sort workflows by total_conditions_satisfied (#3342)

* sort workflows by total_conditions_satisfied

* for wk -> for _, wk

* add test

Author: antgonza

qiita_db/metadata_template/prep_template.py

@@ -853,33 +853,32 @@ def _get_predecessors(workflow, node):
 
         all_workflows = [wk for wk in qdb.software.DefaultWorkflow.iter()]
         # are there any workflows with parameters?
-        check_requirements = False
-        default_parameters = {'prep': {}, 'sample': {}}
-        if [wk for wk in all_workflows if wk.parameters != default_parameters]:
-            check_requirements = True
         ST = qdb.metadata_template.sample_template.SampleTemplate
         workflows = []
         for wk in all_workflows:
             if wk.artifact_type == pt_artifact and pt_dt in wk.data_type:
-                if check_requirements and wk.parameters == default_parameters:
-                    continue
                 wk_params = wk.parameters
                 reqs_satisfied = True
+                total_conditions_satisfied = 0
 
                 if wk_params['sample']:
                     df = ST(self.study_id).to_dataframe(samples=list(self))
                     for k, v in wk_params['sample'].items():
                         if k not in df.columns or v not in df[k].unique():
                             reqs_satisfied = False
+                        else:
+                            total_conditions_satisfied += 1
 
                 if wk_params['prep']:
                     df = self.to_dataframe()
                     for k, v in wk_params['prep'].items():
                         if k not in df.columns or v not in df[k].unique():
                             reqs_satisfied = False
+                        else:
+                            total_conditions_satisfied += 1
 
                 if reqs_satisfied:
-                    workflows.append(wk)
+                    workflows.append((total_conditions_satisfied, wk))
 
         if not workflows:
             # raises option a.
@@ -888,8 +887,12 @@ def _get_predecessors(workflow, node):
                    'could be due to required parameters, please check the '
                    'available workflows.')
             raise ValueError(msg)
+
+        # let's just keep one, let's give it preference to the one with the
+        # most total_conditions_satisfied
+        workflows = sorted(workflows, key=lambda x: x[0], reverse=True)[:1]
         missing_artifacts = dict()
-        for wk in workflows:
+        for _, wk in workflows:
             missing_artifacts[wk] = dict()
             for node, degree in wk.graph.out_degree():
                 if degree != 0:

qiita_db/metadata_template/test/test_prep_template.py

@@ -1447,6 +1447,30 @@ def test_artifact_setter(self):
              'Pick closed-reference OTUs', 'Pick closed-reference OTUs',
              'Pick closed-reference OTUs'])
 
+        # at this point we can error all the previous steps, add a new smaller
+        # workflow and make sure you get the same one as before because it will
+        # have a higher match than the new one
+        for pj in wk.graph.nodes:
+            pj._set_error('Killed')
+        sql = """UPDATE qiita.default_workflow_data_type
+                 SET data_type_id = 1
+                 WHERE default_workflow_id = 2"""
+        qdb.sql_connection.perform_as_transaction(sql)
+        wk = pt.add_default_workflow(qdb.user.User('test@foo.bar'))
+        self.assertEqual(len(wk.graph.nodes), 5)
+        self.assertEqual(len(wk.graph.edges), 3)
+        self.assertCountEqual(
+            [x.command.name for x in wk.graph.nodes],
+            # we should have 2 split libraries and 3 close reference
+            ['Split libraries FASTQ', 'Split libraries FASTQ',
+             'Pick closed-reference OTUs', 'Pick closed-reference OTUs',
+             'Pick closed-reference OTUs'])
+        # let's return it back
+        sql = """UPDATE qiita.default_workflow_data_type
+                 SET data_type_id = 2
+                 WHERE default_workflow_id = 2"""
+        qdb.sql_connection.perform_as_transaction(sql)
+
         # now let's try to generate again and it should fail cause the jobs
         # are already created
         with self.assertRaisesRegex(ValueError, "Cannot create job because "