Merge pull request #21489 from mvdbeek/add-sample-sheet-to-landing-requests

Enable attaching sample sheet to landing requests

Author: mvdbeek

client/src/api/schema/schema.ts

@@ -10461,6 +10461,8 @@ export interface components {
             class: "Collection";
             /** Collection Type */
             collection_type: string;
+            /** Column Definitions */
+            column_definitions?: components["schemas"]["SampleSheetColumnDefinition"][] | null;
             /**
              * Deferred
              * @default false
@@ -10473,6 +10475,10 @@ export interface components {
             )[];
             /** Name */
             name?: string | null;
+            /** Rows */
+            rows?: {
+                [key: string]: (number | boolean | string | null)[];
+            } | null;
             /** Src */
             src?: null;
         };
@@ -12742,6 +12748,10 @@ export interface components {
             items_from?: components["schemas"]["ElementsFromType"] | null;
             /** Name */
             name?: string | null;
+            /** Rows */
+            rows?: {
+                [key: string]: (number | boolean | string | null)[];
+            } | null;
             /**
              * Src
              * @constant
@@ -14471,6 +14481,10 @@ export interface components {
             name?: string | null;
             /** Path */
             path?: string | null;
+            /** Rows */
+            rows?: {
+                [key: string]: (number | boolean | string | null)[];
+            } | null;
             /** Server Dir */
             server_dir?: string | null;
             src: components["schemas"]["ItemsFromSrc"];
@@ -14509,6 +14523,10 @@ export interface components {
             )[];
             /** Name */
             name?: string | null;
+            /** Rows */
+            rows?: {
+                [key: string]: (number | boolean | string | null)[];
+            } | null;
             /** Tags */
             tags?: string[] | null;
         };

lib/galaxy/job_execution/output_collect.py

@@ -125,6 +125,15 @@ def collect_dynamic_outputs(
         destination = unnamed_output_dict["destination"]
         elements = unnamed_output_dict["elements"]
 
+        # If rows are specified at the collection level, add them to individual elements
+        # This is a defensive check in case rows weren't already distributed in data_fetch.py
+        if "rows" in unnamed_output_dict:
+            rows_dict = unnamed_output_dict["rows"]
+            for element in elements:
+                element_name = element.get("name")
+                if element_name and element_name in rows_dict and "row" not in element:
+                    element["row"] = rows_dict[element_name]
+
         assert "type" in destination
         destination_type = destination["type"]
         assert destination_type in ["library_folder", "hdca", "hdas"]

lib/galaxy/managers/landing.py

@@ -15,13 +15,18 @@
     ItemAlreadyClaimedException,
     ItemMustBeClaimed,
     ObjectNotFound,
+    RequestParameterInvalidException,
     RequestParameterMissingException,
 )
 from galaxy.managers.workflows import WorkflowContentsManager
 from galaxy.model import (
     ToolLandingRequest as ToolLandingRequestModel,
     WorkflowLandingRequest as WorkflowLandingRequestModel,
 )
+from galaxy.model.dataset_collections.types.sample_sheet_util import (
+    validate_column_definitions,
+    validate_row,
+)
 from galaxy.model.scoped_session import galaxy_scoped_session
 from galaxy.schema.schema import (
     ClaimLandingPayload,
@@ -41,7 +46,10 @@
     LandingRequestInternalToolState,
     LandingRequestToolState,
 )
-from galaxy.tool_util_models.parameters import DataOrCollectionRequestAdapter
+from galaxy.tool_util_models.parameters import (
+    DataOrCollectionRequestAdapter,
+    DataRequestCollectionUri,
+)
 from galaxy.util import safe_str_cmp
 from .context import ProvidesUserContext
 from .tools import (
@@ -86,6 +94,38 @@ def create_tool_landing_request(self, payload: CreateToolLandingRequestPayload,
                 input_state=landing_request_state.input_state
             )
 
+        # Validate sample sheet metadata in request_state for __DATA_FETCH__ tool
+        if tool.id == "__DATA_FETCH__" and request_state:
+
+            # Check each item in request_state for sample sheet metadata
+            for item in landing_request_state.input_state.get("request_state", []):
+                # Try to parse as DataRequestCollectionUri to access sample sheet fields
+                if isinstance(item, dict) and item.get("class") == "Collection":
+                    column_definitions = item.get("column_definitions")
+                    rows = item.get("rows")
+                    collection_type = item.get("collection_type", "")
+
+                    if column_definitions is not None or rows is not None:
+                        # Validate that sample sheet metadata is only used with sample_sheet collection types
+                        if not collection_type.startswith("sample_sheet"):
+                            raise RequestParameterInvalidException(
+                                f"Sample sheet metadata (column_definitions, rows) can only be used with collection_type 'sample_sheet' or 'sample_sheet:<type>', not '{collection_type}'"
+                            )
+
+                        # Validate column definitions structure
+                        if column_definitions is not None:
+                            validate_column_definitions(column_definitions)
+
+                        # Validate rows against column definitions and element identifiers
+                        if rows:
+                            element_identifiers = [elem.get("identifier") for elem in item.get("elements", [])]
+                            for identifier, row in rows.items():
+                                if identifier not in element_identifiers:
+                                    raise RequestParameterInvalidException(
+                                        f"Row identifier '{identifier}' not found in collection elements"
+                                    )
+                                validate_row(row, column_definitions, element_identifiers)
+
         model = ToolLandingRequestModel()
         model.tool_id = tool_id
         model.tool_version = tool_version
@@ -126,9 +166,35 @@ def validate_workflow_request_state(self, request_state: Optional[dict]) -> Opti
                 if isinstance(value, dict):
                     try:
                         # persist values after model validators and aliases have been applied
-                        request_state[key] = DataOrCollectionRequestAdapter.validate_python(value).model_dump(
-                            by_alias=True, exclude_unset=True, mode="json"
-                        )
+                        validated_value = DataOrCollectionRequestAdapter.validate_python(value)
+
+                        # Validate sample sheet metadata for collections
+                        if isinstance(validated_value, DataRequestCollectionUri):
+                            has_sample_sheet_metadata = (
+                                validated_value.column_definitions is not None or validated_value.rows is not None
+                            )
+                            if has_sample_sheet_metadata:
+                                collection_type = validated_value.collection_type
+                                if not collection_type.startswith("sample_sheet"):
+                                    raise RequestParameterInvalidException(
+                                        f"Sample sheet metadata (column_definitions, rows) can only be used with collection_type 'sample_sheet' or 'sample_sheet:<type>', not '{collection_type}'"
+                                    )
+
+                                # Validate column definitions structure
+                                if validated_value.column_definitions is not None:
+                                    validate_column_definitions(validated_value.column_definitions)
+
+                                # Validate rows against column definitions and element identifiers
+                                if validated_value.rows:
+                                    element_identifiers = [elem.identifier for elem in validated_value.elements]
+                                    for identifier, row in validated_value.rows.items():
+                                        if identifier not in element_identifiers:
+                                            raise RequestParameterInvalidException(
+                                                f"Row identifier '{identifier}' not found in collection elements"
+                                            )
+                                        validate_row(row, validated_value.column_definitions, element_identifiers)
+
+                        request_state[key] = validated_value.model_dump(by_alias=True, exclude_unset=True, mode="json")
                     except ValidationError:
                         pass
         return request_state

lib/galaxy/model/__init__.py

@@ -182,15 +182,17 @@
     DatasetValidatedState,
     InvocationsStateCounts,
     JobState,
-    SampleSheetColumnDefinitions,
-    SampleSheetRow,
     ToolRequestState,
 )
 from galaxy.schema.workflow.comments import WorkflowCommentModel
 from galaxy.security import get_permitted_actions
 from galaxy.security.idencoding import IdEncodingHelper
 from galaxy.security.validate_user_input import validate_password_str
 from galaxy.tool_util.output_checker import AnyJobMessage
+from galaxy.tool_util_models.sample_sheet import (
+    SampleSheetColumnDefinitions,
+    SampleSheetRow,
+)
 from galaxy.util import (
     directory_hash_id,
     enum_values,

lib/galaxy/model/dataset_collections/builder.py

@@ -20,7 +20,7 @@
         BaseDatasetCollectionType,
         DatasetInstanceMapping,
     )
-    from galaxy.schema.schema import SampleSheetRow
+    from galaxy.tool_util_models.sample_sheet import SampleSheetRow
     from galaxy.tool_util_models.tool_source import FieldDict
 
 

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

@@ -14,14 +14,15 @@
 from typing_extensions import Self
 
 from galaxy.exceptions import RequestParameterInvalidException
-from galaxy.schema.schema import (
+from galaxy.tool_util_models.parameter_validators import AnySafeValidatorModel
+from galaxy.tool_util_models.sample_sheet import (
     SampleSheetColumnDefinition,
     SampleSheetColumnDefinitions,
     SampleSheetColumnType,
     SampleSheetColumnValueT,
     SampleSheetRow,
 )
-from galaxy.tool_util_models.parameter_validators import AnySafeValidatorModel
+from galaxy.util import strip_control_characters
 
 SampleSheetRows = dict[str, SampleSheetRow]
 OptionalSampleSheetRows = Optional[SampleSheetRows]
@@ -98,6 +99,10 @@ def validate_row(
 ):
     if column_definitions is None:
         return
+    if row is None:
+        raise RequestParameterInvalidException(
+            "Sample sheet row is missing. Ensure all element names in 'elements' have corresponding entries in 'rows'."
+        )
     if len(row) != len(column_definitions):
         raise RequestParameterInvalidException(
             "Sample sheet row validation failed, incorrect number of columns specified."
@@ -141,7 +146,7 @@ def validate_column_value(
     elif column_type == "string":
         if not isinstance(column_value, (str,)):
             raise RequestParameterInvalidException(f"{column_value} was not a string as expected")
-        validate_no_special_characters(column_value)
+        strip_control_characters(column_value)
     elif column_type == "boolean":
         if not isinstance(column_value, (bool,)):
             raise RequestParameterInvalidException(f"{column_value} was not a boolean as expected")

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

@@ -47,7 +47,7 @@
     ReadOnlyWorkbook,
     set_column_width,
 )
-from galaxy.schema.schema import SampleSheetColumnValueT
+from galaxy.tool_util_models.sample_sheet import SampleSheetColumnValueT
 from galaxy.util import (
     string_as_bool,
     string_as_bool_or_none,

lib/galaxy/model/dereference.py

@@ -16,6 +16,7 @@
     Session,
 )
 
+from galaxy.exceptions import RequestParameterInvalidException
 from galaxy.model import (
     Dataset,
     DatasetCollection,
@@ -29,6 +30,10 @@
     REQUESTED_TRANSFORM_ACTIONS,
     User,
 )
+from galaxy.model.dataset_collections.types.sample_sheet_util import (
+    validate_column_definitions,
+    validate_row,
+)
 from galaxy.model.scoped_session import galaxy_scoped_session
 from galaxy.tool_util_models.parameters import (
     CollectionElementCollectionRequestUri,
@@ -37,6 +42,7 @@
     DataRequestUri,
     FileRequestUri,
 )
+from galaxy.tool_util_models.sample_sheet import SampleSheetRow
 
 log = logging.getLogger(__name__)
 
@@ -97,13 +103,21 @@ def derefence_collection_element(
     element: CollectionElementCollectionRequestUri,
     parent_dataset_collection: DatasetCollection,
     element_index: int,
+    rows: Optional[dict[str, SampleSheetRow]] = None,
 ):
     child_dataset_collection = DatasetCollection(collection_type=element.collection_type)
+
+    # Extract row for this element if present
+    columns = None
+    if rows and element.identifier in rows:
+        columns = rows[element.identifier]
+
     DatasetCollectionElement(
         collection=parent_dataset_collection,
         element=child_dataset_collection,
         element_identifier=element.identifier,
         element_index=element_index,
+        columns=columns,
     )
     sa_session.add(child_dataset_collection)
     for index, child_element in enumerate(element.elements):
@@ -126,39 +140,93 @@ def dereference_collection_dataset_element(
     element: CollectionElementDataRequestUri,
     parent_dataset_collection: DatasetCollection,
     element_index: int,
+    rows: Optional[dict[str, SampleSheetRow]] = None,
 ):
     hda = dereference_to_model(sa_session, user, history, element, add_to_history=False, visible=False)
     history.stage_addition(hda)
+
+    # Extract row for this element if present
+    columns = None
+    if rows and element.identifier in rows:
+        columns = rows[element.identifier]
+
     dce = DatasetCollectionElement(
         collection=parent_dataset_collection,
         element=hda,
         element_identifier=element.identifier,
         element_index=element_index,
+        columns=columns,
     )
     parent_dataset_collection.elements.append(dce)
 
 
+def _validate_sample_sheet_metadata(
+    data_request_uri: DataRequestCollectionUri,
+):
+    """Validate sample sheet metadata for landing requests."""
+    # Extract metadata from data request
+    collection_type = data_request_uri.collection_type
+    column_definitions = data_request_uri.column_definitions
+    rows = data_request_uri.rows
+
+    # Validate that sample sheet metadata is only used with sample_sheet collection types
+    is_sample_sheet = collection_type.startswith("sample_sheet")
+    has_sample_sheet_metadata = column_definitions is not None or rows is not None
+
+    if has_sample_sheet_metadata and not is_sample_sheet:
+        raise RequestParameterInvalidException(
+            f"Sample sheet metadata (column_definitions, rows) can only be used with collection_type 'sample_sheet' or 'sample_sheet:<type>', not '{collection_type}'"
+        )
+
+    # Validate column definitions structure
+    if column_definitions is not None:
+        validate_column_definitions(column_definitions)
+
+    # Validate each row
+    if rows:
+        # Get element identifiers for validation
+        element_identifiers = [elem.identifier for elem in data_request_uri.elements]
+
+        for identifier, row in rows.items():
+            if identifier not in element_identifiers:
+                raise RequestParameterInvalidException(
+                    f"Row identifier '{identifier}' not found in collection elements"
+                )
+            validate_row(row, column_definitions, element_identifiers)
+
+
 def derefence_collection_to_model(
     sa_session: galaxy_scoped_session,
     user: User,
     history: History,
     data_request_uri: DataRequestCollectionUri,
     collection_name: str = "Collection",
 ) -> HistoryDatasetCollectionAssociation:
+    # Validate sample sheet metadata before creating any objects
+    _validate_sample_sheet_metadata(data_request_uri)
+
     name = data_request_uri.name or collection_name
     hdca = HistoryDatasetCollectionAssociation(
         name=name,
         history=history,
     )
     sa_session.add(hdca)
-    dc = DatasetCollection(collection_type=data_request_uri.collection_type)
+    dc = DatasetCollection(
+        collection_type=data_request_uri.collection_type,
+        column_definitions=data_request_uri.column_definitions,
+    )
     sa_session.add(dc)
     hdca.collection = dc
+
+    # Extract rows for passing to element creation
+    rows = data_request_uri.rows
+
     for i, element in enumerate(data_request_uri.elements):
         if element.class_ == "File":
-            dereference_collection_dataset_element(sa_session, user, history, element, dc, element_index=i)
+            dereference_collection_dataset_element(sa_session, user, history, element, dc, element_index=i, rows=rows)
         elif element.class_ == "Collection":
-            derefence_collection_element(sa_session, user, history, element, dc, i)
+            derefence_collection_element(sa_session, user, history, element, dc, i, rows=rows)
+
     dc.populated_state = "ok"
     dc.element_count = len(data_request_uri.elements)
     history.stage_addition(hdca)