12483: Merge branch 'master' into 12483-writable-fs
[arvados.git] / sdk / cwl / tests / test_submit.py
index 731b1d561f277b82c8a7a8a668393db146c6b4bb..059b47275c9207279dc3aa8e0933980268e70512 100644 (file)
@@ -1,3 +1,7 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 import copy
 import cStringIO
 import functools
@@ -14,7 +18,7 @@ import arvados_cwl
 import arvados_cwl.runner
 import arvados.keep
 
-from .matcher import JsonDiffMatcher
+from .matcher import JsonDiffMatcher, StripYAMLComments
 from .mock_discovery import get_rootDesc
 
 import ruamel.yaml as yaml
@@ -34,7 +38,6 @@ def stubs(func):
         stubs.events = events
         stubs.keepdocker = keepdocker
 
-
         def putstub(p, **kwargs):
             return "%s+%i" % (hashlib.md5(p).hexdigest(), len(p))
         keep_client1().put.side_effect = putstub
@@ -53,33 +56,43 @@ def stubs(func):
             "uuid": stubs.fake_user_uuid,
         }
         stubs.api.collections().list().execute.return_value = {"items": []}
-        stubs.api.collections().create().execute.side_effect = ({
-            "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz1",
-            "portable_data_hash": "99999999999999999999999999999991+99",
-            "manifest_text": ""
-        }, {
-            "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2",
-            "portable_data_hash": "99999999999999999999999999999992+99",
-            "manifest_text": "./tool 00000000000000000000000000000000+0 0:0:submit_tool.cwl 0:0:blub.txt"
-        },
-        {
-            "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz4",
-            "portable_data_hash": "99999999999999999999999999999994+99",
-            "manifest_text": ""
-        },
-        {
-            "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz5",
-            "portable_data_hash": "99999999999999999999999999999995+99",
-            "manifest_text": ""
-        },
-        {
-            "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz6",
-            "portable_data_hash": "99999999999999999999999999999996+99",
-            "manifest_text": ""
+
+        class CollectionExecute(object):
+            def __init__(self, exe):
+                self.exe = exe
+            def execute(self, num_retries=None):
+                return self.exe
+
+        def collection_createstub(created_collections, body, ensure_unique_name=None):
+            mt = body["manifest_text"]
+            uuid = "zzzzz-4zz18-zzzzzzzzzzzzzz%d" % len(created_collections)
+            pdh = "%s+%i" % (hashlib.md5(mt).hexdigest(), len(mt))
+            created_collections[uuid] = {
+                "uuid": uuid,
+                "portable_data_hash": pdh,
+                "manifest_text": mt
+            }
+            return CollectionExecute(created_collections[uuid])
+
+        def collection_getstub(created_collections, uuid):
+            for v in created_collections.itervalues():
+                if uuid in (v["uuid"], v["portable_data_hash"]):
+                    return CollectionExecute(v)
+
+        created_collections = {
+            "99999999999999999999999999999998+99": {
+                "uuid": "",
+                "portable_data_hash": "99999999999999999999999999999998+99",
+                "manifest_text": ". 99999999999999999999999999999998+99 0:0:file1.txt"
+            },
+            "99999999999999999999999999999994+99": {
+                "uuid": "",
+                "portable_data_hash": "99999999999999999999999999999994+99",
+                "manifest_text": ". 99999999999999999999999999999994+99 0:0:expect_arvworkflow.cwl"
+            }
         }
-        )
-        stubs.api.collections().get().execute.return_value = {
-            "portable_data_hash": "99999999999999999999999999999993+99", "manifest_text": "./tool 00000000000000000000000000000000+0 0:0:submit_tool.cwl 0:0:blub.txt"}
+        stubs.api.collections().create.side_effect = functools.partial(collection_createstub, created_collections)
+        stubs.api.collections().get.side_effect = functools.partial(collection_getstub, created_collections)
 
         stubs.expect_job_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
         stubs.api.jobs().create().execute.return_value = {
@@ -106,7 +119,7 @@ def stubs(func):
             'script_parameters': {
                 'x': {
                     'basename': 'blorp.txt',
-                    'location': 'keep:99999999999999999999999999999992+99/blorp.txt',
+                    'location': 'keep:169f39d466a5438ac4a90e779bf750c7+53/blorp.txt',
                     'class': 'File'
                 },
                 'y': {
@@ -123,8 +136,7 @@ def stubs(func):
                     }],
                     'class': 'Directory'
                 },
-                'cwl:tool':
-                '99999999999999999999999999999994+99/workflow.cwl#main'
+                'cwl:tool': '3fffdeaa75e018172e1b583425f4ebff+60/workflow.cwl#main'
             },
             'repository': 'arvados',
             'script_version': 'master',
@@ -141,12 +153,21 @@ def stubs(func):
                     'runtime_constraints': {'docker_image': 'arvados/jobs:'+arvados_cwl.__version__, 'min_ram_mb_per_node': 1024},
                     'script_parameters': {
                         'y': {"value": {'basename': '99999999999999999999999999999998+99', 'location': 'keep:99999999999999999999999999999998+99', 'class': 'Directory'}},
-                        'x': {"value": {'basename': 'blorp.txt', 'class': 'File', 'location': 'keep:99999999999999999999999999999992+99/blorp.txt'}},
+                        'x': {"value": {
+                            'basename': 'blorp.txt',
+                            'class': 'File',
+                            'location': 'keep:169f39d466a5438ac4a90e779bf750c7+53/blorp.txt',
+                            "size": 16
+                        }},
                         'z': {"value": {'basename': 'anonymous', 'class': 'Directory',
                               'listing': [
-                                  {'basename': 'renamed.txt', 'class': 'File', 'location': 'keep:99999999999999999999999999999998+99/file1.txt'}
+                                  {
+                                      'basename': 'renamed.txt',
+                                      'class': 'File', 'location':
+                                      'keep:99999999999999999999999999999998+99/file1.txt'
+                                  }
                               ]}},
-                        'cwl:tool': '99999999999999999999999999999994+99/workflow.cwl#main',
+                        'cwl:tool': '3fffdeaa75e018172e1b583425f4ebff+60/workflow.cwl#main',
                         'arv:enable_reuse': True,
                         'arv:on_error': 'continue'
                     },
@@ -190,10 +211,21 @@ def stubs(func):
                 '/var/lib/cwl/cwl.input.json': {
                     'kind': 'json',
                     'content': {
-                        'y': {'basename': '99999999999999999999999999999998+99', 'location': 'keep:99999999999999999999999999999998+99', 'class': 'Directory'},
-                        'x': {'basename': u'blorp.txt', 'class': 'File', 'location': u'keep:99999999999999999999999999999992+99/blorp.txt'},
+                        'y': {
+                            'basename': '99999999999999999999999999999998+99',
+                            'location': 'keep:99999999999999999999999999999998+99',
+                            'class': 'Directory'},
+                        'x': {
+                            'basename': u'blorp.txt',
+                            'class': 'File',
+                            'location': u'keep:169f39d466a5438ac4a90e779bf750c7+53/blorp.txt',
+                            "size": 16
+                        },
                         'z': {'basename': 'anonymous', 'class': 'Directory', 'listing': [
-                            {'basename': 'renamed.txt', 'class': 'File', 'location': 'keep:99999999999999999999999999999998+99/file1.txt'}
+                            {'basename': 'renamed.txt',
+                             'class': 'File',
+                             'location': 'keep:99999999999999999999999999999998+99/file1.txt'
+                            }
                         ]}
                     },
                     'kind': 'json'
@@ -213,7 +245,8 @@ def stubs(func):
                 'vcpus': 1,
                 'ram': 1024*1024*1024
             },
-            "properties": {}
+            'use_existing': True,
+            'properties': {}
         }
 
         stubs.expect_workflow_uuid = "zzzzz-7fd4e-zzzzzzzzzzzzzzz"
@@ -245,21 +278,24 @@ class TestSubmit(unittest.TestCase):
         self.assertEqual(exited, 0)
 
         stubs.api.collections().create.assert_has_calls([
-            mock.call(),
             mock.call(body=JsonDiffMatcher({
                 'manifest_text':
                 '. 5bcc9fe8f8d5992e6cf418dc7ce4dbb3+16 0:16:blub.txt\n',
                 'replication_desired': None,
                 'name': 'submit_tool.cwl dependencies',
             }), ensure_unique_name=True),
-            mock.call().execute(num_retries=4),
             mock.call(body=JsonDiffMatcher({
                 'manifest_text':
                 '. 979af1245a12a1fed634d4222473bfdc+16 0:16:blorp.txt\n',
                 'replication_desired': None,
                 'name': 'submit_wf.cwl input',
             }), ensure_unique_name=True),
-            mock.call().execute(num_retries=4)])
+            mock.call(body=JsonDiffMatcher({
+                'manifest_text':
+                '. 61df2ed9ee3eb7dd9b799e5ca35305fa+1217 0:1217:workflow.cwl\n',
+                'replication_desired': None,
+                'name': 'submit_wf.cwl',
+            }), ensure_unique_name=True)        ])
 
         arvdock.assert_has_calls([
             mock.call(stubs.api, {"class": "DockerRequirement", "dockerPull": "debian:8"}, True, None),
@@ -285,6 +321,7 @@ class TestSubmit(unittest.TestCase):
 
         expect_pipeline = copy.deepcopy(stubs.expect_pipeline_instance)
         expect_pipeline["components"]["cwl-runner"]["script_parameters"]["arv:enable_reuse"] = {"value": False}
+        expect_pipeline["properties"] = {"run_options": {"enable_job_reuse": False}}
 
         stubs.api.pipeline_instances().create.assert_called_with(
             body=JsonDiffMatcher(expect_pipeline))
@@ -428,21 +465,18 @@ class TestSubmit(unittest.TestCase):
             logging.exception("")
 
         stubs.api.collections().create.assert_has_calls([
-            mock.call(),
             mock.call(body=JsonDiffMatcher({
                 'manifest_text':
                 '. 5bcc9fe8f8d5992e6cf418dc7ce4dbb3+16 0:16:blub.txt\n',
                 'replication_desired': None,
                 'name': 'submit_tool.cwl dependencies',
             }), ensure_unique_name=True),
-            mock.call().execute(num_retries=4),
             mock.call(body=JsonDiffMatcher({
                 'manifest_text':
                 '. 979af1245a12a1fed634d4222473bfdc+16 0:16:blorp.txt\n',
                 'replication_desired': None,
                 'name': 'submit_wf.cwl input',
-            }), ensure_unique_name=True),
-            mock.call().execute(num_retries=4)])
+            }), ensure_unique_name=True)])
 
         expect_container = copy.deepcopy(stubs.expect_container_spec)
         stubs.api.container_requests().create.assert_called_with(
@@ -463,9 +497,41 @@ class TestSubmit(unittest.TestCase):
             logging.exception("")
 
         expect_container = copy.deepcopy(stubs.expect_container_spec)
-        expect_container["command"] = ['arvados-cwl-runner', '--local', '--api=containers', '--no-log-timestamps',
-                                                  '--disable-reuse', '--on-error=continue',
-                                                  '/var/lib/cwl/workflow.json#main', '/var/lib/cwl/cwl.input.json']
+        expect_container["command"] = [
+            'arvados-cwl-runner', '--local', '--api=containers', '--no-log-timestamps',
+            '--disable-reuse', '--on-error=continue',
+            '/var/lib/cwl/workflow.json#main', '/var/lib/cwl/cwl.input.json']
+        expect_container["use_existing"] = False
+
+        stubs.api.container_requests().create.assert_called_with(
+            body=JsonDiffMatcher(expect_container))
+        self.assertEqual(capture_stdout.getvalue(),
+                         stubs.expect_container_request_uuid + '\n')
+
+
+    @stubs
+    def test_submit_container_reuse_disabled_by_workflow(self, stubs):
+        capture_stdout = cStringIO.StringIO()
+
+        exited = arvados_cwl.main(
+            ["--submit", "--no-wait", "--api=containers", "--debug",
+             "tests/wf/submit_wf_no_reuse.cwl", "tests/submit_test_job.json"],
+            capture_stdout, sys.stderr, api_client=stubs.api, keep_client=stubs.keep_client)
+        self.assertEqual(exited, 0)
+
+        expect_container = copy.deepcopy(stubs.expect_container_spec)
+        expect_container["command"] = [
+            'arvados-cwl-runner', '--local', '--api=containers', '--no-log-timestamps',
+            '--disable-reuse', '--on-error=continue',
+            '/var/lib/cwl/workflow.json#main', '/var/lib/cwl/cwl.input.json']
+        expect_container["use_existing"] = False
+        expect_container["name"] = "submit_wf_no_reuse.cwl"
+        expect_container["mounts"]["/var/lib/cwl/workflow.json"]["content"]["$graph"][1]["hints"] = [
+            {
+                "class": "http://arvados.org/cwl#ReuseRequirement",
+                "enableReuse": False,
+            },
+        ]
 
         stubs.api.container_requests().create.assert_called_with(
             body=JsonDiffMatcher(expect_container))
@@ -544,6 +610,29 @@ class TestSubmit(unittest.TestCase):
         self.assertEqual(capture_stdout.getvalue(),
                          stubs.expect_container_request_uuid + '\n')
 
+    @stubs
+    def test_submit_container_trash_intermediate(self, stubs):
+        capture_stdout = cStringIO.StringIO()
+        try:
+            exited = arvados_cwl.main(
+                ["--submit", "--no-wait", "--api=containers", "--debug", "--trash-intermediate",
+                 "tests/wf/submit_wf.cwl", "tests/submit_test_job.json"],
+                capture_stdout, sys.stderr, api_client=stubs.api, keep_client=stubs.keep_client)
+            self.assertEqual(exited, 0)
+        except:
+            logging.exception("")
+
+        expect_container = copy.deepcopy(stubs.expect_container_spec)
+        expect_container["command"] = ['arvados-cwl-runner', '--local', '--api=containers', '--no-log-timestamps',
+                                       '--enable-reuse', '--on-error=continue',
+                                       "--trash-intermediate",
+                                       '/var/lib/cwl/workflow.json#main', '/var/lib/cwl/cwl.input.json']
+
+        stubs.api.container_requests().create.assert_called_with(
+            body=JsonDiffMatcher(expect_container))
+        self.assertEqual(capture_stdout.getvalue(),
+                         stubs.expect_container_request_uuid + '\n')
+
     @stubs
     def test_submit_container_output_tags(self, stubs):
         output_tags = "tag0,tag1,tag2"
@@ -650,7 +739,8 @@ class TestSubmit(unittest.TestCase):
                 'vcpus': 1,
                 'ram': 1073741824
             },
-            "properties": {}
+            'use_existing': True,
+            'properties': {}
         }
 
         stubs.api.container_requests().create.assert_called_with(
@@ -765,7 +855,8 @@ class TestSubmit(unittest.TestCase):
                 'vcpus': 1,
                 'ram': 1073741824
             },
-            "properties": {
+            'use_existing': True,
+            'properties': {
                 "template_uuid": "962eh-7fd4e-gkbzl62qqtfig37"
             }
         }
@@ -797,6 +888,31 @@ class TestSubmit(unittest.TestCase):
                          stubs.expect_container_request_uuid + '\n')
 
 
+    @stubs
+    def test_submit_container_project(self, stubs):
+        project_uuid = 'zzzzz-j7d0g-zzzzzzzzzzzzzzz'
+        capture_stdout = cStringIO.StringIO()
+        try:
+            exited = arvados_cwl.main(
+                ["--submit", "--no-wait", "--api=containers", "--debug", "--project-uuid="+project_uuid,
+                 "tests/wf/submit_wf.cwl", "tests/submit_test_job.json"],
+                capture_stdout, sys.stderr, api_client=stubs.api, keep_client=stubs.keep_client)
+            self.assertEqual(exited, 0)
+        except:
+            logging.exception("")
+
+        expect_container = copy.deepcopy(stubs.expect_container_spec)
+        expect_container["owner_uuid"] = project_uuid
+        expect_container["command"] = ['arvados-cwl-runner', '--local', '--api=containers', '--no-log-timestamps',
+                                       '--enable-reuse', '--on-error=continue', '--project-uuid='+project_uuid,
+                                       '/var/lib/cwl/workflow.json#main', '/var/lib/cwl/cwl.input.json']
+
+        stubs.api.container_requests().create.assert_called_with(
+            body=JsonDiffMatcher(expect_container))
+        self.assertEqual(capture_stdout.getvalue(),
+                         stubs.expect_container_request_uuid + '\n')
+
+
     @stubs
     def test_submit_job_runner_image(self, stubs):
         capture_stdout = cStringIO.StringIO()
@@ -878,7 +994,7 @@ class TestCreateTemplate(unittest.TestCase):
             'dataclass': 'File',
             'required': True,
             'type': 'File',
-            'value': '99999999999999999999999999999992+99/blorp.txt',
+            'value': '169f39d466a5438ac4a90e779bf750c7+53/blorp.txt',
         }
         expect_component['script_parameters']['y'] = {
             'dataclass': 'Collection',
@@ -997,7 +1113,8 @@ class TestCreateTemplate(unittest.TestCase):
 
 class TestCreateWorkflow(unittest.TestCase):
     existing_workflow_uuid = "zzzzz-7fd4e-validworkfloyml"
-    expect_workflow = open("tests/wf/expect_packed.cwl").read()
+    expect_workflow = StripYAMLComments(
+        open("tests/wf/expect_packed.cwl").read())
 
     @stubs
     def test_create(self, stubs):
@@ -1130,6 +1247,37 @@ class TestCreateWorkflow(unittest.TestCase):
                          self.existing_workflow_uuid + '\n')
 
 
+    @stubs
+    def test_create_collection_per_tool(self, stubs):
+        project_uuid = 'zzzzz-j7d0g-zzzzzzzzzzzzzzz'
+
+        capture_stdout = cStringIO.StringIO()
+
+        exited = arvados_cwl.main(
+            ["--create-workflow", "--debug",
+             "--api=containers",
+             "--project-uuid", project_uuid,
+             "tests/collection_per_tool/collection_per_tool.cwl"],
+            capture_stdout, sys.stderr, api_client=stubs.api)
+        self.assertEqual(exited, 0)
+
+        toolfile = "tests/collection_per_tool/collection_per_tool_packed.cwl"
+        expect_workflow = StripYAMLComments(open(toolfile).read())
+
+        body = {
+            "workflow": {
+                "owner_uuid": project_uuid,
+                "name": "collection_per_tool.cwl",
+                "description": "",
+                "definition": expect_workflow,
+            }
+        }
+        stubs.api.workflows().create.assert_called_with(
+            body=JsonDiffMatcher(body))
+
+        self.assertEqual(capture_stdout.getvalue(),
+                         stubs.expect_workflow_uuid + '\n')
+
 class TestTemplateInputs(unittest.TestCase):
     expect_template = {
         "components": {
@@ -1140,7 +1288,7 @@ class TestTemplateInputs(unittest.TestCase):
                 },
                 'script_parameters': {
                     'cwl:tool':
-                    '99999999999999999999999999999992+99/workflow.cwl#main',
+                    '6c5ee1cd606088106d9f28367cde1e41+60/workflow.cwl#main',
                     'optionalFloatInput': None,
                     'fileInput': {
                         'type': 'File',
@@ -1200,8 +1348,8 @@ class TestTemplateInputs(unittest.TestCase):
         expect_template = copy.deepcopy(self.expect_template)
         params = expect_template[
             "components"]["inputs_test.cwl"]["script_parameters"]
-        params["fileInput"]["value"] = '99999999999999999999999999999992+99/blorp.txt'
-        params["cwl:tool"] = '99999999999999999999999999999994+99/workflow.cwl#main'
+        params["fileInput"]["value"] = '169f39d466a5438ac4a90e779bf750c7+53/blorp.txt'
+        params["cwl:tool"] = '6c5ee1cd606088106d9f28367cde1e41+60/workflow.cwl#main'
         params["floatInput"]["value"] = 1.234
         params["boolInput"]["value"] = True