Merge pull request #254 from sennetconsortium/tjmadonna/add-vitessce-…

…config

Adding basic validation and error handling to Vitessce config endpoint

ingest-api-spec.yaml

@@ -35,6 +35,44 @@ components:
 
     file_uuid:
       type: string
+
+    File:
+      type: object
+      properties:
+        description:
+          type: string
+          description: Free text description of the file
+        edam_term:
+          type: string
+          description: The EDAM ontology term for the file type
+        is_qa_qc:
+          type: boolean
+          description: True if the file is a QA file
+        rel_path:
+          type: string
+          description: The path on the local file system, relative to the base data directory.
+        size:
+          type: integer
+          description: The size of the file in bytes
+        type:
+          type: string
+          description: The file type
+
+    DagProvenanceItem:
+      type: object
+      properties:
+        hash:
+          type: string
+          description: The 7 character hash of the commit
+          example: 1994f0f
+        name:
+          type: string
+          description: The name of the cwl file
+          example: anndata-to-ui.cwl
+        origin:
+          type: string
+          description: The git origin of the cwl file
+          example: https://github.com/hubmapconsortium/portal-containers
 
     Sources:
       type: object
@@ -761,3 +799,94 @@ paths:
           description: User's token is not valid
         "500":
           description: An unexpected error occured
+  /vitessce/config:
+    post:
+      tags:
+        - vitessce
+      summary: Return a Vitessce configuration for a given Dataset
+      requestBody:
+        content:
+          application/json:
+            schema:
+              type: object
+              properties:
+                uuid:
+                  type: string
+                  description: The SenNet unique identifier, intended for internal software use only. This is a 32 digit hexadecimal UUID e.g. 461bbfdc353a2673e381f632510b0f17
+                  example: 461bbfdc353a2673e381f632510b0f17
+                status:
+                  type: string
+                  enum:
+                    - New
+                    - Processing
+                    - QA
+                    - Published
+                    - Error
+                    - Hold
+                    - Invalid
+                  description: 'One of: New|Processing|QA|Published|Error|Hold|Invalid'
+                entity_type:
+                  type: string
+                  enum:
+                    - Dataset
+                  description: 'Dataset entity type'
+                data_types:
+                  type: array
+                  items:
+                    type: string
+                    enum:
+                      - 10x-multiome
+                      - bulk-RNA
+                      - CITE-Seq
+                      - CODEX
+                      - codex_cytokit
+                      - codex_cytokit_v1
+                      - CosMX (RNA)
+                      - DBiT-seq
+                      - FACS - Fluorescence-activated Cell Sorting
+                      - GeoMX (RNA)
+                      - image_pyramid
+                      - LC-MS
+                      - Lightsheet
+                      - MIBI
+                      - mibi_deepcell
+                      - Mint-ChIP
+                      - publication
+                      - publication_ancillary
+                      - salmon_rnaseq_10x
+                      - salmon_rnaseq_bulk
+                      - salmon_sn_rnaseq_10x
+                      - SASP
+                      - scRNA-seq
+                      - sn_atac_seq
+                      - snATAC-seq
+                      - snRNA-seq
+                      - snRNAseq-10xGenomics-v3
+                      - Stained Slides
+                      - Visium
+                  description: The data or assay types contained in this dataset as a json array of strings. Each is an assay code from [assay types](https://ontology.api.hubmapconsortium.org/datasets?application_context=Sennet).
+                files:
+                  type: array
+                  items:
+                    $ref: '#/components/schemas/File'
+                  description: The files associated with this Dataset
+                metadata:
+                  type: object
+                  description: The metadata associated with this Dataset
+                  properties:
+                    dag_provenance_list:
+                      type: array
+                      items:
+                        $ref: '#/components/schemas/DagProvenanceItem'
+                      description: The list of dag provenance items associated with this Dataset
+      responses:
+        "200":
+          description: The [Vitessce JSON configuration](http://vitessce.io/docs/view-config-json) for the given Dataset
+        "400":
+          description: Request body schema was invalid or was for an invalid Dataset
+        "401":
+          description: The user's token has expired or the user did not have authorization
+        "404":
+          description: The Vitesce configuration for the given Dataset could not be found
+        "500":
+          description: Internal error

src/routes/vitessce/__init__.py

@@ -1,6 +1,7 @@
 import logging
+from uuid import UUID
 
-from flask import Blueprint, Response, current_app, jsonify, request
+from flask import Blueprint, current_app, jsonify, request
 from hubmap_commons.hm_auth import AuthHelper
 from portal_visualization.builder_factory import get_view_config_builder
 
@@ -14,6 +15,9 @@
 @vitessce_blueprint.route("/vitessce/config", methods=["POST"])
 @require_json(param="entity")
 def vitessce_config(entity: dict):
+    if errors := validate_vitessce_entity(entity) :
+        return jsonify({"error": errors}), 400
+
     def get_assaytype(entity):
         if "metadata" in entity.get("metadata", {}):
             metadata = entity["metadata"]["metadata"]
@@ -38,7 +42,43 @@ def get_assaytype(entity):
             entity, groups_token, current_app.config["ASSETS_WEBSERVICE_URL"]
         )
         vitessce_conf = builder.get_conf_cells(marker=None)
-        return jsonify(vitessce_conf[0] or {}), 200
+        if len(vitessce_conf) < 1 or not vitessce_conf[0]:
+            raise ValueError("empty vitessce config")
+        return jsonify(vitessce_conf[0]), 200
     except Exception as e:
-        logger.error(e)
-        return Response("Unexpected error while retrieving Vitessce config", 500)
+        logger.error(f"Error while retrieving Vitessce config for uuid {entity['uuid']}:", e)
+        return jsonify({"error": "404 Not Found: Entity or filepath not found"}), 404
+
+
+def validate_vitessce_entity(entity):
+    """Add basic validation for vitessce config request body
+
+    Args:
+        entity (dict): vitessce config request body
+
+    Returns:
+        list: list of errors
+    """
+    errors = []
+    try:
+        UUID(entity.get("uuid"))
+    except Exception:
+        errors.append("uuid must be a valid UUID")
+
+    if not entity.get("status") or not isinstance(entity["status"], str):
+        errors.append("'status' string is required")
+
+    if not entity.get("data_types") or not isinstance(entity["data_types"], list):
+        errors.append("'data_types' array is required")
+
+    if not entity.get("files") or not isinstance(entity["files"], list):
+        errors.append("'files' array is required")
+
+    if entity.get("entity_type") != "Dataset":
+        errors.append("'entity_type' 'Dataset' is required")
+
+    prov_list = entity.get("metadata", {}).get("dag_provenance_list")
+    if not prov_list or not isinstance(prov_list, list):
+        errors.append("'metadata.dag_provenance_list' array is required")
+
+    return errors