cmap

#!/usr/bin/env python3

"""
[c]olumn [map]per --- applies a subcommand to each column of a delimited stream. The provided subcommand must necessarily provide one line of stdout for each line consumed over stdin without buffering.

Usage:
    paste <(echo "1+1") <(echo "2+2") <(echo "3+3") | cmap bc

Original author: Elijah Rippeth (@erip)
"""

import subprocess

from signal import signal, SIGPIPE, SIG_DFL
from argparse import ArgumentParser, FileType, REMAINDER

# Silence broken pipes when downstream stops reading; e.g., piping to `head`
signal(SIGPIPE, SIG_DFL)

def setup_argparse():
    parser = ArgumentParser()
    parser.add_argument("-f", "--fields", default=None, required=False, help="The comma-delimited field numbers (1-indexed) to which to apply the subcommand")
    parser.add_argument("-d", "--delimiter", default="\t", help="The delimiter on which columns are read and written")
    parser.add_argument("-i", "--input", type=FileType("r"), default="-", help="The input stream")
    parser.add_argument("-o", "--output", type=FileType("w"), default="-", help="The output stream")
    parser.add_argument("command", nargs=REMAINDER, help="The subcommand to be applied to all columns. Must necessarily produce one line of stdout for each line consumed over stdin without buffering.")
    return parser

if __name__ == "__main__":
    args = setup_argparse().parse_args()
    if args.fields:
        rel_fields = [int(e)-1 for e in args.fields.split(",")]

    proc = subprocess.Popen(args.command, stdin=subprocess.PIPE, stdout=subprocess.PIPE, bufsize=0)
    with args.input as fin, args.output as fout:
        for line in map(str.strip, fin):
            columns = line.split(args.delimiter)
            if args.fields:
                columns = [columns[f] for f in rel_fields]
            output = []
            for i, col in enumerate(columns):
                proc.stdin.write((col+"\n").encode())
                proc.stdin.flush()
                stdout = proc.stdout.readline()
                output.append(stdout.decode().strip())
            print(args.delimiter.join(output), file=fout)