folio-migration-tools 1.9.0rc8__tar.gz → 1.9.0rc10__tar.gz

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: folio_migration_tools
-Version: 1.9.0rc8
+Version: 1.9.0rc10
 Summary: A tool allowing you to migrate data from legacy ILS:s (Library systems) into FOLIO LSP
 License: MIT
 Keywords: FOLIO,ILS,LSP,Library Systems,MARC21,Library data

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "folio_migration_tools"
-version = "1.9.0rc8"
+version = "1.9.0rc10"
 description =  "A tool allowing you to migrate data from legacy ILS:s (Library systems) into FOLIO LSP"
 authors = [
 	{name = "Theodor Tolstoy", email = "github.teddes@tolstoy.se"},

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/mapper_base.py

@@ -272,7 +272,7 @@ class MapperBase:
             [
                 object_type == FOLIONamespaces.instances,
                 (not getattr(self.task_configuration, "data_import_marc", False)),
-                getattr(self.task_configuration, "create_source_records", True),
+                getattr(self, "create_source_records", True),
             ]
         ):
             return (legacy_id, folio_record["id"], folio_record["hrid"])

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/marc_rules_transformation/marc_file_processor.py

@@ -3,7 +3,7 @@ import os
 import sys
 import time
 import traceback
-from typing import List
+from typing import BinaryIO, Dict, List, Set, TextIO
 
 import i18n
 from folio_uuid.folio_namespaces import FOLIONamespaces
@@ -24,24 +24,26 @@ from folio_migration_tools.migration_report import MigrationReport
 
 class MarcFileProcessor:
     def __init__(
-        self, mapper: RulesMapperBase, folder_structure: FolderStructure, created_objects_file
+        self, mapper: RulesMapperBase, folder_structure: FolderStructure, created_objects_file: TextIO
     ):
         self.object_type: FOLIONamespaces = folder_structure.object_type
         self.folder_structure: FolderStructure = folder_structure
         self.mapper: RulesMapperBase = mapper
-        self.created_objects_file = created_objects_file
-        if mapper.task_configuration.create_source_records:
-            self.srs_records_file = open(self.folder_structure.srs_records_path, "w+")
+        self.created_objects_file: TextIO = created_objects_file
+        if mapper.create_source_records and any(
+            x.create_source_records for x in mapper.task_configuration.files
+        ):
+            self.srs_records_file: TextIO = open(self.folder_structure.srs_records_path, "w+")
         if getattr(mapper.task_configuration, "data_import_marc", False):
-            self.data_import_marc_file = open(self.folder_structure.data_import_marc_path, "wb+")
-        self.unique_001s: set = set()
+            self.data_import_marc_file: BinaryIO = open(self.folder_structure.data_import_marc_path, "wb+")
+        self.unique_001s: Set[str] = set()
         self.failed_records_count: int = 0
         self.records_count: int = 0
         self.start: float = time.time()
-        self.legacy_ids: set = set()
+        self.legacy_ids: Set[str] = set()
         if (
             self.object_type == FOLIONamespaces.holdings
-            and self.mapper.task_configuration.create_source_records
+            and self.mapper.create_source_records
         ):
             logging.info("Loading Parent HRID map for SRS creation")
             self.parent_hrids = {entity[1]: entity[2] for entity in mapper.parent_id_map.values()}
@@ -78,7 +80,7 @@ class MarcFileProcessor:
 
                     if (
                         file_def.create_source_records
-                        and self.mapper.task_configuration.create_source_records
+                        and self.mapper.create_source_records
                     ):
                         self.save_srs_record(
                             marc_record,
@@ -132,7 +134,7 @@ class MarcFileProcessor:
     def save_marc_record(
         self,
         marc_record: Record,
-        folio_rec: dict,
+        folio_rec: Dict,
         object_type: FOLIONamespaces
     ):
         self.mapper.save_data_import_marc_record(
@@ -146,14 +148,10 @@ class MarcFileProcessor:
         self,
         marc_record: Record,
         file_def: FileDefinition,
-        folio_rec,
+        folio_rec: Dict,
         legacy_ids: List[str],
         object_type: FOLIONamespaces,
     ):
-        if not all(
-            [file_def.create_source_records, self.mapper.task_configuration.create_source_records]
-        ):
-            return
         if object_type in [FOLIONamespaces.holdings]:
             if "008" in marc_record and len(marc_record["008"].data) > 32:
                 remain, rest = (
@@ -188,7 +186,7 @@ class MarcFileProcessor:
         )
         self.mapper.migration_report.add_general_statistics(i18n.t("SRS records written to disk"))
 
-    def add_mapped_location_code_to_record(self, marc_record, folio_rec):
+    def add_mapped_location_code_to_record(self, marc_record: Record, folio_rec: Dict):
         location_code = next(
             (
                 location["code"]
@@ -225,9 +223,9 @@ class MarcFileProcessor:
 
     @staticmethod
     def get_valid_folio_record_ids(
-        legacy_ids, folio_record_identifiers, migration_report: MigrationReport
-    ):
-        new_ids = set()
+        legacy_ids: List[str], folio_record_identifiers: Set[str], migration_report: MigrationReport
+    ) -> List[str]:
+        new_ids: Set[str] = set()
         for legacy_id in legacy_ids:
             if legacy_id not in folio_record_identifiers:
                 new_ids.add(legacy_id)
@@ -266,12 +264,12 @@ class MarcFileProcessor:
                 self.mapper.mapped_folio_fields,
                 self.mapper.mapped_legacy_fields,
             )
-        if self.mapper.task_configuration.create_source_records:
+        if hasattr(self, "srs_records_file"):
             self.srs_records_file.seek(0)
             if not self.srs_records_file.seek(0):
                 os.remove(self.srs_records_file.name)
             self.srs_records_file.close()
-        if getattr(self.mapper.task_configuration, "data_import_marc", False):
+        if hasattr(self, "data_import_marc_file"):
             self.data_import_marc_file.seek(0)
             if not self.data_import_marc_file.read(1):
                 os.remove(self.data_import_marc_file.name)
@@ -281,7 +279,7 @@ class MarcFileProcessor:
         logging.info("Transformation report written to %s", report_file.name)
         logging.info("Processor is done.")
 
-    def add_legacy_ids_to_map(self, folio_rec, filtered_legacy_ids):
+    def add_legacy_ids_to_map(self, folio_rec: Dict, filtered_legacy_ids: List[str]):
         for legacy_id in filtered_legacy_ids:
             self.legacy_ids.add(legacy_id)
             if legacy_id not in self.mapper.id_map:

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/marc_rules_transformation/rules_mapper_base.py

@@ -51,6 +51,9 @@ class RulesMapperBase(MapperBase):
         self.item_json_schema = ""
         self.mappings: dict = {}
         self.schema_properties = None
+        self.create_source_records = all(
+            [self.task_configuration.create_source_records, (not getattr(self.task_configuration, "data_import_marc", False))]
+        )
         if hasattr(self.task_configuration, "hrid_handling"):
             self.hrid_handler = HRIDHandler(
                 folio_client,

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/marc_rules_transformation/rules_mapper_bibs.py

@@ -7,7 +7,7 @@ import time
 import typing
 import uuid
 from pathlib import Path
-from typing import Generator, List
+from typing import Dict, Generator, List
 
 import i18n
 import pymarc
@@ -16,6 +16,7 @@ from folio_uuid.folio_namespaces import FOLIONamespaces
 from folio_uuid.folio_uuid import FolioUUID
 from folioclient import FolioClient
 from pymarc.record import Leader, Record
+from pymarc.field import Field
 
 from folio_migration_tools.custom_exceptions import (
     TransformationProcessError,
@@ -32,6 +33,7 @@ from folio_migration_tools.marc_rules_transformation.conditions import Condition
 from folio_migration_tools.marc_rules_transformation.rules_mapper_base import (
     RulesMapperBase,
 )
+from folio_migration_tools.migration_tasks.migration_task_base import MarcTaskConfigurationBase
 
 
 class BibsRulesMapper(RulesMapperBase):
@@ -40,9 +42,9 @@ class BibsRulesMapper(RulesMapperBase):
 
     def __init__(
         self,
-        folio_client,
+        folio_client: FolioClient,
         library_configuration: LibraryConfiguration,
-        task_configuration,
+        task_configuration: MarcTaskConfigurationBase,
     ):
         super().__init__(
             folio_client,
@@ -59,15 +61,12 @@ class BibsRulesMapper(RulesMapperBase):
         self.instance_relationships: dict = {}
         self.instance_relationship_types: dict = {}
         self.other_mode_of_issuance_id = get_unspecified_mode_of_issuance(self.folio_client)
-        self.create_source_records = all(
-            [self.task_configuration.create_source_records, (not getattr(self.task_configuration, "data_import_marc", False))]
-        )
         self.data_import_marc = self.task_configuration.data_import_marc
         if self.data_import_marc:
             self.hrid_handler.deactivate035_from001 = True
         self.start = time.time()
 
-    def perform_initial_preparation(self, marc_record: pymarc.Record, legacy_ids):
+    def perform_initial_preparation(self, file_def: FileDefinition, marc_record: Record, legacy_ids: List[str]):
         folio_instance = {}
         folio_instance["id"] = str(
             FolioUUID(
@@ -76,7 +75,10 @@ class BibsRulesMapper(RulesMapperBase):
                 str(legacy_ids[-1]),
             )
         )
-        if self.create_source_records or self.hrid_handler.handling == HridHandling.preserve001:
+        if (
+            all([self.create_source_records, file_def.create_source_records])
+            or self.hrid_handler.handling == HridHandling.preserve001
+        ):
             self.hrid_handler.handle_hrid(
                 FOLIONamespaces.instances,
                 folio_instance,
@@ -90,7 +92,7 @@ class BibsRulesMapper(RulesMapperBase):
 
         return folio_instance
 
-    def handle_leader_05(self, marc_record, legacy_ids):
+    def handle_leader_05(self, marc_record: Record, legacy_ids: List[str]):
         leader_05 = marc_record.leader[5] or "Empty"
         self.migration_report.add("RecordStatus", i18n.t("Original value") + f": {leader_05}")
         if leader_05 not in ["a", "c", "d", "n", "p"]:
@@ -102,14 +104,14 @@ class BibsRulesMapper(RulesMapperBase):
             Helper.log_data_issue(legacy_ids, "d in leader. Is this correct?", marc_record.leader)
 
     def parse_record(
-        self, marc_record: pymarc.Record, file_def: FileDefinition, legacy_ids: List[str]
+        self, marc_record: Record, file_def: FileDefinition, legacy_ids: List[str]
     ) -> list[dict]:
         """Parses a bib recod into a FOLIO Inventory instance object
         Community mapping suggestion: https://bit.ly/2S7Gyp3
          This is the main function
 
         Args:
-            marc_record (pymarc.Record): _description_
+            marc_record (Record): _description_
             file_def (FileDefinition): _description_
             legacy_ids (List[str]): List of legacy ids in record
 
@@ -119,7 +121,7 @@ class BibsRulesMapper(RulesMapperBase):
         self.print_progress()
         ignored_subsequent_fields: set = set()
         bad_tags = set(self.task_configuration.tags_to_delete)  # "907"
-        folio_instance = self.perform_initial_preparation(marc_record, legacy_ids)
+        folio_instance = self.perform_initial_preparation(file_def, marc_record, legacy_ids)
         if self.data_import_marc:
             self.simple_bib_map(folio_instance, marc_record, ignored_subsequent_fields, legacy_ids)
         else:
@@ -178,7 +180,7 @@ class BibsRulesMapper(RulesMapperBase):
             legacy_ids (List[str]): _description_
             file_def (FileDefinition): _description_
         """
-        if file_def.create_source_records and self.task_configuration.create_source_records:
+        if file_def.create_source_records and self.create_source_records:
             folio_instance["source"] = "MARC"
         else:
             folio_instance["source"] = "FOLIO"
@@ -198,7 +200,7 @@ class BibsRulesMapper(RulesMapperBase):
             del folio_instance["succeedingTitles"]
             self.migration_report.add("PrecedingSuccedingTitles", f"{len(succ_titles)}")
 
-    def handle_languages(self, folio_instance, marc_record, legacy_ids):
+    def handle_languages(self, folio_instance: Dict, marc_record: Record, legacy_ids: List[str]):
         if "languages" in folio_instance:
             orig_languages = {lang: None for lang in folio_instance["languages"]}
             orig_languages.update(
@@ -255,7 +257,7 @@ class BibsRulesMapper(RulesMapperBase):
             if self.task_configuration.update_hrid_settings:
                 self.hrid_handler.store_hrid_settings()
 
-    def get_instance_type_id(self, marc_record, legacy_id):
+    def get_instance_type_id(self, marc_record: Record, legacy_ids: List[str]) -> str:
         return_id = ""
 
         def get_folio_id_by_name(f336a: str):
@@ -283,7 +285,7 @@ class BibsRulesMapper(RulesMapperBase):
                     + f" ({f336a})",
                 )
                 Helper.log_data_issue(
-                    legacy_id,
+                    legacy_ids,
                     "instance type name (336$a) -Unsuccessful matching",
                     f336a,
                 )
@@ -316,7 +318,7 @@ class BibsRulesMapper(RulesMapperBase):
                     ),
                 )
                 Helper.log_data_issue(
-                    legacy_id,
+                    legacy_ids,
                     i18n.t("instance type code (%{code}) not found in FOLIO", code="336$b"),
                     f336_b,
                 )
@@ -337,7 +339,7 @@ class BibsRulesMapper(RulesMapperBase):
             return_id = t[0]
         return return_id
 
-    def get_instance_format_id_by_code(self, legacy_id: str, code: str):
+    def get_instance_format_id_by_code(self, legacy_ids: List[str], code: str):
         try:
             match = next(f for f in self.folio_client.instance_formats if f["code"] == code)
             self.migration_report.add(
@@ -347,14 +349,14 @@ class BibsRulesMapper(RulesMapperBase):
             return match["id"]
         except Exception:
             # TODO: Distinguish between generated codes and proper 338bs
-            Helper.log_data_issue(legacy_id, "Instance format Code not found in FOLIO", code)
+            Helper.log_data_issue(legacy_ids, "Instance format Code not found in FOLIO", code)
             self.migration_report.add(
                 "InstanceFormat",
                 i18n.t("Code '%{code}' not found in FOLIO", code=code),
             )
             return ""
 
-    def get_instance_format_id_by_name(self, f337a: str, f338a: str, legacy_id: str):
+    def get_instance_format_id_by_name(self, f337a: str, f338a: str, legacy_ids: List[str]):
         f337a = f337a.lower().strip()
         f338a = f338a.lower().strip()
         match_template = f"{f337a} -- {f338a}"
@@ -376,7 +378,7 @@ class BibsRulesMapper(RulesMapperBase):
             return match["id"]
         except Exception:
             Helper.log_data_issue(
-                legacy_id,
+                legacy_ids,
                 "Unsuccessful matching on 337$a and 338$a",
                 match_template,
             )
@@ -391,7 +393,7 @@ class BibsRulesMapper(RulesMapperBase):
             )
             return ""
 
-    def f338_source_is_rda_carrier(self, field: pymarc.Field):
+    def f338_source_is_rda_carrier(self, field: Field):
         if "2" not in field:
             self.migration_report.add(
                 "InstanceFormat",
@@ -407,7 +409,7 @@ class BibsRulesMapper(RulesMapperBase):
         return False
 
     def get_instance_format_ids_from_a(
-        self, field_index, f_338: pymarc.Field, all_337s, legacy_id
+        self, field_index: int, f_338: Field, all_337s: List[Field], legacy_id: List[str]
     ):
         self.migration_report.add(
             "InstanceFormat",
@@ -421,7 +423,7 @@ class BibsRulesMapper(RulesMapperBase):
                 ):
                     yield fmt_id
 
-    def get_instance_format_ids(self, marc_record, legacy_id):
+    def get_instance_format_ids(self, marc_record: Record, legacy_id: List[str]):
         all_337s = marc_record.get_fields("337")
         all_338s = marc_record.get_fields("338")
         for fidx, f_338 in enumerate(all_338s):
@@ -466,7 +468,7 @@ class BibsRulesMapper(RulesMapperBase):
                                         legacy_id, combined_code
                                     )
 
-    def get_mode_of_issuance_id(self, marc_record: Record, legacy_id: List[str]) -> str:
+    def get_mode_of_issuance_id(self, marc_record: Record, legacy_ids: List[str]) -> str:
         level = marc_record.leader[7]
         try:
             name = "unspecified"
@@ -496,7 +498,7 @@ class BibsRulesMapper(RulesMapperBase):
             return ret
         except IndexError:
             self.migration_report.add(
-                "PossibleCleaningTasks", i18n.t("No Leader[7] in") + f" {legacy_id}"
+                "PossibleCleaningTasks", i18n.t("No Leader[7] in") + f" {legacy_ids}"
             )
 
             return self.other_mode_of_issuance_id
@@ -512,7 +514,7 @@ class BibsRulesMapper(RulesMapperBase):
             return "".join(marc_record["008"].data[35:38])
         return ""
 
-    def get_languages_041(self, marc_record, legacy_id):
+    def get_languages_041(self, marc_record: Record, legacy_id: List[str]) -> Dict[str, None]:
         languages = dict()
         lang_fields = marc_record.get_fields("041")
         if not any(lang_fields):
@@ -543,21 +545,21 @@ class BibsRulesMapper(RulesMapperBase):
         }
         return languages
 
-    def get_languages(self, marc_record: Record, legacy_id: str) -> List[str]:
+    def get_languages(self, marc_record: Record, legacy_id: List[str]) -> List[str]:
         """Get languages and tranforms them to correct codes
 
         Args:
-            marc_record (Record): _description_
-            legacy_id (str): _description_
+            marc_record (Record): A pymarc Record object
+            legacy_id (List[str]): A list of legacy ids from the legacy record
 
         Returns:
-            List[str]: _description_
+            List[str]: List of language codes
         """
         languages = self.get_languages_041(marc_record, legacy_id)
         languages[self.get_languages_008(marc_record)] = None
-        for lang in languages.keys():
+        for lang in languages:
             self.migration_report.add("LanguagesInRecords", lang)
-        return list(languages.keys())
+        return list(languages)
 
     def fetch_language_codes(self) -> Generator[str, None, None]:
         """Loads the  list of standardized language codes from LoC
@@ -575,7 +577,7 @@ class BibsRulesMapper(RulesMapperBase):
             yield code.text
 
     def filter_langs(
-        self, language_values: List[str], marc_record: Record, index_or_legacy_id
+        self, language_values: List[str], marc_record: Record, index_or_legacy_id: List[str]
     ) -> typing.Generator:
         forbidden_values = ["###", "zxx", "n/a", "N/A", "|||"]
         for language_value in language_values:
@@ -629,7 +631,7 @@ class BibsRulesMapper(RulesMapperBase):
         else:
             raise TransformationProcessError("", f"ILS {ils_flavour} not configured")
 
-    def get_aleph_bib_id(self, marc_record: Record):
+    def get_aleph_bib_id(self, marc_record: Record) -> List[str]:
         res = {f["b"].strip(): None for f in marc_record.get_fields("998") if "b" in f}
         if any(res):
             self.migration_report.add_general_statistics(
@@ -651,7 +653,7 @@ class BibsRulesMapper(RulesMapperBase):
                 ) from e
 
 
-def get_unspecified_mode_of_issuance(folio_client: FolioClient):
+def get_unspecified_mode_of_issuance(folio_client: FolioClient) -> str:
     m_o_is = list(folio_client.modes_of_issuance)
     if not any(m_o_is):
         logging.critical("No Modes of issuance set up in tenant. Quitting...")
@@ -684,7 +686,7 @@ def get_custom_bib_id(marc_record: Record, field_string: str):
         )
 
 
-def get_iii_bib_id(marc_record: Record):
+def get_iii_bib_id(marc_record: Record) -> List[str]:
     try:
         return [marc_record["907"]["a"]]
     except Exception as e:

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/marc_rules_transformation/rules_mapper_holdings.py

@@ -1,7 +1,7 @@
 import copy
 import json
 import logging
-from typing import List
+from typing import Dict, List, Set
 
 import i18n
 from folio_uuid.folio_namespaces import FOLIONamespaces
@@ -235,10 +235,10 @@ class RulesMapperHoldings(RulesMapperBase):
 
     def process_marc_field(
         self,
-        folio_holding: dict,
+        folio_holding: Dict,
         marc_field: Field,
-        ignored_subsequent_fields,
-        index_or_legacy_ids,
+        ignored_subsequent_fields: Set,
+        index_or_legacy_ids: List[str],
     ):
         """This overwrites the implementation for Auth and instances
 
@@ -261,7 +261,7 @@ class RulesMapperHoldings(RulesMapperBase):
                 ignored_subsequent_fields.add(marc_field.tag)
 
     def perform_additional_mapping(
-        self, marc_record: Record, folio_holding, legacy_ids: List[str], file_def: FileDefinition
+        self, marc_record: Record, folio_holding: Dict, legacy_ids: List[str], file_def: FileDefinition
     ):
         """_summary_
 
@@ -279,11 +279,8 @@ class RulesMapperHoldings(RulesMapperBase):
         self.pick_first_location_if_many(folio_holding, legacy_ids)
         self.parse_coded_holdings_statements(marc_record, folio_holding, legacy_ids)
         HoldingsHelper.handle_notes(folio_holding)
-        create_source_records = all(
-            [file_def.create_source_records, self.task_configuration.create_source_records]
-        )
         if (
-            create_source_records
+            all([file_def.create_source_records, self.create_source_records])
             or self.task_configuration.hrid_handling == HridHandling.preserve001
         ):
             self.hrid_handler.handle_hrid(
@@ -298,9 +295,9 @@ class RulesMapperHoldings(RulesMapperBase):
                 "",
             )
         self.handle_suppression(folio_holding, file_def, True)
-        self.set_source_id(self.task_configuration, folio_holding, self.holdingssources, file_def)
+        self.set_source_id(self.create_source_records, folio_holding, self.holdingssources, file_def)
 
-    def pick_first_location_if_many(self, folio_holding, legacy_ids: List[str]):
+    def pick_first_location_if_many(self, folio_holding: Dict, legacy_ids: List[str]):
         if " " in folio_holding.get("permanentLocationId", ""):
             Helper.log_data_issue(
                 legacy_ids,
@@ -312,14 +309,14 @@ class RulesMapperHoldings(RulesMapperBase):
             ]
 
     @staticmethod
-    def set_source_id(task_configuration, folio_rec, holdingssources, file_def):
-        if file_def.create_source_records and task_configuration.create_source_records:
+    def set_source_id(create_source_records: bool, folio_rec: Dict, holdingssources: Dict, file_def: FileDefinition):
+        if file_def.create_source_records and create_source_records:
             folio_rec["sourceId"] = holdingssources.get("MARC")
         else:
             folio_rec["sourceId"] = holdingssources.get("FOLIO")
 
     def parse_coded_holdings_statements(
-        self, marc_record: Record, folio_holding, legacy_ids: List[str]
+        self, marc_record: Record, folio_holding: Dict, legacy_ids: List[str]
     ):
         # TODO: Should one be able to switch these things off?
         a = {
@@ -351,7 +348,7 @@ class RulesMapperHoldings(RulesMapperBase):
             x.create_source_records for x in self.task_configuration.files
         ]
         if all(source_file_create_source_records):
-            create_source_records = self.task_configuration.create_source_records
+            create_source_records = self.create_source_records
         else:
             logging.info(
                 "If all source files have create_source_records set to false, "
@@ -369,7 +366,7 @@ class RulesMapperHoldings(RulesMapperBase):
         logging.info("Fetching HoldingsRecord schema...")
         return folio_client.get_holdings_schema()
 
-    def set_holdings_type(self, marc_record: Record, folio_holding, legacy_ids: List[str]):
+    def set_holdings_type(self, marc_record: Record, folio_holding: Dict, legacy_ids: List[str]):
         # Holdings type mapping
         ldr06 = marc_record.leader[6]
         # TODO: map this better
@@ -425,7 +422,7 @@ class RulesMapperHoldings(RulesMapperBase):
                     ldr06,
                 )
 
-    def set_default_call_number_type_if_empty(self, folio_holding):
+    def set_default_call_number_type_if_empty(self, folio_holding: Dict):
         if not folio_holding.get("callNumberTypeId", ""):
             folio_holding["callNumberTypeId"] = self.conditions.default_call_number_type["id"]
 
@@ -462,7 +459,7 @@ class RulesMapperHoldings(RulesMapperBase):
             )
         return results
 
-    def verity_boundwith_map_entry(self, entry):
+    def verity_boundwith_map_entry(self, entry: Dict):
         if "MFHD_ID" not in entry or not entry.get("MFHD_ID", ""):
             raise TransformationProcessError(
                 "", "Column MFHD_ID missing from Boundwith relationship map", ""
@@ -472,7 +469,7 @@ class RulesMapperHoldings(RulesMapperBase):
                 "", "Column BIB_ID missing from Boundwith relationship map", ""
             )
 
-    def setup_boundwith_relationship_map(self, boundwith_relationship_map):
+    def setup_boundwith_relationship_map(self, boundwith_relationship_map: List[Dict]):
         """
         Creates a map of MFHD_ID to BIB_ID for boundwith relationships.
 
@@ -503,7 +500,7 @@ class RulesMapperHoldings(RulesMapperBase):
                 self.handle_transformation_record_failed_error(idx, trfe)
         return new_map
 
-    def get_bw_instance_id_map_tuple(self, entry):
+    def get_bw_instance_id_map_tuple(self, entry: Dict):
         try:
             return self.parent_id_map[entry["BIB_ID"]]
         except KeyError:

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/migration_tasks/bibs_transformer.py

@@ -18,35 +18,11 @@ from folio_migration_tools.marc_rules_transformation.marc_file_processor import
 from folio_migration_tools.marc_rules_transformation.rules_mapper_bibs import (
     BibsRulesMapper,
 )
-from folio_migration_tools.migration_tasks.migration_task_base import MigrationTaskBase
-from folio_migration_tools.task_configuration import AbstractTaskConfiguration
+from folio_migration_tools.migration_tasks.migration_task_base import MarcTaskConfigurationBase, MigrationTaskBase
 
 
 class BibsTransformer(MigrationTaskBase):
-    class TaskConfiguration(AbstractTaskConfiguration):
-        name: Annotated[
-            str,
-            Field(
-                description=(
-                    "Name of this migration task. The name is being used to call the specific "
-                    "task, and to distinguish tasks of similar types"
-                )
-            ),
-        ]
-        migration_task_type: Annotated[
-            str,
-            Field(
-                title="Migration task type",
-                description=("The type of migration task you want to perform."),
-            ),
-        ]
-        files: Annotated[
-            List[FileDefinition],
-            Field(
-                title="Source files",
-                description=("List of MARC21 files with bibliographic records."),
-            ),
-        ]
+    class TaskConfiguration(MarcTaskConfigurationBase):
         ils_flavour: Annotated[
             IlsFlavour,
             Field(
@@ -87,16 +63,6 @@ class BibsTransformer(MigrationTaskBase):
                 ),
             ),
         ] = []
-        create_source_records: Annotated[
-            bool,
-            Field(
-                title="Create source records",
-                description=(
-                    "Controls wheter or not to retain the MARC records in "
-                    "Source Record Storage."
-                ),
-            ),
-        ] = True
         data_import_marc: Annotated[
             bool,
             Field(
@@ -107,7 +73,7 @@ class BibsTransformer(MigrationTaskBase):
                     "of FOLIO instance records (and optional SRS records) will be generated."
                 ),
             )
-        ] = False
+        ] = True
         parse_cataloged_date: Annotated[
             bool,
             Field(
@@ -118,17 +84,6 @@ class BibsTransformer(MigrationTaskBase):
                 ),
             ),
         ] = False
-        hrid_handling: Annotated[
-            HridHandling,
-            Field(
-                title="HRID Handling",
-                description=(
-                    "Setting to default will make FOLIO generate HRIDs and move the existing "
-                    "001:s into a 035, concatenated with the 003. Choosing preserve001 means "
-                    "the 001:s will remain in place, and that they will also become the HRIDs"
-                ),
-            ),
-        ] = HridHandling.default
         reset_hrid_settings: Annotated[
             bool,
             Field(
@@ -146,16 +101,6 @@ class BibsTransformer(MigrationTaskBase):
                 description="At the end of the run, update FOLIO with the HRID settings",
             ),
         ] = True
-        deactivate035_from001: Annotated[
-            bool,
-            Field(
-                title="Create 035 from 001 and 003",
-                description=(
-                    "This deactivates the FOLIO default functionality of moving the previous 001 "
-                    "into a 035, prefixed with the value from 003"
-                ),
-            ),
-        ] = False
 
     @staticmethod
     def get_object_type() -> FOLIONamespaces:

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/migration_tasks/holdings_marc_transformer.py

@@ -19,12 +19,11 @@ from folio_migration_tools.library_configuration import (
 from folio_migration_tools.marc_rules_transformation.rules_mapper_holdings import (
     RulesMapperHoldings,
 )
-from folio_migration_tools.migration_tasks.migration_task_base import MigrationTaskBase
-from folio_migration_tools.task_configuration import AbstractTaskConfiguration
+from folio_migration_tools.migration_tasks.migration_task_base import MarcTaskConfigurationBase, MigrationTaskBase
 
 
 class HoldingsMarcTransformer(MigrationTaskBase):
-    class TaskConfiguration(AbstractTaskConfiguration):
+    class TaskConfiguration(MarcTaskConfigurationBase):
         name: Annotated[
             str,
             Field(
@@ -59,16 +58,6 @@ class HoldingsMarcTransformer(MigrationTaskBase):
                 ),
             ),
         ] = HridHandling.default
-        deactivate035_from001: Annotated[
-            bool,
-            Field(
-                title="Create 035 from 001 and 003",
-                description=(
-                    "This deactivates the FOLIO default functionality of moving the previous 001 "
-                    "into a 035, prefixed with the value from 003"
-                ),
-            ),
-        ] = False
         holdings_type_uuid_for_boundwiths: Annotated[
             str,
             Field(
@@ -89,16 +78,6 @@ class HoldingsMarcTransformer(MigrationTaskBase):
                 ),
             ),
         ] = ""
-        create_source_records: Annotated[
-            bool,
-            Field(
-                title="Create source records",
-                description=(
-                    "Controls wheter or not to retain the MARC records in "
-                    "Source Record Storage."
-                ),
-            ),
-        ] = True
         update_hrid_settings: Annotated[
             bool,
             Field(

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/migration_tasks/migration_task_base.py

@@ -9,11 +9,12 @@ from abc import abstractmethod
 from datetime import datetime, timezone
 from genericpath import isfile
 from pathlib import Path
-from typing import Optional
+from typing import Annotated, List, Optional
 
 import folioclient
 from folio_uuid.folio_namespaces import FOLIONamespaces
 from folioclient import FolioClient
+from pydantic import Field
 
 from folio_migration_tools import library_configuration, task_configuration
 from folio_migration_tools.custom_exceptions import (
@@ -163,6 +164,7 @@ class MigrationTaskBase:
         the central tenant instance ids. This is a bit of a hack, but it works for now.
         """
         map_files = []
+        instance_id_map = {}
         if self.library_configuration.is_ecs and self.central_folder_structure:
             logging.info(
                 "Loading ECS central tenant instance id map from %s", self.central_folder_structure.instance_id_map_path
@@ -459,6 +461,68 @@ class MigrationTaskBase:
             return None
 
 
+class MarcTaskConfigurationBase(task_configuration.AbstractTaskConfiguration):
+    """
+    Base class for MARC task configurations.
+
+    Attributes:
+        files (List[library_configuration.FileDefinition]):
+            List of MARC21 files to be processed.
+
+        create_source_records (bool):
+            Controls whether or not to retain the MARC records in Source Record Storage.
+            Default is False, meaning MARC records will not be retained.
+
+        hrid_handling (library_configuration.HridHandling):
+            Determines how HRIDs are handled.
+            - 'default': FOLIO generates HRIDs and moves existing 001 fields into a 035 field, concatenated with the 003 field.
+            - 'preserve001': Keeps the 001 fields in place and uses them as HRIDs.
+            Default is 'default'.
+
+        deactivate035_from001 (bool):
+            Disables the default FOLIO functionality of moving the previous 001 field into a 035 field, prefixed with the value from 003.
+            Default is False, meaning the functionality remains active.
+    """
+
+    files: Annotated[
+        List[library_configuration.FileDefinition],
+        Field(
+            title="Source files",
+            description=("List of MARC21 files with bibliographic records."),
+        ),
+    ]
+    create_source_records: Annotated[
+        bool,
+        Field(
+            title="Create source records",
+            description=(
+                "Controls whether or not to retain the MARC records in "
+                "Source Record Storage."
+            ),
+        ),
+    ] = False
+    hrid_handling: Annotated[
+        library_configuration.HridHandling,
+        Field(
+            title="HRID Handling",
+            description=(
+                "Setting to default will make FOLIO generate HRIDs and move the existing "
+                "001:s into a 035, concatenated with the 003. Choosing preserve001 means "
+                "the 001:s will remain in place, and that they will also become the HRIDs"
+            ),
+        ),
+    ] = library_configuration.HridHandling.default
+    deactivate035_from001: Annotated[
+        bool,
+        Field(
+            title="Create 035 from 001 and 003",
+            description=(
+                "This deactivates the FOLIO default functionality of moving the previous 001 "
+                "into a 035, prefixed with the value from 003"
+            ),
+        ),
+    ] = False
+
 class ExcludeLevelFilter(logging.Filter):
     def __init__(self, level):
         super().__init__()

{folio_migration_tools-1.9.0rc8 → folio_migration_tools-1.9.0rc10}/src/folio_migration_tools/task_configuration.py

@@ -10,7 +10,24 @@ def to_camel(string):
 
 
 class AbstractTaskConfiguration(BaseModel):
-    name: str
+    """Abstract class for task configuration."""
+
+    name: Annotated[
+        str,
+        Field(
+            description=(
+                "Name of this migration task. The name is being used to call the specific "
+                "task, and to distinguish tasks of similar types"
+            )
+        ),
+    ]
+    migration_task_type: Annotated[
+        str,
+        Field(
+            title="Migration task type",
+            description=("The type of migration task you want to perform."),
+        ),
+    ]
     ecs_tenant_id: Annotated[
         str,
         Field(