Merge pull request #22926 from mvdbeek/fix-dce-submission

[26.1] Accept dce collection references in the tool request model

lib/galaxy/tool_util_models/parameters.py

@@ -501,15 +501,14 @@ class FloatParameterModel(BaseGalaxyToolParameterModelDefinition):
         return False
 
 
-DataSrcT = Literal["hda", "ldda"]
-MultiDataSrcT = Literal["hda", "ldda", "hdca", "dce"]
-# @jmchilton you meant CollectionSrcT - fix that at some point please.
-CollectionStrT = Literal["hdca"]
+# External collection source type. ``dce`` is accepted because a job that maps
+# over a nested collection (e.g. subcollection mapping over a ``list:paired``)
+# records its input as a ``DatasetCollectionElement``; rerunning such a job
+# resubmits that ``dce`` reference through the request model.
+CollectionSrcT = Literal["hdca", "dce"]
 # Internal collection source type - includes dce for subcollection mapping
 CollectionInternalSrcT = Literal["hdca", "dce"]
 
-TestCaseDataSrcT = Literal["File"]
-
 
 class LegacyRequestModelAttributes(StrictModel):
     # Here for bioblend's sake, should be stripped
@@ -1267,12 +1266,12 @@ class DataParameterModel(BaseGalaxyToolParameterModelDefinition):
 
 
 class DataCollectionRequest(StrictModel):
-    src: CollectionStrT
+    src: CollectionSrcT
     id: StrictStr
 
 
 class BatchCollectionInstance(StrictModel):
-    src: CollectionStrT
+    src: CollectionSrcT
     id: StrictStr
     map_over_type: Optional[str] = None
 

lib/galaxy_test/api/test_tools.py

@@ -3194,6 +3194,33 @@ class TestToolsApi(ApiTestCase, TestsTools):
             assert len(response_object["jobs"]) == 2
             assert len(response_object["implicit_collections"]) == 1
 
+    @skip_without_tool("collection_paired_test")
+    def test_request_paired_collection_input_with_dce(self):
+        """Regression for https://github.com/galaxyproject/galaxy/issues/22923
+
+        A job that maps over a ``list:paired`` records its paired-collection
+        input as a ``DatasetCollectionElement`` (``src: dce``). Rerunning such a
+        job resubmits that ``dce`` reference through the structured tool-request
+        path, which must validate against the request model.
+        """
+        with self.dataset_populator.test_history() as history_id:
+            pair_ids = []
+            for _ in range(2):
+                pair_id = self.dataset_collection_populator.create_pair_in_history(
+                    history_id, contents=["forward", "reverse"], wait=True
+                ).json()["outputs"][0]["id"]
+                pair_ids.append(pair_id)
+            list_hdca = self.dataset_collection_populator.create_list_from_pairs(history_id, pair_ids)
+            # The element of a list:paired is itself a paired collection, referenced via a dce.
+            dce_id = list_hdca.json()["elements"][0]["id"]
+
+            inputs = {"f1": {"src": "dce", "id": dce_id}}
+            response = self.dataset_populator.tool_request_raw("collection_paired_test", inputs, history_id)
+            self._assert_status_code_is(response, 200)
+            tool_request_id = response.json()["tool_request_id"]
+            assert self.dataset_populator.wait_on_tool_request(tool_request_id)
+            self.dataset_populator.wait_for_history(history_id, assert_ok=True)
+
     @skip_without_tool("identifier_source")
     def test_default_identifier_source_map_over(self):
         with self.dataset_populator.test_history() as history_id:

test/unit/tool_util/parameter_specification.yml

@@ -1591,9 +1591,16 @@ gx_data_multiple_optional:
 gx_data_collection:
   request_valid:
    - parameter: {src: hdca, id: abcdabcd}
+   # dce subcollection reference - e.g. rerunning a job that mapped over a list:paired
+   - parameter: {src: dce, id: abcdabcd}
+   # batching (multirun) over dce subcollections
+   - parameter: {__class__: "Batch", values: [{src: dce, id: abcdabcd}]}
+   - parameter: {__class__: "Batch", values: [{src: dce, id: abcdabcd, map_over_type: paired}]}
    - parameter: {class: "Collection", collection_type: "list", elements: []}
   request_invalid:
    - parameter: {src: hdca, id: 7}
+   # external request ids must be encoded strings, not decoded ints - for hdca and dce alike
+   - parameter: {src: dce, id: 7}
    - parameter: null
    - parameter: {src: fooda, id: abcdabcd}
    - parameter: {id: abcdabcd}
@@ -1605,9 +1612,12 @@ gx_data_collection:
    - parameter: {class: "CollectionX", collection_type: "list", elements: []}
   request_internal_valid:
    - parameter: {src: hdca, id: 5}
+   - parameter: {src: dce, id: 5}
+   - parameter: {__class__: "Batch", values: [{src: dce, id: 5}]}
    - parameter: {class: "Collection", collection_type: "list", elements: []}
   request_internal_invalid:
    - parameter: {src: hdca, id: abcdabcd}
+   - parameter: {src: dce, id: abcdabcd}
    - parameter: null
    - parameter: {src: fooda, id: abcdabcd}
    - parameter: {id: abcdabcd}
@@ -1619,6 +1629,7 @@ gx_data_collection:
    - parameter: {class: "CollectionX", collection_type: "list", elements: []}
   request_internal_dereferenced_valid:
    - parameter: {src: hdca, id: 5}
+   - parameter: {src: dce, id: 5}
   request_internal_dereferenced_invalid:
    - parameter: {src: hdca, id: abcdabcd}
    # At thsi point the collection should have been dereferenced into datasets,
@@ -1759,6 +1770,7 @@ gx_data_collection:
 gx_data_collection_optional:
   request_valid:
    - parameter: {src: hdca, id: abcdabcd}
+   - parameter: {src: dce, id: abcdabcd}
    - parameter: null
    - {}
   request_invalid:
@@ -1771,6 +1783,7 @@ gx_data_collection_optional:
    - parameter: {}
   request_internal_valid:
    - parameter: {src: hdca, id: 5}
+   - parameter: {src: dce, id: 5}
    - parameter: null
    - {}
   request_internal_invalid: