TypeError: tuple indices must be integers or slices, not str

[MeisterP]

After updating to v0.2.3 (September 2022)
I get the following error when calling correct_rr(rr):

Cleaning the RR interval time series.
... correcting 3 missed interval(s).
... correcting 1 extra interval(s).
... correcting 89 ectopic interval(s).
... correcting 5 short interval(s).
... correcting 5 long interval(s).
Traceback (most recent call last):
  File "<string>", line 79, in <module>
TypeError: tuple indices must be integers or slices, not str

While trying to disable things one by one to debug (eg. correct_rr(rr, missed_correction=False) I got:

Cleaning the RR interval time series.
... correcting 1 extra interval(s).
... correcting 89 ectopic interval(s).
... correcting 5 short interval(s).
... correcting 5 long interval(s).
Traceback (most recent call last):
  File "<string>", line 77, in <module>
  File "/usr/lib/python3.10/site-packages/systole/correction.py", line 363, in correct_rr
    return _correct_rr(
  File "/usr/lib/python3.10/site-packages/systole/correction.py", line 263, in _correct_rr
    return clean_rr, (nMissed, nExtra, nEctopic, nShort, nLong)
UnboundLocalError: local variable 'nMissed' referenced before assignment

correct_rr(rr, n_iterations=2) doesn't work either:

Traceback (most recent call last):
  File "<string>", line 77, in <module>
TypeError: correct_rr() got an unexpected keyword argument 'n_iterations'

With version 0.2.2 correct_rr(rr, n_iterations=2, missed_correction=False) works as expected without errors.

[LegrandNico]

Thank you for reporting this, I suspect this might be due to artefacts close to each other, can you share the data you are using so I can reproduce the error?

[MeisterP]

My code is failing because the output of the correct_rr() function changed.

import pandas as pd
from systole.correction import correct_rr

df = pd.read_csv('data.csv')
correct = correct_rr(df['R-R'])

with version 0.2.2:

print(correct)
{'clean_rr': array([565., 560., 560., ..., 507., 512., 509.]), 'ectopic': 66, 'short': 40, 'long': 11, 'extra': 17, 'missed': 0}

print(correct['clean_rr'])
[565. 560. 560. ... 507. 512. 509.]

with version 0.2.3:

print(correct)
(array([565., 560., 560., ..., 507., 512., 509.]), (0, 17, 21, 30, 10))

print(correct['clean_rr'])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: tuple indices must be integers or slices, not str

[LegrandNico]

It looks like the function is working correctly but the way it returns the results have changed:
https://embodied-computation-group.github.io/systole/generated/correction/systole.correction.correct_rr.html#systole.correction.correct_rr

It is now returning a tuple like:

clean_rr, (nMissed, nExtra, nEctopic, nShort, Long) = correct_rr(df['R-R'])

If you use the code above it should work. Sorry for this refactoring, the function has been completely rewritten in this release.

[MeisterP]

There's no need to be sorry. Thanks for your work and support. Much appreciated. 👍

Not sure whether I'm misreading the docs... but I can't get arguments to work.

correct = correct_rr(df['R-R'], extra_correction=False)
Cleaning the RR interval time series.
... correcting 21 ectopic interval(s).
... correcting 30 short interval(s).
... correcting 12 long interval(s).
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.10/site-packages/systole/correction.py", line 363, in correct_rr
    return _correct_rr(
  File "/usr/lib/python3.10/site-packages/systole/correction.py", line 263, in _correct_rr
    return clean_rr, (nMissed, nExtra, nEctopic, nShort, nLong)
UnboundLocalError: local variable 'nExtra' referenced before assignment

correct = correct_rr(df['R-R'], n_iterations=2)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: correct_rr() got an unexpected keyword argument 'n_iterations'

[LegrandNico]

I see the error comes from the absence of extra correction that is not correctly accounted for in the code, I will fix that as soon as possible. Thank you for pointing that out.

For the second example, the n_iteration argument has been removed in the new API to better fit with the original paper, so the function will not allow running multiple iterations of the algorithm.