Initial implementation of legacyId normalization for III identifiers …

…for bibs and holdings

src/folio_migration_tools/mapping_file_transformation/holdings_mapper.py

@@ -1,7 +1,8 @@
 import ast
-import i18n
 
+import i18n
 from folio_uuid.folio_uuid import FOLIONamespaces
+from folio_uuid.folio_uuid import FolioUUID
 from folioclient import FolioClient
 
 from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
@@ -126,28 +127,22 @@ def get_instance_ids(self, legacy_value: str, index_or_id: str):
             i18n.t("Number of bib records referenced in item") + f": {len(legacy_bib_ids)}",
         )
         for legacy_instance_id in legacy_bib_ids:
-            new_legacy_value = (
-                f".{legacy_instance_id}"
-                if legacy_instance_id.startswith("b")
-                else legacy_instance_id
-            )
             if (
-                new_legacy_value not in self.instance_id_map
-                and legacy_instance_id not in self.instance_id_map
-            ):
-                self.migration_report.add_general_statistics(
-                    i18n.t("Records not matched to Instances")
-                )
-                s = "Bib id not in instance id map."
-                raise TransformationRecordFailedError(index_or_id, s, new_legacy_value)
-            else:
+                new_legacy_value := FolioUUID.clean_iii_identifiers(legacy_instance_id)
+            ) and new_legacy_value in self.instance_id_map:
                 self.migration_report.add_general_statistics(
                     i18n.t("Records matched to Instances")
                 )
                 entry = self.instance_id_map.get(new_legacy_value, "") or self.instance_id_map.get(
                     legacy_instance_id
                 )
                 return_ids.append(entry[1])
+            else:
+                self.migration_report.add_general_statistics(
+                    i18n.t("Records not matched to Instances")
+                )
+                s = "Bib id not in instance id map."
+                raise TransformationRecordFailedError(index_or_id, s, new_legacy_value)
         if any(return_ids):
             return return_ids
         else:

src/folio_migration_tools/mapping_file_transformation/item_mapper.py

@@ -5,9 +5,10 @@
 from datetime import timezone
 from typing import Set
 from uuid import uuid4
-import i18n
 
+import i18n
 from folio_uuid.folio_uuid import FOLIONamespaces
+from folio_uuid.folio_uuid import FolioUUID
 from folioclient import FolioClient
 
 from folio_migration_tools.custom_exceptions import TransformationProcessError
@@ -232,6 +233,7 @@ def get_prop(self, legacy_item, folio_prop_name, index_or_id, schema_default_val
                     self.unique_barcodes.add(barcode)
                 return barcode
         elif folio_prop_name == "holdingsRecordId":
+            mapped_value = FolioUUID.clean_iii_identifiers(mapped_value)
             if mapped_value in self.holdings_id_map:
                 return self.holdings_id_map[mapped_value][1]
             elif f"{self.bib_id_template}{mapped_value}" in self.holdings_id_map:

src/folio_migration_tools/marc_rules_transformation/marc_file_processor.py

@@ -6,6 +6,7 @@
 
 import i18n
 from folio_uuid.folio_namespaces import FOLIONamespaces
+from folio_uuid.folio_uuid import FolioUUID
 from pymarc import Field
 from pymarc import Record
 from pymarc import Subfield
@@ -254,6 +255,7 @@ def wrap_up(self):
 
     def add_legacy_ids_to_map(self, folio_rec, filtered_legacy_ids):
         for legacy_id in filtered_legacy_ids:
+            legacy_id = FolioUUID.clean_iii_identifiers(legacy_id)
             self.legacy_ids.add(legacy_id)
             if legacy_id not in self.mapper.id_map:
                 self.mapper.id_map[legacy_id] = self.mapper.get_id_map_tuple(

src/folio_migration_tools/marc_rules_transformation/rules_mapper_holdings.py

@@ -169,7 +169,7 @@ def set_instance_id_by_map(self, legacy_ids: list, folio_holding: dict, marc_rec
                 ("No 004 in record. The tools only support bib-mfhd linking throuh 004"),
                 legacy_ids,
             )
-        legacy_instance_id = marc_record["004"].data.strip()
+        legacy_instance_id = FolioUUID.clean_iii_identifiers(marc_record["004"].data.strip())
         folio_holding["formerIds"].append(f"{self.bib_id_template}{legacy_instance_id}")
         if legacy_instance_id in self.parent_id_map:
             folio_holding["instanceId"] = self.parent_id_map[legacy_instance_id][1]

src/folio_migration_tools/migration_tasks/holdings_csv_transformer.py

@@ -5,12 +5,13 @@
 import sys
 import time
 import traceback
-import i18n
 from typing import Annotated
 from typing import List
 from typing import Optional
 
+import i18n
 from folio_uuid.folio_namespaces import FOLIONamespaces
+from folio_uuid.folio_uuid import FolioUUID
 from httpx import HTTPError
 from pydantic import Field
 
@@ -249,9 +250,9 @@ def wrap_up(self):
                         # Prevent the first item in a boundwith to be overwritten
                         # TODO: Find out why not
                         # if legacy_id not in self.holdings_id_map:
-                        self.holdings_id_map[legacy_id] = self.mapper.get_id_map_tuple(
-                            legacy_id, holding, self.object_type
-                        )
+                        self.holdings_id_map[
+                            FolioUUID.clean_iii_identifiers(legacy_id)
+                        ] = self.mapper.get_id_map_tuple(legacy_id, holding, self.object_type)
                     Helper.write_to_file(holdings_file, holding)
                     self.mapper.migration_report.add_general_statistics(
                         i18n.t("Holdings Records Written to disk")