Filtering failure while trying to index

[jchutrue]

See error and stack trace below:

---

IndexError Traceback (most recent call last)
in ()
4 fs=30000,
5 fl=600,
----> 6 fh=6000)
7 filt_data_ch0 = np.fromfile('/home/jchu/data/install/long-recording/experiments/07-09-2017/filt-ch.00.raw',
8 np.int16)

~/code/jagular/jagular/filtering.py in filtfilt_mmap(timestamps, finname, foutname, fs, fl, fh, gpass, gstop, dtype, ftype, buffer_len, overlap_len, max_len, **kwargs)
109 overlap_len=overlap_len,
110 max_len=max_len,
--> 111 **kwargs)
112 return y
113

~/code/jagular/jagular/filtering.py in filtfilt_within_epochs_mmap(timestamps, finname, foutname, dtype, sos, buffer_len, overlap_len, max_len, filter_epochs, **kwargs)
178 assume_sorted=assume_sorted,
179 step=step,
--> 180 index=True)
181
182 for (start, stop) in filter_epochs:

~/code/jagular/jagular/utils.py in _get_contiguous_segments_fast(data, step, assume_sorted, index, inclusive)
277 starts = np.insert(breaks+1, 0, 0)
278 stops = np.append(breaks, len(data)-1)
--> 279 bdries = np.vstack((data[starts], data[stops] + step)).T
280 if index:
281 if inclusive:

IndexError: arrays used as indices must be of integer (or boolean) type

[eackermann]

@jchutrue can you inspect starts and / or stops ? that would be the useful thing to see here...

[eackermann]

Could be related to #28 ? What do your timestamps look like?

[jchutrue]

starts: [ 0.]
stops: [ 59860882.]

They are floats. However, the timestamps are uint32

[eackermann]

Should be fixed now. Could you test again after pulling the latest? And close the issue if fixed.

[jchutrue]

filtfilt_within_epochs_mmap() is supposed to call get_contiguous_segments() but for some reason the stack trace ends up with an error from _get_contiguous_segments_fast(), so it still fails

[eackermann]

Ok, astype(int) was applied in the wrong place. Now fixed.