1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
5 from __future__ import print_function
6 from __future__ import absolute_import
7 from builtins import range
8 from past.builtins import basestring
9 from builtins import object
11 import arvados.commands.ws as ws
23 import arvados.commands._util as arv_cmd
24 import arvados.collection
26 from arvados._version import __version__
28 logger = logging.getLogger('arvados.arv-run')
29 logger.setLevel(logging.INFO)
31 arvrun_parser = argparse.ArgumentParser(parents=[arv_cmd.retry_opt])
32 arvrun_parser.add_argument('--dry-run', action="store_true",
33 help="Print out the pipeline that would be submitted and exit")
34 arvrun_parser.add_argument('--local', action="store_true",
35 help="Run locally using arv-run-pipeline-instance")
36 arvrun_parser.add_argument('--docker-image', type=str,
37 help="Docker image to use, otherwise use instance default.")
38 arvrun_parser.add_argument('--ignore-rcode', action="store_true",
39 help="Commands that return non-zero return codes should not be considered failed.")
40 arvrun_parser.add_argument('--no-reuse', action="store_true",
41 help="Do not reuse past jobs.")
42 arvrun_parser.add_argument('--no-wait', action="store_true",
43 help="Do not wait and display logs after submitting command, just exit.")
44 arvrun_parser.add_argument('--project-uuid', type=str,
45 help="Parent project of the pipeline")
46 arvrun_parser.add_argument('--git-dir', type=str, default="",
47 help="Git repository passed to arv-crunch-job when using --local")
48 arvrun_parser.add_argument('--repository', type=str, default="arvados",
49 help="repository field of component, default 'arvados'")
50 arvrun_parser.add_argument('--script-version', type=str, default="master",
51 help="script_version field of component, default 'master'")
52 arvrun_parser.add_argument('--version', action='version',
53 version="%s %s" % (sys.argv[0], __version__),
54 help='Print version and exit.')
55 arvrun_parser.add_argument('args', nargs=argparse.REMAINDER)
57 class ArvFile(object):
58 def __init__(self, prefix, fn):
63 return (self.prefix+self.fn).__hash__()
65 def __eq__(self, other):
66 return (self.prefix == other.prefix) and (self.fn == other.fn)
68 class UploadFile(ArvFile):
71 # Determine if a file is in a collection, and return a tuple consisting of the
72 # portable data hash and the path relative to the root of the collection.
73 # Return None if the path isn't with an arv-mount collection or there was is error.
74 def is_in_collection(root, branch):
78 fn = os.path.join(root, ".arvados#collection")
79 if os.path.exists(fn):
80 with file(fn, 'r') as f:
82 return (c["portable_data_hash"], branch)
84 sp = os.path.split(root)
85 return is_in_collection(sp[0], os.path.join(sp[1], branch))
86 except (IOError, OSError):
89 # Determine the project to place the output of this command by searching upward
90 # for arv-mount psuedofile indicating the project. If the cwd isn't within
91 # an arv-mount project or there is an error, return current_user.
92 def determine_project(root, current_user):
96 fn = os.path.join(root, ".arvados#project")
97 if os.path.exists(fn):
98 with file(fn, 'r') as f:
100 if 'writable_by' in c and current_user in c['writable_by']:
105 sp = os.path.split(root)
106 return determine_project(sp[0], current_user)
107 except (IOError, OSError):
110 # Determine if string corresponds to a file, and if that file is part of a
111 # arv-mounted collection or only local to the machine. Returns one of
112 # ArvFile() (file already exists in a collection), UploadFile() (file needs to
113 # be uploaded to a collection), or simply returns prefix+fn (which yields the
114 # original parameter string).
115 def statfile(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False):
116 absfn = os.path.abspath(fn)
119 sp = os.path.split(absfn)
120 (pdh, branch) = is_in_collection(sp[0], sp[1])
122 if stat.S_ISREG(st.st_mode):
123 return ArvFile(prefix, fnPattern % (pdh, branch))
124 elif stat.S_ISDIR(st.st_mode):
125 return ArvFile(prefix, dirPattern % (pdh, branch))
127 raise Exception("%s is not a regular file or directory" % absfn)
129 # trim leading '/' for path prefix test later
130 return UploadFile(prefix, absfn[1:])
132 if e.errno == errno.ENOENT and not raiseOSError:
139 def write_file(collection, pathprefix, fn, flush=False):
140 with open(os.path.join(pathprefix, fn)) as src:
141 dst = collection.open(fn, "w")
142 r = src.read(1024*128)
145 r = src.read(1024*128)
146 dst.close(flush=flush)
148 def uploadfiles(files, api, dry_run=False, num_retries=0,
150 fnPattern="$(file %s/%s)",
154 # Find the smallest path prefix that includes all the files that need to be uploaded.
155 # This starts at the root and iteratively removes common parent directory prefixes
156 # until all file paths no longer have a common parent.
166 # no parent directories left
169 # path step takes next directory
170 pathstep = sp[0] + "/"
172 # check if pathstep is common prefix for all files
173 if not c.fn.startswith(pathstep):
177 # pathstep is common parent directory for all files, so remove the prefix
179 pathprefix += pathstep
181 c.fn = c.fn[len(pathstep):]
183 logger.info("Upload local files: \"%s\"", '" "'.join([c.fn for c in files]))
186 logger.info("$(input) is %s", pathprefix.rstrip('/'))
189 files = sorted(files, key=lambda x: x.fn)
190 if collection is None:
191 collection = arvados.collection.Collection(api_client=api, num_retries=num_retries)
194 localpath = os.path.join(pathprefix, f.fn)
195 if prev and localpath.startswith(prev+"/"):
196 # If this path is inside an already uploaded subdirectory,
197 # don't redundantly re-upload it.
198 # e.g. we uploaded /tmp/foo and the next file is /tmp/foo/bar
199 # skip it because it starts with "/tmp/foo/"
202 if os.path.isfile(localpath):
203 write_file(collection, pathprefix, f.fn, not packed)
204 elif os.path.isdir(localpath):
205 for root, dirs, iterfiles in os.walk(localpath):
206 root = root[len(pathprefix):]
207 for src in iterfiles:
208 write_file(collection, pathprefix, os.path.join(root, src), not packed)
210 filters=[["portable_data_hash", "=", collection.portable_data_hash()]]
212 filters.append(["name", "like", name+"%"])
214 filters.append(["owner_uuid", "=", project])
216 exists = api.collections().list(filters=filters, limit=1).execute(num_retries=num_retries)
219 item = exists["items"][0]
220 pdh = item["portable_data_hash"]
221 logger.info("Using collection %s (%s)", pdh, item["uuid"])
222 elif len(collection) > 0:
223 collection.save_new(name=name, owner_uuid=project, ensure_unique_name=True)
224 pdh = collection.portable_data_hash()
225 logger.info("Uploaded to %s (%s)", pdh, collection.manifest_locator())
228 c.keepref = "%s/%s" % (pdh, c.fn)
229 c.fn = fnPattern % (pdh, c.fn)
232 def main(arguments=None):
233 args = arvrun_parser.parse_args(arguments)
235 if len(args.args) == 0:
236 arvrun_parser.print_help()
239 starting_args = args.args
243 # Parse the command arguments into 'slots'.
244 # All words following '>' are output arguments and are collected into slots[0].
245 # All words following '<' are input arguments and are collected into slots[1].
246 # slots[2..] store the parameters of each command in the pipeline.
248 # e.g. arv-run foo arg1 arg2 '|' bar arg3 arg4 '<' input1 input2 input3 '>' output.txt
249 # will be parsed into:
251 # ['input1', 'input2', 'input3'],
252 # ['foo', 'arg1', 'arg2'],
253 # ['bar', 'arg3', 'arg4']]
256 if c.startswith('>'):
259 slots[reading_into].append(c[1:])
260 elif c.startswith('<'):
263 slots[reading_into].append(c[1:])
265 reading_into = len(slots)
268 slots[reading_into].append(c)
270 if slots[0] and len(slots[0]) > 1:
271 logger.error("Can only specify a single stdout file (run-command substitutions are permitted)")
275 api = arvados.api('v1')
276 if args.project_uuid:
277 project = args.project_uuid
279 project = determine_project(os.getcwd(), api.users().current().execute()["uuid"])
281 # Identify input files. Look at each parameter and test to see if there is
282 # a file by that name. This uses 'patterns' to look for within
283 # command line arguments, such as --foo=file.txt or -lfile.txt
284 patterns = [re.compile("([^=]+=)(.*)"),
285 re.compile("(-[A-Za-z])(.+)")]
286 for j, command in enumerate(slots[1:]):
287 for i, a in enumerate(command):
289 # j == 0 is stdin, j > 0 is commands
290 # always skip program executable (i == 0) in commands
292 elif a.startswith('\\'):
293 # if it starts with a \ then don't do any interpretation
296 # See if it looks like a file
297 command[i] = statfile('', a)
299 # If a file named command[i] was found, it would now be an
300 # ArvFile or UploadFile. If command[i] is a basestring, that
301 # means it doesn't correspond exactly to a file, so do some
303 if isinstance(command[i], basestring):
307 command[i] = statfile(m.group(1), m.group(2))
310 files = [c for command in slots[1:] for c in command if isinstance(c, UploadFile)]
312 uploadfiles(files, api, dry_run=args.dry_run, num_retries=args.retries, project=project)
314 for i in range(1, len(slots)):
315 slots[i] = [("%s%s" % (c.prefix, c.fn)) if isinstance(c, ArvFile) else c for c in slots[i]]
318 "script": "run-command",
319 "script_version": args.script_version,
320 "repository": args.repository,
321 "script_parameters": {
323 "runtime_constraints": {}
326 if args.docker_image:
327 component["runtime_constraints"]["docker_image"] = args.docker_image
330 group_parser = argparse.ArgumentParser()
331 group_parser.add_argument('-b', '--batch-size', type=int)
332 group_parser.add_argument('args', nargs=argparse.REMAINDER)
334 for s in range(2, len(slots)):
335 for i in range(0, len(slots[s])):
336 if slots[s][i] == '--':
337 inp = "input%i" % (s-2)
338 groupargs = group_parser.parse_args(slots[2][i+1:])
339 if groupargs.batch_size:
340 component["script_parameters"][inp] = {"value": {"batch":groupargs.args, "size":groupargs.batch_size}}
341 slots[s] = slots[s][0:i] + [{"foreach": inp, "command": "$(%s)" % inp}]
343 component["script_parameters"][inp] = groupargs.args
344 slots[s] = slots[s][0:i] + ["$(%s)" % inp]
345 task_foreach.append(inp)
347 if slots[s][i] == '\--':
351 component["script_parameters"]["task.stdout"] = slots[0][0]
353 task_foreach.append("stdin")
354 component["script_parameters"]["stdin"] = slots[1]
355 component["script_parameters"]["task.stdin"] = "$(stdin)"
358 component["script_parameters"]["task.foreach"] = task_foreach
360 component["script_parameters"]["command"] = slots[2:]
361 if args.ignore_rcode:
362 component["script_parameters"]["task.ignore_rcode"] = args.ignore_rcode
365 "name": "arv-run " + " | ".join([s[0] for s in slots[2:]]),
366 "description": "@" + " ".join(starting_args) + "@",
370 "state": "RunningOnClient" if args.local else "RunningOnServer"
374 print(json.dumps(pipeline, indent=4))
376 pipeline["owner_uuid"] = project
377 pi = api.pipeline_instances().create(body=pipeline, ensure_unique_name=True).execute()
378 logger.info("Running pipeline %s", pi["uuid"])
381 subprocess.call(["arv-run-pipeline-instance", "--instance", pi["uuid"], "--run-jobs-here"] + (["--no-reuse"] if args.no_reuse else []))
382 elif not args.no_wait:
383 ws.main(["--pipeline", pi["uuid"]])
385 pi = api.pipeline_instances().get(uuid=pi["uuid"]).execute()
386 logger.info("Pipeline is %s", pi["state"])
387 if "output_uuid" in pi["components"]["command"]:
388 logger.info("Output is %s", pi["components"]["command"]["output_uuid"])
390 logger.info("No output")
392 if __name__ == '__main__':