Merge pull request #53 from datmo/issue-36

Added check for status if task is already run

datmo/core/controller/task.py

@@ -185,7 +185,7 @@ def run(self, task_id, snapshot_dict=None, task_dict=None):
             of whether the user provides another value for `visible`.
         task_dict : dict
             set of parameters to characterize the task run
-            (default is None, which translate to {}, see datmo.core.entity.task for more details on inputs)
+            (default is None, which translate to {}, see datmo.core.entity.task.Task for more details on inputs)
 
         Returns
         -------
@@ -211,6 +211,13 @@ def run(self, task_id, snapshot_dict=None, task_dict=None):
         # Obtain Task to run
         task_obj = self.dal.task.get_by_id(task_id)
 
+        if task_obj.status==None:
+            task_obj.status = 'RUNNING'
+        else:
+            raise TaskRunException(__("error",
+                                      "cli.task.run.already_running",
+                                      task_obj.id))
+
         # Create Task directory for user during run
         task_dirpath = os.path.join("datmo_tasks", task_obj.id)
         try:
@@ -249,7 +256,7 @@ def run(self, task_id, snapshot_dict=None, task_dict=None):
         # Set the parameters set in the task
         environment_run_options = {
             "command": task_obj.command,
-            "ports": task_obj.ports,
+            "ports": [] if task_obj.ports is None else task_obj.ports,
             "gpu": task_obj.gpu,
             "name": "datmo-task-" + task_obj.id,
             "volumes": {

datmo/core/controller/test/test_task.py

@@ -16,7 +16,7 @@
 from datmo.core.controller.environment.environment import EnvironmentController
 from datmo.core.controller.task import TaskController
 from datmo.core.util.exceptions import EntityNotFound, \
-    EnvironmentExecutionException
+    EnvironmentExecutionException, TaskRunException
 
 
 class TestTaskController():
@@ -173,7 +173,7 @@ def test_run(self):
         assert task_obj.id == updated_task_obj.id
 
         assert updated_task_obj.before_snapshot_id
-        assert updated_task_obj.ports == []
+        assert updated_task_obj.ports == None
         assert updated_task_obj.gpu == False
         assert updated_task_obj.interactive == False
         assert updated_task_obj.task_dirpath
@@ -194,7 +194,7 @@ def test_run(self):
         failed = False
         try:
              self.task.run(task_obj.id)
-        except EnvironmentExecutionException:
+        except TaskRunException:
             failed = True
         assert failed
 
@@ -217,38 +217,55 @@ def test_run(self):
         try:
             self.task.run(task_obj.id,
                           snapshot_dict=snapshot_dict)
-        except EnvironmentExecutionException:
+        except TaskRunException:
+            failed = True
+        assert failed
+
+        # Test when the specific task id is already RUNNING
+        # Create task in the project
+        task_obj_1 = self.task.create(input_dict)
+        self.task.dal.task.update({"id": task_obj_1.id, "status": "RUNNING"})
+        # Create environment_driver definition
+        env_def_path = os.path.join(self.project.home,
+                                    "Dockerfile")
+        with open(env_def_path, "w") as f:
+            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))
+
+        failed = False
+        try:
+            self.task.run(task_obj_1.id)
+        except TaskRunException:
             failed = True
         assert failed
 
         # 4) Test option 4
 
         # Create a new task in the project
-        task_obj_1 = self.task.create(input_dict)
+        task_obj_2 = self.task.create(input_dict)
 
         # Run another task in the project
-        updated_task_obj_1 = self.task.run(task_obj_1.id,
+        updated_task_obj_2 = self.task.run(task_obj_2.id,
                                            snapshot_dict=snapshot_dict)
 
-        assert task_obj_1.id == updated_task_obj_1.id
-
-        assert updated_task_obj_1.before_snapshot_id
-        assert updated_task_obj_1.ports == []
-        assert updated_task_obj_1.gpu == False
-        assert updated_task_obj_1.interactive == False
-        assert updated_task_obj_1.task_dirpath
-        assert updated_task_obj_1.log_filepath
-        assert updated_task_obj_1.start_time
-
-        assert updated_task_obj_1.after_snapshot_id
-        assert updated_task_obj_1.run_id
-        assert updated_task_obj_1.logs
-        assert "accuracy" in updated_task_obj_1.logs
-        assert updated_task_obj_1.results
-        assert updated_task_obj_1.results == {"accuracy": "0.45"}
-        assert updated_task_obj_1.status == "SUCCESS"
-        assert updated_task_obj_1.end_time
-        assert updated_task_obj_1.duration
+        assert task_obj_2.id == updated_task_obj_2.id
+
+        assert updated_task_obj_2.before_snapshot_id
+        assert updated_task_obj_2.ports == None
+        assert updated_task_obj_2.gpu == False
+        assert updated_task_obj_2.interactive == False
+        assert updated_task_obj_2.task_dirpath
+        assert updated_task_obj_2.log_filepath
+        assert updated_task_obj_2.start_time
+
+        assert updated_task_obj_2.after_snapshot_id
+        assert updated_task_obj_2.run_id
+        assert updated_task_obj_2.logs
+        assert "accuracy" in updated_task_obj_2.logs
+        assert updated_task_obj_2.results
+        assert updated_task_obj_2.results == {"accuracy": "0.45"}
+        assert updated_task_obj_2.status == "SUCCESS"
+        assert updated_task_obj_2.end_time
+        assert updated_task_obj_2.duration
 
         # 5) Test option 5
 
@@ -281,7 +298,7 @@ def test_run(self):
         updated_task_obj_2 = self.task.run(task_obj_2.id)
 
         assert updated_task_obj_2.before_snapshot_id
-        assert updated_task_obj_2.ports == []
+        assert updated_task_obj_2.ports == None
         assert updated_task_obj_2.gpu == False
         assert updated_task_obj_2.interactive == False
         assert updated_task_obj_2.task_dirpath
@@ -291,7 +308,7 @@ def test_run(self):
         assert updated_task_obj_2.after_snapshot_id
         assert updated_task_obj_2.run_id
         assert updated_task_obj_2.logs
-        assert "accuracy" in updated_task_obj_1.logs
+        assert "accuracy" in updated_task_obj_2.logs
         assert updated_task_obj_2.results
         assert updated_task_obj_2.results == {"accuracy": "0.56"}
         assert updated_task_obj_2.status == "SUCCESS"

datmo/core/entity/task.py

@@ -24,11 +24,11 @@ class Task():
             command that is used by the task
         before_snapshot_id : str, optional
             snapshot created before the task is run
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         ports : list, optional
             list of string mappings from host system (left) to environment (right)
             (e.g. ["9999:9999", "8888:8888"])
-            (default is [], which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         gpu : bool, optional
             boolean to signify if run requires gpu
             (default is False, which means no gpu unless specified)
@@ -37,28 +37,28 @@ class Task():
             (default is False, which means no interactive mode unless specified)
         task_dirpath : str, optional
             task directory path relative to the project root
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         log_filepath : str, optional
             log filepath relative to the project root
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         start_time : datetime.datetime
             timestamp for the beginning time of the task
             (default is None, which means it isn't set yet)
         after_snapshot_id : str, optional
             snapshot created after the task is run
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         run_id : str, optional
             run id for the run (different from environment id and task id)
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         logs : str, optional
             string output of logs
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         status : str, optional
             status of the current task
-            (default is "", which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         results : dict, optional
             dictionary containing output results from the task
-            (default is {}, which means it isn't set yet)
+            (default is None, which means it isn't set yet)
         end_time : datetime.datetime, optional
             timestamp for the beginning time of the task
             (default is None, which means it isn't set yet)
@@ -122,20 +122,20 @@ def __init__(self, dictionary):
         self.command = dictionary['command']
 
         # Pre-Execution
-        self.before_snapshot_id = dictionary.get('before_snapshot_id', "")
-        self.ports = dictionary.get('ports', [])
+        self.before_snapshot_id = dictionary.get('before_snapshot_id', None)
+        self.ports = dictionary.get('ports', None)
         self.gpu = dictionary.get('gpu', False)
         self.interactive = dictionary.get('interactive', False)
-        self.task_dirpath = dictionary.get('task_dirpath', "")
-        self.log_filepath = dictionary.get('log_filepath', "")
+        self.task_dirpath = dictionary.get('task_dirpath', None)
+        self.log_filepath = dictionary.get('log_filepath', None)
         self.start_time = dictionary.get('start_time', None)
 
         # Post-Execution
-        self.after_snapshot_id = dictionary.get('after_snapshot_id', "")
-        self.run_id = dictionary.get('run_id', "")
-        self.logs = dictionary.get('logs', "")
-        self.status = dictionary.get('status', "")
-        self.results = dictionary.get('results', {})
+        self.after_snapshot_id = dictionary.get('after_snapshot_id', None)
+        self.run_id = dictionary.get('run_id', None)
+        self.logs = dictionary.get('logs', None)
+        self.status = dictionary.get('status', None)
+        self.results = dictionary.get('results', None)
         self.end_time = dictionary.get('end_time', None)
         self.duration = dictionary.get('duration', None)
 

datmo/core/entity/test/test_task.py

@@ -18,20 +18,20 @@ def test_init_no_id(self):
         for k, v in self.input_dict.items():
             assert getattr(task_entity, k) == v
         assert task_entity.id == None
-        assert task_entity.before_snapshot_id == ""
-        assert task_entity.ports == []
+        assert task_entity.before_snapshot_id == None
+        assert task_entity.ports == None
         assert task_entity.gpu == False
         assert task_entity.interactive == False
-        assert task_entity.task_dirpath == ""
-        assert task_entity.log_filepath == ""
+        assert task_entity.task_dirpath == None
+        assert task_entity.log_filepath == None
         assert task_entity.start_time == None
 
         # Post-Execution
-        assert task_entity.after_snapshot_id == ""
-        assert task_entity.run_id == ""
-        assert task_entity.logs == ""
-        assert task_entity.status == ""
-        assert task_entity.results == {}
+        assert task_entity.after_snapshot_id == None
+        assert task_entity.run_id == None
+        assert task_entity.logs == None
+        assert task_entity.status == None
+        assert task_entity.results == None
         assert task_entity.end_time == None
         assert task_entity.duration == None
         assert task_entity.created_at
@@ -43,20 +43,20 @@ def test_init_with_id(self):
 
         for k, v in self.input_dict.items():
             assert getattr(task_entity, k) == v
-        assert task_entity.before_snapshot_id == ""
-        assert task_entity.ports == []
+        assert task_entity.before_snapshot_id == None
+        assert task_entity.ports == None
         assert task_entity.gpu == False
         assert task_entity.interactive == False
-        assert task_entity.task_dirpath == ""
-        assert task_entity.log_filepath == ""
+        assert task_entity.task_dirpath == None
+        assert task_entity.log_filepath == None
         assert task_entity.start_time == None
 
         # Post-Execution
-        assert task_entity.after_snapshot_id == ""
-        assert task_entity.run_id == ""
-        assert task_entity.logs == ""
-        assert task_entity.status == ""
-        assert task_entity.results == {}
+        assert task_entity.after_snapshot_id == None
+        assert task_entity.run_id == None
+        assert task_entity.logs == None
+        assert task_entity.status == None
+        assert task_entity.results == None
         assert task_entity.end_time == None
         assert task_entity.duration == None
         assert task_entity.created_at

datmo/core/util/lang/en.py

@@ -28,6 +28,7 @@
         "cli.general.method.not_found": "Method %s.%s not found",
         "cli.project": "No project found in the current directory: %s",
         "cli.task.run": "Error while running the task: %s",
+        "cli.task.run.already_running": "Already task running with id: %s",
         "cli.task.stop": "Error while stopping the task: %s",
         "util.misc_functions.get_filehash": "Filepath does not point to a valid file: %s",
         "util.misc_functions.mutually_exclusive": "Mutually exclusive arguments passed: %s",

datmo/task.py

@@ -28,17 +28,17 @@ class Task():
         id of session associated with task
     command : str
         command that is used by the task
-    status : str
+    status : str or None
         status of the current task
-    start_time : datetime.datetime
+    start_time : datetime.datetime or None
         timestamp for the beginning time of the task
-    end_time : datetime.datetime
+    end_time : datetime.datetime or None
         timestamp for the end time of the task
-    duration : datetime.timedelta
+    duration : datetime.timedelta or None
         delta between start and end times
-    logs : str
+    logs : str or None
         string output of logs
-    results : dict
+    results : dict or None
         dictionary containing output results from the task
 
     Methods

datmo/test/test_task.py

@@ -43,12 +43,12 @@ def test_task_entity_instantiate(self):
 
         for k, v in input_dict.items():
             assert getattr(task_entity, k) == v
-        assert task_entity.status == ""
+        assert task_entity.status == None
         assert task_entity.start_time == None
         assert task_entity.end_time == None
         assert task_entity.duration == None
-        assert task_entity.logs == ""
-        assert task_entity.results == {}
+        assert task_entity.logs == None
+        assert task_entity.results == None
 
     def test_run(self):
         # 1) Run task with no commit or code available (cannot save states before), string command