Merge branch 'hotfix/3857_harvest_errors'

DOAJ · Jan 13, 2025 · a5d1d64 · a5d1d64
2 parents 288f6f7 + 9eccc49
commit a5d1d64
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 6 deletions.
diff --git a/portality/bll/services/article.py b/portality/bll/services/article.py
@@ -126,13 +126,13 @@ def _prepare_update_admin(article, duplicate, update_article_id, merge_duplicate
 
         is_update = 0
         if duplicate is not None:
-            if duplicate.id != update_article_id:
+            if update_article_id is not None and duplicate.id != update_article_id:
                 # it means that doi or ft url has been changed so that it duplicates existing article
                 raise exceptions.DuplicateArticleException(Messages.EXCEPTION_IDENTIFIER_CHANGE_CLASH)
             elif merge_duplicate:
                 is_update += 1
                 article.merge(duplicate)
-        elif merge_duplicate:  # requested to update article has both url and doi changed to new values - no duplicate detected
+        elif update_article_id is not None and merge_duplicate:  # requested to update article has both url and doi changed to new values - no duplicate detected
             is_update += 1
             art = models.Article.pull(update_article_id)
             article.merge(art)
@@ -227,7 +227,7 @@ def create_article(self, article, account, duplicate_check=True, merge_duplicate
             # ~~!ArticleCreate:Feature->ArticleDeduplication:Feature~~
             duplicate = self.get_duplicate(article)
             try:
-                if account.has_role("admin") and update_article_id is not None:     # is update_article_id is None then treat as normal publisher upload
+                if account.has_role("admin"):     # is update_article_id is None then treat as normal publisher upload
                                                                                     # for testing by admin
                     is_update = self._prepare_update_admin(article, duplicate, update_article_id, merge_duplicate)
                 else:

diff --git a/portality/tasks/harvester_helpers/workflow.py b/portality/tasks/harvester_helpers/workflow.py
@@ -7,6 +7,16 @@
 from portality.lib.dataobj import DataObjException
 
 
+SYSTEM_ACCOUNT = {
+    "email": app.config['ADMIN_EMAIL'],
+    "name": "harvest_background_job",
+    "role": ['admin'],
+    "id": "harvest_background_job"
+}
+
+sys_acc = Account(**SYSTEM_ACCOUNT)
+
+
 class DefaultLogger(object):
     def __init__(self):
         self._log = []
@@ -110,12 +120,13 @@ def process_article(self, account_id, article):
             Report.record_error((article.get_identifier("doi") or "< DOI MISSING >") + " - " + str(e))
             return False
 
-        acc = Account.pull(account_id)
         try:
-            id = ArticlesCrudApi.create(article.data, acc).id
+            # Use the admin account to have sufficient rights to update full text or doi.
+            # Related github issue #3857
+            id = ArticlesCrudApi.create(article.data, sys_acc).id
         except Exception as e:
             self._write_to_logger("Article caused DOAJException: {m} ... skipping".format(m=e))
             Report.record_error((article.get_identifier("doi") or "< DOI MISSING >"))
             return False
-        self._write_to_logger("Created article in DOAJ for Account:{x} with ID: {y}".format(x=account_id, y=id))
+        self._write_to_logger("Created article in DOAJ for Account:{x} with ID: {y}".format(x=sys_acc.id, y=id))
         return True