Wrong dtype when reading old TDT files (tev) with new tdtio? #198
Description
Hello,
I'm trying to read a TDT file with the 0.3.X branch (which seems to be the default when install neo with pip, I have 0.3.3). Neo does not crash but the data does not look at all correct. I think the data is read as float64 whatever the actual dtype is (the same file could be read with the master branch).
Part of the problem might come from in tdtio.py L225:
In [12]: print anaSig.lazy_dtype
int16
In [13]: print anaSig.dtype
float64
In [14]: v[channel] = anaSig.duplicate_with_new_array(np.zeros((anaSig.lazy_shape) ,
dtype = anaSig.lazy_dtype)*pq.V )
In [15]: print v[channel].dtype
float64v[channel].dtype is not a anaSig.lazy_dtype. this seems to be because of the pq.V:
In [17]: v[channel] = anaSig.duplicate_with_new_array(np.zeros((anaSig.lazy_shape) ,
dtype = anaSig.lazy_dtype) )
In [18]: print v[channel].dtype
int16I tried to simply remove the pq.V just to try, but it's obviously not the solution since I run into some problem downstream (traceback below). What should be the dtype of v[channel]? Is it linked to the tev file? (does it work with a sev file)?
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-6-6f5fc22ee99c> in <module>()
----> 1 bl = tdtreader.read_block()
/home/blota/anaconda/envs/ip_notebook/lib/python2.7/site-packages/neo/io/tdtio.py in read_block(self, lazy, cascade)
285 s = int((h['size']*4-40)/dt.itemsize)
286 a.fid.seek(h['eventoffset'])
--> 287 a[ a.pos:a.pos+s ] = np.fromstring( a.fid.read( s*dt.itemsize ), dtype = a.dtype)
288 a.pos += s
289
ValueError: could not broadcast input array from shape (2048) into shape (512)