Check status of AVL combined QC records and transcripts' existence

avl_qc_transcript_status.py

# execute it from within /data/predict1/data_from_nda/Prescient/PHOENIX/GENERAL

from os.path import dirname, join as pjoin
import pandas as pd
from glob import glob

files=glob('*/processed/*/interviews/open/*_combinedQCRecords.csv')

j=0
for file in files:
    df=pd.read_csv(file)
    day=df.loc[0,'day']
    template=pjoin(dirname(file),f'transcripts/*_interviewAudioTranscript_open_day{day:04}_session*_REDACTED.txt')
    transcript_file=glob(template)

    if len(transcript_file)==1:
        # print(transcript_file[0])
        j+=1
    elif len(transcript_file)>1:
         print(transcript_file)
    else:
        print(template, 'does not exist')

print('Total combinedQCRecords.csv', len(files))
print('Total transcript files',j)