deposit: handling missing recids

* Missing record IDs are created in reserved state and deposit is
  marked as unsubmitted.

Signed-off-by: Krzysztof Nowak <k.nowak@cern.ch>

invenio_migrator/tasks/deposit.py

@@ -79,7 +79,7 @@ def create_record_and_pid(data):
 def create_files_and_sip(deposit, dep_pid):
     """Create deposit Bucket, Files and SIPs."""
     from invenio_pidstore.errors import PIDDoesNotExistError
-    from invenio_pidstore.models import PersistentIdentifier
+    from invenio_pidstore.models import PersistentIdentifier, PIDStatus
     from invenio_sipstore.errors import SIPUserDoesNotExist
     from invenio_sipstore.models import SIP, RecordSIP, SIPFile
     from invenio_files_rest.models import Bucket, FileInstance, ObjectVersion
@@ -162,7 +162,23 @@ def create_files_and_sip(deposit, dep_pid):
                 logger.exception('Record {recid} referred in '
                                  'Deposit {depid} does not exists.'.format(
                                      recid=recid, depid=dep_pid.pid_value))
-                raise DepositRecidDoesNotExist(dep_pid.pid_value, recid)
+                if deposit['_p']['submitted'] == True:
+                    logger.exception('Pair {recid}/{depid} was submitted,'
+                                     ' setting to unpublished.'.format(
+                                         recid=recid, depid=dep_pid.pid_value))
+                    deposit['_p']['submitted'] = False
+                else:
+                    logger.exception('Pair {recid}/{depid} was not submitted.'.
+                        format(recid=recid, depid=dep_pid.pid_value))
+
+                # Reserve recid
+                pid = PersistentIdentifier.create(
+                    pid_type='recid',
+                    pid_value=str(recid),
+                    object_type='rec',
+                    object_uuid=str(deposit.id),
+                    status=PIDStatus.RESERVED
+                )
         if idx == 0:
             for fp, fi in dep_file_instances:
                 sipf = SIPFile(sip_id=sip.id, filepath=fp, file_id=fi.id)

tests/unit/test_deposit_load.py

@@ -28,14 +28,14 @@
 
 import pytest
 from invenio_pidstore.resolver import Resolver
+from invenio_pidstore.models import PersistentIdentifier, PIDStatus
 from invenio_records.api import Record
 from invenio_records.models import RecordMetadata
 from invenio_records_files.api import Record as RecordFiles
 from invenio_sipstore.models import SIP, RecordSIP, SIPFile
 
 from invenio_migrator.tasks.deposit import load_deposit
-from invenio_migrator.tasks.errors import DepositMultipleRecids, \
-    DepositRecidDoesNotExist
+from invenio_migrator.tasks.errors import DepositMultipleRecids
 
 
 def test_deposit_load(dummy_location, deposit_user, deposit_record_pid):
@@ -47,7 +47,7 @@ def test_deposit_load(dummy_location, deposit_user, deposit_record_pid):
     dep2 = dict(sips=[dict(metadata=dict(recid='50'),
                            agents=[dict(user_id=1), ],
                            package='Content2'), ],
-                _p=dict(id='2'))
+                _p=dict(id='2', submitted=True))
     dep3 = dict(sips=[dict(metadata=dict(recid='10'),
                            agents=[dict(user_id=5), ],
                            package='Content3'), ],
@@ -60,13 +60,21 @@ def test_deposit_load(dummy_location, deposit_user, deposit_record_pid):
                            package='Content5'), ],
                 _p=dict(id='4'))
     load_deposit(dep1)
-    pytest.raises(DepositRecidDoesNotExist, load_deposit, dep2)
     pytest.raises(DepositMultipleRecids, load_deposit, dep4)
 
     # Should set user to null because user_id does not exist
     load_deposit(dep3)
     assert SIP.query.filter_by(content="Content3").one().user_id is None
 
+    # Should create reserved recid and leave deposit unsubmitted
+    load_deposit(dep2)
+    assert PersistentIdentifier.query.filter_by(
+        pid_type='recid', pid_value='50').one().status == PIDStatus.RESERVED
+
+    pid, dep = Resolver(pid_type='depid', object_type='rec',
+                        getter=Record.get_record).resolve('2')
+    assert not dep['_p']['submitted']
+
 
 def test_deposit_load_task(dummy_location, deposit_dump, deposit_user,
                            deposit_record_pid):