TypeError when using Matplotlib's strpdate2num with Python 3.2

Question

In my current project I want to read some experimental data from a text file into Python using the following code:

import numpy as np
from matplotlib.dates import strpdate2num

data = np.recfromtxt('example.txt',
                 comments='#',
                 delimiter=';',
                 names=('time', 't_ref', 't_s', 't_amb1', 't_amb2', 't_amb3')
                 ,converters={'time': strpdate2num('"%d.%m.%Y %H:%M:%S"')}
                 )

with example.txt looking like

"04.10.2012 08:15:27";14.4;16;12.78;12.65;12.52
"04.10.2012 08:15:37";14.4;16;12.78;12.65;12.5
"04.10.2012 08:15:47";14.4;16;12.78;12.62;12.5
"04.10.2012 08:15:57";14.4;15.9;12.78;12.65;12.52
...

In Python 2.7, all is well, but when I try to transfer the code in 3.2, I get a TypeError from strpdate2num() saying

TypeError: strptime() argument 0 must be str, not <class 'bytes'>

I am fairly new to Python, but my theory is that NumPy somehow stores the time-array internally as byte instead of string, which collides with the stricter handling of both since Python 3.

Long story short, do you have any idea what might be the cause how to fix that?

Answer 1

Here is a workaround:

import numpy as np
import matplotlib.dates as mdates

def bytedate2num(fmt):
    def converter(b):
        return mdates.strpdate2num(fmt)(b.decode('ascii'))
    return converter

date_converter = bytedate2num("%d.%m.%Y %H:%M:%S")

data = np.recfromtxt('example.txt',
                     comments='#',
                     delimiter=';',
                     names=('time', 't_ref', 't_s', 't_amb1', 't_amb2', 't_amb3'),
                     converters={'time': date_converter})