Parse CPT codes and case info from a SIS Complete CSV export

sis-parser.py

#!/usr/local/bin/python3

import csv
import re
import argparse

CASES_OUT = 'sis-parser-out-cases.csv'
CPTS_OUT = 'sis-parser-out-cpts.csv'

parser = argparse.ArgumentParser(prog="sis-parser.py", description='Parse cases and CPT codes from a SIS Complete CSV export.')
parser.add_argument('--input', help='CSV file to parse', required=True)
args = parser.parse_args()

cpts = set()
rows = []

with open(args.input, newline='') as data:
    reader = csv.DictReader(data)
    for row in reader:
        rows.append(row)

for row in rows:
    # Extract CPTs from Procedures col
    procedures = row['Procedures']
    row_cpts = re.findall('[0-9]{5}', procedures)

    # Add CPTs from this case to set
    [cpts.add(cpt) for cpt in row_cpts]

    # Add CPTs col
    row['cpts'] = ', '.join(row_cpts)

# Write data to a new CSV
with open(CASES_OUT, 'w', newline='') as data:
    fieldnames = ['date', 'time', 'physician', 'cpts', 'patient', 'status']
    writer = csv.DictWriter(data, fieldnames=fieldnames)
    
    writer.writeheader()
    [writer.writerow({
        'date': row['\ufeffDateofSurgery'],
        'time': row['CaseStartTime'],
        'physician': row['Physician'],
        'cpts': row['cpts'],
        'patient': row['PatientName'],
        'status': row['CaseStatus']
    }) for row in rows]

# Write CPT codes to their own CSV
with open(CPTS_OUT, 'w', newline='') as data:
    fieldnames = ['code']
    writer = csv.DictWriter(data, fieldnames=fieldnames)

    writer.writeheader()
    [writer.writerow({
        'code': cpt
    }) for cpt in cpts]

print('SUMMARY')
print(f'Rows: {len(rows)}')
print(f'CPTs: {len(cpts)}')
print(f'Wrote data to {CASES_OUT}')
print(f'Wrote CPTs to {CPTS_OUT}')