loinc_search.py

#!/usr/bin/env python
"""
Goal: search multiple LOINC codes in a reference file
"""
import argparse
import sys
import pandas as pd
# from contextlib import redirect_stdout
# import io
pd.set_option('display.width', 2500)

REF_FILE = "/inbox/processing/loinc/loinc_261_core/LoincTableCore.csv"
# USECOLS = ["LOINC_NUM", "COMPONENT","PROPERTY","TIME_ASPCT","SYSTEM",
#"SCALE_TYP","METHOD_TYP","CLASS","CLASSTYPE","LONG_COMMON_NAME","SHORTNAME"]
USECOLS = ["LOINC_NUM", "COMPONENT", "PROPERTY", "TIME_ASPCT", "SYSTEM",
           "SCALE_TYP", "METHOD_TYP", "LONG_COMMON_NAME"]


def run(args):
    """ Merge two frames """
    df_ref = pd.read_csv(args.ref_file, sep=",", usecols=USECOLS)
    # print("Using reference: {}".format(args.ref_file))
    # print("Reference header: \n{}".format(df_ref.head()))
    df_ref.fillna('', inplace=True)
    df = pd.read_csv(args.input_file, sep=",",
                     header=None, names=["LOINC_NUM"])

    merged = pd.merge(df, df_ref, how="inner")
    merged.to_csv(args.output_file, sep="\t", index=False, encoding="utf-8")

    if sys.stdout != args.output_file:
        print("Result was saved to: {}".format(args.output_file))

    # with io.StringIO() as buf, redirect_stdout(buf):
    #     data = buf.getvalue()


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="cat data | %prog [args]")

    parser.add_argument("-r", "--reference", dest="ref_file",
                        default=REF_FILE)
    parser.add_argument("-i", "--input", dest="input_file",
                        default=sys.stdin)
    parser.add_argument("-o", "--output", dest="output_file",
                        default=sys.stdout)

    parsed_args = parser.parse_args()

    if sys.stdin.isatty():
        parser.print_usage()
        print("for more help use --help")
        sys.exit(1)
    run(parsed_args)