weko#25440 - fixbug harvest permission on OAI-PMH

Author: weko3-dev040

modules/invenio-oaiserver/invenio_oaiserver/query.py

@@ -146,16 +146,19 @@ def add_condition_doi_and_future_date(query):
             search = search.filter('range', **{'_updated': time_range})
 
         search = search.query('match', **{'relation_version_is_last': 'true'})
-        indexes = Indexes.get_unharvested_indexes()
-        for index in indexes:
-            search = search.query(
-                'bool',
-                **{'must_not': [
-                    {'wildcard': {'path': str(index.id)}}]})
-            search = search.query(
-                'bool',
-                **{'must_not': [
-                    {'wildcard': {'path': '*/' + str(index.id)}}]})
+        index_paths = Indexes.get_harverted_index_list()
+        query_filter = [
+            # script get deleted items.
+            {"bool": {"must_not": {"exists": {"field": "path"}}}}
+        ]
+        for index_path in index_paths:
+            query_filter.append({
+                "wildcard": {
+                    "path": index_path
+                }
+            })
+        search = search.query(
+            'bool', **{'must': [{'bool': {'should': query_filter}}]})
         add_condition_doi_and_future_date(search)
         response = search.execute().to_dict()
     else:

modules/invenio-oaiserver/invenio_oaiserver/response.py

@@ -281,28 +281,7 @@ def is_private_workflow(record):
 
 def is_private_index(record):
     """Check index of workflow is private."""
-    def is_future_date(public_date):
-        """Check public date (Index Tree) is future date."""
-        cur_date = datetime.now()
-        return public_date and public_date > cur_date
-
-    # Get current and parent indexes of this record.
-    list_index = record.get("path")
-    index_lst = []
-    if list_index:
-        index_id_lst = []
-        for index in list_index:
-            indexes = str(index).split('/')
-            index_id_lst.append(indexes[-1])
-        index_lst = index_id_lst
-    # check private and future date of all(current and parent) indexes.
-    indexes = Indexes.get_path_list(index_lst)
-    publish_state = 6
-    publish_date = 7
-    for index in indexes:
-        if not index[publish_state] or is_future_date(index[publish_date]):
-            return True
-    return False
+    return not Indexes.is_public_state(copy.deepcopy(record.get("path")))
 
 
 def set_identifier(param_record, param_rec):
@@ -351,17 +330,18 @@ def get_error_code_msg():
     e_record = SubElement(e_getrecord, etree.QName(NS_OAIPMH, 'record'))
     set_identifier(record, record)
     # Harvest is private
+    _is_private_index = is_private_index(record)
     if not harvest_public_state or\
             (identify and not identify.outPutSetting) or \
-            (is_private_index(record)
+            (_is_private_index
                 and harvest_public_state and is_exists_doi(record)):
         return error(get_error_code_msg(), **kwargs)
     # Item is deleted
     # or Harvest is public & Item is private
     # or Harvest is public & Index is private
     elif is_deleted_workflow(pid) or (
         harvest_public_state and is_private_workflow(record)) or (
-            harvest_public_state and is_private_index(record)):
+            harvest_public_state and _is_private_index):
         header(
             e_record,
             identifier=pid.pid_value,
@@ -446,8 +426,6 @@ def append_deleted_record(e_listrecords, pid_object, rec):
             pid = oaiid_fetcher(record['id'], record['json']['_source'])
             pid_object = OAIIDProvider.get(pid_value=pid.pid_value).pid
             rec = WekoRecord.get_record(record['id'])
-            harvest_public_state, r = \
-                WekoRecord.get_record_with_hps(pid_object.object_uuid)
             set_identifier(record, rec)
             # Check output delete, noRecordsMatch
             if not is_private_index(rec):
@@ -474,7 +452,7 @@ def append_deleted_record(e_listrecords, pid_object, rec):
             e_metadata.append(record_dumper(pid, etree_record))
         except Exception:
             current_app.logger.error(traceback.print_exc())
-            current_app.logger.error('Error when exporting item id'
+            current_app.logger.error('Error when exporting item id '
                                      + str(record['id']))
     # Check <record> tag not exist.
     if len(e_listrecords) == 0:

modules/weko-index-tree/weko_index_tree/api.py

@@ -21,7 +21,7 @@
 """API for weko-index-tree."""
 
 from copy import deepcopy
-from datetime import datetime
+from datetime import date, datetime
 from functools import partial
 
 from flask import current_app, json
@@ -1187,6 +1187,34 @@ def get_harvest_public_state(cls, path):
             current_app.logger.debug(se)
             return False
 
+    @classmethod
+    def is_public_state(cls, paths):
+        """Check have public state."""
+        def _query(path):
+            return db.session. \
+                query(func.every(db.and_(
+                    Index.public_state,
+                    db.or_(
+                        Index.public_date == None,
+                        Index.public_date <= date.today()
+                    ))).label('parent_state')
+                ).filter(Index.id.in_(path))
+
+        try:
+            last_path = paths.pop(-1).split('/')
+            qry = _query(last_path)
+            for i in range(len(paths)):
+                paths[i] = paths[i].split('/')
+                paths[i] = _query(paths[i])
+            smt = qry.union_all(*paths).subquery()
+            result = db.session.query(
+                func.bool_or(
+                    smt.c.parent_state).label('parent_state')).one()
+            return result.parent_state
+        except Exception as se:
+            current_app.logger.debug(se)
+            return False
+
     @classmethod
     def set_item_sort_custom(cls, index_id, sort_json={}):
         """Set custom sort."""
@@ -1496,3 +1524,39 @@ def get_full_path(cls, index_id=0):
             obj = db.session.query(*qlst). \
                 order_by(recursive_t.c.pid).first()
             return obj.path if obj else ''
+
+    @classmethod
+    def get_harverted_index_list(cls):
+        """Get full path of index.
+
+        :return: path.
+        """
+        recursive_t = db.session.query(
+            Index.parent.label("pid"),
+            Index.id.label("cid"),
+            func.cast(Index.id, db.Text).label("path")
+        ).filter(
+            Index.parent == 0,
+            Index.harvest_public_state == True
+        ).cte(name="recursive_t", recursive=True)
+
+        rec_alias = aliased(recursive_t, name="rec")
+        test_alias = aliased(Index, name="t")
+        recursive_t = recursive_t.union_all(
+            db.session.query(
+                test_alias.parent,
+                test_alias.id,
+                rec_alias.c.path + '/' + func.cast(test_alias.id, db.Text)
+            ).filter(
+                test_alias.parent == rec_alias.c.cid,
+                test_alias.harvest_public_state == True)
+        )
+
+        paths = []
+        with db.session.begin_nested():
+            qlst = [recursive_t.c.path]
+            indexes = db.session.query(*qlst). \
+                order_by(recursive_t.c.pid).all()
+            for idx in indexes:
+                paths.append(idx.path)
+        return paths