netCDF
compatibility, you can use these features by setting ``invalid_netcdf=True``
when creating a file:

.. code-block:: python

  # avoid the .nc extension for non-netcdf files
  f = h5netcdf.File("mydata.h5", invalid_netcdf=True)
  ...

  # works with the legacy API, too, though compression options are not exposed
  ds = h5netcdf.legacyapi.Dataset("mydata.h5", invalid_netcdf=True)
  ...

In such cases the `_NCProperties` attribute will not be saved to the file or be removed
from an existing file. A warning will be issued if the file has `.nc`-extension.

.. rubric:: Footnotes

.. [#] h5netcdf we will raise ``h5netcdf.CompatibilityError``.

Decoding variable length strings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

h5py 3.0 introduced `new behavior`_ for handling variable length string.
Instead of being automatically decoded with UTF-8 into NumPy arrays of ``str``,
they are required as arrays of ``bytes``.

The legacy API preserves the old behavior of h5py (which matches netCDF4),
and automatically decodes strings.

The new API matches h5py behavior. Explicitly set ``decode_vlen_strings=True``
in the ``h5netcdf.File`` constructor to opt-in to automatic decoding.

.. _new behavior: https://docs.h5py.org/en/stable/strings.html

.. _phony dims:

Datasets with missing dimension scales
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

By default [#]_ h5netcdf raises a ``ValueError`` if variables with no dimension
scale associated with one of their axes are accessed.
You can set ``phony_dims='sort'`` when opening a file to let h5netcdf invent
phony dimensions according to `netCDF`_ behaviour.

.. code-block:: python

  # mimic netCDF-behaviour for non-netcdf files
  f = h5netcdf.File("mydata.h5", mode="r", phony_dims="sort")
  ...

Note, that this iterates once over the whole group-hierarchy. This has affects
on performance in case you rely on laziness of group access.
You can set ``phony_dims='access'`` instead to defer phony dimension creation
to group access time. The created phony dimension naming will differ from
`netCDF`_ behaviour.

.. code-block:: python

  f = h5netcdf.File("mydata.h5", mode="r", phony_dims="access")
  ...

.. rubric:: Footnotes

.. [#] Keyword default setting ``phony_dims=None`` for backwards compatibility.

.. _netCDF: https://docs.unidata.ucar.edu/netcdf-c/current/interoperability_hdf5.html

Track Order
~~~~~~~~~~~

As of h5netcdf 1.1.0, if h5py 3.7.0 or greater is detected, the ``track_order``
parameter is set to ``True`` enabling `order tracking`_ for newly created
netCDF4 files. This helps ensure that files created with the h5netcdf library
can be modified by the netCDF4-c and netCDF4-python implementation used in
other software stacks. Since this change should be transparent to most users,
it was made without deprecation.

Since track_order is set at creation time, any dataset that was created with
``track_order=False`` (h5netcdf version 1.0.2 and older except for 0.13.0) will
continue to opened with order tracker disabled.

The following describes the behavior of h5netcdf with respect to order tracking
for a few key versions:

- Version 0.12.0 and earlier, the ``track_order`` parameter`order was missing
  and thus order tracking was implicitely set to ``False``.
- Version 0.13.0 enabled order tracking by setting the parameter
  ``track_order`` to ``True`` by default without deprecation.
- Versions 0.13.1 to 1.0.2 set ``track_order`` to ``False`` due to a bug in a
  core dependency of h5netcdf, h5py `upstream bug`_ which was resolved in h5py
  3.7.0 with the help of the h5netcdf team.
- In version 1.1.0, if h5py 3.7.0 or above is detected, the ``track_order``
  parameter is set to ``True`` by default.


.. _order tracking: https://docs.unidata.ucar.edu/netcdf-c/current/file_format_specifications.html#creation_order
.. _upstream bug: https://github.com/h5netcdf/h5netcdf/issues/136
.. _[*]: https://github.com/h5netcdf/h5netcdf/issues/128

.. changelog

Changelog
---------

`Changelog`_

.. _Changelog: https://github.com/h5netcdf/h5netcdf/blob/main/CHANGELOG.rst

.. license

License
-------

`3-clause BSD`_

.. _3-clause BSD: https://github.com/h5netcdf/h5netcdf/blob/main/LICENSE

        

---

Owner metadata

• Name: h5netcdf
• Login: h5netcdf
• Email:
• Kind: organization
• Description:
• Website:
• Location:
• Twitter:
• Company:
• Icon url: https://avatars.githubusercontent.com/u/74427192?v=4
• Repositories: 1
• Last ynced at: 2023-03-01T05:25:22.286Z
• Profile URL: https://github.com/h5netcdf

---

GitHub Events

Total

• Create event: 4
• Release event: 4
• Issues event: 9
• Watch event: 15
• Delete event: 1
• Issue comment event: 70
• Push event: 13
• Pull request review comment event: 4
• Pull request event: 27
• Pull request review event: 7
• Fork event: 2

Last Year

• Create event: 4
• Release event: 4
• Issues event: 9
• Watch event: 15
• Delete event: 1
• Issue comment event: 70
• Push event: 13
• Pull request review comment event: 4
• Pull request event: 27
• Pull request review event: 7
• Fork event: 2

---

Committers metadata

Last synced: 2 days ago

Name	Email	Commits
Kai Mühlbauer	k****r@u****e	91
Stephan Hoyer	s****r@c****m	87
Aleksandar Jelenak	a****k@g****e	20
Lion Krischer	l****r@g****m	16
Mark Harfouche	m****e@g****m	11
Aleksandar Jelenak	a****k@h****g	7
Scott Henderson	s****q@g****m	5
dependabot[bot]	4****]	5
Martin Raspaud	m****d@s****e	3
Brett Naul	b****l@g****m	2
Frederic Laliberte	l****c@g****m	2
Ghislain Antony Vaillant	g****l	2
Rickard Holmberg	r****g@n****m	2
Dion Häfner	d****r@n****k	2
Tom Augspurger	t****r@m****m	2
Drew Parsons	d****s@e****m	1
Ezequiel Cimadevilla Alvarez	e****a@u****s	1
John Readey	j****y@h****g	1
Ryan Grout	g****r	1
Thomas Kluyver	t****l@g****m	1
paugier	p****r@u****r	1

Committer domains:

• hdfgroup.org: 2
• univ-grenoble-alpes.fr: 1
• unican.es: 1
• emerall.com: 1
• microsoft.com: 1
• nbi.ku.dk: 1
• novatronfusion.com: 1
• smhi.se: 1
• gharial.home: 1
• climate.com: 1
• uni-bonn.de: 1

---

Issue and Pull Request metadata

Last synced: 1 day ago

More stats: https://issues.ecosyste.ms/repositories/lookup?url=https://github.com/h5netcdf/h5netcdf

---

---

---

Package metadata

• Total packages: 4
• Total downloads:
  • pypi: 1,294,961 last-month
• Total docker downloads: 310,858
• Total dependent packages: 177 (may contain duplicates)
• Total dependent repositories: 786 (may contain duplicates)
• Total versions: 78
• Total maintainers: 3

---

Dependencies

Score: 22.70957967608488