futhark-data 1.0.2

Reading and writing Futhark data files

Python implementation of the Futhark data format

This is a small Python library that implements reading and writing of the textual and binary data format used by Futhark executables and test tools. It is intended to make it easy to write Python scripts that use Python libraries to convert complex formats (e.g. images, audio) into Futhark test data. It provides functions that convert between Numpy values and textual/binary representations in the Futhark formats. Comments are supported. Beware: reading and writing the textual data format is very slow. Use the binary format for all arrays of more than a few hundred elements.

The following Numpy types are supported: np.int8, np.int16, np.in32, np.int64, np.uint8, np.uint16, np.uint32, np.uint64, np.float32, np.float64, np.bool_, as well as up to 255-dimensional arrays containing elements of these sizes.

Installation

$ pip install --user futhark-data

API

Serialising

• dump(v, f, binary=None): Dump v, which must be a Numpy value, to the file-like object f. The parameter binary indicates whether to use binary data format. If None, decide based on the type of f.

• dumps(v): Returns the argument in the textual data format.

• dumpb(v): Returns the argument in the binary data format.

Deserialising

• load(f): Load all values from the file-like object f.

• loads(s): Load all values from the string s.

• loadb(b): Load all values from the byte sequence b.

Since a file (or str, or bytes) can contain any number of Futhark values, the functions above all return generators. Use next if you know for sure there is just a single value, and you want it (see example below).

The functions automatically detect whether the data is encoded using the binary or textual format, so there is no need for the caller to specify.

Examples

>>> import futhark_data
>>> for x in futhark_data.loads('[1,2,3] [4,5,6]'):
...     print(x)
...
[1 2 3]
[4 5 6]
>>> futhark_data.dumpb(next(futhark_data.loads('[1,2,3]')))
b'b\x02\x01 i32\x03\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x02\x00\x00\x00\x03\x00\x00\x00'