Hey @fabbra, I have to look into this further but you can use this for now:

record = wfdb.rdrecord(recs[0].split('/')[1], pn_dir='mimic3wdb/'+recs[0])

or for numerics:

record = wfdb.rdrecord(recs[0].split('/')[1]+'n', pn_dir='mimic3wdb/'+recs[0])

This may not work for all the records in recs since some only have a numerics file but this should be a good start. You could do this:

import wfdb
recs = wfdb.get_record_list('mimic3wdb')
for r in recs:
    try:
        record = wfdb.rdrecord(r.split('/')[1], pn_dir='mimic3wdb/'+r)
    except:
        pass
    try:
        record_n = wfdb.rdrecord(r.split('/')[1]+'n', pn_dir='mimic3wdb/'+r)
    except:
        pass

This is ugly, so I'll work on finding a better solution.

Hey @Lucas-Mc, thanks for your prompt reply!
Unfortunately neither of your two suggestions work for me since I always get the following error in wfdb\io\record.py:

   1248     contents = [line.decode('utf-8').strip() for line in response.content.splitlines()]
   1249     version_number = [v for v in contents if 'Version:' in v]
-> 1250     version_number = version_number[0].split(':')[-1].strip().split('<')[0]
   1251
   1252     return version_number

IndexError: list index out of range

Looking forward for a better solution. Thanks a lot in advance.

@fabbra Strange, both work for me on the latest version. I got that error when I misspelled mimic3wdb though. What version are you on?

So far I was using wfdb version 3.1.0 which I obtained via pip install wfdb.
I have now switched to the latest git using pip install git+https://github.com/MIT-LCP/wfdb-python.git (now pip list tells me that I am using version 3.1.1). However this did not solve my problem.

Executing the following lines results in the same IndexError mentioned above:

import wfdb
recs = wfdb.get_record_list('mimic3wdb')
record = wfdb.rdrecord(recs[0].split('/')[1], pn_dir='mimic3wdb/'+recs[0])

when i run this:

`cwd = os.getcwd()
dl_dir = os.path.join(cwd, 'tmp_dl_dir')

wfdb.dl_database('mitdb', dl_dir=dl_dir)`

got
ValueError: The database https://physionet.org/files/mitdb/1.0.0/1.0.0 has no WFDB files to download

using wfdb3.1.0: 404 error, there is a %5C problem
update to 3.1.1 , got ValueError above.

Sounds like a Windows thing, I really need to get my Windows computer set up so I can test on both OSs!! Anyway, I'll try and find it ... Do you have a stack trace? What is recs[0]?

Sounds like a Windows thing,

I confirm! When running the same commands in the Unix subsystem (Ubuntu) on my Win64 machine this works without errors.

Do you have a stack trace?

In [1]: import wfdb

In [2]: recs = wfdb.get_record_list('mimic3wdb')

In [3]: record = wfdb.rdrecord(recs[0].split('/')[1], pn_dir='mimic3wdb/'+recs[0])
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-3-1b0d25728761> in <module>
----> 1 record = wfdb.rdrecord(recs[0].split('/')[1], pn_dir='mimic3wdb/'+recs[0])

~\venv\Lib\site-packages\wfdb\io\record.py in rdrecord(record_name, sampfrom, sampto, channels, physical, pn_dir, m2s, smooth_frames, ignore_skew, return_res, force_channels, channel_names, warn_empty)
   2740     if (pn_dir is not None) and ('.' not in pn_dir):
   2741         dir_list = pn_dir.split(os.sep)
-> 2742         pn_dir = posixpath.join(dir_list[0], get_version(dir_list[0]), *dir_list[1:])
   2743
   2744     if record_name.endswith('.edf'):

~\venv\Lib\site-packages\wfdb\io\record.py in get_version(pn_dir)
   1248     contents = [line.decode('utf-8').strip() for line in response.content.splitlines()]
   1249     version_number = [v for v in contents if 'Version:' in v]
-> 1250     version_number = version_number[0].split(':')[-1].strip().split('<')[0]
   1251
   1252     return version_number

IndexError: list index out of range

What is recs[0]?

In [4]: recs[0]
Out[4]: '30/3000003/'

Hope that helps to further debug the thing...

Hey @fabbra, I just got my Windows machine to reproduce the error... let the debugging begin!

Hey @fabbra, the issue was very subtle and not picked up on the stack trace with this line:

Silly me using os.sep for a URL instead of file path.. update coming soon!

@Lucas-Mc, thanks that works for rdrecord(), however, aren't there are other occurences of dir_list = pn_dir.split(os.sep) which need to be fixed, e.g. in rdheader()?

Hey @fabbra, yes I have to look around at other instances and will fix them in a separate pull request if they occur. Thanks for reminding me!

@Lucas-Mc, thanks that works for rdrecord(), however, aren't there are other occurences of dir_list = pn_dir.split(os.sep) which need to be fixed, e.g. in rdheader()?

wfdb-python/wfdb/io/record.py

Line 2571 in 4d4c354

dir_list = pn_dir.split(os.sep)

It's not the only occurence but just one example...

just found another occurence which might need a fix:

Similar question on the same subject, is there a way to iterate through all records of the MIMIC-III Waveform Database Matched Subset?

The method suggested by you above does not work since the filename of the header has a postfix, i.e. it does not correspond exactly to the name in the RECORDS file, e.g. for p000020 the header file is not called p000020.hea but p000020-2183-04-28-17-47.hea

A potential way might be to read a different records file, e.g. RECORDS-waveforms but that doesn't work using wfdb.get_record_list(), does it?

Hi,
Can this issue be kept open till it is totally resolved? I just came across the error again while using wfdb.rdann(). I think if I understand this right, I see atleast 8 direct instances just by doing a search of pn_dir.split and only one of them has been changed in the master right now.