SChunk.prefilter#

SChunk.prefilter(input_dtype, output_dtype=None)#

Decorator to set a function as a prefilter.

This function will be executed each time before compressing the data. It will receive three parameters: * the actual data as a ndarray from which to read, * the ndarray to be filled * the offset inside the SChunk instance where the corresponding block begins

(see example below).

Parameters:

input_dtype¶ (np.dtype) – Data type of the input that will receive the prefilter function.
output_dtype¶ (np.dtype) – Data type of the output that will receive and fill the prefilter function. If None (default) it will be input_dtype.

Returns:

out

Return type:

None

Notes

nthreads must be 1 when compressing.
The input_dtype itemsize must be the same as the output_dtype itemsize.

Examples

# Set the compression and decompression parameters
input_dtype = np.dtype(np.int32)
output_dtype = np.dtype(np.float32)
cparams = {"typesize": output_dtype.itemsize, "nthreads": 1}
# Create schunk
schunk = blosc2.SChunk(chunksize=200 * 1000 * input_dtype.itemsize,
                       cparams=cparams)

# Set prefilter with decorator
@schunk.prefilter(input_dtype, output_dtype)
def prefilter(input, output, offset):
    output[:] = input - np.pi