folio-migration-tools 1.9.0a2__py3-none-any.whl → 1.9.0a4__py3-none-any.whl

folio_migration_tools/__init__.py

@@ -0,0 +1,3 @@
+import importlib.metadata
+
+__version__ = importlib.metadata.version("folio_migration_tools")

folio_migration_tools/__main__.py

@@ -8,6 +8,7 @@ import httpx
 import humps
 import i18n
 from argparse_prompt import PromptParser
+from folioclient import FolioClient
 from pydantic import ValidationError
 
 from folio_migration_tools.config_file_load import merge_load
@@ -38,14 +39,16 @@ def parse_args(args):
     )
     parser.add_argument(
         "--okapi_password",
-        help="pasword for the tenant in the configuration file",
+        help="password for the tenant in the configuration file",
         prompt="FOLIO_MIGRATION_TOOLS_OKAPI_PASSWORD" not in environ,
         default=environ.get("FOLIO_MIGRATION_TOOLS_OKAPI_PASSWORD"),
         secure=True,
     )
     parser.add_argument(
         "--base_folder_path",
-        help=("path to the base folder for this library. Built on migration_repo_template"),
+        help=(
+            "path to the base folder for this library. Built on migration_repo_template"
+        ),
         prompt="FOLIO_MIGRATION_TOOLS_BASE_FOLDER_PATH" not in environ,
         default=environ.get("FOLIO_MIGRATION_TOOLS_BASE_FOLDER_PATH"),
     )
@@ -107,10 +110,17 @@ def main():
             )
             sys.exit("Task Type Not Found")
         try:
-            task_config = task_class.TaskConfiguration(**migration_task_config)
-            task_obj = task_class(task_config, library_config)
-            task_obj.do_work()
-            task_obj.wrap_up()
+            logging.getLogger("httpx").setLevel(logging.WARNING) # Exclude info messages from httpx
+            with FolioClient(
+                library_config.okapi_url,
+                library_config.tenant_id,
+                library_config.okapi_username,
+                library_config.okapi_password,
+            ) as folio_client:
+                task_config = task_class.TaskConfiguration(**migration_task_config)
+                task_obj = task_class(task_config, library_config, folio_client)
+                task_obj.do_work()
+                task_obj.wrap_up()
         except TransformationProcessError as tpe:
             logging.critical(tpe.message)
             print(f"\n{tpe.message}: {tpe.data_value}")

folio_migration_tools/folder_structure.py

@@ -101,6 +101,9 @@ class FolderStructure:
         self.srs_records_path = (
             self.results_folder / f"folio_srs_{object_type_string}{self.file_template}.json"
         )
+        self.data_import_marc_path = (
+            self.results_folder / f"folio_marc_{object_type_string}{self.file_template}.mrc"
+        )
         self.organizations_id_map_path = (
             self.results_folder / f"{str(FOLIONamespaces.organizations.name).lower()}_id_map.json"
         )

folio_migration_tools/library_configuration.py

@@ -1,9 +1,7 @@
 from enum import Enum
-from typing import Annotated
-from typing import Optional
+from typing import Annotated, Optional
 
-from pydantic import BaseModel
-from pydantic import Field
+from pydantic import BaseModel, Field
 from pydantic.types import DirectoryPath
 
 
@@ -102,6 +100,9 @@ class LibraryConfiguration(BaseModel):
     failed_percentage_threshold: Annotated[
         int, Field(description=("Percentage of failed records until the process shuts down"))
     ] = 20
+    generic_exception_threshold: Annotated[
+        int, Field(description=("Number of generic exceptions until the process shuts down"))
+    ] = 50
     library_name: str
     log_level_debug: bool
     folio_release: FolioRelease = Field(
@@ -112,6 +113,6 @@ class LibraryConfiguration(BaseModel):
         )
     )
     iteration_identifier: str
-    add_time_stamp_to_file_names: Annotated[bool, Field(title="Add time stamp to file names")] = (
-        False
-    )
+    add_time_stamp_to_file_names: Annotated[
+        bool, Field(title="Add time stamp to file names")
+    ] = False

folio_migration_tools/mapper_base.py

@@ -4,18 +4,20 @@ import json
 import logging
 import sys
 import uuid
-import i18n
-from datetime import datetime
-from datetime import timezone
+from datetime import datetime, timezone
 from pathlib import Path
+from typing import List
 
+import i18n
 from folio_uuid.folio_namespaces import FOLIONamespaces
 from folio_uuid.folio_uuid import FolioUUID
 from folioclient import FolioClient
 
-from folio_migration_tools.custom_exceptions import TransformationFieldMappingError
-from folio_migration_tools.custom_exceptions import TransformationProcessError
-from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
+from folio_migration_tools.custom_exceptions import (
+    TransformationFieldMappingError,
+    TransformationProcessError,
+    TransformationRecordFailedError,
+)
 from folio_migration_tools.extradata_writer import ExtradataWriter
 from folio_migration_tools.library_configuration import LibraryConfiguration
 from folio_migration_tools.mapping_file_transformation.ref_data_mapping import (
@@ -44,7 +46,7 @@ class MapperBase:
         self.mapped_folio_fields: dict = {}
         self.migration_report: MigrationReport = MigrationReport()
         self.num_criticalerrors = 0
-        self.num_exeptions = 0
+        self.num_exceptions = 0
         self.mapped_legacy_fields: dict = {}
         self.schema_properties = None
 
@@ -269,18 +271,18 @@ class MapperBase:
             return (legacy_id, folio_record["id"], folio_record["hrid"])
         return (legacy_id, folio_record["id"])
 
-    def handle_generic_exception(self, idx, excepion: Exception):
-        self.num_exeptions += 1
+    def handle_generic_exception(self, idx, exception: Exception):
+        self.num_exceptions += 1
         print("\n=======ERROR===========")
         print(
-            f"Row {idx:,} failed with the following unhandled Exception: {excepion}  "
-            f"of type {type(excepion).__name__}"
+            f"Row {idx:,} failed with the following unhandled Exception: {exception}  "
+            f"of type {type(exception).__name__}"
         )
-        logging.error(excepion, exc_info=True)
-        if self.num_exeptions > 50:
+        logging.error(exception, exc_info=True)
+        if self.num_exceptions > self.library_configuration.generic_exception_threshold:
             logging.fatal(
                 "Stopping. More than %s unhandled exceptions. Code needs fixing",
-                self.num_exeptions,
+                self.num_exceptions,
             )
             sys.exit(1)
 
@@ -345,7 +347,8 @@ class MapperBase:
                 "One or many required properties empty",
                 f"{json.dumps(missing)}",
             )
-        cleaned_folio_object.pop("type", None)
+        if object_type not in [FOLIONamespaces.users]:
+            cleaned_folio_object.pop("type", None)
         return cleaned_folio_object
 
     @staticmethod
@@ -428,8 +431,14 @@ class MapperBase:
 
             if call_number := folio_holding.get("callNumber", None):
                 if "[" in call_number:
-                    call_numbers = ast.literal_eval(str(folio_holding["callNumber"]))
-                    bound_with_holding["callNumber"] = call_numbers[bwidx]
+                    try:
+                        call_numbers: List = ast.literal_eval(str(folio_holding["callNumber"]))
+                        bound_with_holding["callNumber"] = call_numbers[bwidx]
+                    except IndexError:
+                        if call_numbers:
+                            bound_with_holding["callNumber"] = call_numbers[0]
+                    except SyntaxError:
+                        bound_with_holding["callNumber"] = call_number
                 else:
                     bound_with_holding["callNumber"] = call_number
             bound_with_holding["holdingsTypeId"] = bound_with_holdings_type_id

folio_migration_tools/mapping_file_transformation/holdings_mapper.py

@@ -2,12 +2,13 @@ import ast
 
 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
-from folio_migration_tools.library_configuration import FileDefinition
-from folio_migration_tools.library_configuration import LibraryConfiguration
+from folio_migration_tools.library_configuration import (
+    FileDefinition,
+    LibraryConfiguration,
+)
 from folio_migration_tools.mapping_file_transformation.mapping_file_mapper_base import (
     MappingFileMapperBase,
 )
@@ -97,6 +98,11 @@ class HoldingsMapper(MappingFileMapperBase):
                 "BoundWithMappings",
                 (f"Number of bib-level callnumbers in record: {len(legacy_value.split(','))}"),
             )
+        if legacy_value.startswith("[") and len(legacy_value.split(",")) == 1:
+            try:
+                legacy_value = ast.literal_eval(str(legacy_value))[0]
+            except SyntaxError:
+                return legacy_value
         return legacy_value
 
     def get_location_id(
@@ -127,9 +133,21 @@ class HoldingsMapper(MappingFileMapperBase):
             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 := FolioUUID.clean_iii_identifiers(legacy_instance_id)
-            ) and new_legacy_value in self.instance_id_map:
+                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:
                 self.migration_report.add_general_statistics(
                     i18n.t("Records matched to Instances")
                 )
@@ -137,12 +155,6 @@ class HoldingsMapper(MappingFileMapperBase):
                     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:

folio_migration_tools/mapping_file_transformation/item_mapper.py

@@ -1,21 +1,23 @@
 import json
 import logging
 import sys
-from datetime import datetime
-from datetime import timezone
+from datetime import datetime, timezone
 from typing import Set
 from uuid import uuid4
 
 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
-from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
+from folio_migration_tools.custom_exceptions import (
+    TransformationProcessError,
+    TransformationRecordFailedError,
+)
 from folio_migration_tools.helper import Helper
-from folio_migration_tools.library_configuration import FileDefinition
-from folio_migration_tools.library_configuration import LibraryConfiguration
+from folio_migration_tools.library_configuration import (
+    FileDefinition,
+    LibraryConfiguration,
+)
 from folio_migration_tools.mapping_file_transformation.mapping_file_mapper_base import (
     MappingFileMapperBase,
 )
@@ -224,16 +226,16 @@ class ItemMapper(MappingFileMapperBase):
             return self.transform_status(mapped_value)
         elif folio_prop_name == "barcode":
             barcode = mapped_value
-            if barcode.strip() and barcode in self.unique_barcodes:
+            normalized_barcode = barcode.strip().lower()
+            if normalized_barcode and normalized_barcode in self.unique_barcodes:
                 Helper.log_data_issue(index_or_id, "Duplicate barcode", mapped_value)
                 self.migration_report.add_general_statistics(i18n.t("Duplicate barcodes"))
                 return f"{barcode}-{uuid4()}"
             else:
-                if barcode.strip():
-                    self.unique_barcodes.add(barcode)
+                if normalized_barcode:
+                    self.unique_barcodes.add(normalized_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:

folio_migration_tools/mapping_file_transformation/mapping_file_mapper_base.py

@@ -6,19 +6,18 @@ import re
 import uuid
 from functools import reduce
 from pathlib import Path
-from typing import Dict
-from typing import List
-from typing import Set
+from typing import Dict, List, Set
 from uuid import UUID
 
 import i18n
-from folio_uuid.folio_uuid import FOLIONamespaces
-from folio_uuid.folio_uuid import FolioUUID
+from folio_uuid.folio_uuid import FOLIONamespaces, FolioUUID
 from folioclient import FolioClient
 
-from folio_migration_tools.custom_exceptions import TransformationFieldMappingError
-from folio_migration_tools.custom_exceptions import TransformationProcessError
-from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
+from folio_migration_tools.custom_exceptions import (
+    TransformationFieldMappingError,
+    TransformationProcessError,
+    TransformationRecordFailedError,
+)
 from folio_migration_tools.library_configuration import LibraryConfiguration
 from folio_migration_tools.mapper_base import MapperBase
 from folio_migration_tools.mapping_file_transformation.ref_data_mapping import (
@@ -787,7 +786,7 @@ class MappingFileMapperBase(MapperBase):
                     missing_keys_in_record,
                 )
             else:
-                logging.info("All maped legacy fields are in the legacy object")
+                logging.info("All mapped legacy fields are in the legacy object")
 
     def get_ref_data_tuple_by_code(self, ref_data, ref_name, code):
         return self.get_ref_data_tuple(ref_data, ref_name, code, "code")
@@ -872,7 +871,7 @@ class MappingFileMapperBase(MapperBase):
 
 def skip_property(property_name, property):
     return bool(
-        property_name in ["metadata", "id", "type", "lastCheckIn"]
+        property_name in ["metadata", "id", "lastCheckIn"]
         or property_name.startswith("effective")
         or property.get("folio:isVirtual", False)
         or property.get("description", "") == "Deprecated"

folio_migration_tools/mapping_file_transformation/order_mapper.py

@@ -5,9 +5,9 @@ import re
 import sys
 import urllib.parse
 import uuid
-import i18n
 
 import httpx
+import i18n
 from folio_uuid.folio_uuid import FOLIONamespaces
 from folioclient import FolioClient
 from httpx import HTTPError
@@ -141,7 +141,7 @@ class CompositeOrderMapper(MappingFileMapperBase):
             }
 
             if os.environ.get("GITHUB_TOKEN"):
-                logging.info("Using GITHB_TOKEN environment variable for Gihub API Access")
+                logging.info("Using GITHUB_TOKEN environment variable for GitHub API Access")
                 github_headers["authorization"] = f"token {os.environ.get('GITHUB_TOKEN')}"
 
             # Start talkign to GitHub...

folio_migration_tools/mapping_file_transformation/organization_mapper.py

@@ -31,7 +31,7 @@ class OrganizationMapper(MappingFileMapperBase):
     ):
         # Build composite organization schema
         if os.environ.get("GITHUB_TOKEN"):
-            logging.info("Using GITHB_TOKEN environment variable for Gihub API Access")
+            logging.info("Using GITHUB_TOKEN environment variable for GitHub API Access")
         organization_schema = OrganizationMapper.get_latest_acq_schemas_from_github(
             "folio-org", "mod-organizations-storage", "mod-orgs", "organization"
         )

folio_migration_tools/mapping_file_transformation/user_mapper.py

@@ -2,14 +2,16 @@ import csv
 import json
 import logging
 import sys
-import i18n
 
+import i18n
 from dateutil.parser import parse
 from folio_uuid.folio_namespaces import FOLIONamespaces
 from folioclient import FolioClient
 
-from folio_migration_tools.custom_exceptions import TransformationProcessError
-from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
+from folio_migration_tools.custom_exceptions import (
+    TransformationProcessError,
+    TransformationRecordFailedError,
+)
 from folio_migration_tools.mapping_file_transformation.mapping_file_mapper_base import (
     MappingFileMapperBase,
 )
@@ -79,7 +81,7 @@ class UserMapper(MappingFileMapperBase):
         )
         if "personal" not in folio_user:
             folio_user["personal"] = {}
-        folio_user["personal"]["preferredContactTypeId"] = "Email"
+        folio_user["personal"]["preferredContactTypeId"] = "email"
         folio_user["active"] = True
         if folio_user.get("requestPreference"):
             folio_user["requestPreference"].update(

folio_migration_tools/marc_rules_transformation/conditions.py

@@ -6,9 +6,11 @@ import pymarc
 from folioclient import FolioClient
 from pymarc import field
 
-from folio_migration_tools.custom_exceptions import TransformationFieldMappingError
-from folio_migration_tools.custom_exceptions import TransformationProcessError
-from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
+from folio_migration_tools.custom_exceptions import (
+    TransformationFieldMappingError,
+    TransformationProcessError,
+    TransformationRecordFailedError,
+)
 from folio_migration_tools.helper import Helper
 from folio_migration_tools.library_configuration import FolioRelease
 from folio_migration_tools.marc_rules_transformation.rules_mapper_base import (
@@ -58,6 +60,7 @@ class Conditions:
         logging.info("%s\tcontributor_types", len(self.folio.contributor_types))
         logging.info("%s\talt_title_types", len(self.folio.alt_title_types))
         logging.info("%s\tidentifier_types", len(self.folio.identifier_types))
+        logging.info("%s\tsubject_types", len(self.folio.subject_types))
         # Raise for empty settings
         if not self.folio.contributor_types:
             raise TransformationProcessError("", "No contributor_types in FOLIO")
@@ -67,6 +70,8 @@ class Conditions:
             raise TransformationProcessError("", "No identifier_types in FOLIO")
         if not self.folio.alt_title_types:
             raise TransformationProcessError("", "No alt_title_types in FOLIO")
+        if not self.folio.subject_types:
+            raise TransformationProcessError("", "No subject_types in FOLIO")
 
         # Set defaults
         logging.info("Setting defaults")
@@ -113,9 +118,7 @@ class Conditions:
         logging.info("Default Callnumber type Name:\t%s", self.default_call_number_type["name"])
 
     def setup_and_validate_holdings_types(self):
-        self.holdings_types = list(
-            self.folio.folio_get_all("/holdings-types", "holdingsTypes", "", 1000)
-        )
+        self.holdings_types = self.folio.holdings_types
         if not self.holdings_types:
             raise TransformationProcessError("", "No holdings_types in FOLIO")
         missing_holdings_types = [
@@ -846,6 +849,19 @@ class Conditions:
             + i18n.t("1 is public, all other values are Staff only")
             + ")",
         )
-        if ind1 != "1":
+        if ind1 == "0":
             return "true"
         return "false"
+
+    def condition_set_subject_type_id(self, legacy_id, value, parameter, marc_field: field.Field):
+        try:
+            t = self.get_ref_data_tuple_by_name(
+                self.folio.subject_types, "subject_types", parameter["name"]
+            )
+            self.mapper.migration_report.add("MappedSubjectTypes", t[1])
+            return t[0]
+        except Exception:
+            raise TransformationProcessError(
+                legacy_id,
+                f"Subject type not found for {parameter['name']} {marc_field}",
+            )

folio_migration_tools/marc_rules_transformation/holdings_statementsparser.py

@@ -2,11 +2,10 @@ import calendar
 import contextlib
 import logging
 import re
-import i18n
 from typing import List
 
-from pymarc import Field
-from pymarc import Record
+import i18n
+from pymarc import Field, Record
 
 from folio_migration_tools.custom_exceptions import TransformationFieldMappingError
 
@@ -72,9 +71,20 @@ class HoldingsStatementsParser:
 
             else:
                 for linked_value_field in linked_value_fields:
-                    parsed_dict = HoldingsStatementsParser.parse_linked_field(
-                        pattern_field, linked_value_field
-                    )
+                    try:
+                        parsed_dict = HoldingsStatementsParser.parse_linked_field(
+                            pattern_field, linked_value_field
+                        )
+                    except KeyError:
+                        raise TransformationFieldMappingError(
+                            legacy_ids,
+                            i18n.t(
+                                "subfield present in %{linked_value_tag} but not in %{pattern_field}",
+                                pattern_field=pattern_field,
+                                linked_value_tag=linked_value_field,
+                            ),
+                            pattern_field,
+                        )
                     if parsed_dict["hlm_stmt"]:
                         return_dict["hlm_stmts"].append(parsed_dict["hlm_stmt"])
                     if parsed_dict["statement"]:
@@ -222,7 +232,7 @@ class HoldingsStatementsParser:
 
     @staticmethod
     def dedupe_list_of_dict(list_of_dict):
-        return [dict(t) for t in {tuple(d.items()) for d in list_of_dict}]
+        return [dict(t) for t in {tuple(d.items()): None for d in list_of_dict}]
 
     @staticmethod
     def g_m(m: int):
@@ -254,8 +264,8 @@ class HoldingsStatementsParser:
         year = False
         is_span = False
         for chron_level in [cl for cl in "ijkl" if cl in linked_value_field]:
-            desc = pattern_field[chron_level] or ""
-            if linked_value_field[chron_level]:
+            desc = pattern_field.get(chron_level, "")
+            if linked_value_field.get(chron_level):
                 if chron_level == "i" and desc == "(year)":
                     hlm_stmt = linked_value_field[chron_level]
                 if desc == "(year)":
@@ -295,9 +305,9 @@ class HoldingsStatementsParser:
         _to = ""
         is_span = False
         for enum_level in [el for el in "abcdef" if el in linked_value_field]:
-            desc = pattern_field[enum_level] or ""
+            desc = pattern_field.get(enum_level, "")
             desc = desc.strip() if "(" not in desc else ""
-            if linked_value_field[enum_level]:
+            if linked_value_field.get(enum_level):
                 val, *val_rest = linked_value_field[enum_level].split("-")
                 is_span = "-" in linked_value_field[enum_level] or is_span
                 _from = f"{_from}{(':' if _from else '')}{desc}{val}"

folio_migration_tools/marc_rules_transformation/marc_file_processor.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import sys
 import time
 import traceback
@@ -6,17 +7,15 @@ from typing import List
 
 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
+from pymarc import Field, Record, Subfield
 
-from folio_migration_tools.custom_exceptions import TransformationProcessError
-from folio_migration_tools.custom_exceptions import TransformationRecordFailedError
+from folio_migration_tools.custom_exceptions import (
+    TransformationProcessError,
+    TransformationRecordFailedError,
+)
 from folio_migration_tools.folder_structure import FolderStructure
 from folio_migration_tools.helper import Helper
-from folio_migration_tools.library_configuration import FileDefinition
-from folio_migration_tools.library_configuration import HridHandling
+from folio_migration_tools.library_configuration import FileDefinition, HridHandling
 from folio_migration_tools.marc_rules_transformation.rules_mapper_base import (
     RulesMapperBase,
 )
@@ -33,6 +32,8 @@ class MarcFileProcessor:
         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+")
+        if mapper.task_configuration.data_import_marc:
+            self.data_import_marc_file = open(self.folder_structure.data_import_marc_path, "wb+")
         self.unique_001s: set = set()
         self.failed_records_count: int = 0
         self.records_count: int = 0
@@ -86,6 +87,12 @@ class MarcFileProcessor:
                             legacy_ids,
                             self.object_type,
                         )
+                    if self.mapper.task_configuration.data_import_marc:
+                        self.save_marc_record(
+                            marc_record,
+                            folio_rec,
+                            self.object_type
+                        )
                 Helper.write_to_file(self.created_objects_file, folio_rec)
                 self.mapper.migration_report.add_general_statistics(
                     i18n.t("Inventory records written to disk")
@@ -122,6 +129,19 @@ class MarcFileProcessor:
                     ):
                         self.mapper.remove_from_id_map(folio_rec.get("formerIds", []))
 
+    def save_marc_record(
+        self,
+        marc_record: Record,
+        folio_rec: dict,
+        object_type: FOLIONamespaces
+    ):
+        self.mapper.save_data_import_marc_record(
+            self.data_import_marc_file,
+            object_type,
+            marc_record,
+            folio_rec,
+        )
+
     def save_srs_record(
         self,
         marc_record: Record,
@@ -247,7 +267,15 @@ class MarcFileProcessor:
                 self.mapper.mapped_legacy_fields,
             )
         if self.mapper.task_configuration.create_source_records:
+            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 self.mapper.task_configuration.data_import_marc:
+            self.data_import_marc_file.seek(0)
+            if not self.data_import_marc_file.read(1):
+                os.remove(self.data_import_marc_file.name)
+            self.data_import_marc_file.close()
         self.mapper.wrap_up()
 
         logging.info("Transformation report written to %s", report_file.name)
@@ -255,7 +283,6 @@ class MarcFileProcessor:
 
     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(