From eca5e9e109010e663fd49f55a13771beabc213f8 Mon Sep 17 00:00:00 2001
From: Harris Tzovanakis <me@drjova.com>
Date: Mon, 26 Feb 2024 16:45:02 +0100
Subject: [PATCH] articles: fixes

---
 scoap3/articles/tests/test_article_views.py | 49 +--------------------
 scoap3/tasks.py                             | 35 +++------------
 2 files changed, 7 insertions(+), 77 deletions(-)

diff --git a/scoap3/articles/tests/test_article_views.py b/scoap3/articles/tests/test_article_views.py
index a1e344951..dd1265b2f 100644
--- a/scoap3/articles/tests/test_article_views.py
+++ b/scoap3/articles/tests/test_article_views.py
@@ -1,14 +1,10 @@
 import json
-import os
-from pathlib import Path
 
 import pytest
-from django.test import TestCase
 from django.urls import reverse
 from rest_framework import status
 
-from scoap3.articles.models import Article, ArticleIdentifier
-from scoap3.users.tests.factories import UserFactory
+from scoap3.articles.models import Article
 
 pytestmark = pytest.mark.django_db
 
@@ -76,9 +72,6 @@ def test_update_article_from_workflow(self, client, user, shared_datadir):
         assert "10.5506/APhysPolB.54.10-A3" in expected_dois
 
 
-pytestmark = pytest.mark.django_db
-
-
 class TestArticleIdentifierViewSet:
     def test_get_article_identifier(self, client):
         url = reverse("api:articleidentifier-list")
@@ -88,43 +81,3 @@ def test_get_article_identifier(self, client):
         url = reverse("api:articleidentifier-detail", kwargs={"pk": 0})
         response = client.get(url)
         assert response.status_code == status.HTTP_404_NOT_FOUND
-
-
-pytestmark = pytest.mark.django_db
-
-
-class TestTheSameArticleCreationTwice(TestCase):
-    @staticmethod
-    def shared_datadir():
-        return os.path.join(
-            Path(__file__).parent.resolve(), "data", "record_failing_on_airflow.json"
-        )
-
-    def test_create_article_from_workflow(self):
-        user = UserFactory()
-        self.client.force_login(user)
-        with open(self.shared_datadir()) as file:
-            contents = file.read()
-        data = json.loads(contents)
-        response = self.client.post(
-            reverse("api:article-workflow-import-list"),
-            data,
-            content_type="application/json",
-        )
-        assert response.status_code == status.HTTP_200_OK
-
-        response = self.client.post(
-            reverse("api:article-workflow-import-list"),
-            data,
-            content_type="application/json",
-        )
-        assert response.status_code == status.HTTP_200_OK
-        doi = data.get("dois")[0].get("value")
-        assert (
-            len(
-                ArticleIdentifier.objects.filter(
-                    identifier_type="DOI", identifier_value=doi
-                )
-            )
-            == 1
-        )
diff --git a/scoap3/tasks.py b/scoap3/tasks.py
index 53785eb68..535a82dc5 100644
--- a/scoap3/tasks.py
+++ b/scoap3/tasks.py
@@ -125,16 +125,8 @@ def _create_article_identifier(data, article):
             "identifier_type": "arXiv",
             "identifier_value": arxiv.get("value"),
         }
-        doi = data.get("dois")[0].get("value")
-        if ArticleIdentifier.objects.filter(
-            identifier_type="arXiv", article_id=article
-        ).exists():
-            article_identifier = ArticleIdentifier.objects.get(
-                article_id=article, identifier_type="arXiv"
-            )
-            article_identifier.__dict__.update(**article_identifier_data)
-        else:
-            ArticleIdentifier.objects.get_or_create(**article_identifier_data)
+
+        ArticleIdentifier.objects.get_or_create(**article_identifier_data)
 
 
 def _create_copyright(data, article):
@@ -145,11 +137,7 @@ def _create_copyright(data, article):
             "holder": copyright.get("holder", ""),
             "year": copyright.get("year"),
         }
-        if Copyright.objects.filter(article_id=article).exists():
-            copyright = Copyright.objects.get(article_id=article)
-            copyright.__dict__.update(**copyright_data)
-        else:
-            Copyright.objects.get_or_create(**copyright_data)
+        Copyright.objects.get_or_create(**copyright_data)
 
 
 def _create_article_arxiv_category(data, article):
@@ -162,15 +150,7 @@ def _create_article_arxiv_category(data, article):
                 "category": arxiv_category,
                 "primary": True if idx == 0 else False,
             }
-            if ArticleArxivCategory.objects.filter(article_id=article).exists():
-                article_arxiv_category = ArticleArxivCategory.objects.get(
-                    article_id=article
-                )
-                article_arxiv_category.__dict__.update(**article_arxiv_category_data)
-            else:
-                ArticleArxivCategory.objects.get_or_create(
-                    **article_arxiv_category_data
-                )
+            ArticleArxivCategory.objects.get_or_create(**article_arxiv_category_data)
 
 
 def _create_publisher(data):
@@ -198,11 +178,8 @@ def _create_publication_info(data, article, publishers):
             "journal_issue_date": publication_info.get("journal_issue_date"),
             "publisher_id": publishers[idx].id,
         }
-        if PublicationInfo.objects.filter(article_id=article).exists():
-            publication_info = PublicationInfo.objects.get(article_id=article)
-            publication_info.__dict__.update(**publication_info_data)
-        else:
-            PublicationInfo.objects.get_or_create(**publication_info_data)
+
+        PublicationInfo.objects.get_or_create(**publication_info_data)
 
 
 def _create_experimental_collaborations(data):