Adding performance test for the chunk example

%% Cell type:markdown id: tags:

## HDF5 Datasets

In the following we show some examples on how to create Datasets in HDF5 (whith h5py) and update values

### First dataset

%% Cell type:code id: tags:

``` python

import h5py

from timeit import timeit  # To measure execution time

import numpy as np # this is the main python numerical library

f = h5py.File("testdata.hdf5",'w')

# We create a test 2-d array filled with 1s and with 10 rows and 6 columns

data = np.ones((10, 6))

f["dataset_one"] = data

# We now retrieve the dataset from file

# We now retrieve the dataset from file (it is still in memory in fact)

dset = f["dataset_one"]

```

%% Cell type:code id: tags:

``` python

#The following instructions show some dataset metadata

print(dset)

print(dset.dtype)

print(dset.shape)

```

%% Cell type:markdown id: tags:

### Dataset slicing

Datasets provide analogous slicing operations as numpy arrays (with h5py). But these selections are translated by h5py to portion of the dataset and then HDF5 read the data form "disk". Slicing into a dataset object returns a NumpPy array.

Datasets provide analogous slicing operations as numpy arrays (with h5py). But these selections are translated by h5py to portion of the dataset and then HDF5 reads the data form "disk". Slicing into a dataset object returns a NumpPy array.

%% Cell type:code id: tags:

``` python

# The ellipses means "as many ':' as needed"

# here we use it to get a numpy array of the

# entire dataset

out = dset[...]

print(out)

type(out)

```

%% Cell type:code id: tags:

``` python

dset[1:5, 1] = 0.0

dset[...]

#but we cannot use negative steps with a dataset

try:

    dset[0,::-1]

except:

    print('No no no!')

```

%% Cell type:code id: tags:

``` python

# random 2d distribution

# random 2d distribution in the range (-1,1)

data = np.random.rand(15, 10)*2 - 1

dset = f.create_dataset('random', data=data)

# print the first 5 even rows and the first two columns

out = dset[0:10:2, :2]

print(out)

# clipping to zero all negative values

# clipping to zero all negative values, using boolean indexing

dset[data<0] = 0

```

%% Cell type:markdown id: tags:

### Resizable datasets

If we don't know in advance the dataset size and we need to append new data several times, we have to create a resizable dataset, then we have to append data in a scalable manner

%% Cell type:code id: tags:

``` python

dset = f.create_dataset('dataset_two', (1,1000), dtype=np.float32,

                        maxshape=(None, 1000))

a = np.ones((1000,1000))

num_rows = dset.shape[0]

dset.resize((num_rows+a.shape[0], 1000))

for row in a:

    dset[num_rows,:] = row

    num_rows +=1

print(dset[1000,:20])

dset[num_rows:] = a

print(dset[1:5,:20])

```

%% Cell type:markdown id: tags:

## Groups

We can directly create nested groups with a single instruction. For instance to create the group 'nisp_frame', then the subgroup 'detectors' and at last its child group 'det11', we can use the instruction below.

%% Cell type:code id: tags:

``` python

grp = f.create_group('nisp_frame/detectors/det11')

grp['sci_image'] = np.zeros((2040,2040))

print(grp.name)     # the group name property

print(grp.parent)   # the parent group property

print(grp.file)     # the file property

print(grp)          # prints some group information. It has one member, the dataset

```

%% Cell type:markdown id: tags:

## Attributes

Attributes can be defined inside a group or in a dataset. Both have the **.attrs** property to access an attribute or define new attributes. With h5py, the attribute type is inferred from the passed value, but it is also possible to explicitly assign a type.

%% Cell type:code id: tags:

``` python

grp = f['nisp_frame']

grp.attrs['telescope'] = 'Euclid'

grp.attrs['instrument'] = 'NISP'

grp.attrs['pointing'] = np.array([8.48223045516, -20.4610801911, 64.8793517547])

grp.attrs.create('detector_id', '11', dtype="|S2")

print(grp.attrs['pointing'])

print(grp.attrs['detector_id'])

```

%% Cell type:code id: tags:

``` python

f.close()

!h5ls -vlr testdata.hdf5

```

%% Cell type:markdown id: tags:

## Tables (compound types)

Tables can be stored as datasets where the elements (rows) have the same compound type.

%% Cell type:code id: tags:

``` python

f = h5py.File("testdata.hdf5")

f = h5py.File("testdata.hdf5",'a')

dt = np.dtype([('source_id', np.uint32), ('ra', np.float32), ('dec', np.float32), ('magnitude', np.float64)])

grp = f.create_group('source_catalog/det11')

dset = grp.create_dataset('star_catalog', (100,), dtype=dt)

dset['source_id', 0] = 1

print(dset['source_id', 'ra', :20])

print(dset[0])

```

%% Cell type:markdown id: tags:

## References and Region references

In the following instruction we create a reference from the detector 11 scientific image to the corresponding star catalog, which is stored in the same file

%% Cell type:code id: tags:

``` python

sci_image = f['/nisp_frame/detectors/det11/sci_image']

sci_image.attrs['star_catalog'] = dset.ref

cat_ref = sci_image.attrs['star_catalog']

print(cat_ref)

dset = f[cat_ref]

print(dset[0])

dt = h5py.special_dtype(ref=h5py.Reference)

# the above data type dt can be used to create a dataset of references or just an attribute

```

%% Cell type:code id: tags:

``` python

roi = sci_image.regionref[15:20, 36:78]

sci_image[roi]

```

%% Cell type:markdown id: tags:

## Chuncking

%% Cell type:code id: tags:

``` python

dset = f.create_dataset('chunked', (10,2048,2048), dtype=np.uint16, chunks=(1,64,64))

rdata = np.random.randint(0,2**16,(100,2048,2048), dtype=np.uint16)

f = h5py.File('image_sequence.hdf5','w')

dset = f.create_dataset('nochunk', data=rdata)

f.flush()

f.close()

f = h5py.File('image_sequence_chunked.hdf5','w')

dset = f.create_dataset('chunked', data=rdata, chunks=(100,64,64))

f.flush()

f.close()

f = h5py.File('image_sequence.hdf5','r')

dset = f['nochunk']

```

%% Cell type:code id: tags:

``` python

%%time

for i in range(32):

    for j in range(32):

        block = dset[:,64*i:64*(i+1), 64*j:64*(j+1)]

        np.median(block, 0)

```

%% Cell type:code id: tags:

``` python

f.close()

f = h5py.File('image_sequence_chunked.hdf5','r')

dset = f['chunked']

```

%% Cell type:code id: tags:

``` python

%%time

for i in range(32):

    for j in range(32):

        block = dset[:,64*i:64*(i+1), 64*j:64*(j+1)]

        np.median(block, 0)

```

%% Cell type:code id: tags:

``` python

f.close()

f = h5py.File('image_sequence_chunked.hdf5','a')

dset = f.require_dataset('auto_chunked', (2048,2048), dtype=np.float32, compression="gzip")

print(dset.compression)

print(dset.compression_opts)

print(dset.chunks)

```

%% Cell type:code id: tags:

``` python

```