Ensure correct publication dates after subsequent publish (#1436)

designsafe/apps/api/projects_v2/operations/datacite_operations.py

@@ -247,3 +247,10 @@ def hide_datacite_doi(doi: str):
         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"]

designsafe/apps/api/projects_v2/operations/project_publish_operations.py

@@ -8,6 +8,7 @@
 import datetime
 from pathlib import Path
 import logging
+import requests
 from django.conf import settings
 from django.db import close_old_connections
 import networkx as nx
@@ -19,6 +20,7 @@
     get_datacite_json,
     publish_datacite_doi,
     upsert_datacite_json,
+    get_doi_publication_date,
 )
 from designsafe.apps.api.projects_v2.operations.project_archive_operations import (
     archive_publication_async,
@@ -319,21 +321,22 @@ def get_publication_subtree(
     return subtree, path_mapping
 
 
-def fix_publication_dates(existing_tree: nx.DiGraph, incoming_tree: nx.DiGraph):
+def fix_publication_dates(incoming_tree: nx.DiGraph):
     """
     Update publication date on versioned pubs to match the initial publication date.
     """
-    initial_pub_dates = {}
-    for published_entity in existing_tree.successors("NODE_ROOT"):
-        published_uuid = existing_tree.nodes[published_entity]["uuid"]
-        initial_pub_dates[published_uuid] = existing_tree.nodes[published_entity][
-            "publicationDate"
-        ]
-    for node in incoming_tree:
-        if incoming_tree.nodes[node]["uuid"] in initial_pub_dates:
-            incoming_tree.nodes[node]["publicationDate"] = initial_pub_dates[
-                incoming_tree.nodes[node]["uuid"]
-            ]
+
+    for incoming_node in incoming_tree.successors("NODE_ROOT"):
+        node_data = incoming_tree.nodes[incoming_node]
+        existing_doi = next(iter(node_data["value"].get("dois", [])), None)
+        if existing_doi:
+            try:
+                existing_pub_date = get_doi_publication_date(existing_doi)
+                incoming_tree.nodes[incoming_node]["publicationDate"] = (
+                    datetime.datetime.fromisoformat(existing_pub_date).isoformat()
+                )
+            except requests.HTTPError:
+                logger.error("Datacite lookup error for DOI %s", existing_doi)
 
     return incoming_tree
 
@@ -356,12 +359,13 @@ def get_publication_full_tree(
 
     full_tree = nx.compose_all(subtrees)
 
+    # Update publication date on versioned/amended pubs to match the initial publication date.
+    full_tree = fix_publication_dates(full_tree)
+
     if version and version > 1:
         existing_pub = Publication.objects.get(project_id=project_id)
         published_tree: nx.DiGraph = nx.node_link_graph(existing_pub.tree)
 
-        # Update publication date on versioned pubs to match the initial publication date.
-        full_tree = fix_publication_dates(published_tree, full_tree)
         full_tree = nx.compose(published_tree, full_tree)
 
     return full_tree, full_path_mapping