sdk/python/arvados/commands/keepdocker.py

   1 #!/usr/bin/env python
   2
   3 import argparse
   4 import datetime
   5 import errno
   6 import json
   7 import os
   8 import subprocess
   9 import sys
  10 import tarfile
  11 import tempfile
  12
  13 from collections import namedtuple
  14 from stat import *
  15
  16 import arvados
  17 import arvados.commands._util as arv_cmd
  18 import arvados.commands.put as arv_put
  19
  20 STAT_CACHE_ERRORS = (IOError, OSError, ValueError)
  21
  22 DockerImage = namedtuple('DockerImage',
  23                          ['repo', 'tag', 'hash', 'created', 'vsize'])
  24
  25 keepdocker_parser = argparse.ArgumentParser(add_help=False)
  26 keepdocker_parser.add_argument(
  27     '-f', '--force', action='store_true', default=False,
  28     help="Re-upload the image even if it already exists on the server")
  29
  30 _group = keepdocker_parser.add_mutually_exclusive_group()
  31 _group.add_argument(
  32     '--pull', action='store_true', default=False,
  33     help="Try to pull the latest image from Docker registry")
  34 _group.add_argument(
  35     '--no-pull', action='store_false', dest='pull',
  36     help="Use locally installed image only, don't pull image from Docker registry (default)")
  37
  38 keepdocker_parser.add_argument(
  39     'image', nargs='?',
  40     help="Docker image to upload, as a repository name or hash")
  41 keepdocker_parser.add_argument(
  42     'tag', nargs='?', default='latest',
  43     help="Tag of the Docker image to upload (default 'latest')")
  44
  45 # Combine keepdocker options listed above with run_opts options of arv-put.
  46 # The options inherited from arv-put include --name, --project-uuid,
  47 # --progress/--no-progress/--batch-progress and --resume/--no-resume.
  48 arg_parser = argparse.ArgumentParser(
  49         description="Upload or list Docker images in Arvados",
  50         parents=[keepdocker_parser, arv_put.run_opts, arv_cmd.retry_opt])
  51
  52 class DockerError(Exception):
  53     pass
  54
  55
  56 def popen_docker(cmd, *args, **kwargs):
  57     manage_stdin = ('stdin' not in kwargs)
  58     kwargs.setdefault('stdin', subprocess.PIPE)
  59     kwargs.setdefault('stdout', sys.stderr)
  60     try:
  61         docker_proc = subprocess.Popen(['docker.io'] + cmd, *args, **kwargs)
  62     except OSError:  # No docker.io in $PATH
  63         docker_proc = subprocess.Popen(['docker'] + cmd, *args, **kwargs)
  64     if manage_stdin:
  65         docker_proc.stdin.close()
  66     return docker_proc
  67
  68 def check_docker(proc, description):
  69     proc.wait()
  70     if proc.returncode != 0:
  71         raise DockerError("docker {} returned status code {}".
  72                           format(description, proc.returncode))
  73
  74 def docker_images():
  75     # Yield a DockerImage tuple for each installed image.
  76     list_proc = popen_docker(['images', '--no-trunc'], stdout=subprocess.PIPE)
  77     list_output = iter(list_proc.stdout)
  78     next(list_output)  # Ignore the header line
  79     for line in list_output:
  80         words = line.split()
  81         size_index = len(words) - 2
  82         repo, tag, imageid = words[:3]
  83         ctime = ' '.join(words[3:size_index])
  84         vsize = ' '.join(words[size_index:])
  85         yield DockerImage(repo, tag, imageid, ctime, vsize)
  86     list_proc.stdout.close()
  87     check_docker(list_proc, "images")
  88
  89 def find_image_hashes(image_search, image_tag=None):
  90     # Given one argument, search for Docker images with matching hashes,
  91     # and return their full hashes in a set.
  92     # Given two arguments, also search for a Docker image with the
  93     # same repository and tag.  If one is found, return its hash in a
  94     # set; otherwise, fall back to the one-argument hash search.
  95     # Returns None if no match is found, or a hash search is ambiguous.
  96     hash_search = image_search.lower()
  97     hash_matches = set()
  98     for image in docker_images():
  99         if (image.repo == image_search) and (image.tag == image_tag):
 100             return set([image.hash])
 101         elif image.hash.startswith(hash_search):
 102             hash_matches.add(image.hash)
 103     return hash_matches
 104
 105 def find_one_image_hash(image_search, image_tag=None):
 106     hashes = find_image_hashes(image_search, image_tag)
 107     hash_count = len(hashes)
 108     if hash_count == 1:
 109         return hashes.pop()
 110     elif hash_count == 0:
 111         raise DockerError("no matching image found")
 112     else:
 113         raise DockerError("{} images match {}".format(hash_count, image_search))
 114
 115 def stat_cache_name(image_file):
 116     return getattr(image_file, 'name', image_file) + '.stat'
 117
 118 def pull_image(image_name, image_tag):
 119     check_docker(popen_docker(['pull', '-t', image_tag, image_name]), "pull")
 120
 121 def save_image(image_hash, image_file):
 122     # Save the specified Docker image to image_file, then try to save its
 123     # stats so we can try to resume after interruption.
 124     check_docker(popen_docker(['save', image_hash], stdout=image_file),
 125                  "save")
 126     image_file.flush()
 127     try:
 128         with open(stat_cache_name(image_file), 'w') as statfile:
 129             json.dump(tuple(os.fstat(image_file.fileno())), statfile)
 130     except STAT_CACHE_ERRORS:
 131         pass  # We won't resume from this cache.  No big deal.
 132
 133 def prep_image_file(filename):
 134     # Return a file object ready to save a Docker image,
 135     # and a boolean indicating whether or not we need to actually save the
 136     # image (False if a cached save is available).
 137     cache_dir = arv_cmd.make_home_conf_dir(
 138         os.path.join('.cache', 'arvados', 'docker'), 0o700)
 139     if cache_dir is None:
 140         image_file = tempfile.NamedTemporaryFile(suffix='.tar')
 141         need_save = True
 142     else:
 143         file_path = os.path.join(cache_dir, filename)
 144         try:
 145             with open(stat_cache_name(file_path)) as statfile:
 146                 prev_stat = json.load(statfile)
 147             now_stat = os.stat(file_path)
 148             need_save = any(prev_stat[field] != now_stat[field]
 149                             for field in [ST_MTIME, ST_SIZE])
 150         except STAT_CACHE_ERRORS + (AttributeError, IndexError):
 151             need_save = True  # We couldn't compare against old stats
 152         image_file = open(file_path, 'w+b' if need_save else 'rb')
 153     return image_file, need_save
 154
 155 def make_link(api_client, num_retries, link_class, link_name, **link_attrs):
 156     link_attrs.update({'link_class': link_class, 'name': link_name})
 157     return api_client.links().create(body=link_attrs).execute(
 158         num_retries=num_retries)
 159
 160 def ptimestamp(t):
 161     s = t.split(".")
 162     if len(s) == 2:
 163         t = s[0] + s[1][-1:]
 164     return datetime.datetime.strptime(t, "%Y-%m-%dT%H:%M:%SZ")
 165
 166 def list_images_in_arv(api_client, num_retries):
 167     existing_links = api_client.links().list(
 168         filters=[['link_class', 'in', ['docker_image_hash', 'docker_image_repo+tag']]]
 169         ).execute(num_retries=num_retries)['items']
 170     images = {}
 171     for link in existing_links:
 172         collection_uuid = link["head_uuid"]
 173         if collection_uuid not in images:
 174             images[collection_uuid]= {"dockerhash": "<none>",
 175                       "repo":"<none>",
 176                       "tag":"<none>",
 177                       "timestamp": ptimestamp("1970-01-01T00:00:01Z")}
 178
 179         if link["link_class"] == "docker_image_hash":
 180             images[collection_uuid]["dockerhash"] = link["name"]
 181
 182         if link["link_class"] == "docker_image_repo+tag":
 183             r = link["name"].split(":")
 184             images[collection_uuid]["repo"] = r[0]
 185             if len(r) > 1:
 186                 images[collection_uuid]["tag"] = r[1]
 187
 188         if "image_timestamp" in link["properties"]:
 189             images[collection_uuid]["timestamp"] = ptimestamp(link["properties"]["image_timestamp"])
 190         else:
 191             images[collection_uuid]["timestamp"] = ptimestamp(link["created_at"])
 192
 193     st = sorted(images.items(), lambda a, b: cmp(b[1]["timestamp"], a[1]["timestamp"]))
 194
 195     fmt = "{:30}  {:10}  {:12}  {:29}  {:20}"
 196     print fmt.format("REPOSITORY", "TAG", "IMAGE ID", "COLLECTION", "CREATED")
 197     for i, j in st:
 198         print(fmt.format(j["repo"], j["tag"], j["dockerhash"][0:12], i, j["timestamp"].strftime("%c")))
 199
 200 def main(arguments=None):
 201     args = arg_parser.parse_args(arguments)
 202     api = arvados.api('v1')
 203
 204     if args.image is None or args.image == 'images':
 205         list_images_in_arv(api, args.retries)
 206         sys.exit(0)
 207
 208     # Pull the image if requested, unless the image is specified as a hash
 209     # that we already have.
 210     if args.pull and not find_image_hashes(args.image):
 211         pull_image(args.image, args.tag)
 212
 213     try:
 214         image_hash = find_one_image_hash(args.image, args.tag)
 215     except DockerError as error:
 216         print >>sys.stderr, "arv-keepdocker:", error.message
 217         sys.exit(1)
 218
 219     image_repo_tag = '{}:{}'.format(args.image, args.tag) if not image_hash.startswith(args.image.lower()) else None
 220
 221     if args.name is None:
 222         if image_repo_tag:
 223             collection_name = 'Docker image {} {}'.format(image_repo_tag, image_hash[0:12])
 224         else:
 225             collection_name = 'Docker image {}'.format(image_hash[0:12])
 226     else:
 227         collection_name = args.name
 228
 229     if not args.force:
 230         # Check if this image is already in Arvados.
 231
 232         # Project where everything should be owned
 233         if args.project_uuid:
 234             parent_project_uuid = args.project_uuid
 235         else:
 236             parent_project_uuid = api.users().current().execute(
 237                 num_retries=args.retries)['uuid']
 238
 239         # Find image hash tags
 240         existing_links = api.links().list(
 241             filters=[['link_class', '=', 'docker_image_hash'],
 242                      ['name', '=', image_hash]]
 243             ).execute(num_retries=args.retries)['items']
 244         if existing_links:
 245             # get readable collections
 246             collections = api.collections().list(
 247                 filters=[['uuid', 'in', [link['head_uuid'] for link in existing_links]]],
 248                 select=["uuid", "owner_uuid", "name", "manifest_text"]
 249                 ).execute(num_retries=args.retries)['items']
 250
 251             if collections:
 252                 # check for repo+tag links on these collections
 253                 existing_repo_tag = (api.links().list(
 254                     filters=[['link_class', '=', 'docker_image_repo+tag'],
 255                              ['name', '=', image_repo_tag],
 256                              ['head_uuid', 'in', collections]]
 257                     ).execute(num_retries=args_retries)['items']) if image_repo_tag else []
 258
 259                 # Filter on elements owned by the parent project
 260                 owned_col = [c for c in collections if c['owner_uuid'] == parent_project_uuid]
 261                 owned_img = [c for c in existing_links if c['owner_uuid'] == parent_project_uuid]
 262                 owned_rep = [c for c in existing_repo_tag if c['owner_uuid'] == parent_project_uuid]
 263
 264                 if owned_col:
 265                     # already have a collection owned by this project
 266                     coll_uuid = owned_col[0]['uuid']
 267                 else:
 268                     # create new collection owned by the project
 269                     coll_uuid = api.collections().create(
 270                         body={"manifest_text": collections[0]['manifest_text'],
 271                               "name": collection_name,
 272                               "owner_uuid": parent_project_uuid},
 273                         ensure_unique_name=True
 274                         ).execute(num_retries=args.retries)['uuid']
 275
 276                 link_base = {'owner_uuid': parent_project_uuid,
 277                              'head_uuid':  coll_uuid }
 278
 279                 if not owned_img:
 280                     # create image link owned by the project
 281                     make_link(api, args.retries,
 282                               'docker_image_hash', image_hash, **link_base)
 283
 284                 if not owned_rep and image_repo_tag:
 285                     # create repo+tag link owned by the project
 286                     make_link(api, args.retries, 'docker_image_repo+tag',
 287                               image_repo_tag, **link_base)
 288
 289                 print(coll_uuid)
 290
 291                 sys.exit(0)
 292
 293     # Open a file for the saved image, and write it if needed.
 294     outfile_name = '{}.tar'.format(image_hash)
 295     image_file, need_save = prep_image_file(outfile_name)
 296     if need_save:
 297         save_image(image_hash, image_file)
 298
 299     # Call arv-put with switches we inherited from it
 300     # (a.k.a., switches that aren't our own).
 301     put_args = keepdocker_parser.parse_known_args(arguments)[1]
 302
 303     if args.name is None:
 304         put_args += ['--name', collection_name]
 305
 306     coll_uuid = arv_put.main(
 307         put_args + ['--filename', outfile_name, image_file.name]).strip()
 308
 309     # Read the image metadata and make Arvados links from it.
 310     image_file.seek(0)
 311     image_tar = tarfile.open(fileobj=image_file)
 312     json_file = image_tar.extractfile(image_tar.getmember(image_hash + '/json'))
 313     image_metadata = json.load(json_file)
 314     json_file.close()
 315     image_tar.close()
 316     link_base = {'head_uuid': coll_uuid, 'properties': {}}
 317     if 'created' in image_metadata:
 318         link_base['properties']['image_timestamp'] = image_metadata['created']
 319     if args.project_uuid is not None:
 320         link_base['owner_uuid'] = args.project_uuid
 321
 322     make_link(api, args.retries, 'docker_image_hash', image_hash, **link_base)
 323     if image_repo_tag:
 324         make_link(api, args.retries,
 325                   'docker_image_repo+tag', image_repo_tag, **link_base)
 326
 327     # Clean up.
 328     image_file.close()
 329     for filename in [stat_cache_name(image_file), image_file.name]:
 330         try:
 331             os.unlink(filename)
 332         except OSError as error:
 333             if error.errno != errno.ENOENT:
 334                 raise
 335
 336 if __name__ == '__main__':
 337     main()