closes #11362 Merge branch '11362-missing-input-sdk/cwl'
[arvados.git] / sdk / cwl / tests / test_job.py
index 93b5d39ffa6eb7e45efba47c64b69fb7a95a33d8..99dd3cb669729f09d6c45d914178a5e56a57ffd6 100644 (file)
@@ -4,71 +4,77 @@ import logging
 import mock
 import os
 import unittest
+import copy
+import StringIO
 
 import arvados
 import arvados_cwl
 import cwltool.process
 from schema_salad.ref_resolver import Loader
+from schema_salad.sourceline import cmap
+from .mock_discovery import get_rootDesc
+from .matcher import JsonDiffMatcher
 
 if not os.getenv('ARVADOS_DEBUG'):
     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
     logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
 
-
 class TestJob(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.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
-        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",
-            "arguments": [{"valueFrom": "$(runtime.outdir)"}]
-        }
-        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
-        arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
-                                                 basedir="", make_fs_access=make_fs_access, loader=Loader({}))
-        arvtool.formatgraph = None
-        for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
-            j.run()
-            runner.api.jobs().create.assert_called_with(
-                body={
-                    'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
-                    'runtime_constraints': {},
-                    'script_parameters': {
-                        'tasks': [{
-                            'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
-                            'command': ['ls', '$(task.outdir)']
-                        }],
-                    },
-                    'script_version': 'master',
-                    'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
-                    'repository': 'arvados',
-                    'script': 'crunchrunner',
-                    'runtime_constraints': {
-                        '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
-                    }
-                },
-                find_or_create=True,
-                filters=[['repository', '=', 'arvados'],
-                         ['script', '=', 'crunchrunner'],
-                         ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
-                         ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]]
-            )
+        for enable_reuse in (True, False):
+            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_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
+
+            tool = cmap({
+                "inputs": [],
+                "outputs": [],
+                "baseCommand": "ls",
+                "arguments": [{"valueFrom": "$(runtime.outdir)"}]
+            })
+            make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
+                                         collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
+            arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
+                                                     basedir="", make_fs_access=make_fs_access, loader=Loader({}))
+            arvtool.formatgraph = None
+            for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
+                j.run(enable_reuse=enable_reuse)
+                runner.api.jobs().create.assert_called_with(
+                    body=JsonDiffMatcher({
+                        'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
+                        'runtime_constraints': {},
+                        'script_parameters': {
+                            'tasks': [{
+                                'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
+                                'command': ['ls', '$(task.outdir)']
+                            }],
+                        },
+                        'script_version': 'master',
+                        'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
+                        'repository': 'arvados',
+                        'script': 'crunchrunner',
+                        'runtime_constraints': {
+                            'docker_image': 'arvados/jobs',
+                            'min_cores_per_node': 1,
+                            'min_ram_mb_per_node': 1024,
+                            'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
+                        }
+                    }),
+                    find_or_create=enable_reuse,
+                    filters=[['repository', '=', 'arvados'],
+                             ['script', '=', 'crunchrunner'],
+                             ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
+                             ['docker_image_locator', 'in docker', 'arvados/jobs']]
+                )
 
     # The test passes some fields in builder.resources
     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
@@ -103,14 +109,15 @@ class TestJob(unittest.TestCase):
             }],
             "baseCommand": "ls"
         }
-        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
+        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({}))
         arvtool.formatgraph = None
         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
             j.run()
         runner.api.jobs().create.assert_called_with(
-            body={
+            body=JsonDiffMatcher({
                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
                 'runtime_constraints': {},
                 'script_parameters': {
@@ -121,22 +128,22 @@ class TestJob(unittest.TestCase):
                     }]
             },
             'script_version': 'master',
-                'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
+                'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
                 'repository': 'arvados',
                 'script': 'crunchrunner',
                 'runtime_constraints': {
-                    'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
+                    'docker_image': 'arvados/jobs',
                     'min_cores_per_node': 3,
-                    'min_ram_mb_per_node': 3000,
+                    'min_ram_mb_per_node': 3512,     # ramMin + keep_cache
                     '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:'+arvados_cwl.__version__]])
+                     ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
+                     ['docker_image_locator', 'in docker', 'arvados/jobs']])
 
     @mock.patch("arvados.collection.CollectionReader")
     def test_done(self, reader):
@@ -148,7 +155,11 @@ class TestJob(unittest.TestCase):
         runner.num_retries = 0
         runner.ignore_docker_for_reuse = False
 
-        reader().open.return_value = []
+        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
+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"}]})
 
@@ -157,6 +168,7 @@ class TestJob(unittest.TestCase):
         arvjob.builder = mock.MagicMock()
         arvjob.output_callback = mock.MagicMock()
         arvjob.collect_outputs = mock.MagicMock()
+        arvjob.collect_outputs.return_value = {"out": "stuff"}
 
         arvjob.done({
             "state": "Complete",
@@ -182,6 +194,8 @@ class TestJob(unittest.TestCase):
                   'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
                   'name': 'Output 9999999 of testjob'})
 
+        arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
+
     @mock.patch("arvados.collection.CollectionReader")
     def test_done_use_existing_collection(self, reader):
         api = mock.MagicMock()
@@ -191,7 +205,12 @@ class TestJob(unittest.TestCase):
         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
         runner.num_retries = 0
 
-        reader().open.return_value = []
+        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
+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"}]},)
 
         arvjob = arvados_cwl.ArvadosJob(runner)
@@ -199,6 +218,7 @@ class TestJob(unittest.TestCase):
         arvjob.builder = mock.MagicMock()
         arvjob.output_callback = mock.MagicMock()
         arvjob.collect_outputs = mock.MagicMock()
+        arvjob.collect_outputs.return_value = {"out": "stuff"}
 
         arvjob.done({
             "state": "Complete",
@@ -216,6 +236,8 @@ class TestJob(unittest.TestCase):
 
         self.assertFalse(api.collections().create.called)
 
+        arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
+
 
 class TestWorkflow(unittest.TestCase):
     # The test passes no builder.resources
@@ -226,7 +248,8 @@ class TestWorkflow(unittest.TestCase):
         arvados_cwl.add_arv_hints()
 
         api = mock.MagicMock()
-        api._rootDesc = arvados.api('v1')._rootDesc
+        api._rootDesc = get_rootDesc()
+
         runner = arvados_cwl.ArvCwlRunner(api)
         self.assertEqual(runner.work_api, 'jobs')
 
@@ -243,7 +266,8 @@ class TestWorkflow(unittest.TestCase):
 
         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
 
-        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
+        make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
+                                         collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
         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)
@@ -256,8 +280,8 @@ class TestWorkflow(unittest.TestCase):
             subwf = f.read()
 
         runner.api.jobs().create.assert_called_with(
-            body={
-                'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
+            body=JsonDiffMatcher({
+                'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
                 'repository': 'arvados',
                 'script_version': 'master',
                 'script': 'crunchrunner',
@@ -274,24 +298,27 @@ class TestWorkflow(unittest.TestCase):
                 'runtime_constraints': {
                     'min_scratch_mb_per_node': 2048,
                     'min_cores_per_node': 1,
-                    'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
+                    'docker_image': 'arvados/jobs',
                     'min_ram_mb_per_node': 1024
                 },
-                'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
+                '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__]],
+                     ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
+                     ['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('{sleeptime: 5}')])
+        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
+        api._rootDesc = copy.deepcopy(get_rootDesc())
         del api._rootDesc.get('resources')['jobs']['methods']['create']
         runner = arvados_cwl.ArvCwlRunner(api)
         self.assertEqual(runner.work_api, 'containers')