IronPython Example of Writing ISO 8601 Strings to HDF5 from .NET

hdf5dotnet_iso8601_ipy64_example.py

# Writing Fixed Length Strings to HDF5 file using HDF5DotNet
# http://hdf5.net/

# * download HDF5 library and install first
# * add shared objects to PATH
#       set PATH=%PATH%;C:\Program Files\HDF Group\HDF5\1.8.9\bin
# * download and extract HDF5DotNet assemblies
# * start IronPython (ipy64.exe), must be 64-bit version!

# help links:
# http://stackoverflow.com/a/29565641/1020470
# https://support.hdfgroup.org/ftp/HDF5/examples/misc-examples/strarray.c
# http://hdf5.net/FAQ.aspx

# main snippet
import clr  # import MS common language runtime (clr)
# add reference to HDF5DotNet.dll any way you can, EG: by file and path
clr.AddReferenceToFileAndPath('C:\\HDF5DotNet-Net4.0-x86_64\\HDF5DotNet.dll')

import HDF5DotNet
from HDF5DotNet import *
import System
from System import Array, Int64, Byte

LEN_DATETIME = 26  # fixed length of datetime strings
DATETIMES = ["2017-05-17T12:00:00.000000", "2017-05-17T13:00:00.000000"]
FILENAME = 'sample.h5'

def write_datetime_to_hdf5(iso8601_datetimes=DATETIMES, datetime_size=LEN_DATETIME, filename=FILENAME):
    for dt in iso8601_datetimes:
        if len(dt) > datetime_size:
            msg = ('length %d of datetime "%s" must be smaller than string size %d'
                   % (len(dt), dt, datetime_size))
            raise Exception(msg)

    num_datetimes = len(iso8601_datetimes)

    # start hdf5
    # ----------
    H5.Open()

    # open hdf5 file
    # --------------
    h5file = H5F.create(filename, H5F.CreateMode.ACC_TRUNC)

    # create dataspace
    # ----------------
    ndims = 1  # number of dimensions of data space
    # dataspace dimensions must be Int64
    dims = Array[Int64]((num_datetimes,))  # length of each dimension, EG: 2 ISO8601 date strings
    dspace = H5S.create_simple(ndims, dims)  # maxdims is dims

    # setting maxdims:
    # maxdims = Array[Int64]((2,))  # must be > dims, NULL is same as dim
    # -1 is unlimited length, but then must use chunked data creation property list
    # https://lists.hdfgroup.org/pipermail/hdf-forum_lists.hdfgroup.org/2011-May/004686.html
    # dspace = H5S.create_simple(ndims, dims, maxdims)

    # create datatype
    # ---------------
    dtype = H5T.copy(H5T.H5Type.C_S1)
    H5T.setSize(dtype, datetime_size)  # set the length of each string
    # check size of fixed string:
    # LOGGER.debug('string size: %d', H5T.getSize(dtype))
    #  26

    # create dataset and link to file
    P_DEFAULT = H5PropertyListId(H5P.Template.DEFAULT)
    dset = H5D.create(h5file, 'dates', dtype, dspace, P_DEFAULT, P_DEFAULT, P_DEFAULT)

    # setting dataspace with unlimited dimensions
    # if any dataspaces are unlimited, then must use chunked
    # https://lists.hdfgroup.org/pipermail/hdf-forum_lists.hdfgroup.org/2011-May/004686.html
    # see example in http://hdf5.net/
    #  plist = H5P.create(H5P.PropertyListClass.DATASET_CREATE)
    #  H5P.setChunk(plist, Array[Int64]((32, 64)))
    #  H5P.setDeflate(plist, 7)
    #  dset = H5D.create(h5file, 'dates', dtype, dspace, P_DEFAULT, plist, P_DEFAULT)

    # create byte arrays from strings, must use ASCII encoding only
    buf = Array.CreateInstance(Byte, num_datetimes, datetime_size)
    for idx, dt in enumerate(iso8601_datetimes):
        dt_buf = System.Text.Encoding.ASCII.GetBytes(dt)
        for jdx in range(datetime_size):
            buf[idx, jdx] = dt_buf[jdx]


    # write buffer to HDF5 file, must be byte array
    H5D.write(dset, dtype, H5Array[Byte](buf))

    # close and clean up
    H5F.close(h5file)
    H5.Close()


if __name__ == "__main__":
    write_datetime_to_hdf5()