Add dataset types

artifacts/download_dataset.yaml

@@ -0,0 +1,21 @@
+---
+- name: Install datahugger and download dataset
+  hosts: localhost
+  vars:
+    ansible_python_interpreter: /usr/bin/python3
+  tasks:
+    - name: Ensure python3-pip is installed
+      ansible.builtin.package:
+        name: python3-pip
+        state: present
+
+    - name: Install datahugger Python module
+      ansible.builtin.pip:
+        name: datahugger
+        state: present
+        executable: pip3
+
+    - name: Download the dataset
+      grycap.dataset.download_dataset:
+        dataset_url: "{{ dataset_url }}"
+        output_dir: "{{ output_dir }}"

custom_types.yaml

@@ -114,6 +114,39 @@ data_types:
         type: string
         required: no
 
+  tosca.datatypes.Dataset:
+    derived_from: tosca.datatypes.Root
+    description: A data type that represents a dataset, including its DOI and other metadata.
+    properties:
+      identifier:
+        type: string
+        description: The Digital Object Identifier of the dataset.
+        required: true
+      title:
+        type: string
+        description: The title of the dataset.
+        required: false
+      description:
+        type: string
+        description: A description of the dataset.
+        required: false
+      creator:
+        type: string
+        description: The creator of the dataset.
+        required: false
+      release_date:
+        type: string
+        description: The publication date of the dataset.
+        required: false
+      downloadURL:
+        type: string
+        description: The URL to download the dataset.
+        required: false
+      license:
+        type: string
+        description: A legal document under which the resource is made available.
+        required: false
+
 capability_types:
 
   tosca.capabilities.indigo.OperatingSystem:
@@ -1894,6 +1927,38 @@ node_types:
         valid_source_types: [tosca.nodes.Container.Application.Docker]
         relationship: tosca.relationships.HostedOn
 
+  tosca.nodes.eosc.Dataset:
+    derived_from: tosca.nodes.Root
+    properties:
+      dataset:
+        type: tosca.datatypes.Dataset
+        description: The dataset associated with this node.
+        required: true
+      filter:
+        type: string
+        description: Regex to express the set of files to get from the Dataset.
+        required: false
+      dest:
+        type: string
+        description: Path to download the set of files from the Dataset.
+        required: false
+    requirements:
+      - host:
+          capability: tosca.capabilities.Compute
+          node: tosca.nodes.Compute
+          relationship: tosca.relationships.HostedOn
+    artifacts:
+      datase_collection:
+        file: git+https://github.com/grycap/ansible-collection-dataset
+        type: tosca.artifacts.AnsibleGalaxy.collection
+    interfaces:
+      Standard:
+        configure:
+          implementation: https://raw.githubusercontent.com/grycap/tosca/datasets/artifacts/lrms/download_dataset.yaml
+          inputs:
+            dataset_url: { get_property: [ SELF, dataset, identifier ] }
+            output_dir: { get_property: [ SELF, dest ] }
+
 policy_types:
 
   tosca.policies.indigo.Placement:

tests/requirements.yaml

@@ -32,5 +32,8 @@ roles:
 
 collections:
   - name: community.crypto
+  - name: grycap.dataset
+    source: https://github.com/grycap/ansible-collection-dataset
+    type: git
   - name: community.docker
     version: 3.12.2