4434: Merge branch 'master' into 4434-collation

[arvados.git] / sdk / python / bin / arv-normalize

#!/usr/bin/env python

import argparse
import hashlib
import os
import re
import string
import sys

parser = argparse.ArgumentParser(
    description='Read manifest on standard input and put normalized manifest on standard output.')

parser.add_argument('--extract', type=str, help="The file to extract from the input manifest")
parser.add_argument('--strip', action='store_true', help="Strip authorization tokens")

args = parser.parse_args()

import arvados

r = sys.stdin.read()

cr = arvados.CollectionReader(r)
cr.normalize()

if args.extract:
    i = args.extract.rfind('/')
    if i == -1:
        stream = '.'
        fn = args.extract
    else:
        stream = args.extract[:i]
        fn = args.extract[(i+1):]
    for s in cr.all_streams():
        if s.name() == stream:
            if fn in s.files():
                sys.stdout.write(s.files()[fn].as_manifest())
else:
    sys.stdout.write(cr.manifest_text(args.strip))