X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/e69333d856baabf8bbbc27602850dcfe86309d23..36360d1f2987cea89ce217a2519c01da7456d533:/sdk/python/arvados/commands/run.py diff --git a/sdk/python/arvados/commands/run.py b/sdk/python/arvados/commands/run.py index be94e7304a..860fc8f805 100644 --- a/sdk/python/arvados/commands/run.py +++ b/sdk/python/arvados/commands/run.py @@ -11,22 +11,38 @@ import put import time import subprocess import logging +import sys import arvados.commands._util as arv_cmd +from arvados._version import __version__ + logger = logging.getLogger('arvados.arv-run') logger.setLevel(logging.INFO) arvrun_parser = argparse.ArgumentParser(parents=[arv_cmd.retry_opt]) -arvrun_parser.add_argument('--dry-run', action="store_true", help="Print out the pipeline that would be submitted and exit") -arvrun_parser.add_argument('--local', action="store_true", help="Run locally using arv-run-pipeline-instance") -arvrun_parser.add_argument('--docker-image', type=str, help="Docker image to use, otherwise use instance default.") -arvrun_parser.add_argument('--ignore-rcode', action="store_true", help="Commands that return non-zero return codes should not be considered failed.") -arvrun_parser.add_argument('--no-reuse', action="store_true", help="Do not reuse past jobs.") -arvrun_parser.add_argument('--no-wait', action="store_true", help="Do not wait and display logs after submitting command, just exit.") -arvrun_parser.add_argument('--project-uuid', type=str, help="Parent project of the pipeline") -arvrun_parser.add_argument('--git-dir', type=str, default="", help="Git repository passed to arv-crunch-job when using --local") -arvrun_parser.add_argument('--repository', type=str, default="arvados", help="repository field of component, default 'arvados'") -arvrun_parser.add_argument('--script-version', type=str, default="master", help="script_version field of component, default 'master'") +arvrun_parser.add_argument('--dry-run', action="store_true", + help="Print out the pipeline that would be submitted and exit") +arvrun_parser.add_argument('--local', action="store_true", + help="Run locally using arv-run-pipeline-instance") +arvrun_parser.add_argument('--docker-image', type=str, + help="Docker image to use, otherwise use instance default.") +arvrun_parser.add_argument('--ignore-rcode', action="store_true", + help="Commands that return non-zero return codes should not be considered failed.") +arvrun_parser.add_argument('--no-reuse', action="store_true", + help="Do not reuse past jobs.") +arvrun_parser.add_argument('--no-wait', action="store_true", + help="Do not wait and display logs after submitting command, just exit.") +arvrun_parser.add_argument('--project-uuid', type=str, + help="Parent project of the pipeline") +arvrun_parser.add_argument('--git-dir', type=str, default="", + help="Git repository passed to arv-crunch-job when using --local") +arvrun_parser.add_argument('--repository', type=str, default="arvados", + help="repository field of component, default 'arvados'") +arvrun_parser.add_argument('--script-version', type=str, default="master", + help="script_version field of component, default 'master'") +arvrun_parser.add_argument('--version', action='version', + version="%s %s" % (sys.argv[0], __version__), + help='Print version and exit.') arvrun_parser.add_argument('args', nargs=argparse.REMAINDER) class ArvFile(object): @@ -34,6 +50,12 @@ class ArvFile(object): self.prefix = prefix self.fn = fn + def __hash__(self): + return (self.prefix+self.fn).__hash__() + + def __eq__(self, other): + return (self.prefix == other.prefix) and (self.fn == other.fn) + class UploadFile(ArvFile): pass @@ -52,7 +74,7 @@ def is_in_collection(root, branch): else: sp = os.path.split(root) return is_in_collection(sp[0], os.path.join(sp[1], branch)) - except IOError, OSError: + except (IOError, OSError): return (None, None) # Determine the project to place the output of this command by searching upward @@ -73,7 +95,7 @@ def determine_project(root, current_user): else: sp = os.path.split(root) return determine_project(sp[0], current_user) - except IOError, OSError: + except (IOError, OSError): return current_user # Determine if string corresponds to a file, and if that file is part of a @@ -81,7 +103,7 @@ def determine_project(root, current_user): # ArvFile() (file already exists in a collection), UploadFile() (file needs to # be uploaded to a collection), or simply returns prefix+fn (which yields the # original parameter string). -def statfile(prefix, fn): +def statfile(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)"): absfn = os.path.abspath(fn) if os.path.exists(absfn): st = os.stat(absfn) @@ -89,7 +111,7 @@ def statfile(prefix, fn): sp = os.path.split(absfn) (pdh, branch) = is_in_collection(sp[0], sp[1]) if pdh: - return ArvFile(prefix, "$(file %s/%s)" % (pdh, branch)) + return ArvFile(prefix, fnPattern % (pdh, branch)) else: # trim leading '/' for path prefix test later return UploadFile(prefix, absfn[1:]) @@ -97,14 +119,14 @@ def statfile(prefix, fn): sp = os.path.split(absfn) (pdh, branch) = is_in_collection(sp[0], sp[1]) if pdh: - return ArvFile(prefix, "$(dir %s/%s/)" % (pdh, branch)) + return ArvFile(prefix, dirPattern % (pdh, branch)) return prefix+fn -def uploadfiles(files, api, dry_run=False, num_retries=0, project=None): +def uploadfiles(files, api, dry_run=False, num_retries=0, project=None, fnPattern="$(file %s/%s)", name=None): # Find the smallest path prefix that includes all the files that need to be uploaded. # This starts at the root and iteratively removes common parent directory prefixes - # until all file pathes no longer have a common parent. + # until all file paths no longer have a common parent. n = True pathprefix = "/" while n: @@ -148,12 +170,29 @@ def uploadfiles(files, api, dry_run=False, num_retries=0, project=None): stream = sp[0] collection.start_new_stream(stream) collection.write_file(f.fn, sp[1]) - item = api.collections().create(body={"owner_uuid": project, "manifest_text": collection.manifest_text()}).execute() + + filters=[["portable_data_hash", "=", collection.portable_data_hash()], + ["name", "like", name+"%"]] + if project: + filters.append(["owner_uuid", "=", project]) + + exists = api.collections().list(filters=filters).execute(num_retries=num_retries) + + if exists["items"]: + item = exists["items"][0] + logger.info("Using collection %s", item["uuid"]) + else: + body = {"owner_uuid": project, "manifest_text": collection.manifest_text()} + if name is not None: + body["name"] = name + item = api.collections().create(body=body, ensure_unique_name=True).execute() + logger.info("Uploaded to %s", item["uuid"]) + pdh = item["portable_data_hash"] - logger.info("Uploaded to %s", item["uuid"]) for c in files: - c.fn = "$(file %s/%s)" % (pdh, c.fn) + c.keepref = "%s/%s" % (pdh, c.fn) + c.fn = fnPattern % (pdh, c.fn) os.chdir(orgdir) @@ -238,7 +277,7 @@ def main(arguments=None): files = [c for command in slots[1:] for c in command if isinstance(c, UploadFile)] if files: - uploadfiles(files, api, dry_run=args.dry_run, num_retries=args.num_retries, project=project) + uploadfiles(files, api, dry_run=args.dry_run, num_retries=args.retries, project=project) for i in xrange(1, len(slots)): slots[i] = [("%s%s" % (c.prefix, c.fn)) if isinstance(c, ArvFile) else c for c in slots[i]]