From d2f368e89cce24230ef16efb305e30d9d102ffa7 Mon Sep 17 00:00:00 2001
From: Jake Tronge <jtronge@lanl.gov>
Date: Thu, 23 May 2024 08:48:17 -0600
Subject: [PATCH] Check for missing inputs and set workflow defaults

---
 beeflow/common/parser/parser.py              | 14 +++++++++-----
 beeflow/tests/cf.cwl                         |  4 +++-
 beeflow/tests/test_parser.py                 | 14 ++++++++++++--
 ci/test_workflows/missing-input/input.yml    |  1 +
 ci/test_workflows/missing-input/workflow.cwl | 20 ++++++++++++++++++++
 5 files changed, 45 insertions(+), 8 deletions(-)
 create mode 100644 ci/test_workflows/missing-input/input.yml
 create mode 100644 ci/test_workflows/missing-input/workflow.cwl

diff --git a/beeflow/common/parser/parser.py b/beeflow/common/parser/parser.py
index 7a03cddd1..fcf9b8f85 100644
--- a/beeflow/common/parser/parser.py
+++ b/beeflow/common/parser/parser.py
@@ -97,12 +97,13 @@ def resolve_input(input_, type_):
             """
             # Use parsed input parameter for input value if it exists
             input_id = _shortname(input_.id)
-            if input_id not in self.params:
-                return None
-            if not isinstance(self.params[input_id], type_map[type_]):
+            value = self.params[input_id] if input_id in self.params else input_.default
+            if value is None:
+                raise CwlParseError(f"input {input_id} is missing from workflow job file")
+            if not isinstance(value, type_map[type_]):
                 raise CwlParseError("Input/param types do not match: "
-                                    f"{input_id}/{self.params[input_id]}")
-            return self.params[input_id]
+                                    f"{input_id}/{value}")
+            return value
 
         workflow_name = os.path.basename(cwl_path).split(".")[0]
         workflow_inputs = {InputParameter(_shortname(input_.id), input_.type,
@@ -245,6 +246,9 @@ def parse_step_outputs(cwl_out, step_outputs, stdout, stderr):
         :type stderr: str or None
         :rtype: list of StepOutput
         """
+        if not cwl_out:
+            return []
+
         out_short = list(map(_shortname, cwl_out))
         short_id = out_short[0].split("/")[0]
         # Inline step outputs already have short_id+"/" prepended
diff --git a/beeflow/tests/cf.cwl b/beeflow/tests/cf.cwl
index 144bfb31b..aa8b34116 100644
--- a/beeflow/tests/cf.cwl
+++ b/beeflow/tests/cf.cwl
@@ -4,7 +4,9 @@ class: Workflow
 cwlVersion: v1.0
 
 inputs:
-  infile: File
+  infile:
+    type: File
+    default: 'infile'
 
 outputs:
   clamr_dir:
diff --git a/beeflow/tests/test_parser.py b/beeflow/tests/test_parser.py
index ed41b58f1..0c9514e9e 100644
--- a/beeflow/tests/test_parser.py
+++ b/beeflow/tests/test_parser.py
@@ -3,7 +3,7 @@
 
 from pathlib import Path
 import unittest
-from beeflow.common.parser import CwlParser
+from beeflow.common.parser import CwlParser, CwlParseError
 from beeflow.common.wf_data import (generate_workflow_id, Workflow, Task, Hint,
                                     StepInput, StepOutput, InputParameter, OutputParameter)
 
@@ -76,6 +76,16 @@ def test_parse_workflow_no_job(self):
         for task in tasks:
             self.assertEqual(task.workflow_id, workflow_id)
 
+    def test_parse_workflow_missing_input(self):
+        """Test parsing a workflow with a missing input value in the input file."""
+        cwl_wf_file = find('ci/test_workflows/missing-input/workflow.cwl')
+        cwl_job_yaml = find('ci/test_workflows/missing-input/input.yml')
+
+        workflow_id = generate_workflow_id()
+
+        with self.assertRaises(CwlParseError):
+            _, _ = self.parser.parse_workflow(workflow_id, cwl_wf_file, cwl_job_yaml)
+
 
 WORKFLOW_GOLD_SCRIPT = Workflow(
     name='clamr_wf',
@@ -247,7 +257,7 @@ def test_parse_workflow_no_job(self):
     name='cf',
     hints=[],
     requirements=[],
-    inputs={InputParameter(id='infile', type='File', value=None)},
+    inputs={InputParameter(id='infile', type='File', value='infile')},
     outputs={OutputParameter(id='ffmpeg_movie', type='File', value=None, source='ffmpeg/outfile'),
              OutputParameter(id='clamr_dir', type='File', value=None, source='clamr/outfile')},
     workflow_id=generate_workflow_id())
diff --git a/ci/test_workflows/missing-input/input.yml b/ci/test_workflows/missing-input/input.yml
new file mode 100644
index 000000000..3589fa97a
--- /dev/null
+++ b/ci/test_workflows/missing-input/input.yml
@@ -0,0 +1 @@
+b: 232
diff --git a/ci/test_workflows/missing-input/workflow.cwl b/ci/test_workflows/missing-input/workflow.cwl
new file mode 100644
index 000000000..e2f1868ad
--- /dev/null
+++ b/ci/test_workflows/missing-input/workflow.cwl
@@ -0,0 +1,20 @@
+class: Workflow
+cwlVersion: v1.2
+
+inputs:
+  a: int
+outputs: {}
+
+steps:
+  step0:
+    run:
+      class: CommandLineTool
+      baseCommand: echo
+      inputs:
+        a:
+          type: int
+          inputBinding: {}
+      outputs: []
+    in:
+      a: a
+    out: []