Am 2/4/25 um 01:26 schrieb Matthias Klumpp:
Am Mo., 3. Feb. 2025 um 18:45 Uhr schrieb Alois Schlögl
<alois.schlo...@gmail.com>:
[...]
Dear Matthias,
thanks for your offer to help. From my point of view, this would be
appreciated.
Please find below my assessment of the state of stimfit in debian:
- It should be possible bring stimfit 0.16.4-1.1 into
bookworm-backports
as is.
- Concerning sid, this might be tricky because of the current
dependency
on the old distutils. At this time, I'm not positive that I could
do the
migration to numpy>=2 and the new distutils within the time frame
of the
release of trixie. If you know how to address this, your
contribution is
most welcome.
Oh wow, this is a lot more than I expected, and that also explains why
the package isn't in unstable yet...
I can have a look though!
- What is possible is disabling the embedded python (using
"./configure
--disable-python ..."). That's what we call "stimfit-lite" on the
project homepage https://github.com/neurodroid/stimfit . What I can
offer is preparing an upload for a stimfit-lite (w/o python
support) for
Debian.
- This would mean that the debian package "python3-stfio" would
not be
available. Some of its functionality could be replaced by
python3-biosig
<https://packages.debian.org/unstable/python3-biosig> . [1] provides
some starting points how to do this. However, full compatibility
is not
available at this time.
Our lab uses python3-stfio specifically to load CED binary CFS files,
which as far as I know can't be reliably done with any other tool. So,
it would be neat to keep that support. Migrating Python build systems
is just about the least fun thing I can imagine, but the Numpy
transition might not be too bad depending on how stimfit uses it (I
have done that in a different project and it was simple, however,
usage of the API was also fairly basic there).
I'll have a look...
Best,
Matthias
--
I welcome VSRE emails. See http://vsre.info/
You should be able to load your CED binary CFS file with the attached
script demo2.py when adapting FILENAME and installing biosig for python
like this:
sudo apt install python3-biosig
or
sudo apt install libbiosig-dev
# set up your python env
pip install biosig
# adapt FILENAME in demo2.py and run
python demo2.py
If it does not work, let me know
Alois
####### Demo for Python interface to BioSig" #####################
###
### Copyright (C) 2009,2016 Alois Schloegl <alois.schlo...@ist.ac.at>
### This file is part of the "BioSig for Python" repository
### at http://biosig.sf.net/
###
##############################################################
# download and extract
# http://www.biosemi.com/download/BDFtestfiles.zip
# into /tmp/
# then run this demo
#
# on linux you can run instead
# make test
import sys
import biosig
import numpy as np
import matplotlib.pyplot as plt
import json
def printf(format, *args):
sys.stdout.write(format % args)
## read header
FILENAME="/home/schloegl/data/test/cfs/BaseDemo/Leak.cfs"
FILENAME="/home/schloegl/data/test/cfs/20190417_A0AA.dat"
FILENAME="/home/schloegl/data/test/cfs/minis.dat"
FILENAME="data/Newtest17-256.bdf"
HDR=biosig.header(FILENAME)
#print HDR
## extracting header fields
H=json.loads(HDR)
print(H["Filename"])
print(H["TYPE"])
print(H["Samplingrate"])
Fs=H["Samplingrate"]
NS=len(H["CHANNEL"]) # number of channels
T0=H["StartOfRecording"]
print(Fs,NS,T0)
### read and display data ###
A=biosig.data(FILENAME)
NS=np.size(A,1) # number of channels
fig = plt.figure()
ax = fig.add_subplot(111)
h = plt.plot(np.arange(np.size(A,0))/H["Samplingrate"],A[:,0]);
ax.set_xlabel('time [s]')
# plt.show()
rec={"dt" : 1000.0/Fs, "xunits": "ms", "channel" : [] }
for chan in list(range(NS)):
name=H["CHANNEL"][chan]["Label"] # name of channel
unit=H["CHANNEL"][chan]["PhysicalUnit"] # units of channel
printf("#%d:\t%s\t[%s]\n", chan, name, unit)
### get all sweeps, breaks-in-recording
selpos=[0]
if hasattr(H,"EVENT"):
for k in list(range(len(H["EVENT"]))):
if (H["EVENT"][k]["TYP"] == '0x7ffe'):
selpos.append(round(H["EVENT"][k]["POS"]*H["Samplingrate"]))
selpos.append(np.size(A,0))
### data from channel m and sweep c can be obtained by:
m=0 # channel
c=0 # segment, trace, sweep number
d = A[selpos[c]:selpos[c+1]-1,m]
### display data
fig = plt.figure()
k=0
for m in list(range(NS)):
for c in list(range(len(selpos)-1)):
k = k+1
ax = fig.add_subplot(NS,len(selpos)-1,k)
d = A[selpos[c]:selpos[c+1]-1,m]
h = plt.plot(np.arange(np.size(d,0))/H["Samplingrate"],d)
# TODO:
# REC[m].append((c,d));
# rec["channel"].append((chan, [{"name": name, "yunits": unit, 'section': []}]))
# rec["channel"][m]["section"].insert(c,d)
plt.show()
