X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/1d77dec0f7bc1cca34d52288bd39ecd0a79be250..2e74236fa27822addd856f194befc28382990ce0:/sdk/cwl/tests/test_pathmapper.py diff --git a/sdk/cwl/tests/test_pathmapper.py b/sdk/cwl/tests/test_pathmapper.py index 3e7918a24a..194092db7a 100644 --- a/sdk/cwl/tests/test_pathmapper.py +++ b/sdk/cwl/tests/test_pathmapper.py @@ -1,3 +1,7 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + import functools import mock import sys @@ -10,13 +14,14 @@ import arvados import arvados.keep import arvados.collection import arvados_cwl +import arvados_cwl.executor from cwltool.pathmapper import MapperEnt from .mock_discovery import get_rootDesc from arvados_cwl.pathmapper import ArvPathMapper -def upload_mock(files, api, dry_run=False, num_retries=0, project=None, fnPattern="$(file %s/%s)", name=None, collection=None): +def upload_mock(files, api, dry_run=False, num_retries=0, project=None, fnPattern="$(file %s/%s)", name=None, collection=None, packed=None): pdh = "99999999999999999999999999999991+99" for c in files: c.keepref = "%s/%s" % (pdh, os.path.basename(c.fn)) @@ -27,10 +32,17 @@ class TestPathmap(unittest.TestCase): self.api = mock.MagicMock() self.api._rootDesc = get_rootDesc() + def tearDown(self): + root_logger = logging.getLogger('') + + # Remove existing RuntimeStatusLoggingHandlers if they exist + handlers = [h for h in root_logger.handlers if not isinstance(h, arvados_cwl.executor.RuntimeStatusLoggingHandler)] + root_logger.handlers = handlers + def test_keepref(self): """Test direct keep references.""" - arvrunner = arvados_cwl.ArvCwlRunner(self.api) + arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api) p = ArvPathMapper(arvrunner, [{ "class": "File", @@ -45,9 +57,9 @@ class TestPathmap(unittest.TestCase): def test_upload(self, statfile, upl): """Test pathmapper uploading files.""" - arvrunner = arvados_cwl.ArvCwlRunner(self.api) + arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api) - def statfile_mock(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)"): + def statfile_mock(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False): st = arvados.commands.run.UploadFile("", "tests/hw.py") return st @@ -62,32 +74,15 @@ class TestPathmap(unittest.TestCase): self.assertEqual({'file:tests/hw.py': MapperEnt(resolved='keep:99999999999999999999999999999991+99/hw.py', target='/test/99999999999999999999999999999991+99/hw.py', type='File', staged=True)}, p._pathmap) - @mock.patch("arvados.commands.run.uploadfiles") - def test_prev_uploaded(self, upl): - """Test pathmapper handling previously uploaded files.""" - - arvrunner = arvados_cwl.ArvCwlRunner(self.api) - arvrunner.add_uploaded('file:tests/hw.py', MapperEnt(resolved='keep:99999999999999999999999999999992+99/hw.py', target='', type='File', staged=True)) - - upl.side_effect = upload_mock - - p = ArvPathMapper(arvrunner, [{ - "class": "File", - "location": "file:tests/hw.py" - }], "", "/test/%s", "/test/%s/%s") - - self.assertEqual({'file:tests/hw.py': MapperEnt(resolved='keep:99999999999999999999999999999992+99/hw.py', target='/test/99999999999999999999999999999992+99/hw.py', type='File', staged=True)}, - p._pathmap) - @mock.patch("arvados.commands.run.uploadfiles") @mock.patch("arvados.commands.run.statfile") def test_statfile(self, statfile, upl): """Test pathmapper handling ArvFile references.""" - arvrunner = arvados_cwl.ArvCwlRunner(self.api) + arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api) # An ArvFile object returned from arvados.commands.run.statfile means the file is located on a # keep mount, so we can construct a direct reference directly without upload. - def statfile_mock(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)"): + def statfile_mock(prefix, fn, fnPattern="$(file %s/%s)", dirPattern="$(dir %s/%s/)", raiseOSError=False): st = arvados.commands.run.ArvFile("", fnPattern % ("99999999999999999999999999999991+99", "hw.py")) return st @@ -101,3 +96,158 @@ class TestPathmap(unittest.TestCase): self.assertEqual({'file:tests/hw.py': MapperEnt(resolved='keep:99999999999999999999999999999991+99/hw.py', target='/test/99999999999999999999999999999991+99/hw.py', type='File', staged=True)}, p._pathmap) + + @mock.patch("os.stat") + def test_missing_file(self, stat): + """Test pathmapper handling missing references.""" + arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api) + + stat.side_effect = OSError(2, "No such file or directory") + + with self.assertRaises(OSError): + p = ArvPathMapper(arvrunner, [{ + "class": "File", + "location": "file:tests/hw.py" + }], "", "/test/%s", "/test/%s/%s") + + def test_needs_new_collection(self): + arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api) + + # Plain file. Don't need a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py" + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + self.assertFalse(p.needs_new_collection(a)) + + # A file that isn't in the pathmap (for some reason). Need a new collection. + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + self.assertTrue(p.needs_new_collection(a)) + + # A file with a secondary file in the same collection. Don't need + # a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py", + "secondaryFiles": [{ + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.pyc", + "basename": "hw.pyc" + }] + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + p._pathmap["keep:99999999999999999999999999999991+99/hw.pyc"] = True + self.assertFalse(p.needs_new_collection(a)) + + # Secondary file is in a different collection from the + # a new collectionprimary. Need a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py", + "secondaryFiles": [{ + "class": "File", + "location": "keep:99999999999999999999999999999992+99/hw.pyc", + "basename": "hw.pyc" + }] + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + p._pathmap["keep:99999999999999999999999999999992+99/hw.pyc"] = True + self.assertTrue(p.needs_new_collection(a)) + + # Secondary file should be staged to a different name than + # path in location. Need a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py", + "secondaryFiles": [{ + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.pyc", + "basename": "hw.other" + }] + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + p._pathmap["keep:99999999999999999999999999999991+99/hw.pyc"] = True + self.assertTrue(p.needs_new_collection(a)) + + # Secondary file is a directory. Do not need a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py", + "secondaryFiles": [{ + "class": "Directory", + "location": "keep:99999999999999999999999999999991+99/hw", + "basename": "hw", + "listing": [{ + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw/h2", + "basename": "h2" + }] + }] + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + p._pathmap["keep:99999999999999999999999999999991+99/hw"] = True + p._pathmap["keep:99999999999999999999999999999991+99/hw/h2"] = True + self.assertFalse(p.needs_new_collection(a)) + + # Secondary file is a renamed directory. Need a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py", + "secondaryFiles": [{ + "class": "Directory", + "location": "keep:99999999999999999999999999999991+99/hw", + "basename": "wh", + "listing": [{ + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw/h2", + "basename": "h2" + }] + }] + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + p._pathmap["keep:99999999999999999999999999999991+99/hw"] = True + p._pathmap["keep:99999999999999999999999999999991+99/hw/h2"] = True + self.assertTrue(p.needs_new_collection(a)) + + # Secondary file is a file literal. Need a new collection. + a = { + "class": "File", + "location": "keep:99999999999999999999999999999991+99/hw.py", + "basename": "hw.py", + "secondaryFiles": [{ + "class": "File", + "location": "_:123", + "basename": "hw.pyc", + "contents": "123" + }] + } + p = ArvPathMapper(arvrunner, [], "", "%s", "%s/%s") + p._pathmap["keep:99999999999999999999999999999991+99/hw.py"] = True + p._pathmap["_:123"] = True + self.assertTrue(p.needs_new_collection(a)) + + def test_is_in_collection(self): + arvrunner = arvados_cwl.executor.ArvCwlExecutor(self.api) + self.maxDiff = 1000000 + + cwd = os.getcwd() + p = ArvPathMapper(arvrunner, [{ + "class": "File", + "location": "file://"+cwd+"/tests/fake-keep-mount/fake_collection_dir/subdir/banana.txt" + }], "", "/test/%s", "/test/%s/%s") + + self.assertEqual({"file://"+cwd+"/tests/fake-keep-mount/fake_collection_dir/subdir/banana.txt": MapperEnt(resolved='keep:99999999999999999999999999999991+99/subdir/banana.txt', target='/test/99999999999999999999999999999991+99/subdir/banana.txt', type='File', staged=True)}, + p._pathmap)