Task/WC-120: Datacite operations and pipeline integration (#1037)

* Rebase datacite operations on latest DRP branch

* integrate datacite utils in pipeline

---------

Co-authored-by: Shayan Khan <skhan@tacc.utexas.edu>

server/portal/apps/projects/workspace_operations/datacite_operations.py

@@ -0,0 +1,130 @@
+import datetime
+from typing import Optional
+import json
+import requests
+import networkx as nx
+from django.conf import settings
+
+
+def get_datacite_json(pub_graph: nx.DiGraph):
+    """
+    Generate datacite payload for a publishable entity. `pub_graph` is the output of
+    either `get_publication_subtree` or `get_publication_full_tree`.
+    """
+
+    datacite_json = {}
+
+    base_meta_node = "NODE_ROOT"
+
+    base_meta = pub_graph.nodes[base_meta_node]["value"]
+
+    author_attr = []
+    institutions = []
+    for author in base_meta.get("authors", []):
+        author_attr.append(
+            {
+                "nameType": "Personal",
+                "givenName": author.get("first_name", ""),
+                "familyName": author.get("last_name", ""),
+            }
+        )
+        institutions.append(author.get("inst", ""))
+
+    datacite_json["contributors"] = [
+        {
+            "contributorType": "HostingInstitution",
+            "nameType": "Organizational",
+            "name": institution,
+        }
+        for institution in list(set(institutions))
+    ]
+    datacite_json["creators"] = author_attr
+    datacite_json["titles"] = [{"title": base_meta["title"]}]
+
+    datacite_json["publisher"] = "Digital Rocks Portal"
+
+    datacite_json["publicationYear"] = datetime.datetime.now().year
+
+    project_id = base_meta["projectId"]
+    datacite_url = f"{settings.PORTAL_PUBLICATION_DATACITE_URL_PREFIX}/{project_id}"
+
+    datacite_json["url"] = datacite_url
+    datacite_json["prefix"] = settings.PORTAL_PUBLICATION_DATACITE_SHOULDER
+
+    return datacite_json
+
+
+def upsert_datacite_json(datacite_json: dict, doi: Optional[str] = None):
+    """
+    Create a draft DOI in datacite with the specified metadata. If a DOI is specified,
+    the metadata for that DOI is updated instead.
+    """
+    if doi:
+        datacite_json.pop("publicationYear", None)
+
+    datacite_payload = {
+        "data": {
+            "type": "dois",
+            "relationships": {
+                "client": {"data": {"type": "clients", "id": "tdl.tacc"}}
+            },
+            "attributes": datacite_json,
+        }
+    }
+    if not doi:
+        res = requests.post(
+            f"{settings.DATACITE_URL.strip('/')}/dois",
+            auth=(settings.DATACITE_USER, settings.DATACITE_PASS),
+            data=json.dumps(datacite_payload),
+            headers={"Content-Type": "application/vnd.api+json"},
+            timeout=30,
+        )
+    else:
+        res = requests.put(
+            f"{settings.DATACITE_URL.strip('/')}/dois/{doi}",
+            auth=(settings.DATACITE_USER, settings.DATACITE_PASS),
+            data=json.dumps(datacite_payload),
+            headers={"Content-Type": "application/vnd.api+json"},
+            timeout=30,
+        )
+
+    return res.json()
+
+
+def publish_datacite_doi(doi: str):
+    """
+    Set a DOI's status to `Findable` in Datacite.
+    """
+    payload = {"data": {"type": "dois", "attributes": {"event": "publish"}}}
+
+    res = requests.put(
+        f"{settings.DATACITE_URL.strip('/')}/dois/{doi}",
+        auth=(settings.DATACITE_USER, settings.DATACITE_PASS),
+        data=json.dumps(payload),
+        headers={"Content-Type": "application/vnd.api+json"},
+        timeout=30,
+    )
+    return res.json()
+
+
+def hide_datacite_doi(doi: str):
+    """
+    Remove a Datacite DOI from public consumption.
+    """
+    payload = {"data": {"type": "dois", "attributes": {"event": "hide"}}}
+
+    res = requests.put(
+        f"{settings.DATACITE_URL.strip('/')}/dois/{doi}",
+        auth=(settings.DATACITE_USER, settings.DATACITE_PASS),
+        data=json.dumps(payload),
+        headers={"Content-Type": "application/vnd.api+json"},
+        timeout=30,
+    )
+    return res.json()
+
+
+def get_doi_publication_date(doi: str) -> str:
+    """Look up the publication date for a DOI"""
+    res = requests.get(f"{settings.DATACITE_URL.strip('/')}/dois/{doi}", timeout=30)
+    res.raise_for_status()
+    return res.json()["data"]["attributes"]["created"]

server/portal/apps/projects/workspace_operations/project_publish_operations.py

@@ -8,6 +8,7 @@
 from portal.apps._custom.drp import constants
 from portal.libs.agave.utils import user_account, service_account
 from portal.apps.publications.models import Publication, PublicationRequest
+from portal.apps.projects.workspace_operations.datacite_operations import get_datacite_json, upsert_datacite_json, publish_datacite_doi
 from django.db import transaction
 from portal.apps.projects.workspace_operations.graph_operations import remove_trash_nodes
 from portal.apps.search.tasks import index_publication
@@ -118,7 +119,12 @@ def publish_project(self, project_id: str, version: Optional[int] = 1):
             value=nx.node_link_data(publication_tree),
         )
 
-        doi = 'test_doi'  # Replace with actual DOI retrieval logic
+        # Mint a DataCite DOI
+        existing_doi = source_project.value.get("doi", None)
+
+        datacite_json = get_datacite_json(publication_tree)
+        datacite_resp = upsert_datacite_json(datacite_json, doi=existing_doi)
+        doi = datacite_resp["data"]["id"]
 
         # Update project metadata with datacite doi
         source_project_id = f'{settings.PORTAL_PROJECTS_SYSTEM_PREFIX}.{project_id}'
@@ -140,6 +146,9 @@ def publish_project(self, project_id: str, version: Optional[int] = 1):
             defaults={"value": published_project.value, "tree": nx.node_link_data(pub_tree), "version": version},
         )
 
+        if not settings.DEBUG:
+            publish_datacite_doi(doi)
+
         index_publication(project_id)
 
         # transfer files 

server/portal/settings/settings.py

@@ -592,6 +592,21 @@
 PORTAL_PUBLICATION_REVIEWERS_GROUP_NAME = settings_custom.\
     _PORTAL_PUBLICATION_REVIEWERS_GROUP_NAME
 
+PORTAL_PUBLICATION_DATACITE_SHOULDER = settings_custom.\
+    _PORTAL_PUBLICATION_DATACITE_SHOULDER
+
+PORTAL_PUBLICATION_DATACITE_URL_PREFIX = settings_custom.\
+    _PORTAL_PUBLICATION_DATACITE_URL_PREFIX
+
+DATACITE_URL = settings_custom.\
+    _DATACITE_URL
+
+DATACITE_USER = settings_secret.\
+    _DATACITE_USER
+
+DATACITE_PASS = settings_secret.\
+    _DATACITE_PASS
+
 PORTAL_PROJECTS_PRIVATE_KEY = settings_secret.\
     _PORTAL_PROJECTS_PRIVATE_KEY
 

server/portal/settings/settings_default.py

@@ -223,6 +223,11 @@
 
 _PORTAL_PUBLICATION_REVIEWERS_GROUP_NAME = 'PROJECT_REVIEWER'
 
+# Datacite
+_PORTAL_PUBLICATION_DATACITE_SHOULDER = "10.80023"
+_PORTAL_PUBLICATION_DATACITE_URL_PREFIX = "https://cep.test/data/tapis/projects/drp.project.published.test"
+_DATACITE_URL = "https://api.test.datacite.org/"
+
 ########################
 # Custom Portal Template Assets
 # Asset path root is static files output dir.

server/portal/settings/settings_secret.example.py

@@ -116,3 +116,6 @@
         "directory": "external-resources"
     }
 }
+
+_DATACITE_USER = "tdl.tacc"
+_DATACITE_PASS = "CHANGEME"