3895 check against oa start date

doajtest/testbook/article_xml_upload/article_doaj_xml_upload.yml

@@ -357,3 +357,24 @@ tests:
         - Related background job is found
         - status is "complete"
         - Outcome Status is "success"
+
+- title: Upload a file before OA start date of the Journal
+  context:
+    role: publisher
+  steps:
+  - step: Make sure the OA start date of the journal in in future
+  - step: Go to the "Upload Article XML" tab in the "Publisher Area"
+  - step: Select "Choose file" and select the test resource file "no_issn.xml"
+    resource: /xml_upload_test_package/DOAJ/successful.xml
+  - step: Click "Upload"
+    results:
+    - 'A flash message appears at the top of the screen indicating a successful upload:
+      File uploaded and waiting to be processed. Check back here for updates.(Dismiss)'
+    - Your file is shown in the "History of uploads" with status "pending"
+  - step: wait a short amount of time for the job to process, then reload the page
+      (do not re-submit the form data).  If the job remains in "pending", reload the
+      page until the status changes.
+    results:
+    - Your file is shown in the "History of uploads" with status "processing failed"
+      and a entry in the "Notes" and reads as 'Article(s) cannot be uploaded before OA start date of the Journal'.  Check that the explanation link goes to
+      a suitable reason and resolution for the problem.

doajtest/unit/test_tasks_ingestDOAJarticles.py

@@ -23,7 +23,7 @@
 from portality.crosswalks import article_doaj_xml
 from portality.tasks import ingestarticles
 from portality.ui.messages import Messages
-
+from portality.lib import dates
 
 class TestIngestArticlesDoajXML(DoajTestCase):
 
@@ -1048,3 +1048,76 @@ def test_61_journal_not_indoaj(self):
 
         assert file_upload.status == "failed"
         assert file_upload.error == Messages.EXCEPTION_ADDING_ARTICLE_TO_WITHDRAWN_JOURNAL
+
+    def test_62_article_before_oa_start(self):
+        journal = article_upload_tester.create_simple_journal("testowner", pissn="1234-5678", eissn="9876-5432")
+        journal.bibjson().oa_start = dates.now().year
+        helpers.save_all_block_last([ journal,
+            article_upload_tester.create_simple_publisher("testowner")
+        ])
+
+        # make both handles, as we want as little gap as possible between requests in a moment
+        handle1 = DoajXmlArticleFixtureFactory.upload_2_issns_correct()
+
+        f1 = FileMockFactory(stream=handle1)
+
+        job1 = ingestarticles.IngestArticlesBackgroundTask.prepare("testowner", schema="doaj", upload_file=f1)
+        id1 = job1.params.get("ingest_articles__file_upload_id")
+        self.cleanup_ids.append(id1)
+
+        # because file upload gets created and saved by prepare
+        time.sleep(1)
+
+        task1 = ingestarticles.IngestArticlesBackgroundTask(job1)
+
+        task1.run()
+
+        # because file upload needs to be re-saved
+        time.sleep(1)
+
+        fu1 = models.FileUpload.pull(id1)
+
+        assert fu1.status == "failed", "received status: {}".format(fu1.status)
+        assert job1.outcome_status == "fail"
+
+        assert any('Articles before OA start date: Imaginaires autochtones contemporains. Introduction' in entry['message'] for entry in
+                   job1.audit), "No message found with 'Articles before OA start date'"
+
+        # check that article not created
+        assert models.Article.count_by_issns(["1234-5678", "9876-5432"]) == 0
+
+    def test_63_article_after_oa_start(self):
+        journal = article_upload_tester.create_simple_journal("testowner", pissn="1234-5678", eissn="9876-5432")
+        journal.bibjson().oa_start = 2013
+        helpers.save_all_block_last([ journal,
+            article_upload_tester.create_simple_publisher("testowner")
+        ])
+
+        # make both handles, as we want as little gap as possible between requests in a moment
+        handle1 = DoajXmlArticleFixtureFactory.upload_2_issns_correct()
+
+        f1 = FileMockFactory(stream=handle1)
+
+        job1 = ingestarticles.IngestArticlesBackgroundTask.prepare("testowner", schema="doaj", upload_file=f1)
+        id1 = job1.params.get("ingest_articles__file_upload_id")
+        self.cleanup_ids.append(id1)
+
+        # because file upload gets created and saved by prepare
+        time.sleep(1)
+
+        task1 = ingestarticles.IngestArticlesBackgroundTask(job1)
+
+        task1.run()
+
+        # because file upload needs to be re-saved
+        time.sleep(1)
+
+        fu1 = models.FileUpload.pull(id1)
+
+        assert fu1.status == "processed", "received status: {}".format(fu1.status)
+
+        assert not any('Articles before OA start date: Imaginaires autochtones contemporains. Introduction' in entry['message'] for entry in
+                   job1.audit), "No message found with 'Articles before OA start date'"
+
+        # check that article not created
+        assert models.Article.count_by_issns(["1234-5678", "9876-5432"]) == 1

portality/bll/exceptions.py

@@ -73,6 +73,12 @@ def __str__(self):
         super(ArticleNotAcceptable, self).__str__()
         return self.message
 
+class ArticleBeforeOAStartDate(ArticleNotAcceptable):
+    """
+    Exception to raise when the article is uploaded before OA start date of the Journal
+    """
+    pass
+
 class ArticleMergeConflict(Exception):
     """
     Exception to raise when it's not clear which article to merge an update with

portality/bll/services/article.py

@@ -55,6 +55,7 @@ def batch_create_articles(self, articles, account, duplicate_check=True, merge_d
         all_shared = set()
         all_unowned = set()
         all_unmatched = set()
+        all_before_oa_start_date = set()
 
         # Hold on to the exception so we can raise it later
         e_not_acceptable = None
@@ -70,6 +71,11 @@ def batch_create_articles(self, articles, account, duplicate_check=True, merge_d
                                              dry_run=True)
             except (exceptions.ArticleMergeConflict, exceptions.ConfigurationException):
                 raise exceptions.IngestException(message=Messages.EXCEPTION_ARTICLE_BATCH_CONFLICT)
+            except exceptions.ArticleBeforeOAStartDate as e:
+                all_before_oa_start_date.add(article.bibjson().title)
+                result = {'fail': 1}
+                e.message = e.message.format(title=",".join(list(all_before_oa_start_date)))
+                e_not_acceptable = e
             except exceptions.ArticleNotAcceptable as e:
                 # The ArticleNotAcceptable exception is a superset of reasons we can't match a journal to this article
                 e_not_acceptable = e
@@ -84,7 +90,7 @@ def batch_create_articles(self, articles, account, duplicate_check=True, merge_d
             all_unmatched.update(result.get("unmatched", set()))
 
         report = {"success": success, "fail": fail, "update": update, "new": new, "shared": all_shared,
-                  "unowned": all_unowned, "unmatched": all_unmatched}
+                  "unowned": all_unowned, "unmatched": all_unmatched, "before_oa_start_date":all_before_oa_start_date}
 
         # if there were no failures in the batch, then we can do the save
         if fail == 0:
@@ -235,8 +241,15 @@ def create_article(self, article, account, duplicate_check=True, merge_duplicate
             except (exceptions.DuplicateArticleException, exceptions.ArticleMergeConflict, exceptions.ConfigurationException) as e:
                 raise e
 
+        # Check if article is uploaded before OA start date of Journal and reject the article
+        journal = article.get_journal()
+        published_year = int(article.bibjson().year)
+        oa_start_date = journal.has_oa_start_date()
+        if oa_start_date and published_year < oa_start_date:
+            raise exceptions.ArticleBeforeOAStartDate(message=Messages.EXCEPTION_ARTICLE_BEFORE_OA_START_DATE)
+
         if add_journal_info:
-            article.add_journal_metadata()
+            article.add_journal_metadata(j=journal)
 
         # finally, save the new article
         if not dry_run:

portality/tasks/helpers/articles_upload_helper.py

@@ -84,6 +84,7 @@ def upload_process(articles_upload: BaseArticlesUpload,
     shared = result.get("shared", [])
     unowned = result.get("unowned", [])
     unmatched = result.get("unmatched", [])
+    before_oa_start_date = result.get("before_oa_start_date", [])
 
     if success == 0 and fail > 0 and not ingest_exception:
         articles_upload.failed("All articles in file failed to import")
@@ -99,6 +100,8 @@ def upload_process(articles_upload: BaseArticlesUpload,
     job.add_audit_message("Shared ISSNs: " + ", ".join(list(shared)))
     job.add_audit_message("Unowned ISSNs: " + ", ".join(list(unowned)))
     job.add_audit_message("Unmatched ISSNs: " + ", ".join(list(unmatched)))
+    if len(before_oa_start_date) > 0:
+        job.add_audit_message("Articles before OA start date: " + ", ".join(list(before_oa_start_date)))
 
     if new:
         ids = [a.id for a in articles]

portality/templates-v2/public/publisher/xml_help.html

@@ -226,6 +226,17 @@ <h2 id="explanations">Explanation of XML errors</h2>
                 Ensure that all your articles have the correct DOIs and full-text links. If it still doesn’t work please <a href="https://github.com/DOAJ/doaj/issues/new/choose" target="_blank" rel="noopener">submit a bug report</a> or <a href="{{ url_for('doaj.contact') }}">contact us</a> with the details; we may need to clean up your existing articles manually.
             </td>
         </tr>
+        <tr>
+            <th>
+                <code>Article(s)  cannot be uploaded before OA start date of the Journal</code>
+            </th>
+            <td>
+                One or more articles in the xml has the <strong>publicationDate</strong> which is before OpenAccess start date of the Journal.
+            </td>
+            <td>
+                Ensure that all your articles are uploaded on or after OpenAccess start date of the Journal.
+            </td>
+        </tr>
         </tbody>
     </table>
 {% endblock %}

portality/ui/messages.py

@@ -64,6 +64,7 @@ class Messages(object):
     EXCEPTION_IDENTICAL_PISSN_AND_EISSN = "The Print and Online ISSNs supplied are identical. If you supply two ISSNs, they must be different."
     EXCEPTION_NO_ISSNS = "Neither the Print ISSN nor Online ISSN have been supplied. DOAJ requires at least one ISSN."
     EXCEPTION_INVALID_BIBJSON = "Invalid article bibjson: "  # + Dataobj exception message
+    EXCEPTION_ARTICLE_BEFORE_OA_START_DATE = "Article(s) '{title}' cannot be uploaded before OA start date of the Journal"
 
     EXCEPTION_IDENTIFIER_CHANGE_CLASH = "DOI or Fulltext URL has been changed to match another article that already exists in DOAJ"
     EXCEPTION_IDENTIFIER_CHANGE = "Either the DOI or Fulltext URL has been changed.  This operation is not permitted; please contact an administrator for help."