X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/16f704326f44fd1e5e5e60b936c9b5895d6a6ff8..864c3b0afd16c77e046f0072d8517d34c5a44792:/sdk/python/arvados/commands/put.py diff --git a/sdk/python/arvados/commands/put.py b/sdk/python/arvados/commands/put.py index afe75b3105..616a94e903 100644 --- a/sdk/python/arvados/commands/put.py +++ b/sdk/python/arvados/commands/put.py @@ -10,6 +10,7 @@ import argparse import arvados import arvados.collection import base64 +import ciso8601 import copy import datetime import errno @@ -234,6 +235,19 @@ _group.add_argument('--no-cache', action='store_false', dest='use_cache', Do not save upload state in a cache file for resuming. """) +_group = upload_opts.add_mutually_exclusive_group() +_group.add_argument('--trash-at', metavar='YYYY-MM-DDTHH:MM', default=None, + help=""" +Set the trash date of the resulting collection to an absolute date in the future. +The accepted format is defined by the ISO 8601 standard. Examples: 20090103, 2009-01-03, 20090103T181505, 2009-01-03T18:15:05.\n +Timezone information can be added. If not, the provided date/time is assumed as being in the local system's timezone. +""") +_group.add_argument('--trash-after', type=int, metavar='DAYS', default=None, + help=""" +Set the trash date of the resulting collection to an amount of days from the +date/time that the upload process finishes. +""") + arg_parser = argparse.ArgumentParser( description='Copy data from the local filesystem to Keep.', parents=[upload_opts, run_opts, arv_cmd.retry_opt]) @@ -430,7 +444,8 @@ class ArvPutUploadJob(object): put_threads=None, replication_desired=None, filename=None, update_time=60.0, update_collection=None, storage_classes=None, logger=logging.getLogger('arvados.arv_put'), dry_run=False, - follow_links=True, exclude_paths=[], exclude_names=None): + follow_links=True, exclude_paths=[], exclude_names=None, + trash_at=None): self.paths = paths self.resume = resume self.use_cache = use_cache @@ -470,6 +485,13 @@ class ArvPutUploadJob(object): self.follow_links = follow_links self.exclude_paths = exclude_paths self.exclude_names = exclude_names + self._trash_at = trash_at + + if self._trash_at is not None: + if type(self._trash_at) not in [datetime.datetime, datetime.timedelta]: + raise TypeError('trash_at should be None, timezone-naive datetime or timedelta') + if type(self._trash_at) == datetime.datetime and self._trash_at.tzinfo is not None: + raise TypeError('provided trash_at datetime should be timezone-naive') if not self.use_cache and self.resume: raise ArvPutArgumentConflict('resume cannot be True when use_cache is False') @@ -610,6 +632,17 @@ class ArvPutUploadJob(object): if self.use_cache: self._cache_file.close() + def _collection_trash_at(self): + """ + Returns the trash date that the collection should use at save time. + Takes into account absolute/relative trash_at values requested + by the user. + """ + if type(self._trash_at) == datetime.timedelta: + # Get an absolute datetime for trash_at + return datetime.datetime.utcnow() + self._trash_at + return self._trash_at + def save_collection(self): if self.update: # Check if files should be updated on the remote collection. @@ -625,7 +658,8 @@ class ArvPutUploadJob(object): # The file already exist on remote collection, skip it. pass self._remote_collection.save(storage_classes=self.storage_classes, - num_retries=self.num_retries) + num_retries=self.num_retries, + trash_at=self._collection_trash_at()) else: if self.storage_classes is None: self.storage_classes = ['default'] @@ -633,7 +667,8 @@ class ArvPutUploadJob(object): name=self.name, owner_uuid=self.owner_uuid, storage_classes=self.storage_classes, ensure_unique_name=self.ensure_unique_name, - num_retries=self.num_retries) + num_retries=self.num_retries, + trash_at=self._collection_trash_at()) def destroy_cache(self): if self.use_cache: @@ -688,6 +723,15 @@ class ArvPutUploadJob(object): self._save_state() except Exception as e: self.logger.error("Unexpected error trying to save cache file: {}".format(e)) + # Keep remote collection's trash_at attribute synced when using relative expire dates + if self._remote_collection is not None and type(self._trash_at) == datetime.timedelta: + try: + self._api_client.collections().update( + uuid=self._remote_collection.manifest_locator(), + body={'trash_at': self._collection_trash_at().strftime("%Y-%m-%dT%H:%M:%S.%fZ")} + ).execute(num_retries=self.num_retries) + except Exception as e: + self.logger.error("Unexpected error trying to update remote collection's expire date: {}".format(e)) else: self.bytes_written = self.bytes_skipped # Call the reporter, if any @@ -823,7 +867,9 @@ class ArvPutUploadJob(object): update_collection): try: self._remote_collection = arvados.collection.Collection( - update_collection, api_client=self._api_client) + update_collection, + api_client=self._api_client, + num_retries=self.num_retries) except arvados.errors.ApiError as error: raise CollectionUpdateError("Cannot read collection {} ({})".format(update_collection, error)) else: @@ -866,7 +912,8 @@ class ArvPutUploadJob(object): self._state['manifest'], replication_desired=self.replication_desired, put_threads=self.put_threads, - api_client=self._api_client) + api_client=self._api_client, + num_retries=self.num_retries) def _cached_manifest_valid(self): """ @@ -958,6 +1005,9 @@ class ArvPutUploadJob(object): def collection_name(self): return self._my_collection().api_response()['name'] if self._my_collection().api_response() else None + def collection_trash_at(self): + return self._my_collection().get_trash_at() + def manifest_locator(self): return self._my_collection().manifest_locator() @@ -1073,6 +1123,44 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr, if install_sig_handlers: arv_cmd.install_signal_handlers() + # Trash arguments validation + trash_at = None + if args.trash_at is not None: + # ciso8601 considers YYYYMM as invalid but YYYY-MM as valid, so here we + # make sure the user provides a complete YYYY-MM-DD date. + if not re.match(r'^\d{4}(?P-?)\d{2}?(?P=dash)\d{2}', args.trash_at): + logger.error("--trash-at argument format invalid, use --help to see examples.") + sys.exit(1) + # Check if no time information was provided. In that case, assume end-of-day. + if re.match(r'^\d{4}(?P-?)\d{2}?(?P=dash)\d{2}$', args.trash_at): + args.trash_at += 'T23:59:59' + try: + trash_at = ciso8601.parse_datetime(args.trash_at) + except: + logger.error("--trash-at argument format invalid, use --help to see examples.") + sys.exit(1) + else: + if trash_at.tzinfo is not None: + # Timezone aware datetime provided. + utcoffset = -trash_at.utcoffset() + else: + # Timezone naive datetime provided. Assume is local. + if time.daylight: + utcoffset = datetime.timedelta(seconds=time.altzone) + else: + utcoffset = datetime.timedelta(seconds=time.timezone) + # Convert to UTC timezone naive datetime. + trash_at = trash_at.replace(tzinfo=None) + utcoffset + + if trash_at <= datetime.datetime.utcnow(): + logger.error("--trash-at argument must be set in the future") + sys.exit(1) + if args.trash_after is not None: + if args.trash_after < 1: + logger.error("--trash-after argument must be >= 1") + sys.exit(1) + trash_at = datetime.timedelta(seconds=(args.trash_after * 24 * 60 * 60)) + # Determine the name to use if args.name: if args.stream or args.raw: @@ -1178,7 +1266,8 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr, dry_run=args.dry_run, follow_links=args.follow_links, exclude_paths=exclude_paths, - exclude_names=exclude_names) + exclude_names=exclude_names, + trash_at=trash_at) except ResumeCacheConflict: logger.error("\n".join([ "arv-put: Another process is already uploading this data.", @@ -1192,7 +1281,7 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr, " --no-resume to start a new resume cache.", " --no-cache to disable resume cache."])) sys.exit(1) - except CollectionUpdateError as error: + except (CollectionUpdateError, PathDoesNotExistError) as error: logger.error("\n".join([ "arv-put: %s" % str(error)])) sys.exit(1) @@ -1202,10 +1291,6 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr, except ArvPutUploadNotPending: # No files pending for upload sys.exit(0) - except PathDoesNotExistError as error: - logger.error("\n".join([ - "arv-put: %s" % str(error)])) - sys.exit(1) if not args.dry_run and not args.update_collection and args.resume and writer.bytes_written > 0: logger.warning("\n".join([ @@ -1234,10 +1319,21 @@ def main(arguments=None, stdout=sys.stdout, stderr=sys.stderr, output = ','.join(writer.data_locators()) else: try: + expiration_notice = "" + if writer.collection_trash_at() is not None: + # Get the local timezone-naive version, and log it with timezone information. + if time.daylight: + local_trash_at = writer.collection_trash_at().replace(tzinfo=None) - datetime.timedelta(seconds=time.altzone) + else: + local_trash_at = writer.collection_trash_at().replace(tzinfo=None) - datetime.timedelta(seconds=time.timezone) + expiration_notice = ". It will expire on {} {}.".format( + local_trash_at.strftime("%Y-%m-%d %H:%M:%S"), time.strftime("%z")) if args.update_collection: - logger.info(u"Collection updated: '{}'".format(writer.collection_name())) + logger.info(u"Collection updated: '{}'{}".format( + writer.collection_name(), expiration_notice)) else: - logger.info(u"Collection saved as '{}'".format(writer.collection_name())) + logger.info(u"Collection saved as '{}'{}".format( + writer.collection_name(), expiration_notice)) if args.portable_data_hash: output = writer.portable_data_hash() else: