file_sieve.py

#!/usr/bin/env python
"Copies/moves files from the list into the separate folder, CC-BY: hasherezade"

__VERSION__ = "0.1"

import sys
import os
import re
import argparse

def get_names(fname):
    names = set()
    with open(fname, 'r') as f:
        for line in f.readlines():
            line = line.strip()
            names.add(line)
    return names

def remove_ext(filename):
    index = filename.find('.')
    if index == (-1):
        return filename
    return filename[0:index]

def get_matching_files(names, dir_content):
    matching = set()

    for fname in dir_content:
        noext = remove_ext(fname)
        if noext in names:
            matching.add(fname)
            continue

        for name in names:
            if fname.find(name) != (-1):
                #print "%s , %s , %d" % (noext, name, index)
                matching.add(fname)
                break
    return matching 

def is_windows():
    from sys import platform as _platform
    if "win" in _platform:
        return True
    return False

def get_move_command():
    command = "mv"
    if is_windows():
        command = "move"
    return command

def get_copy_command():
    command = "cp"
    if is_windows():
        command = "copy"
    return command

def copy_found(found, indir, outdir, is_move=False):
    command = get_copy_command()
    if is_move == True:
        command = get_move_command()
        print "moving files!"
    for filen in found:
        path1 = indir + "/" + filen
        path2 = outdir + "/" + filen    
        os.system(command + " " + path1 + " " + path2)

def main():
    parser = argparse.ArgumentParser(description="File sieve "+ __VERSION__)
    parser.add_argument('--files', dest="files", default=None, help="Input file with list of files to be copied/moved", required=True)
    parser.add_argument('--indir', dest="indir", default=None, help="Input directory with list of files to be copied/moved", required=True)
    parser.add_argument('--outdir', dest="outdir", default=None, help="Output directory where the files should be copied/moved", required=True)
    parser.add_argument('--move', dest="move", default="False", action='store_true', help="Move instead of coping?")
    args = parser.parse_args()

    names = get_names(args.files)

    dir_content = set(os.listdir(args.indir))
    matching = get_matching_files(names, dir_content)
    copy_found(matching, args.indir, args.outdir, args.move)

    print "Names: %d" % len(names)
    print "Files: %d" % len(dir_content)
    print "Found: %d" % len(matching)
    print "----"

if __name__ == "__main__":
    sys.exit(main())