"""
This example shows how to read and write data to a dataset using
LZF compression.  The program first checks if LZF compression is
available, then if it is it writes integers to a dataset using LZF,
then closes the file.  Next, it reopens the file, reads back the data,
and outputs the type of compression and the maximum value in the dataset
to the screen.
"""
import sys

import numpy as np
import h5py

FILE = "h5ex_d_lzf.h5"
DATASET = "DS1"

# Strings are handled very differently between python2 and python3.
if sys.hexversion >= 0x03000000:
    FILE = FILE.encode()
    DATASET = DATASET.encode()

DIM0 = 32
DIM1 = 64
CHUNK0 = 4
CHUNK1 = 8

def run():

    # Check if LZF compression is available and can be used for
    # both compression and decompression.  Normally we do not perform
    # error checking in these examples for the sake of clarity, but
    # in this case we will make an exception because this filter is
    # an optional part of the hdf5 library.
    if not h5py.h5z.filter_avail(h5py.h5z.FILTER_LZF):
        raise RuntimeError("LZF filter not available.")

    filter_info = h5py.h5z.get_filter_info(h5py.h5z.FILTER_LZF)
    if ((filter_info & h5py.h5z.FILTER_CONFIG_ENCODE_ENABLED) & 
        (filter_info & h5py.h5z.FILTER_CONFIG_DECODE_ENABLED)):
        msg = "LZF filter not available for encoding and decoding."
        raise RuntimeError(msg)

    # Initialize the data.
    wdata = np.zeros((DIM0, DIM1), dtype=np.int32)
    for i in range(DIM0):
        for j in range(DIM1):
            wdata[i][j] = i * j - j

    # Create a new file using the default properties.
    fid = h5py.h5f.create(FILE)

    # Create the dataspace.  No maximum size parameter needed.
    dims = (DIM0, DIM1)
    space_id = h5py.h5s.create_simple(dims)

    # Create the dataset creation property list, add the fletcher32 filter
    # and set a chunk size.
    chunk = (CHUNK0, CHUNK1)
    dcpl = h5py.h5p.create(h5py.h5p.DATASET_CREATE)

    # Have to "manually" set LZF  compression.  The flags argument is 1,
    # as LZF is considered "optional".
    dcpl.set_filter(h5py.h5z.FILTER_LZF, 1)
    dcpl.set_chunk(chunk)

    # Create the datasets using the dataset creation property list.
    dset = h5py.h5d.create(fid, DATASET, h5py.h5t.STD_I32LE, space_id, dcpl)

    # Write the data to the dataset.
    dset.write(h5py.h5s.ALL, h5py.h5s.ALL, wdata)

    # Close and release resources.
    del dcpl
    del dset
    del space_id
    del fid

    # Reopen the file and dataset using default properties.
    fid = h5py.h5f.open(FILE)
    dset = h5py.h5d.open(fid, DATASET)
    dcpl = dset.get_create_plist()

    # Retrieve and print the filter type.  We know there is only one filter,
    # so the index is zero.
    filter_type, flags, vals, name = dcpl.get_filter(0)

    # No NBIT or SCALEOFFSET filter, but there is something new, LZF.
    ddict = {h5py.h5z.FILTER_DEFLATE: "DEFLATE",
             h5py.h5z.FILTER_SHUFFLE: "SHUFFLE",
             h5py.h5z.FILTER_FLETCHER32: "FLETCHER32",
             h5py.h5z.FILTER_SZIP: "SZIP",
             h5py.h5z.FILTER_LZF: "LZF"}
    print("Filter type for %s is H5Z_%s" % (DATASET, ddict[filter_type])) 

    rdata = np.zeros((DIM0, DIM1))
    dset.read(h5py.h5s.ALL, h5py.h5s.ALL, rdata)

    # Verify that the dataset was read correctly.
    np.testing.assert_array_equal(rdata, wdata)
    print("Maximum value in DS1 is:  %d" % rdata.max())

if __name__ == "__main__":
    run()