From 5fa648aa92e4eef4dc1ebf42b50816bcba995024 Mon Sep 17 00:00:00 2001
From: mvdbeek <m.vandenbeek@gmail.com>
Date: Mon, 7 Apr 2025 13:11:54 +0200
Subject: [PATCH] Ensure we copy the cached jobs' extension

The cached job might be skipped and null, in which case we should not
override the `expression.json` extension, because in that case the
dataset would not be considered as a null.
---
 lib/galaxy/tools/actions/__init__.py  |  1 +
 lib/galaxy_test/api/test_workflows.py | 69 +++++++++++++++++++++++++++
 lib/galaxy_test/base/populators.py    |  3 ++
 3 files changed, 73 insertions(+)

diff --git a/lib/galaxy/tools/actions/__init__.py b/lib/galaxy/tools/actions/__init__.py
index 4da5b270545..f322219d2c8 100644
--- a/lib/galaxy/tools/actions/__init__.py
+++ b/lib/galaxy/tools/actions/__init__.py
@@ -546,6 +546,7 @@ class DefaultToolAction(ToolAction):
                         if output_dataset.name == name:
                             create_datasets = False
                             completed_data = output_dataset.dataset
+                            ext = completed_data.extension
                             dataset = output_dataset.dataset.dataset
                             break
 
diff --git a/lib/galaxy_test/api/test_workflows.py b/lib/galaxy_test/api/test_workflows.py
index e4e4c87a473..53a0695a4e2 100644
--- a/lib/galaxy_test/api/test_workflows.py
+++ b/lib/galaxy_test/api/test_workflows.py
@@ -7997,3 +7997,72 @@ should_run:
             job_details = self.dataset_populator.get_job_details(job["id"], full=True).json()
             assert job_details["state"] == "skipped"
             assert job_details["copied_from_job_id"]
+
+    def test_run_workflow_use_cached_job_format_source_pick_param(self):
+        wf = """class: GalaxyWorkflow
+inputs: []
+steps:
+  create_2:
+    tool_id: create_2
+  cat_data_and_sleep:
+    # cat_data_and_sleep has format_source
+    tool_id: cat_data_and_sleep
+    in:
+      input1: create_2/out_file1
+    when: $(false)
+  pick_value:
+    tool_id: pick_value
+    tool_state:
+      style_cond:
+        pick_style: first
+        type_cond:
+          param_type: data
+          pick_from:
+          - __index__: 0
+            value:
+              __class__: RuntimeValue
+          - __index__: 1
+            value:
+              __class__: RuntimeValue
+    in:
+      style_cond|type_cond|pick_from_0|value:
+        source: cat_data_and_sleep/out_file1
+      style_cond|type_cond|pick_from_1|value:
+        source: create_2/out_file2
+  """
+        with self.dataset_populator.test_history() as history_id:
+            # Run once without cache
+            summary = self._run_workflow(wf, history_id=history_id, copy_inputs_to_history=True, wait=True)
+            # Run again, now with cache
+            summary = self._run_workflow(
+                wf,
+                history_id=history_id,
+                use_cached_job=True,
+                copy_inputs_to_history=True,
+            )
+            invocation_details = self.workflow_populator.get_invocation(summary.invocation_id, step_details=True)
+            conditional_step = invocation_details["steps"][1]
+            job_details = self.dataset_populator.get_job_details(conditional_step["jobs"][0]["id"], full=True).json()
+            assert job_details["state"] == "skipped"
+            assert job_details["copied_from_job_id"]
+            hda_id = job_details["outputs"]["out_file1"]["id"]
+            hda = self.dataset_populator.get_history_dataset_details(
+                history_id=job_details["history_id"], content_id=hda_id
+            )
+            assert hda["extension"] == "expression.json"
+            pick_value_step = invocation_details["steps"][2]
+            pick_value_details = self.dataset_populator.get_job_details(
+                pick_value_step["jobs"][0]["id"], full=True
+            ).json()
+            assert pick_value_details["state"] == "ok"
+            pick_value_hda_id = pick_value_details["outputs"]["data_param"]["id"]
+            pick_value_hda = self.dataset_populator.get_history_dataset_details(
+                history_id=job_details["history_id"], content_id=pick_value_hda_id
+            )
+            assert pick_value_hda["extension"] == "txt"
+            assert (
+                self.dataset_populator.get_history_dataset_content(
+                    history_id=summary.history_id, content_id=pick_value_hda_id
+                ).strip()
+                == "2"
+            )
diff --git a/lib/galaxy_test/base/populators.py b/lib/galaxy_test/base/populators.py
index c2e18e9797e..019571559c6 100644
--- a/lib/galaxy_test/base/populators.py
+++ b/lib/galaxy_test/base/populators.py
@@ -2179,6 +2179,7 @@ class BaseWorkflowPopulator(BasePopulator):
         invocations: int = 1,
         raw_yaml: bool = False,
         use_cached_job: bool = False,
+        copy_inputs_to_history: bool = False,
     ):
         """High-level wrapper around workflow API, etc. to invoke format 2 workflows."""
         workflow_populator = self
@@ -2221,6 +2222,8 @@ class BaseWorkflowPopulator(BasePopulator):
         )
         if use_cached_job:
             workflow_request["use_cached_job"] = True
+        if copy_inputs_to_history:
+            workflow_request["copy_inputs_to_history"] = True
         workflow_request["inputs"] = json.dumps(label_map)
         workflow_request["inputs_by"] = "name"
         if parameters: