[24.1] Determine expression tool output extension when input terminal

lib/galaxy/tools/actions/__init__.py

@@ -22,6 +22,7 @@
     AuthenticationRequired,
     ItemAccessibilityException,
     RequestParameterInvalidException,
+    ToolInputsNotReadyException,
 )
 from galaxy.job_execution.actions.post import ActionBox
 from galaxy.managers.context import ProvidesHistoryContext
@@ -1156,8 +1157,14 @@ def get_ext_or_implicit_ext(hda):
         # objects, and their type is the target_ext, so this should be correct even if there
         # are multiple ImplicitlyConvertedDatasetAssociation objects (meaning 2 datasets had been converted
         # to produce a dataset with the required datatype)
-        return hda.implicitly_converted_parent_datasets[0].type
-    return hda.ext
+        ext = hda.implicitly_converted_parent_datasets[0].type
+    else:
+        ext = hda.ext
+    if ext == "expression.json" and hda.dataset.state not in model.Dataset.terminal_states:
+        raise ToolInputsNotReadyException(
+            "Tool uses expression tool output to determine extension, can only succeed once input is terminal."
+        )
+    return ext
 
 
 def determine_output_format(
@@ -1192,6 +1199,8 @@ def determine_output_format(
         try:
             input_dataset = input_datasets[output.format_source]
             ext = get_ext_or_implicit_ext(input_dataset)
+        except ToolInputsNotReadyException:
+            raise
         except Exception:
             pass
     elif format_source is not None:
@@ -1231,6 +1240,8 @@ def determine_output_format(
                                 break
                     input_dataset = input_element.element_object
                 ext = get_ext_or_implicit_ext(input_dataset)
+            except ToolInputsNotReadyException:
+                raise
             except Exception as e:
                 log.debug("Exception while trying to determine format_source: %s", e)
 

lib/galaxy_test/api/test_workflows.py

@@ -2157,6 +2157,38 @@ def test_run_workflow_pick_value_bam_pja(self):
         assert dataset_details["metadata_bam_index"]
         assert dataset_details["file_ext"] == "bam"
 
+    def test_expression_tool_output_in_format_source(self):
+        with self.dataset_populator.test_history() as history_id:
+            self._run_workflow(
+                """class: GalaxyWorkflow
+inputs:
+  input:
+    type: data
+steps:
+  skip:
+    tool_id: cat_data_and_sleep
+    in:
+      input1: input
+    when: $(false)
+  pick_larger:
+    tool_id: expression_pick_larger_file
+    in:
+      input1: skip/out_file1
+      input2: input
+  format_source:
+    tool_id: cat_data_and_sleep
+    in:
+      input1: pick_larger/larger_file
+test_data:
+  input:
+    value: 1.fastqsanger.gz
+    type: File
+    file_type: fastqsanger.gz
+""",
+                history_id=history_id,
+            )
+            self.dataset_populator.wait_for_history(history_id=history_id, assert_ok=True)
+
     def test_run_workflow_simple_conditional_step(self):
         with self.dataset_populator.test_history() as history_id:
             summary = self._run_workflow(

test/functional/tools/expression_pick_larger_file.xml

@@ -20,7 +20,7 @@
         <param name="input2" type="data" optional="true" label="Second file" />
     </inputs>
     <outputs>
-        <output name="larger_file" type="data" from="output" />
+        <output name="larger_file" type="data" from="output" format_source="input1" />
     </outputs>
     <tests>
         <test>