convert matlab to python for reading a binary file

Question

Im trying to convert a matlab script into python. The script read a binary file and reshape into a number of columns. The matlab script is:

fid=fopen(binary_file,'rb');
[inpar,ic]=fread(fid,4,'int');
if (ic<4) ; idata=[];return;end
nmagic=inpar(1);
nh=inpar(2);
nrpar=inpar(3);
nipar=inpar(4);


[rdata,ic]=fread(binary_file,[nh,nrpar],'float');
if (ic<nh*nrpar) ; return;end
[idata,ic]=fread(fid,[nh,nipar],'int');
if (ic<nh*nipar) ; return;end

The python code I tried is:

import numpy as np 

inpar = np.fromfile(fid, dtype=np.int32)
nmagic, nh, nrpar, nipar = inpar

rdata = np.fromfile(fid, dtype=np.float32, count=nh * nrpar).reshape(nh, nrpar)
idata = np.fromfile(fid, dtype=np.int32, count=nh * nipar).reshape(nh, nipar)

I don't exactly know how Matlab is reshaping the data. Can anyone help me translate the code to achieve the same result in Python. A sample data is given here

Answer 1

Lets try to read the data with open in rb mod so it will be read in binary mode. Then we will use np.fromfile to read from the file.

import numpy as np

with open(binary_file, 'rb') as fid:
    inpar = np.fromfile(fid, dtype=np.int32, count=4)
    if inpar.size < 4:
        idata = []
    else:
        nmagic, nh, nrpar, nipar = inpar

        rdata = np.fromfile(fid, dtype=np.float32, count=nh * nrpar)
        if rdata.size < nh * nrpar:
            # handle error
            pass
        else:
            rdata = rdata.reshape(nh, nrpar)

        idata = np.fromfile(fid, dtype=np.int32, count=nh * nipar)
        if idata.size < nh * nipar:
            # handle error
            pass
        else:
            idata = idata.reshape(nh, nipar)