X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/f848ccbeb40ee87f7072b7399545424ecc0369e6..ef661ebdb4cb7ae6e0bc79da5e27fe226ca70bfa:/sdk/cwl/tests/test_job.py diff --git a/sdk/cwl/tests/test_job.py b/sdk/cwl/tests/test_job.py index 5a6e6dc7e3..f08e14f7cd 100644 --- a/sdk/cwl/tests/test_job.py +++ b/sdk/cwl/tests/test_job.py @@ -2,6 +2,11 @@ # # SPDX-License-Identifier: Apache-2.0 +from future import standard_library +standard_library.install_aliases() +from builtins import str +from builtins import next + import functools import json import logging @@ -9,16 +14,20 @@ import mock import os import unittest import copy -import StringIO +import io +import argparse import arvados import arvados_cwl +import arvados_cwl.executor import cwltool.process from arvados.errors import ApiError from schema_salad.ref_resolver import Loader from schema_salad.sourceline import cmap from .mock_discovery import get_rootDesc from .matcher import JsonDiffMatcher, StripYAMLComments +from .test_container import CollectionMock +from arvados_cwl.arvdocker import arv_docker_clear_cache if not os.getenv('ARVADOS_DEBUG'): logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN) @@ -26,16 +35,41 @@ if not os.getenv('ARVADOS_DEBUG'): class TestJob(unittest.TestCase): + def setUp(self): + cwltool.process._names = set() + + def helper(self, runner, enable_reuse=True): + document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.1") + + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) + loadingContext = arvados_cwl.context.ArvLoadingContext( + {"avsc_names": avsc_names, + "basedir": "", + "make_fs_access": make_fs_access, + "loader": Loader({}), + "metadata": {"cwlVersion": "v1.1", "http://commonwl.org/cwltool#original_cwlVersion": "v1.0"}, + "makeTool": runner.arv_make_tool}) + runtimeContext = arvados_cwl.context.ArvRuntimeContext( + {"work_api": "jobs", + "basedir": "", + "name": "test_run_job_"+str(enable_reuse), + "make_fs_access": make_fs_access, + "enable_reuse": enable_reuse, + "priority": 500}) + + return loadingContext, runtimeContext + # The test passes no builder.resources # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} @mock.patch('arvados.commands.keepdocker.list_images_in_arv') def test_run(self, list_images_in_arv): for enable_reuse in (True, False): + arv_docker_clear_cache() runner = mock.MagicMock() runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.ignore_docker_for_reuse = False runner.num_retries = 0 - document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"} @@ -53,15 +87,16 @@ class TestJob(unittest.TestCase): "outputs": [], "baseCommand": "ls", "arguments": [{"valueFrom": "$(runtime.outdir)"}], - "id": "#" + "id": "#", + "class": "CommandLineTool" }) - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, - collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) - arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names, - basedir="", make_fs_access=make_fs_access, loader=Loader({})) + + loadingContext, runtimeContext = self.helper(runner, enable_reuse) + + arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, loadingContext) arvtool.formatgraph = None - for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access): - j.run(enable_reuse=enable_reuse) + for j in arvtool.job({}, mock.MagicMock(), runtimeContext): + j.run(runtimeContext) runner.api.jobs().create.assert_called_with( body=JsonDiffMatcher({ 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', @@ -102,9 +137,8 @@ class TestJob(unittest.TestCase): # sharing link on the job runner.api.links().create.side_effect = ApiError( mock.MagicMock(return_value={'status': 403}), - 'Permission denied') - j.run(enable_reuse=enable_reuse) - j.output_callback.assert_called_with({}, 'success') + bytes(b'Permission denied')) + j.run(runtimeContext) else: assert not runner.api.links().create.called @@ -119,10 +153,7 @@ class TestJob(unittest.TestCase): arvados_cwl.add_arv_hints() list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] - runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} - - document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") - + runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"} tool = { "inputs": [], @@ -144,15 +175,16 @@ class TestJob(unittest.TestCase): "enableReuse": False }], "baseCommand": "ls", - "id": "#" + "id": "#", + "class": "CommandLineTool" } - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, - collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) - arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names, - make_fs_access=make_fs_access, loader=Loader({})) + + loadingContext, runtimeContext = self.helper(runner) + + arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, loadingContext) arvtool.formatgraph = None - for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access): - j.run(enable_reuse=True) + for j in arvtool.job({}, mock.MagicMock(), runtimeContext): + j.run(runtimeContext) runner.api.jobs().create.assert_called_with( body=JsonDiffMatcher({ 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', @@ -192,19 +224,24 @@ class TestJob(unittest.TestCase): runner.num_retries = 0 runner.ignore_docker_for_reuse = False - reader().open.return_value = StringIO.StringIO( - """2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir + reader().keys.return_value = "log.txt" + reader().open.return_value = io.StringIO( + str(u"""2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.outdir)=/tmp/crunch-job-task-work/compute3.1/outdir 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep - """) + """)) api.collections().list().execute.side_effect = ({"items": []}, {"items": [{"manifest_text": "XYZ"}]}, {"items": []}, {"items": [{"manifest_text": "ABC"}]}) - arvjob = arvados_cwl.ArvadosJob(runner) - arvjob.name = "testjob" - arvjob.builder = mock.MagicMock() + arvjob = arvados_cwl.ArvadosJob(runner, + mock.MagicMock(), + {}, + None, + [], + [], + "testjob") arvjob.output_callback = mock.MagicMock() arvjob.collect_outputs = mock.MagicMock() arvjob.collect_outputs.return_value = {"out": "stuff"} @@ -261,20 +298,25 @@ class TestJob(unittest.TestCase): runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.num_retries = 0 - reader().open.return_value = StringIO.StringIO( - """2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir + reader().keys.return_value = "log.txt" + reader().open.return_value = io.StringIO( + str(u"""2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.outdir)=/tmp/crunch-job-task-work/compute3.1/outdir 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep - """) + """)) api.collections().list().execute.side_effect = ( {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]}, {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]}, ) - arvjob = arvados_cwl.ArvadosJob(runner) - arvjob.name = "testjob" - arvjob.builder = mock.MagicMock() + arvjob = arvados_cwl.ArvadosJob(runner, + mock.MagicMock(), + {}, + None, + [], + [], + "testjob") arvjob.output_callback = mock.MagicMock() arvjob.collect_outputs = mock.MagicMock() arvjob.collect_outputs.return_value = {"out": "stuff"} @@ -306,51 +348,85 @@ class TestJob(unittest.TestCase): class TestWorkflow(unittest.TestCase): + + def setUp(self): + cwltool.process._names = set() + + def helper(self, runner, enable_reuse=True): + document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.1") + + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) + + document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=runner.api, fs_access=make_fs_access("")) + document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session) + document_loader.fetch_text = document_loader.fetcher.fetch_text + document_loader.check_exists = document_loader.fetcher.check_exists + + loadingContext = arvados_cwl.context.ArvLoadingContext( + {"avsc_names": avsc_names, + "basedir": "", + "make_fs_access": make_fs_access, + "loader": document_loader, + "metadata": {"cwlVersion": "v1.1", "http://commonwl.org/cwltool#original_cwlVersion": "v1.0"}, + "construct_tool_object": runner.arv_make_tool}) + runtimeContext = arvados_cwl.context.ArvRuntimeContext( + {"work_api": "jobs", + "basedir": "", + "name": "test_run_wf_"+str(enable_reuse), + "make_fs_access": make_fs_access, + "enable_reuse": enable_reuse, + "priority": 500}) + + return loadingContext, runtimeContext + # The test passes no builder.resources # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} @mock.patch("arvados.collection.CollectionReader") @mock.patch("arvados.collection.Collection") @mock.patch('arvados.commands.keepdocker.list_images_in_arv') def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader): + arv_docker_clear_cache() arvados_cwl.add_arv_hints() api = mock.MagicMock() api._rootDesc = get_rootDesc() - runner = arvados_cwl.ArvCwlRunner(api) + runner = arvados_cwl.executor.ArvCwlExecutor(api, argparse.Namespace(work_api="jobs", + output_name=None, + output_tags=None, + thread_count=1, + collection_cache_size=None)) self.assertEqual(runner.work_api, 'jobs') list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] - runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} - runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]} + runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"} + runner.api.collections().list().execute.return_value = {"items": [{ + "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzzz", + "portable_data_hash": "99999999999999999999999999999993+99"}]} runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.ignore_docker_for_reuse = False runner.num_retries = 0 - document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") - - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, - collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) - document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access("")) - document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session) - document_loader.fetch_text = document_loader.fetcher.fetch_text - document_loader.check_exists = document_loader.fetcher.check_exists - tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl") + loadingContext, runtimeContext = self.helper(runner) + runner.fs_access = runtimeContext.make_fs_access(runtimeContext.basedir) + tool, metadata = loadingContext.loader.resolve_ref("tests/wf/scatter2.cwl") metadata["cwlVersion"] = tool["cwlVersion"] - mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118" + mockc = mock.MagicMock() + mockcollection.side_effect = lambda *args, **kwargs: CollectionMock(mockc, *args, **kwargs) + mockcollectionreader().find.return_value = arvados.arvfile.ArvadosFile(mock.MagicMock(), "token.txt") - arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names, - basedir="", make_fs_access=make_fs_access, loader=document_loader, - makeTool=runner.arv_make_tool, metadata=metadata) + arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, loadingContext) arvtool.formatgraph = None - it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access) - it.next().run() - it.next().run() + it = arvtool.job({}, mock.MagicMock(), runtimeContext) + + next(it).run(runtimeContext) + next(it).run(runtimeContext) with open("tests/wf/scatter2_subwf.cwl") as f: - subwf = StripYAMLComments(f.read()) + subwf = StripYAMLComments(f.read().rstrip()) runner.api.jobs().create.assert_called_with( body=JsonDiffMatcher({ @@ -363,8 +439,8 @@ class TestWorkflow(unittest.TestCase): 'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'}, 'task.vwd': { - 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl', - 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml' + 'workflow.cwl': '$(task.keep)/99999999999999999999999999999996+99/workflow.cwl', + 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999996+99/cwl.input.yml' }, 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'], 'task.stdout': 'cwl.output.json'}]}, @@ -381,16 +457,17 @@ class TestWorkflow(unittest.TestCase): ['docker_image_locator', 'in docker', 'arvados/jobs']], find_or_create=True) - mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)]) - mockcollection().open().__enter__().write.assert_has_calls([mock.call( -'''{ + mockc.open().__enter__().write.assert_has_calls([mock.call(subwf)]) + mockc.open().__enter__().write.assert_has_calls([mock.call( +bytes(b'''{ "fileblub": { "basename": "token.txt", "class": "File", - "location": "/keep/99999999999999999999999999999999+118/token.txt" + "location": "/keep/99999999999999999999999999999999+118/token.txt", + "size": 0 }, "sleeptime": 5 -}''')]) +}'''))]) # The test passes no builder.resources # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} @@ -398,42 +475,42 @@ class TestWorkflow(unittest.TestCase): @mock.patch("arvados.collection.Collection") @mock.patch('arvados.commands.keepdocker.list_images_in_arv') def test_overall_resource_singlecontainer(self, list_images_in_arv, mockcollection, mockcollectionreader): + arv_docker_clear_cache() arvados_cwl.add_arv_hints() api = mock.MagicMock() api._rootDesc = get_rootDesc() - runner = arvados_cwl.ArvCwlRunner(api) + runner = arvados_cwl.executor.ArvCwlExecutor(api, argparse.Namespace(work_api="jobs", + output_name=None, + output_tags=None, + thread_count=1, + collection_cache_size=None)) self.assertEqual(runner.work_api, 'jobs') list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] - runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} - runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]} + runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"} + runner.api.collections().list().execute.return_value = {"items": [{ + "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzzz", + "portable_data_hash": "99999999999999999999999999999993+99"}]} runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.ignore_docker_for_reuse = False runner.num_retries = 0 - document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") - - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, - collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) - document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access("")) - document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session) - document_loader.fetch_text = document_loader.fetcher.fetch_text - document_loader.check_exists = document_loader.fetcher.check_exists - tool, metadata = document_loader.resolve_ref("tests/wf/echo-wf.cwl") - metadata["cwlVersion"] = tool["cwlVersion"] + loadingContext, runtimeContext = self.helper(runner) + loadingContext.do_update = True + runner.fs_access = runtimeContext.make_fs_access(runtimeContext.basedir) + tool, metadata = loadingContext.loader.resolve_ref("tests/wf/echo-wf.cwl") - mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118" + mockcollection.side_effect = lambda *args, **kwargs: CollectionMock(mock.MagicMock(), *args, **kwargs) - arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names, - basedir="", make_fs_access=make_fs_access, loader=document_loader, - makeTool=runner.arv_make_tool, metadata=metadata) + arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, loadingContext) arvtool.formatgraph = None - it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access) - it.next().run() - it.next().run() + it = arvtool.job({}, mock.MagicMock(), runtimeContext) + + next(it).run(runtimeContext) + next(it).run(runtimeContext) with open("tests/wf/echo-subwf.cwl") as f: subwf = StripYAMLComments(f.read()) @@ -449,8 +526,8 @@ class TestWorkflow(unittest.TestCase): 'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'}, 'task.vwd': { - 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl', - 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml' + 'workflow.cwl': '$(task.keep)/99999999999999999999999999999996+99/workflow.cwl', + 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999996+99/cwl.input.yml' }, 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'], 'task.stdout': 'cwl.output.json'}]}, @@ -473,5 +550,5 @@ class TestWorkflow(unittest.TestCase): api = mock.MagicMock() api._rootDesc = copy.deepcopy(get_rootDesc()) del api._rootDesc.get('resources')['jobs']['methods']['create'] - runner = arvados_cwl.ArvCwlRunner(api) + runner = arvados_cwl.executor.ArvCwlExecutor(api) self.assertEqual(runner.work_api, 'containers')