argparse_fofn.py

import argparse

def get_parser():
    parser = argparse.ArgumentParser(version=__version__,
                                     formatter_class=argparse.ArgumentDefaultsHelpFormatter)

    parser.add_argument('input_fofn', help="Path to bas.fofn",
                        type=validate_fofn)

    return parser

fofn_utils.py

import os
import sys
import logging
import functools
import time

log = logging.getLogger(__name__)


def _validate_resource(func, resource):
    """Validate the existence of a file/dir"""
    if func(resource):
        return os.path.abspath(resource)
    else:
        raise IOError("Unable to find {f}".format(f=resource))

validate_file = functools.partial(_validate_resource, os.path.isfile)
validate_dir = functools.partial(_validate_resource, os.path.isdir)
validate_output_dir = functools.partial(_validate_resource, os.path.isdir)


def validate_fofn(fofn):
    """Validate existence of FOFN and files within the FOFN.

    :param fofn: (str) Path to File of file names.
    :raises: IOError if any file is not found.
    :return: (str) abspath of the input fofn
    """
    if os.path.isfile(fofn):
        file_names = fofn_to_files(os.path.abspath(fofn))
        log.debug("Found {n} files in FOFN {f}.".format(n=len(file_names), f=fofn))
        return os.path.abspath(fofn)
    else:
        raise IOError("Unable to find {f}".format(f=fofn))


def fofn_to_files(fofn):
    """Util func to convert a bas/bax fofn file to a list of bas/bax files."""
    if os.path.exists(fofn):
        with open(fofn, 'r') as f:
            bas_files = [line.strip() for line in f.readlines()]

        for bas_file in bas_files:
            if not os.path.isfile(bas_file):
                raise IOError("Unable to find bas/bax file '{f}'".format(f=bas_file))

        if len(set(bas_files)) != len(bas_files):
            raise IOError("Detected duplicate files in fofn. {n} unique, {m} total files in {f}".format(f=fofn, m=len(bas_files), n=len(set(bas_files))))

        return bas_files
    else:
        raise IOError("Unable to find FOFN {f}".format(f=fofn))