Implement paired_or_unpaired collections...

lib/galaxy/model/dataset_collections/registry.py

@@ -2,13 +2,15 @@
 from .types import (
     list,
     paired,
+    paired_or_unpaired,
     record,
 )
 
 PLUGIN_CLASSES = [
     list.ListDatasetCollectionType,
     paired.PairedDatasetCollectionType,
     record.RecordDatasetCollectionType,
+    paired_or_unpaired.PairedOrUnpairedDatasetCollectionType,
 ]
 
 

lib/galaxy/model/dataset_collections/types/__init__.py

@@ -21,3 +21,9 @@ def generate_elements(self, dataset_instances: dict, **kwds):
 class BaseDatasetCollectionType(DatasetCollectionType):
     def _validation_failed(self, message):
         raise exceptions.ObjectAttributeInvalidException(message)
+
+    def _ensure_dataset_with_identifier(self, dataset_instances: dict, name: str):
+        dataset_instance = dataset_instances.get(name)
+        if dataset_instance is None:
+            raise exceptions.ObjectAttributeInvalidException(f"An element with the identifier {name} is required to create this collection type")
+        return dataset_instance

lib/galaxy/model/dataset_collections/types/paired.py

@@ -1,3 +1,4 @@
+from galaxy.exceptions import RequestParameterInvalidException
 from galaxy.model import (
     DatasetCollectionElement,
     HistoryDatasetAssociation,
@@ -16,13 +17,17 @@ class PairedDatasetCollectionType(BaseDatasetCollectionType):
     collection_type = "paired"
 
     def generate_elements(self, dataset_instances, **kwds):
-        if forward_dataset := dataset_instances.get(FORWARD_IDENTIFIER):
+        num_datasets = len(dataset_instances)
+        if num_datasets != 2:
+            raise RequestParameterInvalidException(f"Incorrect number of datasets - 2 datasets exactly are required to create a single_or_paired collection")
+
+        if forward_dataset := self._ensure_dataset_with_identifier(dataset_instances, FORWARD_IDENTIFIER):
             left_association = DatasetCollectionElement(
                 element=forward_dataset,
                 element_identifier=FORWARD_IDENTIFIER,
             )
             yield left_association
-        if reverse_dataset := dataset_instances.get(REVERSE_IDENTIFIER):
+        if reverse_dataset := self._ensure_dataset_with_identifier(dataset_instances, REVERSE_IDENTIFIER):
             right_association = DatasetCollectionElement(
                 element=reverse_dataset,
                 element_identifier=REVERSE_IDENTIFIER,

lib/galaxy/model/dataset_collections/types/paired_or_unpaired.py

@@ -0,0 +1,45 @@
+from galaxy.exceptions import RequestParameterInvalidException
+from galaxy.model import (
+    DatasetCollectionElement,
+    HistoryDatasetAssociation,
+)
+from . import BaseDatasetCollectionType
+from .paired import (
+    FORWARD_IDENTIFIER,
+    REVERSE_IDENTIFIER,
+)
+
+SINGLETON_IDENTIFIER = "unpaired"
+
+
+class PairedOrUnpairedDatasetCollectionType(BaseDatasetCollectionType):
+    """
+    """
+
+    collection_type = "paired_or_unpaired"
+
+    def generate_elements(self, dataset_instances, **kwds):
+        num_datasets = len(dataset_instances)
+        if num_datasets > 2 or num_datasets < 1:
+            raise RequestParameterInvalidException(f"Incorrect number of datasets - 1 or 2 datasets is required to create a paired_or_unpaired collection")
+
+        if num_datasets == 2:
+            if forward_dataset := self._ensure_dataset_with_identifier(dataset_instances, FORWARD_IDENTIFIER):
+                left_association = DatasetCollectionElement(
+                    element=forward_dataset,
+                    element_identifier=FORWARD_IDENTIFIER,
+                )
+                yield left_association
+            if reverse_dataset := self._ensure_dataset_with_identifier(dataset_instances, REVERSE_IDENTIFIER):
+                right_association = DatasetCollectionElement(
+                    element=reverse_dataset,
+                    element_identifier=REVERSE_IDENTIFIER,
+                )
+                yield right_association
+        else:
+            if single_datasets := self._ensure_dataset_with_identifier(dataset_instances, SINGLETON_IDENTIFIER):
+                single_association = DatasetCollectionElement(
+                    element=single_datasets,
+                    element_identifier=SINGLETON_IDENTIFIER,
+                )
+                yield single_association

lib/galaxy/schema/schema.py

@@ -33,6 +33,8 @@
 from typing_extensions import (
     Annotated,
     Literal,
+    NotRequired,
+    TypedDict,
 )
 
 from galaxy.schema import partial_model

lib/galaxy/tools/__init__.py

@@ -3534,6 +3534,62 @@ def produce_outputs(self, trans, out_data, output_collections, incoming, history
         )
 
 
+class SplitPairedAndUnpairedTool(DatabaseOperationTool):
+    tool_type = "split_paired_and_unpaired"
+    require_terminal_states = False
+    require_dataset_ok = False
+
+    def produce_outputs(self, trans, out_data, output_collections, incoming, history, **kwds):
+        has_collection = incoming["input"]
+        if hasattr(has_collection, "element_type"):
+            # It is a DCE
+            collection = has_collection.element_object
+        else:
+            # It is an HDCA
+            collection = has_collection.collection
+
+        collection_type = collection.collection_type
+        assert collection_type in ["list", "list:paired", "list:paired_or_unpaired"]
+
+        unpaired_dce_copies = {}
+        paired_dce_copies = {}
+        paired_datasets = []
+
+        def _handle_unpaired(dce):
+            element_identifier = dce.element_identifier
+            assert getattr(dce.element_object, "history_content_type", None) == "dataset"
+            copied_value = dce.element_object.copy(copy_tags=dce.element_object.tags, flush=False)
+            unpaired_dce_copies[element_identifier] = copied_value
+
+        def _handle_paired(dce):
+            element_identifier = dce.element_identifier
+            copied_value = dce.element_object.copy(flush=False)
+            paired_dce_copies[element_identifier] = copied_value
+            paired_datasets.append(copied_value.elements[0].element_object)
+            paired_datasets.append(copied_value.elements[1].element_object)
+
+        if collection_type == "list":
+            for element in collection.elements:
+                _handle_unpaired(element)
+        elif collection_type == "list:paired":
+            for element in collection.elements:
+                _handle_paired(element)
+        elif collection_type == "list:paired_or_unpaired":
+            for element in collection.elements:
+                if getattr(element.element_object, "history_content_type", None) == "dataset":
+                    _handle_unpaired(element)
+                else:
+                    _handle_paired(element)
+
+        self._add_datasets_to_history(history, unpaired_dce_copies.values())
+        self._add_datasets_to_history(history, paired_datasets)
+        output_collections.create_collection(
+            self.outputs["output_unpaired"], "output_unpaired", elements=unpaired_dce_copies, propagate_hda_tags=False
+        )
+        output_collections.create_collection(
+            self.outputs["output_paired"], "output_paired", elements=paired_dce_copies, propagate_hda_tags=False
+        )
+
 class ExtractDatasetCollectionTool(DatabaseOperationTool):
     tool_type = "extract_dataset"
     require_terminal_states = False

lib/galaxy/tools/split_paired_and_unpaired.xml

@@ -0,0 +1,132 @@
+<tool id="__SPLIT_PAIRED_AND_UNPAIRED__"
+      name="Split Paired and Unpaired"
+      version="1.0.0"
+      tool_type="split_paired_and_unpaired">
+  <description></description>
+  <type class="SplitPairedAndUnpairedTool" module="galaxy.tools" />
+  <action module="galaxy.tools.actions.model_operations"
+          class="ModelOperationToolAction"/>
+  <edam_operations>
+    <edam_operation>operation_2409</edam_operation>
+  </edam_operations>
+  <inputs>
+    <param name="input" type="data_collection" label="Input Collection" collection_type="list:paired,list,list:paired_or_unpaired" />
+  </inputs>
+  <outputs>
+    <collection name="output_unpaired" format_source="input" type="list" label="${on_string} (unpaired)" >
+    </collection>
+    <collection name="output_paired" format_source="input" type="list:paired" label="${on_string} (paired)" >
+    </collection>
+  </outputs>
+  <tests>
+    <test>
+      <param name="input">
+        <collection type="list">
+          <element name="el1" value="simple_line.txt" />
+          <element name="el2" value="simple_line_alternative.txt" />
+        </collection>
+      </param>
+      <output_collection name="output_unpaired" type="list" count="2">
+        <element name="el1" ftype="txt">
+          <assert_contents>
+            <has_line line="This is a line of text." />
+          </assert_contents>
+        </element>
+        <element name="el2" ftype="txt">
+          <assert_contents>
+            <has_line line="This is a different line of text." />
+          </assert_contents>
+        </element>
+      </output_collection>
+      <output_collection name="output_paired" type="list:paired" count="0">
+      </output_collection>
+    </test>
+    <test>
+      <param name="input">
+        <collection type="list:paired">
+          <element name="el1">
+            <collection type="paired">
+              <element name="forward" value="simple_line.txt" />
+              <element name="reverse" value="simple_line_alternative.txt" />
+            </collection>
+          </element>
+        </collection>
+      </param>
+      <output_collection name="output_unpaired" type="list" count="0">
+      </output_collection>
+      <output_collection name="output_paired" type="list:paired" count="1">
+        <element name="el1">
+          <element name="forward">
+            <assert_contents>
+              <has_line line="This is a line of text." />
+            </assert_contents>
+          </element>
+          <element name="reverse">
+            <assert_contents>
+              <has_line line="This is a different line of text." />
+            </assert_contents>
+          </element>
+        </element>
+      </output_collection>
+    </test>
+    <test>
+      <param name="input">
+        <collection type="list:paired_or_unpaired">
+          <element name="el1">
+            <collection type="paired">
+              <element name="forward" value="simple_line.txt" />
+              <element name="reverse" value="simple_line_alternative.txt" />
+            </collection>
+          </element>
+          <element name="el2" value="simple_line.txt">
+          </element>
+          <element name="el3" value="simple_line_alternative.txt">
+          </element>
+        </collection>
+      </param>
+      <output_collection name="output_unpaired" type="list" count="2">
+        <element name="el2" ftype="txt">
+          <assert_contents>
+            <has_line line="This is a line of text." />
+          </assert_contents>
+        </element>
+        <element name="el3" ftype="txt">
+          <assert_contents>
+            <has_line line="This is a different line of text." />
+          </assert_contents>
+        </element>
+      </output_collection>
+      <output_collection name="output_paired" type="list:paired" count="1">
+        <element name="el1">
+          <element name="forward">
+            <assert_contents>
+              <has_line line="This is a line of text." />
+            </assert_contents>
+          </element>
+          <element name="reverse">
+            <assert_contents>
+              <has_line line="This is a different line of text." />
+            </assert_contents>
+          </element>
+        </element>
+      </output_collection>
+    </test>
+  </tests>
+  <help><![CDATA[
+
+========
+Synopsis
+========
+
+
+===========
+Description
+===========
+
+
+.. class:: infomark
+
+This tool will create new history datasets for your collection but your quota usage will not increase.
+
+  ]]></help>
+</tool>

lib/galaxy/tools/wrappers.py

@@ -766,6 +766,10 @@ def serialize(
             include_collection_name=include_collection_name,
         )
 
+    @property
+    def has_single_item(self) -> bool:
+        return self.__input_supplied and len(self.__element_instance_list) == 1
+
     @property
     def is_input_supplied(self) -> bool:
         return self.__input_supplied

lib/galaxy_test/api/test_dataset_collections.py

@@ -101,6 +101,25 @@ def test_create_list_of_new_pairs(self):
             pair_1_element_1 = pair_elements[0]
             assert pair_1_element_1["element_index"] == 0
 
+    def test_create_paried_or_unpaired(self, history_id):
+        collection_name = "a singleton in a paired_or_unpaired collection"
+        contents = [
+            ("unpaired", "1\t2\t3"),
+        ]
+        single_identifier = self.dataset_collection_populator.list_identifiers(history_id, contents)
+        payload = dict(
+            name=collection_name,
+            instance_type="history",
+            history_id=history_id,
+            element_identifiers=single_identifier,
+            collection_type="paired_or_unpaired",
+        )
+        create_response = self._post("dataset_collections", payload, json=True)
+        dataset_collection = self._check_create_response(create_response)
+        assert dataset_collection["collection_type"] == "paired_or_unpaired"
+        returned_collections = dataset_collection["elements"]
+        assert len(returned_collections) == 1, dataset_collection
+
     def test_create_record(self, history_id):
         contents = [
             ("condition", "1\t2\t3"),

test/functional/tools/collection_paired_or_unpaired.xml

@@ -0,0 +1,51 @@
+<tool id="collection_paired_or_unpaired" name="collection_paired_or_unpaired" version="0.1.0">
+  <command>
+    #if $f1.has_single_item:
+        cat $f1.unpaired >> $out1;
+        echo "Single item"
+    #else
+        cat $f1.forward $f1['reverse'] >> $out1;
+        echo "Paired items"
+    #end if
+  </command>
+  <inputs>
+    <param name="f1" type="data_collection" collection_type="paired_or_unpaired" label="Input" />
+  </inputs>
+  <outputs>
+    <data format="txt" name="out1" />
+  </outputs>
+  <tests>
+    <test>
+      <param name="f1">
+        <collection type="paired_or_unpaired" name="collection name">
+          <element name="forward" value="simple_line.txt" />
+          <element name="reverse" value="simple_line_alternative.txt" />
+        </collection>
+      </param>
+      <output name="out1">
+        <assert_contents>
+          <has_line line="This is a line of text." />
+          <has_line line="This is a different line of text." />
+        </assert_contents>
+      </output>
+      <assert_stdout>
+          <has_line line="Paired items" />
+      </assert_stdout>
+    </test>
+    <test>
+      <param name="f1">
+        <collection type="paired_or_unpaired" name="collection name">
+          <element name="single" value="simple_line.txt" />
+        </collection>
+      </param>
+      <output name="out1">
+        <assert_contents>
+          <has_line line="This is a line of text." />
+        </assert_contents>
+      </output>
+      <assert_stdout>
+          <has_line line="Single item" />
+      </assert_stdout>
+    </test>
+  </tests>
+</tool>