Scottx611x/derived node attributes bugfix (#2076)

* Update Galaxy mock data

* Fix `help`

* Add constants

* Dynamically create analysis group numbers based on the structure of our workflow

* Fix docstring

* Use temp var for easier readability

* Add mechanism to derive which Refinery file a workflow output was derived from, and in turn gain information about which analysis group said output belongs to.

* Fix tests

* Fix typo

* Add docstrings

* Don't overuse `@property`. Make method private

* Remove statements from debugging

* Change to default value

* Fix styling

* Remove comment stating the obvious

* Flip if/else logic to be more clear

* Add constant

* Add test coverage for `Workflowtool._has_dataset_collection_input()`

* Fix typo

* Add logic to determine analysis group numbering, and test coverage

* Fix spelling

* Add tests coverage for an edge case in the recursive calls to `WorkflowTool._get_refinery_input_file_id()`

* Fix bug where incorrect AnnotatedNode information was being indexed in solr. Add test coverage.

* Add test coverage for `_get_analysis_group_number`

* Simplify logic

* Tidy mock glaxy data

* DRY: condense repeated calls in tests

* Fix tests

* Address issue while creating expanded workflow graph where Dataset Collections were never considered

* Update AnalysisNodeConnection input creation and update tests

* Assert that we assign the correct name to input AnalysisNodeConnections

refinery/galaxy_connector/galaxy_workflow.py

@@ -13,6 +13,7 @@
 import networkx as nx
 
 from core.utils import get_aware_local_time
+import tool_manager
 
 logger = logging.getLogger(__name__)
 
@@ -529,6 +530,11 @@ def configure_workflow(workflow_dict, ret_list):
 def create_expanded_workflow_graph(dictionary):
     graph = nx.MultiDiGraph()
     steps = dictionary["steps"]
+    galaxy_input_types = [
+        'data_input',
+        tool_manager.models.WorkflowTool.DATA_COLLECTION_INPUT
+    ]
+
     # iterate over steps to create nodes
     for current_node_id, step in steps.iteritems():
         # ensure node id is an integer
@@ -553,9 +559,10 @@ def create_expanded_workflow_graph(dictionary):
             parent_node_id = input_connection["id"]
             # test if parent node is a tool node or an input node to pick the
             # right name for the outgoing edge
-            if graph.node[parent_node_id]['type'] == 'data_input':
-                parent_node_output_name = \
+            if graph.node[parent_node_id]['type'] in galaxy_input_types:
+                parent_node_output_name = (
                     steps[str(parent_node_id)]['inputs'][0]['name']
+                )
             else:
                 parent_node_output_name = input_connection['output_name']
 

refinery/tool_manager/models.py

@@ -466,6 +466,7 @@ class WorkflowTool(Tool):
     GALAXY_WORKFLOW_INVOCATION_DATA = "galaxy_workflow_invocation_data"
     GALAXY_TO_REFINERY_MAPPING_LIST = "galaxy_to_refinery_mapping_list"
     HISTORY_DATASET_COLLECTION_ASSOCIATION = "hdca"
+    INPUT_DATASET_COLLECTION = "Input Dataset Collection"
     LIST = "list"
     PAIRED = "paired"
     REVERSE = "reverse"
@@ -581,9 +582,9 @@ def create_analysis_input_node_connections(self):
                 analysis=self.analysis,
                 node=node,
                 direction=INPUT_CONNECTION,
-                name=node.name,
+                name=file_store_item.datafile.name,
                 step=0,
-                filename=file_store_item.datafile.name,
+                filename=self.INPUT_DATASET_COLLECTION,
                 is_refinery_file=file_store_item.is_local()
             )
 

refinery/tool_manager/tests.py

@@ -1429,10 +1429,22 @@ def test_analysis_node_connections_are_created_for_all_input_nodes(self):
         ]
         self.assertEqual(len(analysis_node_connections), len(tool_nodes))
         for analysis_node_connection in analysis_node_connections:
+            file_store_item = (
+                analysis_node_connection.node.get_file_store_item()
+            )
             self.assertEqual(
                 analysis_node_connection.direction,
                 INPUT_CONNECTION
             )
+            self.assertEqual(
+                analysis_node_connection.filename,
+                WorkflowTool.INPUT_DATASET_COLLECTION
+            )
+            self.assertEqual(analysis_node_connection.step, 0)
+            self.assertEqual(
+                analysis_node_connection.name,
+                file_store_item.datafile.name
+            )
 
     def test_galaxy_parameter_dict_creation(self):
         self.create_valid_tool(