sdk/cwl/arvados_cwl/arvdocker.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import logging
   6 import sys
   7 import threading
   8 import copy
   9 import re
  10 import subprocess
  11
  12 from schema_salad.sourceline import SourceLine
  13
  14 import cwltool.docker
  15 from cwltool.errors import WorkflowException
  16 import arvados.commands.keepdocker
  17
  18 logger = logging.getLogger('arvados.cwl-runner')
  19
  20 def determine_image_id(dockerImageId):
  21     for line in (
  22         subprocess.check_output(  # nosec
  23             ["docker", "images", "--no-trunc", "--all"]
  24         )
  25         .decode("utf-8")
  26         .splitlines()
  27     ):
  28         try:
  29             match = re.match(r"^([^ ]+)\s+([^ ]+)\s+([^ ]+)", line)
  30             split = dockerImageId.split(":")
  31             if len(split) == 1:
  32                 split.append("latest")
  33             elif len(split) == 2:
  34                 #  if split[1] doesn't  match valid tag names, it is a part of repository
  35                 if not re.match(r"[\w][\w.-]{0,127}", split[1]):
  36                     split[0] = split[0] + ":" + split[1]
  37                     split[1] = "latest"
  38             elif len(split) == 3:
  39                 if re.match(r"[\w][\w.-]{0,127}", split[2]):
  40                     split[0] = split[0] + ":" + split[1]
  41                     split[1] = split[2]
  42                     del split[2]
  43
  44             # check for repository:tag match or image id match
  45             if match and (
  46                 (split[0] == match.group(1) and split[1] == match.group(2))
  47                 or dockerImageId == match.group(3)
  48             ):
  49                 return match.group(3)
  50         except ValueError:
  51             pass
  52
  53     return None
  54
  55
  56 def arv_docker_get_image(api_client, dockerRequirement, pull_image, runtimeContext):
  57     """Check if a Docker image is available in Keep, if not, upload it using arv-keepdocker."""
  58
  59     project_uuid = runtimeContext.project_uuid
  60     force_pull = runtimeContext.force_docker_pull
  61     tmp_outdir_prefix = runtimeContext.tmp_outdir_prefix
  62     match_local_docker = runtimeContext.match_local_docker
  63     copy_deps = runtimeContext.copy_deps
  64     cached_lookups = runtimeContext.cached_docker_lookups
  65
  66     if "http://arvados.org/cwl#dockerCollectionPDH" in dockerRequirement:
  67         return dockerRequirement["http://arvados.org/cwl#dockerCollectionPDH"]
  68
  69     if "dockerImageId" not in dockerRequirement and "dockerPull" in dockerRequirement:
  70         dockerRequirement = copy.deepcopy(dockerRequirement)
  71         dockerRequirement["dockerImageId"] = dockerRequirement["dockerPull"]
  72         if hasattr(dockerRequirement, 'lc'):
  73             dockerRequirement.lc.data["dockerImageId"] = dockerRequirement.lc.data["dockerPull"]
  74
  75     if dockerRequirement["dockerImageId"] in cached_lookups:
  76         return cached_lookups[dockerRequirement["dockerImageId"]]
  77
  78     with SourceLine(dockerRequirement, "dockerImageId", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
  79         sp = dockerRequirement["dockerImageId"].split(":")
  80         image_name = sp[0]
  81         image_tag = sp[1] if len(sp) > 1 else "latest"
  82
  83         out_of_project_images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
  84                                                                 image_name=image_name,
  85                                                                 image_tag=image_tag,
  86                                                                 project_uuid=None)
  87
  88         if copy_deps:
  89             # Only images that are available in the destination project
  90             images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
  91                                                                     image_name=image_name,
  92                                                                     image_tag=image_tag,
  93                                                                     project_uuid=project_uuid)
  94         else:
  95             images = out_of_project_images
  96
  97         if match_local_docker:
  98             local_image_id = determine_image_id(dockerRequirement["dockerImageId"])
  99             if local_image_id:
 100                 # find it in the list
 101                 found = False
 102                 for i in images:
 103                     if i[1]["dockerhash"] == local_image_id:
 104                         found = True
 105                         images = [i]
 106                         break
 107                 if not found:
 108                     # force re-upload.
 109                     images = []
 110
 111                 for i in out_of_project_images:
 112                     if i[1]["dockerhash"] == local_image_id:
 113                         found = True
 114                         out_of_project_images = [i]
 115                         break
 116                 if not found:
 117                     # force re-upload.
 118                     out_of_project_images = []
 119
 120         if not images:
 121             if not out_of_project_images:
 122                 # Fetch Docker image if necessary.
 123                 try:
 124                     dockerjob = cwltool.docker.DockerCommandLineJob(None, None, None, None, None, None)
 125                     result = dockerjob.get_image(dockerRequirement, pull_image,
 126                                                                   force_pull, tmp_outdir_prefix)
 127                     if not result:
 128                         raise WorkflowException("Docker image '%s' not available" % dockerRequirement["dockerImageId"])
 129                 except OSError as e:
 130                     raise WorkflowException("While trying to get Docker image '%s', failed to execute 'docker': %s" % (dockerRequirement["dockerImageId"], e))
 131
 132             # Upload image to Arvados
 133             args = []
 134             if project_uuid:
 135                 args.append("--project-uuid="+project_uuid)
 136             args.append(image_name)
 137             args.append(image_tag)
 138             logger.info("Uploading Docker image %s:%s", image_name, image_tag)
 139             try:
 140                 arvados.commands.put.api_client = api_client
 141                 arvados.commands.keepdocker.main(args, stdout=sys.stderr, install_sig_handlers=False, api=api_client)
 142             except SystemExit as e:
 143                 # If e.code is None or zero, then keepdocker exited normally and we can continue
 144                 if e.code:
 145                     raise WorkflowException("keepdocker exited with code %s" % e.code)
 146
 147             images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
 148                                                                     image_name=image_name,
 149                                                                     image_tag=image_tag,
 150                                                                     project_uuid=project_uuid)
 151
 152         if not images:
 153             raise WorkflowException("Could not find Docker image %s:%s" % (image_name, image_tag))
 154
 155         pdh = api_client.collections().get(uuid=images[0][0]).execute()["portable_data_hash"]
 156
 157         cached_lookups[dockerRequirement["dockerImageId"]] = pdh
 158
 159     return pdh