diff --git a/CHANGES.rst b/CHANGES.rst
index 54b4c88c983f79b055485be017cc1e4ab009285e..b6ac56efff7bc7751b9da8c7563112947f1f5225 100644
--- a/CHANGES.rst
+++ b/CHANGES.rst
@@ -8,7 +8,7 @@
Changes
=======
-Version 2024.4 (released 2024-11-26, updated 2024-12-05)
+Version 2024.4 (released 2024-11-26, updated 2024-12-12)
- Pin `Flask-Menu`` dependency
- Add `Invenio-DAMAP` to the dependencies
@@ -16,6 +16,7 @@ Version 2024.4 (released 2024-11-26, updated 2024-12-05)
- Cast TISS ID into string for the Invenio-DAMAP integration
- Send out notification emails when records get published
- Rework curation menu item registration and unpin `Flask-Menu`
+- Fix TISS name vocabulary synchronization
Version 2024.3 (released 2024-10-01, updated 2024-11-13)
diff --git a/invenio_config_tuw/__init__.py b/invenio_config_tuw/__init__.py
index 3882381b54509764e814684ca45a7ef4b844a502..8b0fddfb986a311e081d9f98a10bc74cd1d10f5c 100644
--- a/invenio_config_tuw/__init__.py
+++ b/invenio_config_tuw/__init__.py
@@ -9,6 +9,6 @@
from .ext import InvenioConfigTUW
-__version__ = "2024.4.4"
+__version__ = "2024.4.5"
__all__ = ("__version__", "InvenioConfigTUW")
diff --git a/invenio_config_tuw/tasks.py b/invenio_config_tuw/tasks.py
index 894cba6b9b1d95217d68fed6617812e2d39c8767..b520b9cbeca83a8b7ac6ae515b0253ff6b704f29 100644
--- a/invenio_config_tuw/tasks.py
+++ b/invenio_config_tuw/tasks.py
@@ -8,6 +8,8 @@
"""Celery tasks running in the background."""
import copy
+from collections import defaultdict
+from difflib import SequenceMatcher
from typing import List, Optional
import requests
@@ -18,13 +20,13 @@ from invenio_accounts.proxies import current_datastore
from invenio_db import db
from invenio_mail.tasks import send_email
from invenio_rdm_records.proxies import current_rdm_records_service as records_service
-from invenio_records_resources.services.uow import UnitOfWork
+from invenio_records_resources.services.uow import RecordIndexOp, UnitOfWork
from invenio_vocabularies.contrib.names.api import Name
from .tiss import Employee, fetch_tiss_data
-def get_tuw_ror_aliases():
+def get_tuw_ror_aliases() -> List[str]:
"""Fetch the aliases of TU Wien known to ROR."""
try:
response = requests.get("https://api.ror.org/organizations/04d836q62")
@@ -46,15 +48,13 @@ def get_tuw_ror_aliases():
]
-def find_orcid_match(employee: Employee, names: List[Name]) -> Optional[Name]:
- """Find the name entry with the same ORCID as the given employee."""
- if not employee.orcid:
+def find_orcid_match(orcid: str, names: List[Name]) -> Optional[Name]:
+ """Find the name entry with the given ORCID."""
+ if not orcid:
return None
for name in names:
- if {"scheme": "orcid", "identifier": employee.orcid} in name.get(
- "identifiers", []
- ):
+ if {"scheme": "orcid", "identifier": orcid} in name.get("identifiers", []):
return name
return None
@@ -68,6 +68,8 @@ def update_name_data(
name = copy.deepcopy(name)
name["given_name"] = employee.first_name
name["family_name"] = employee.last_name
+ if "name" in name:
+ name["name"] = f"{employee.last_name}, {employee.first_name}"
# normalize & deduplicate affilations, and make sure that TU Wien is one of them
# NOTE: sorting is done to remove indeterminism and prevent unnecessary updates
@@ -92,10 +94,23 @@ def update_name_data(
return name
+def _calc_name_distance(
+ employee: Optional[Employee], name_voc: Optional[Name]
+) -> float:
+ """Calculate the distance between the employee name and the vocabulary entry."""
+ if employee is None or name_voc is None:
+ return 0
+
+ fn, ln = name_voc.get("given_name", ""), name_voc.get("family_name", "")
+ fn_dist = SequenceMatcher(a=fn, b=employee.first_name).ratio()
+ ln_dist = SequenceMatcher(a=ln, b=employee.last_name).ratio()
+ return fn_dist + ln_dist
+
+
@shared_task(ignore_result=True)
-def sync_names_from_tiss():
+def sync_names_from_tiss() -> dict:
"""Look up TU Wien employees via TISS and update the names vocabulary."""
- results = {"created": 0, "updated": 0}
+ results = {"created": 0, "updated": 0, "failed": 0}
tuw_ror_aliases = get_tuw_ror_aliases()
svc = current_app.extensions["invenio-vocabularies"].names_service
@@ -104,35 +119,75 @@ def sync_names_from_tiss():
for model in svc.record_cls.model_cls.query.all()
if not model.is_deleted and model.data
]
-
_, employees = fetch_tiss_data()
- employees_with_orcid = [e for e in employees if not e.pseudoperson and e.orcid]
+
+ # it can happen that 2+ TISS profiles have the same ORCID listed
+ orcid_employees = defaultdict(list)
+ for employee in employees:
+ if not employee.pseudoperson and employee.orcid:
+ orcid_employees[employee.orcid].append(employee)
with UnitOfWork(db.session) as uow:
- for employee in employees_with_orcid:
- matching_name = find_orcid_match(employee, all_names)
-
- if matching_name:
- # if we found a match via ORCID, we update it according to the TISS data
- name = svc.read(identity=system_identity, id_=matching_name["id"])
- new_name_data = update_name_data(name.data, employee, tuw_ror_aliases)
-
- # only update the entry if it actually differs somehow
- if name.data != new_name_data:
- svc.update(
- identity=system_identity,
- id_=name.id,
- data=new_name_data,
- uow=uow,
+ for orcid, employees in orcid_employees.items():
+ matching_name = find_orcid_match(orcid, all_names)
+ if len(employees) == 1:
+ (employee,) = employees
+ elif len(employees) > 1:
+ # if we several TISS profiles with the same ORCID, we use the one
+ # with the closest match in name in our names vocabulary
+ employee = sorted(
+ employees,
+ key=lambda e: _calc_name_distance(employee, matching_name),
+ )[-1]
+ else:
+ continue
+
+ try:
+ if matching_name:
+ name_voc_id = matching_name.pid.pid_value
+ old_name = matching_name.get(
+ "name",
+ f"{matching_name.get('family_name')}, {matching_name.get('given_name')}",
)
- results["updated"] += 1
+ new_name = f"{employee.last_name}, {employee.first_name}"
- else:
- # if we couldn't find a match via ORCID, that's a new entry
- svc.create(
- identity=system_identity, data=employee.to_name_entry(), uow=uow
+ # if we found a match via ORCID, we update it according to the TISS data
+ name = svc.read(identity=system_identity, id_=name_voc_id)
+ new_name_data = update_name_data(
+ name.data, employee, tuw_ror_aliases
+ )
+
+ # only update the entry if it actually differs somehow
+ if name.data != new_name_data:
+ # note: it seems like `svc.update()` is currently broken because
+ # some fields that get added to the record (e.g. "pid") isn't
+ # expected in the JSONSchema used for validation
+ # we avoid this by popping "$schema" here
+ # app-rdm v12, vocabularies v3.4.2
+ name_rec = name._record
+ name_rec.update(new_name_data)
+ name_rec.pop("$schema")
+ name_rec.commit()
+ uow.register(RecordIndexOp(name_rec))
+ results["updated"] += 1
+
+ current_app.logger.info(
+ f"TISS sync: updated name '{name_voc_id}' from "
+ f"'{old_name}' to '{new_name}'"
+ )
+
+ else:
+ # if we couldn't find a match via ORCID, that's a new entry
+ svc.create(
+ identity=system_identity, data=employee.to_name_entry(), uow=uow
+ )
+ results["created"] += 1
+
+ except Exception as e:
+ results["failed"] += 1
+ current_app.logger.warn(
+ f"TISS sync: failed for '{employee}', with error: {e}"
)
- results["created"] += 1
uow.commit()
diff --git a/invenio_config_tuw/tiss/models.py b/invenio_config_tuw/tiss/models.py
index 0efad7b5fefa37ddc52e11b8aed4f1a291d38782..bb8168f6abef0d320e6a6788154b92ea76a1800c 100644
--- a/invenio_config_tuw/tiss/models.py
+++ b/invenio_config_tuw/tiss/models.py
@@ -75,6 +75,7 @@ class Employee:
ids.append({"scheme": "orcid", "identifier": self.orcid})
return {
+ "id": self.orcid or str(self.tiss_id),
"given_name": self.first_name,
"family_name": self.last_name,
"identifiers": ids,