add(staff-association): new model and transformation rules

cds_migrator_kit/rdm/records/transform/models/staff_association.py

@@ -0,0 +1,103 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2026 CERN.
+#
+# CDS-RDM is free software; you can redistribute it and/or modify it under
+# the terms of the MIT License; see LICENSE file for more details.
+
+"""CDS-RDM Staff Association model."""
+from cds_migrator_kit.rdm.records.transform.models.bulletin_issue import (
+    bull_issue_model,
+)
+from cds_migrator_kit.transform.overdo import CdsOverdo
+
+
+class StaffAssociationModel(CdsOverdo):
+    """Translation model for Staff Association."""
+
+    __query__ = """
+        (
+            980__:BULLETINSTAFF
+            -980__:CERN_BULLETIN_ARTICLE
+            -980__:CERN_BULLETIN_ISSUE
+        )
+        OR
+        (
+            980__:STAFFASSOCIATION
+            594__:PUB
+        )
+    """
+
+    # Copy-pasted from bulletin issue
+    __ignore_keys__ = {
+        "0248_a",
+        "0248_p",
+        "0248_q",
+        "100__m",  # email of contributor
+        "110__a",  # corporate author, always CERN, safe to ignore
+        "300__a",  # number of pages
+        "336__a",  # DM metadata
+        "5831_2",  # DM tags 1054836
+        "5831_5",  # DM tags
+        "5831_a",  # DM tags
+        "5831_c",  # DM tags
+        "5831_f",  # DM tags
+        "5831_i",  # DM tags
+        "5831_k",  # DM tags
+        "5831_u",  # DM tags
+        "5831_3",  # DM tags
+        "5831_6",  # DM tags
+        "5831_n",  # DM tags
+        "5831_b",  # DM tags
+        "5831_o",  # DM tags
+        "583__a",  # DM tags
+        "583__c",  # DM tags
+        "583__z",  # DM tags
+        "594__a",  # values: "no", "pub"
+        "650172",  # scheme of subjects
+        "6531_9",  # scheme of keywords
+        "691__a",  # draft/online values, redundant
+        "700__m",  # email of contributor
+        "773__p",  # title of the "CERN Bulletin" series
+        "773__t",  # CERN Bulletin value, redundant
+        "773__y",  # year, duplicate of 260
+        "8560_f",  # contact email
+        "8564_8",  # file id
+        "8564_s",  # bibdoc id
+        "8564_x",  # icon thumbnails sizes
+        "8564_y",  # file description - done by files dump
+        "8564_2",  # DM metadata
+        "8564_q",  # DM metadata
+        "8564_w",  # DM metadata
+        "8564_z",  # DM metadata
+        "8567_2",  # DM tags
+        "8567_q",  # DM tags
+        "8567_w",  # DM tags
+        "8567_d",  # DM tags
+        "859__a",  # TODO: Implement rule for this, 2595/3306 records have this field
+        "906__m",  # edit rights, will be granted by the community
+        "937__c",  # last modified by
+        "937__s",  # last modification date
+        "960__a",  # base number
+        "961__a",  # CDS modification tag # TODO
+        "961__b",  # CDS modification tag # TODO
+        "961__c",  # CDS modification tag # TODO
+        "961__h",  # CDS modification tag # TODO
+        "961__l",  # CDS modification tag # TODO
+        "961__x",  # CDS modification tag # TODO
+        "981__a",  # duplicate record id
+        # "246_1a",
+        # "690C_a",
+    }
+
+    _default_fields = {
+        # TODO should we keep this custom field?
+        "custom_fields": {"journal:journal": {"title": "CERN Bulletin"}},
+        "creators": [{"person_or_org": {"type": "organizational", "name": "CERN"}}],
+    }
+
+
+staff_association_model = StaffAssociationModel(
+    bases=(bull_issue_model,),
+    entry_point_group="cds_migrator_kit.migrator.rules.staff_association",
+)

cds_migrator_kit/rdm/records/transform/xml_processing/rules/base.py

@@ -819,11 +819,33 @@ def related_identifiers_787(self, key, value):
         "article":{
             "relation_type": {"id": "references"},
             "resource_type": {"id": "publication-article"},
-        }
+        },
+        "corresponding video": {
+            "relation_type": {"id": "references"},
+            "resource_type": {"id": "audio"},
+        },
+        "manuscript": {
+            "relation_type": {"id": "isderivedfrom"},
+            "resource_type": {"id": "publication-preprint"},
+        },
+        "bulletin article": {
+            "relation_type": {"id": "references"},
+            "resource_type": {"id": "publication-periodicalarticle"},
+        },
+        "report": {
+            "relation_type": {"id": "isderivedfrom"},
+            "resource_type": {"id": "publication-report"},
+        },
     }
 
     if recid:
         if description:
+            if description not in relation_map.keys():
+                raise UnexpectedValue(
+                    f"Unexpected relation description {description}",
+                    field=key,
+                    value=value,
+                )
             new_id = {
                 "identifier": recid,
                 "scheme": "cds",

cds_migrator_kit/rdm/records/transform/xml_processing/rules/bulletin_issue.py

@@ -229,7 +229,7 @@ def urls_bulletin_bis(self, key, value):
 
 
 @model.over("custom_fields_journal", "(^916__)", override=True)
-def issue_number(self, key, value):
+def custom_fields_journal(self, key, value):
     _custom_fields = self.get("custom_fields", {})
 
     issue = value.get("z")

cds_migrator_kit/rdm/records/transform/xml_processing/rules/staff_association.py

@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2026 CERN.
+#
+# CDS-RDM is free software; you can redistribute it and/or modify it under
+# the terms of the MIT License; see LICENSE file for more details.
+
+"""CDS-RDM Staff Association rules."""
+
+from dojson.errors import IgnoreKey
+
+from cds_migrator_kit.errors import UnexpectedValue
+from cds_migrator_kit.rdm.records.transform.xml_processing.rules.base import (
+    additional_titles,
+)
+from cds_migrator_kit.transform.xml_processing.quality.decorators import for_each_value
+
+from ...models.staff_association import staff_association_model as model
+from .bulletin_issue import collection
+from .publications import internal_notes
+
+model.over("internal_notes", "^562__")(internal_notes)
+model.over("additional_titles", "(^242__)")(additional_titles)
+
+
+@model.over("resource_type", "^980__", override=True)
+def resource_type(self, key, value):
+    """Translates resource_type."""
+    value = value.get("a") if "a" in value else value.get("b")
+    if value:
+        value = value.lower()
+    if value in ["bulletinstaff", "staffassociation"]:
+        # TODO what is the resource type?
+        return {"id": "publication-periodicalarticle"}
+    raise UnexpectedValue(
+        "Unknown resource type (STAFF ASSOCIATION)", field=key, value=value
+    )
+
+
+@model.over("collection", "^690C_", override=True)
+@for_each_value
+def staff_association_collection(self, key, value):
+    """Translates collection field."""
+    collection_a = value.get("a", "").strip().lower()
+    # Drop sa documents
+    if collection_a == "sa documents":
+        raise IgnoreKey("collection")
+    collection(self, key, value)

cds_migrator_kit/rdm/streams.yaml

@@ -472,3 +472,14 @@ records:
       missing_users: cds_migrator_kit/rdm/data/users
       communities_ids:
         - ""
+  staff_association:
+    data_dir: cds_migrator_kit/rdm/data/staff_association
+    tmp_dir: cds_migrator_kit/rdm/tmp/staff_association
+    log_dir: cds_migrator_kit/rdm/log/staff_association
+    extract:
+      dirpath: cds_migrator_kit/rdm/data/staff_association/dump/
+    transform:
+      files_dump_dir: cds_migrator_kit/rdm/data/staff_association/files/
+      missing_users: cds_migrator_kit/rdm/data/users
+      communities_ids:
+        - "9ab1f6bd-b213-4bb7-9249-13b9665e453a"

setup.cfg

@@ -86,7 +86,7 @@ cds_migrator_kit.migrator.models =
     te = cds_migrator_kit.rdm.records.transform.models.te:te_model
     en = cds_migrator_kit.rdm.records.transform.models.en:en_model
     annual_rep = cds_migrator_kit.rdm.records.transform.models.annual_report:annual_rep_model
-
+    staff_association = cds_migrator_kit.rdm.records.transform.models.staff_association:staff_association_model
 cds_migrator_kit.migrator.rules.base =
     base = cds_migrator_kit.transform.xml_processing.rules.base
 cds_migrator_kit.migrator.rdm.rules.base =
@@ -168,6 +168,10 @@ cds_migrator_kit.migrator.rules.fap =
     base = cds_migrator_kit.transform.xml_processing.rules.base
     base_records = cds_migrator_kit.rdm.records.transform.xml_processing.rules.base
     fap = cds_migrator_kit.rdm.records.transform.xml_processing.rules.fap
+cds_migrator_kit.migrator.rules.staff_association =
+    base = cds_migrator_kit.transform.xml_processing.rules.base
+    base_records = cds_migrator_kit.rdm.records.transform.xml_processing.rules.base
+    staff_association = cds_migrator_kit.rdm.records.transform.xml_processing.rules.staff_association
 cds_migrator_kit.migrator.rules.people =
     people = cds_migrator_kit.rdm.users.transform.xml_processing.rules.people
 invenio_pidstore.minters =