How to read a file with fortran formatted numbers into python?

Question

I have a fortran formatted txt file with numbers like this 0.755473D-08.

if I use np.genfromtxt(data,dtype=None), it shows b'0.755473D-08'. How can I can I get rid of b?

See https://stackoverflow.com/questions/5989351/reading-fortran-double-precision-format-into-python Combine with `converters` in `genfromtxt`. — Vladimir F Героям слава, Apr 25 '18 at 19:51
And this one https://stackoverflow.com/questions/1959210/python-scientific-notation-using-d-instead-of-e — Simon Johansson, Apr 25 '18 at 19:54
Yes, but my problem is I cannot load the fortran formatted numbers into python because it shows `nan`, not unable to replace `D`. — kinder chen, Apr 25 '18 at 20:32
this seems to be specifically about using `genfromtxt` ? Off hand I'd say you need to set the type to string and convert after reading. — agentp, Apr 25 '18 at 20:54
The `b'...'` denotes a bytestring in py3. If you change the 'D' to 'e' ordinary float parsing should work. — hpaulj, Apr 25 '18 at 22:15
@hpaulj I'm a newbie about bytestring, how exactly I should do? I tried `data.replace("D","e")`, but it doesn't work. — kinder chen, Apr 25 '18 at 22:47

score 1 · Answer 1 · answered Apr 26 '18 at 00:10

In [315]: txt = '''0.755473D-08
     ...: 0.755473D-08
     ...: 0.755473D-08'''
In [316]: np.genfromtxt(txt.splitlines())
Out[316]: array([nan, nan, nan])
In [317]: np.genfromtxt(txt.splitlines(),dtype=None)
...
Out[317]: array([b'0.755473D-08', b'0.755473D-08', b'0.755473D-08'], dtype='|S12')

Using String replace:

In [319]: txt.replace('D','e')
Out[319]: '0.755473e-08\n0.755473e-08\n0.755473e-08'
In [320]: np.genfromtxt(_.splitlines())
Out[320]: array([7.55473e-09, 7.55473e-09, 7.55473e-09])

That could also be done line by line (when reading from a file).

With a converter, applied to each element (column per line):

In [326]: np.genfromtxt(txt.splitlines(), converters={0:lambda x:float(x.replace('D','e'))}, encoding=None)
Out[326]: array([7.55473e-09, 7.55473e-09, 7.55473e-09])

It doesn't try to do any further dtype conversion after the converter, so I had to include the float() in the converter.

How to read a file with fortran formatted numbers into python?

1 Answers1