Merge branch 'master' into 10194-cwl-version-skew
[arvados.git] / sdk / cwl / tests / test_job.py
index dba65b0f8a884c207c29656d9f5d0107479f0b7a..93b5d39ffa6eb7e45efba47c64b69fb7a95a33d8 100644 (file)
@@ -1,9 +1,14 @@
-import arvados_cwl
+import functools
+import json
 import logging
 import mock
-import unittest
 import os
+import unittest
+
+import arvados
+import arvados_cwl
 import cwltool.process
+from schema_salad.ref_resolver import Loader
 
 if not os.getenv('ARVADOS_DEBUG'):
     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
@@ -14,20 +19,28 @@ class TestJob(unittest.TestCase):
 
     # The test passes no builder.resources
     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
-    def test_run(self):
+    @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
+    def test_run(self, list_images_in_arv):
         runner = mock.MagicMock()
         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
         runner.ignore_docker_for_reuse = False
-        document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3")
+        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_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
 
         tool = {
             "inputs": [],
             "outputs": [],
-            "baseCommand": "ls"
+            "baseCommand": "ls",
+            "arguments": [{"valueFrom": "$(runtime.outdir)"}]
         }
-        arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, avsc_names=avsc_names, basedir="")
+        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
+        arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
+                                                 basedir="", make_fs_access=make_fs_access, loader=Loader({}))
         arvtool.formatgraph = None
-        for j in arvtool.job({}, mock.MagicMock(), basedir=""):
+        for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
             j.run()
             runner.api.jobs().create.assert_called_with(
                 body={
@@ -35,8 +48,8 @@ class TestJob(unittest.TestCase):
                     'runtime_constraints': {},
                     'script_parameters': {
                         'tasks': [{
-                            'task.env': {'TMPDIR': '$(task.tmpdir)'},
-                            'command': ['ls']
+                            'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
+                            'command': ['ls', '$(task.outdir)']
                         }],
                     },
                     'script_version': 'master',
@@ -44,7 +57,7 @@ class TestJob(unittest.TestCase):
                     'repository': 'arvados',
                     'script': 'crunchrunner',
                     'runtime_constraints': {
-                        'docker_image': 'arvados/jobs',
+                        'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
                         'min_cores_per_node': 1,
                         'min_ram_mb_per_node': 1024,
                         'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
@@ -54,16 +67,24 @@ class TestJob(unittest.TestCase):
                 filters=[['repository', '=', 'arvados'],
                          ['script', '=', 'crunchrunner'],
                          ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
-                         ['docker_image_locator', 'in docker', 'arvados/jobs']]
+                         ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]]
             )
 
     # The test passes some fields in builder.resources
     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
-    def test_resource_requirements(self):
+    @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
+    def test_resource_requirements(self, list_images_in_arv):
         runner = mock.MagicMock()
         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
         runner.ignore_docker_for_reuse = False
-        document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3")
+        runner.num_retries = 0
+        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")
+
 
         tool = {
             "inputs": [],
@@ -73,12 +94,20 @@ class TestJob(unittest.TestCase):
                 "coresMin": 3,
                 "ramMin": 3000,
                 "tmpdirMin": 4000
+            }, {
+                "class": "http://arvados.org/cwl#RuntimeConstraints",
+                "keep_cache": 512,
+                "outputDirType": "keep_output_dir"
+            }, {
+                "class": "http://arvados.org/cwl#APIRequirement",
             }],
             "baseCommand": "ls"
         }
-        arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, avsc_names=avsc_names)
+        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
+        arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
+                                                 make_fs_access=make_fs_access, loader=Loader({}))
         arvtool.formatgraph = None
-        for j in arvtool.job({}, mock.MagicMock(), basedir=""):
+        for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
             j.run()
         runner.api.jobs().create.assert_called_with(
             body={
@@ -86,7 +115,8 @@ class TestJob(unittest.TestCase):
                 'runtime_constraints': {},
                 'script_parameters': {
                     'tasks': [{
-                        'task.env': {'TMPDIR': '$(task.tmpdir)'},
+                        'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
+                        'task.keepTmpOutput': True,
                         'command': ['ls']
                     }]
             },
@@ -95,20 +125,21 @@ class TestJob(unittest.TestCase):
                 'repository': 'arvados',
                 'script': 'crunchrunner',
                 'runtime_constraints': {
-                    'docker_image': 'arvados/jobs',
+                    'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
                     'min_cores_per_node': 3,
                     'min_ram_mb_per_node': 3000,
-                    'min_scratch_mb_per_node': 5024 # tmpdirSize + outdirSize
+                    'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
+                    'keep_cache_mb_per_task': 512
                 }
             },
             find_or_create=True,
             filters=[['repository', '=', 'arvados'],
                      ['script', '=', 'crunchrunner'],
                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
-                     ['docker_image_locator', 'in docker', 'arvados/jobs']])
+                     ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]])
 
-    @mock.patch("arvados.collection.Collection")
-    def test_done(self, col):
+    @mock.patch("arvados.collection.CollectionReader")
+    def test_done(self, reader):
         api = mock.MagicMock()
 
         runner = mock.MagicMock()
@@ -117,7 +148,7 @@ class TestJob(unittest.TestCase):
         runner.num_retries = 0
         runner.ignore_docker_for_reuse = False
 
-        col().open.return_value = []
+        reader().open.return_value = []
         api.collections().list().execute.side_effect = ({"items": []},
                                                         {"items": [{"manifest_text": "XYZ"}]})
 
@@ -151,8 +182,8 @@ class TestJob(unittest.TestCase):
                   'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
                   'name': 'Output 9999999 of testjob'})
 
-    @mock.patch("arvados.collection.Collection")
-    def test_done_use_existing_collection(self, col):
+    @mock.patch("arvados.collection.CollectionReader")
+    def test_done_use_existing_collection(self, reader):
         api = mock.MagicMock()
 
         runner = mock.MagicMock()
@@ -160,7 +191,7 @@ class TestJob(unittest.TestCase):
         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
         runner.num_retries = 0
 
-        col().open.return_value = []
+        reader().open.return_value = []
         api.collections().list().execute.side_effect = ({"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},)
 
         arvjob = arvados_cwl.ArvadosJob(runner)
@@ -184,3 +215,83 @@ class TestJob(unittest.TestCase):
             mock.call().execute(num_retries=0)])
 
         self.assertFalse(api.collections().create.called)
+
+
+class TestWorkflow(unittest.TestCase):
+    # The test passes no builder.resources
+    # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
+    @mock.patch("arvados.collection.Collection")
+    @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
+    def test_run(self, list_images_in_arv, mockcollection):
+        arvados_cwl.add_arv_hints()
+
+        api = mock.MagicMock()
+        api._rootDesc = arvados.api('v1')._rootDesc
+        runner = arvados_cwl.ArvCwlRunner(api)
+        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.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")
+
+        tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
+        metadata["cwlVersion"] = tool["cwlVersion"]
+
+        mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
+
+        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
+        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.formatgraph = None
+        it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
+        it.next().run()
+        it.next().run()
+
+        with open("tests/wf/scatter2_subwf.cwl") as f:
+            subwf = f.read()
+
+        runner.api.jobs().create.assert_called_with(
+            body={
+                'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
+                'repository': 'arvados',
+                'script_version': 'master',
+                'script': 'crunchrunner',
+                'script_parameters': {
+                    'tasks': [{'task.env': {
+                        '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'
+                               },
+                    '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'}]},
+                'runtime_constraints': {
+                    'min_scratch_mb_per_node': 2048,
+                    'min_cores_per_node': 1,
+                    'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
+                    'min_ram_mb_per_node': 1024
+                },
+                'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
+            filters=[['repository', '=', 'arvados'],
+                     ['script', '=', 'crunchrunner'],
+                     ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
+                     ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]],
+            find_or_create=True)
+
+        mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
+        mockcollection().open().__enter__().write.assert_has_calls([mock.call('{sleeptime: 5}')])
+
+    def test_default_work_api(self):
+        arvados_cwl.add_arv_hints()
+
+        api = mock.MagicMock()
+        api._rootDesc = arvados.api('v1')._rootDesc
+        del api._rootDesc.get('resources')['jobs']['methods']['create']
+        runner = arvados_cwl.ArvCwlRunner(api)
+        self.assertEqual(runner.work_api, 'containers')