cnum_provider: fix issues with wrong cnums when holes in cnums sequence

 * INSPIR-3236

Author: Pawel Zembrzuski

backend/inspirehep/pidstore/providers/cnum.py

@@ -16,6 +16,7 @@
 
 class InspireCNUMProvider(BaseProvider):
     """CNUM identifier provider."""
+
     pid_type = None
     pid_provider = "cnum"
     default_status = PIDStatus.RESERVED
@@ -50,15 +51,27 @@ def next(cls, data):
         opening_date = data.get("opening_date")
         if not opening_date:
             return
-
         adjusted_date = datetime.strptime(opening_date, "%Y-%m-%d").strftime("%y-%m-%d")
         cnum = f"C{adjusted_date}"
-        cnums_count = PersistentIdentifier.query\
-            .filter(PersistentIdentifier.pid_value.like(f"{cnum}%"))\
-            .filter(PersistentIdentifier.pid_type == "cnum")\
-            .count()
+        all_postfixes = []
+        all_cnums = (
+            PersistentIdentifier.query.with_for_update()
+            .filter(PersistentIdentifier.pid_value.like(f"{cnum}%"))
+            .filter(PersistentIdentifier.pid_type == "cnum")
+            .all()
+        )
+        if not all_cnums:
+            return cnum
+
+        for cnum_entry in all_cnums:
+            cnum_value = cnum_entry.pid_value.split(".")
+            if len(cnum_value) > 1:
+                all_postfixes.append(int(cnum_value[-1]))
+        if all_postfixes:
+            postfix = sorted(all_postfixes)[-1] + 1
+        else:
+            postfix = 1
 
-        if cnums_count > 0:
-            cnum = f"{cnum}.{cnums_count}"
+        full_cnum = f"{cnum}.{postfix}"
 
-        return cnum
+        return full_cnum

backend/tests/integration/pidstore/minters/test_minters_cnum.py

@@ -189,3 +189,43 @@ def test_minter_mints_cnum_on_update_when_cnum_is_missing_in_db(
     expected_cnum = "C05-01-01"
     assert 1 == len(record_cnums)
     assert record_cnums[0].pid_value == expected_cnum
+
+
+def test_generate_cnum_when_holes_in_cnums_sequence(base_app, db, create_record):
+    data = {"opening_date": "2020-01-01"}
+    expected_cnum = "C20-01-01.2"
+    rec1 = create_record("con", data)
+    rec2 = create_record("con", data)
+    rec1.hard_delete()
+    rec3 = create_record("con", data)
+    assert rec3.get("cnum") == expected_cnum
+
+
+def test_generate_cnum_when_holes_in_cnums_sequence_and_weird_creation_order(
+    base_app, db, create_record
+):
+    data = {"opening_date": "2020-01-01", "cnum": "C20-01-01.2"}
+    rec1 = create_record("con", data)
+    data["cnum"] = "C20-01-01.1"
+    rec2 = create_record("con", data)
+    data["cnum"] = "C20-01-01"
+    rec3 = create_record("con", data)
+    del data["cnum"]
+    rec2.hard_delete()
+    rec4 = create_record("con", data)
+    assert rec4["cnum"] == "C20-01-01.3"
+
+
+def test_generate_cnum_when_holes_in_cnums_sequence_and_big_holes(
+    base_app, db, create_record
+):
+    data = {"opening_date": "2020-01-01", "cnum": "C20-01-01.20"}
+    rec1 = create_record("con", data)
+    data["cnum"] = "C20-01-01.31"
+    rec2 = create_record("con", data)
+    data["cnum"] = "C20-01-01.3"
+    rec3 = create_record("con", data)
+    del data["cnum"]
+    rec2.hard_delete()
+    rec4 = create_record("con", data)
+    assert rec4["cnum"] == "C20-01-01.21"