1 # Copyright (C) The Arvados Authors. All rights reserved.
2 # Copyright (C) 2018 Genome Research Ltd.
4 # SPDX-License-Identifier: Apache-2.0
6 # Licensed under the Apache License, Version 2.0 (the "License");
7 # you may not use this file except in compliance with the License.
8 # You may obtain a copy of the License at
10 # http://www.apache.org/licenses/LICENSE-2.0
12 # Unless required by applicable law or agreed to in writing, software
13 # distributed under the License is distributed on an "AS IS" BASIS,
14 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15 # See the License for the specific language governing permissions and
16 # limitations under the License.
19 import arvados.commands.ws as ws
31 import arvados.commands._util as arv_cmd
32 import arvados.collection
33 import arvados.config as config
35 from arvados._version import __version__
37 logger = logging.getLogger('arvados.arv-run')
38 logger.setLevel(logging.INFO)
40 class ArvFile(object):
41 def __init__(self, prefix, fn):
46 return (self.prefix+self.fn).__hash__()
48 def __eq__(self, other):
49 return (self.prefix == other.prefix) and (self.fn == other.fn)
51 class UploadFile(ArvFile):
54 # Determine if a file is in a collection, and return a tuple consisting of the
55 # portable data hash and the path relative to the root of the collection.
56 # Return None if the path isn't with an arv-mount collection or there was is error.
57 def is_in_collection(root, branch):
61 fn = os.path.join(root, ".arvados#collection")
62 if os.path.exists(fn):
63 with open(fn, 'r') as f:
65 return (c["portable_data_hash"], branch)
67 sp = os.path.split(root)
68 return is_in_collection(sp[0], os.path.join(sp[1], branch))
69 except (IOError, OSError):
72 # Determine the project to place the output of this command by searching upward
73 # for arv-mount psuedofile indicating the project. If the cwd isn't within
74 # an arv-mount project or there is an error, return current_user.
75 def determine_project(root, current_user):
79 fn = os.path.join(root, ".arvados#project")
80 if os.path.exists(fn):
81 with file(fn, 'r') as f:
83 if 'writable_by' in c and current_user in c['writable_by']:
88 sp = os.path.split(root)
89 return determine_project(sp[0], current_user)
90 except (IOError, OSError):
93 # Determine if string corresponds to a file, and if that file is part of a
94 # arv-mounted collection or only local to the machine. Returns one of
95 # ArvFile() (file already exists in a collection), UploadFile() (file needs to
96 # be uploaded to a collection), or simply returns prefix+fn (which yields the
97 # original parameter string).
98 def statfile(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False):
99 absfn = os.path.abspath(fn)
102 sp = os.path.split(absfn)
103 (pdh, branch) = is_in_collection(sp[0], sp[1])
105 if stat.S_ISREG(st.st_mode):
106 return ArvFile(prefix, fnPattern % (pdh, branch))
107 elif stat.S_ISDIR(st.st_mode):
108 return ArvFile(prefix, dirPattern % (pdh, branch))
110 raise Exception("%s is not a regular file or directory" % absfn)
112 # trim leading '/' for path prefix test later
113 return UploadFile(prefix, absfn[1:])
115 if e.errno == errno.ENOENT and not raiseOSError:
122 def write_file(collection, pathprefix, fn, flush=False):
123 with open(os.path.join(pathprefix, fn), "rb") as src:
124 dst = collection.open(fn, "wb")
125 r = src.read(1024*128)
128 r = src.read(1024*128)
129 dst.close(flush=flush)
131 def uploadfiles(files, api, dry_run=False, num_retries=0,
133 fnPattern="$(file %s/%s)",
137 # Find the smallest path prefix that includes all the files that need to be uploaded.
138 # This starts at the root and iteratively removes common parent directory prefixes
139 # until all file paths no longer have a common parent.
149 # no parent directories left
152 # path step takes next directory
153 pathstep = sp[0] + "/"
155 # check if pathstep is common prefix for all files
156 if not c.fn.startswith(pathstep):
160 # pathstep is common parent directory for all files, so remove the prefix
162 pathprefix += pathstep
164 c.fn = c.fn[len(pathstep):]
166 logger.info("Upload local files: \"%s\"", '" "'.join([c.fn for c in files]))
169 logger.info("$(input) is %s", pathprefix.rstrip('/'))
172 files = sorted(files, key=lambda x: x.fn)
173 if collection is None:
174 collection = arvados.collection.Collection(api_client=api, num_retries=num_retries)
177 localpath = os.path.join(pathprefix, f.fn)
178 if prev and localpath.startswith(prev+"/"):
179 # If this path is inside an already uploaded subdirectory,
180 # don't redundantly re-upload it.
181 # e.g. we uploaded /tmp/foo and the next file is /tmp/foo/bar
182 # skip it because it starts with "/tmp/foo/"
185 if os.path.isfile(localpath):
186 write_file(collection, pathprefix, f.fn, not packed)
187 elif os.path.isdir(localpath):
188 for root, dirs, iterfiles in os.walk(localpath):
189 root = root[len(pathprefix):]
190 for src in iterfiles:
191 write_file(collection, pathprefix, os.path.join(root, src), not packed)
194 if len(collection) > 0:
195 # non-empty collection
196 filters = [["portable_data_hash", "=", collection.portable_data_hash()]]
197 name_pdh = "%s (%s)" % (name, collection.portable_data_hash())
199 filters.append(["name", "=", name_pdh])
201 filters.append(["owner_uuid", "=", project])
203 # do the list / create in a loop with up to 2 tries as we are using `ensure_unique_name=False`
204 # and there is a potential race with other workflows that may have created the collection
205 # between when we list it and find it does not exist and when we attempt to create it.
207 while pdh is None and tries > 0:
208 exists = api.collections().list(filters=filters, limit=1).execute(num_retries=num_retries)
211 item = exists["items"][0]
212 pdh = item["portable_data_hash"]
213 logger.info("Using collection %s (%s)", pdh, item["uuid"])
216 collection.save_new(name=name_pdh, owner_uuid=project, ensure_unique_name=False)
217 pdh = collection.portable_data_hash()
218 logger.info("Uploaded to %s (%s)", pdh, collection.manifest_locator())
219 except arvados.errors.ApiError as ae:
222 # Something weird going on here, probably a collection
223 # with a conflicting name but wrong PDH. We won't
224 # able to reuse it but we still need to save our
225 # collection, so so save it with unique name.
226 logger.info("Name conflict on '%s', existing collection has an unexpected portable data hash", name_pdh)
227 collection.save_new(name=name_pdh, owner_uuid=project, ensure_unique_name=True)
228 pdh = collection.portable_data_hash()
229 logger.info("Uploaded to %s (%s)", pdh, collection.manifest_locator())
232 pdh = collection.portable_data_hash()
233 assert (pdh == config.EMPTY_BLOCK_LOCATOR), "Empty collection portable_data_hash did not have expected locator, was %s" % pdh
234 logger.debug("Using empty collection %s", pdh)
237 c.keepref = "%s/%s" % (pdh, c.fn)
238 c.fn = fnPattern % (pdh, c.fn)
241 def main(arguments=None):
242 raise Exception("Legacy arv-run removed.")
244 if __name__ == '__main__':