X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/935c8f81f0a041a58bbb45602ef44cdd6673163a..0561bd0c3c07257fd58ded6c7cfa5feeae97af57:/sdk/python/arvados/commands/run.py?ds=sidebyside diff --git a/sdk/python/arvados/commands/run.py b/sdk/python/arvados/commands/run.py index 860fc8f805..831e496a29 100644 --- a/sdk/python/arvados/commands/run.py +++ b/sdk/python/arvados/commands/run.py @@ -1,5 +1,12 @@ -#!/usr/bin/env python - +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + +from __future__ import print_function +from __future__ import absolute_import +from builtins import range +from past.builtins import basestring +from builtins import object import arvados import arvados.commands.ws as ws import argparse @@ -7,12 +14,14 @@ import json import re import os import stat -import put +from . import put import time import subprocess import logging import sys +import errno import arvados.commands._util as arv_cmd +import arvados.collection from arvados._version import __version__ @@ -103,99 +112,121 @@ 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, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)"): +def statfile(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False): absfn = os.path.abspath(fn) - if os.path.exists(absfn): + try: st = os.stat(absfn) - if stat.S_ISREG(st.st_mode): - sp = os.path.split(absfn) - (pdh, branch) = is_in_collection(sp[0], sp[1]) - if pdh: + sp = os.path.split(absfn) + (pdh, branch) = is_in_collection(sp[0], sp[1]) + if pdh: + if stat.S_ISREG(st.st_mode): return ArvFile(prefix, fnPattern % (pdh, branch)) - else: - # trim leading '/' for path prefix test later - return UploadFile(prefix, absfn[1:]) - if stat.S_ISDIR(st.st_mode): - sp = os.path.split(absfn) - (pdh, branch) = is_in_collection(sp[0], sp[1]) - if pdh: + elif stat.S_ISDIR(st.st_mode): return ArvFile(prefix, dirPattern % (pdh, branch)) + else: + raise Exception("%s is not a regular file or directory" % absfn) + else: + # trim leading '/' for path prefix test later + return UploadFile(prefix, absfn[1:]) + except OSError as e: + if e.errno == errno.ENOENT and not raiseOSError: + pass + else: + raise return prefix+fn -def uploadfiles(files, api, dry_run=False, num_retries=0, project=None, fnPattern="$(file %s/%s)", name=None): +def write_file(collection, pathprefix, fn): + with open(os.path.join(pathprefix, fn)) as src: + dst = collection.open(fn, "w") + r = src.read(1024*128) + while r: + dst.write(r) + r = src.read(1024*128) + dst.close(flush=False) + +def uploadfiles(files, api, dry_run=False, num_retries=0, + project=None, + fnPattern="$(file %s/%s)", + name=None, + collection=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 paths no longer have a common parent. - n = True - pathprefix = "/" - while n: - pathstep = None - for c in files: - if pathstep is None: - sp = c.fn.split('/') - if len(sp) < 2: - # no parent directories left - n = False - break - # path step takes next directory - pathstep = sp[0] + "/" - else: - # check if pathstep is common prefix for all files - if not c.fn.startswith(pathstep): - n = False - break - if n: - # pathstep is common parent directory for all files, so remove the prefix - # from each path - pathprefix += pathstep + if files: + n = True + pathprefix = "/" + while n: + pathstep = None for c in files: - c.fn = c.fn[len(pathstep):] - - orgdir = os.getcwd() - os.chdir(pathprefix) - - logger.info("Upload local files: \"%s\"", '" "'.join([c.fn for c in files])) + if pathstep is None: + sp = c.fn.split('/') + if len(sp) < 2: + # no parent directories left + n = False + break + # path step takes next directory + pathstep = sp[0] + "/" + else: + # check if pathstep is common prefix for all files + if not c.fn.startswith(pathstep): + n = False + break + if n: + # pathstep is common parent directory for all files, so remove the prefix + # from each path + pathprefix += pathstep + for c in files: + c.fn = c.fn[len(pathstep):] + + logger.info("Upload local files: \"%s\"", '" "'.join([c.fn for c in files])) if dry_run: logger.info("$(input) is %s", pathprefix.rstrip('/')) pdh = "$(input)" else: files = sorted(files, key=lambda x: x.fn) - collection = arvados.CollectionWriter(api, num_retries=num_retries) - stream = None + if collection is None: + collection = arvados.collection.Collection(api_client=api, num_retries=num_retries) + prev = "" for f in files: - sp = os.path.split(f.fn) - if sp[0] != stream: - stream = sp[0] - collection.start_new_stream(stream) - collection.write_file(f.fn, sp[1]) - - filters=[["portable_data_hash", "=", collection.portable_data_hash()], - ["name", "like", name+"%"]] + localpath = os.path.join(pathprefix, f.fn) + if prev and localpath.startswith(prev+"/"): + # If this path is inside an already uploaded subdirectory, + # don't redundantly re-upload it. + # e.g. we uploaded /tmp/foo and the next file is /tmp/foo/bar + # skip it because it starts with "/tmp/foo/" + continue + prev = localpath + if os.path.isfile(localpath): + write_file(collection, pathprefix, f.fn) + elif os.path.isdir(localpath): + for root, dirs, iterfiles in os.walk(localpath): + root = root[len(pathprefix):] + for src in iterfiles: + write_file(collection, pathprefix, os.path.join(root, src)) + + filters=[["portable_data_hash", "=", collection.portable_data_hash()]] + if name: + filters.append(["name", "like", name+"%"]) if project: filters.append(["owner_uuid", "=", project]) - exists = api.collections().list(filters=filters).execute(num_retries=num_retries) + exists = api.collections().list(filters=filters, limit=1).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"] + pdh = item["portable_data_hash"] + logger.info("Using collection %s (%s)", pdh, item["uuid"]) + elif len(collection) > 0: + collection.save_new(name=name, owner_uuid=project, ensure_unique_name=True) + pdh = collection.portable_data_hash() + logger.info("Uploaded to %s (%s)", pdh, collection.manifest_locator()) for c in files: c.keepref = "%s/%s" % (pdh, c.fn) c.fn = fnPattern % (pdh, c.fn) - os.chdir(orgdir) - def main(arguments=None): args = arvrun_parser.parse_args(arguments) @@ -279,7 +310,7 @@ def main(arguments=None): if files: uploadfiles(files, api, dry_run=args.dry_run, num_retries=args.retries, project=project) - for i in xrange(1, len(slots)): + for i in range(1, len(slots)): slots[i] = [("%s%s" % (c.prefix, c.fn)) if isinstance(c, ArvFile) else c for c in slots[i]] component = { @@ -299,8 +330,8 @@ def main(arguments=None): group_parser.add_argument('-b', '--batch-size', type=int) group_parser.add_argument('args', nargs=argparse.REMAINDER) - for s in xrange(2, len(slots)): - for i in xrange(0, len(slots[s])): + for s in range(2, len(slots)): + for i in range(0, len(slots[s])): if slots[s][i] == '--': inp = "input%i" % (s-2) groupargs = group_parser.parse_args(slots[2][i+1:])