msg_compare.py

import os
from hl7apy.parser import parse_message


input_folder_prod = ''
input_folder_test = ''


def get_fnames(folder):
    ret = []
    for f in os.listdir(folder):
        f_path = os.path.join(folder, f)
        if os.path.isfile(f_path):
            ret.append(f_path)
    return ret

input_files_prod = get_fnames(input_folder_prod)
input_files_test = get_fnames(input_folder_test)

def parse_msg(f_path):
    with open(f_path, 'r') as f:
        raw_hl7 = f.read()
    hl7_msg = parse_message(raw_hl7, find_groups=False)
    return hl7_msg

def get_ids(input_files):
    ret = []
    for f_path in input_files:
        hl7_msg = parse_msg(f_path)
        msg_id = hl7_msg.msh.msh_10.value
        ret.append(msg_id)
    return ret

input_ids_prod = get_ids(input_files_prod)
input_ids_test = get_ids(input_files_test)

def compare_msg_ids(ids_prod, ids_test):
    only_prod = []
    only_test = []
    both = []
    
    for id_prod in ids_prod:
        if id_prod in ids_test:
            both.append(id_prod)
        else:
            only_prod.append(id_prod)
    
    for id_test in ids_test:
        if id_test in ids_prod:
            if id_test not in both:
                both.append(id_test)
        else:
            only_test.append(id_test)
    
    return both, only_prod, only_test

results = compare_msg_ids(input_ids_prod, input_ids_test)
both, only_prod, only_test = results

def output(f_path, ids):
    with open(f_path, 'w+') as f:
        for _id in ids:
            f.write(_id + '\n')

output_folder = './'
file_pre = 'wdp_'
output_file_both = os.path.join(output_folder, file_pre + 'both.csv')
output_file_only_prod = os.path.join(output_folder, file_pre + 'only_prod.csv')
output_file_only_test = os.path.join(output_folder, file_pre + 'only_test.csv')

output(output_file_both, both)
output(output_file_only_prod, only_prod)
output(output_file_only_test, only_test)