annotation-ai/gonpy

Read and write Numpy binary files (.npy files) in Golang

gonpy : Read Numpy binary files (.npy files) in Golang

gonpy reads and writes Numpy binary array data to/from Go slices

The npy file specification is here:

https://www.numpy.org/devdocs/reference/generated/numpy.lib.format.html

The documentation for this package can be found here:

https://godoc.org/github.com/kshedden/gonpy

When reading a multidimensional array, the data are returned as a one-dimensional slice that has been flattened as specified by the ColumnMajor field in the NpyReader struct. If ColumnMajor is true, the array is flattened in column major order. If ColumnMajor is false, the array is flattened in row major order.

The writer defaults to writing the data as a vector. The Shape field of the NpyWriter struct can be used to set other shapes.

Unsigned numeric data types, fixed-width string types, compound dtypes, record arrays, and time dtypes are not supported (file an issue if any of these are needed). Python object types are of course not supported.

The following example shows how to read a npy file into a slice of float64 values (for clarity error handling is omitted).

r, _ := gonpy.NewFileReader("data.npy")
data, _ := r.GetFloat64()

The reader value has fields such sas r.Shape that provide additional information about the array.

The following example shows how to write a slice of int32 values to a npy file (again, error handling is omitted).

w, _ := gonpy.NewFileWriter("data.npy")
_ = w.WriteFloat64(data)

To specify a shape or other attributes, modify the writer object before writing the array, for example:

w, _ := gonpy.NewFileWriter("data.npy")
w.Shape = []int32{50, 2}
w.Version = 2
_ = w.WriteFloat64(data)

To write to a stream, say a gzip stream, use the following:

f, _ := os.Create("data.npy.gz")
g := gzip.NewWriter(f)
w, _ := gonpy.NewWriter(g)
_ = w.WriteFloat64(data)
f.Close()