def search_serial_by_title(title):
"""Return serial search by title."""
series_search = current_app_ils.series_search_cls()
search = series_search\
.query("query_string", query=f'title:"{title}"')\
.filter("term", mode_of_issuance="SERIAL")
return search
def link_documents_and_serials():
"""Link documents/multiparts and serials."""
document_class = current_app_ils.document_record_cls
document_search = current_app_ils.document_search_cls()
series_class = current_app_ils.series_record_cls
series_search = current_app_ils.series_search_cls()
def link_records_and_serial(record_cls, search):
for hit in search.scan():
# Skip linking if the hit doesn't have a legacy recid since it
# means it's a volume of a multipart
if "legacy_recid" not in hit:
continue
record = record_cls.get_record_by_pid(hit.pid)
check_for_special_series(record)
for serial in get_serials_by_child_recid(hit.legacy_recid):
volume = get_migrated_volume_by_serial_title(
record, serial["title"])
create_parent_child_relation(serial, record, SERIAL_RELATION,
volume)
RecordRelationIndexer().index(record, serial)
def link_record_and_journal(record_cls, search):
for hit in search.scan():
if "legacy_recid" not in hit:
continue
record = record_cls.get_record_by_pid(hit.pid)
for journal in hit["_migration"]["journal_record_legacy_recids"]:
serial = get_record_by_legacy_recid(series_class,
journal["recid"])
create_parent_child_relation(serial, record, SERIAL_RELATION,
journal["volume"])
del record["publication_info"]
record.commit()
db.session.commit()
click.echo("Creating serial relations...")
link_records_and_serial(
document_class,
document_search.filter("term", _migration__has_serial=True),
)
link_records_and_serial(
series_class,
series_search.filter(
"bool",
filter=[
Q("term", mode_of_issuance="MULTIPART_MONOGRAPH"),
Q("term", _migration__has_serial=True),
],
),
)
link_record_and_journal(
document_class,
document_search.filter("term", _migration__has_journal=True),
)
def search_series_with_relations():
"""Return series with relations."""
series_search = current_app_ils.series_search_cls()
search = series_search.filter(
"bool",
filter=[
Q("term", _migration__has_related=True),
],
)
return search
def search_series_by_issn(issn):
"""Find series by ISSN."""
series_search = current_app_ils.series_search_cls()
search = series_search.query(
"bool",
must=[
Q("term", identifiers__scheme="ISSN"),
Q("term", identifiers__value=issn),
],
)
return search
def get_serials_by_child_recid(recid):
"""Search serials by children recid."""
series_search = current_app_ils.series_search_cls()
series_class = current_app_ils.series_record_cls
search = series_search.query(
"bool",
filter=[
Q("term", mode_of_issuance="SERIAL"),
Q("term", _migration__children=recid),
],
)
for hit in search.params(scroll='1h').scan():
yield series_class.get_record_by_pid(hit.pid)
def get_multipart_by_multipart_id(multipart_id):
"""Search multiparts by its identifier."""
series_search = current_app_ils.series_search_cls()
series_cls = current_app_ils.series_record_cls
# f.e. multipart id = vol234
search = series_search.query("match",
_migration__multipart_id=multipart_id).filter(
"match",
mode_of_issuance="MULTIPART_MONOGRAPH")
result = search.execute()
hits_total = result.hits.total.value
if hits_total == 1:
return series_cls.get_record_by_pid(result.hits[0].pid)
if hits_total == 0:
click.secho("no multipart found with id {}".format(multipart_id),
fg="red")
else:
raise MultipartMigrationError(
"found more than one multipart id {}".format(multipart_id))
def validate_serial_records():
"""Validate that serials were migrated successfully.
Performs the following checks:
* Find duplicate serials
* Ensure all children of migrated serials were migrated
"""
def validate_serial_relation(serial, recids):
document_cls = current_app_ils.document_record_cls
relations = serial.relations.get().get("serial", [])
if len(recids) != len(relations):
click.echo("[Serial {}] Incorrect number of children: {} "
"(expected {})".format(serial["pid"], len(relations),
len(recids)))
for relation in relations:
child = document_cls.get_record_by_pid(
relation["pid"], pid_type=relation["pid_type"])
if "legacy_recid" in child and child["legacy_recid"] not in recids:
click.echo("[Serial {}] Unexpected child with legacy "
"recid: {}".format(serial["pid"],
child["legacy_recid"]))
titles = set()
series_search = current_app_ils.series_search_cls()
search = series_search.filter("term", mode_of_issuance="SERIAL")
for serial_hit in search.params(scroll='1h').scan():
# Store titles and check for duplicates
if "title" in serial_hit:
title = serial_hit.title
if title in titles:
current_app.logger.warning(
'Serial title "{}" already exists'.format(title))
else:
titles.add(title)
# Check if any children are missing
children = serial_hit._migration.children
serial = Series.get_record_by_pid(serial_hit.pid)
validate_serial_relation(serial, children)
click.echo("Serial validation check done!")
def search_series_by_title(title):
"""Find series by title."""
series_search = current_app_ils.series_search_cls()
title = title.lower()
search = series_search.filter("term", title__normalized_keyword=title)
return search
def link_documents_and_serials():
"""Link documents/multiparts and serials."""
document_class = current_app_ils.document_record_cls
document_search = current_app_ils.document_search_cls()
series_class = current_app_ils.series_record_cls
series_search = current_app_ils.series_search_cls()
journal_legacy_pid_type =\
current_app.config["CDS_ILS_SERIES_LEGACY_PID_TYPE"]
def link_records_and_serial(record_cls, search):
click.echo(f"FOUND {search.count()} serial related records.")
for hit in search.params(scroll='1h').scan():
try:
click.echo(f"Processing record {hit.pid}.")
# Skip linking if the hit doesn't have a legacy recid since it
# means it's a volume of a multipart
if "legacy_recid" not in hit:
continue
record = record_cls.get_record_by_pid(hit.pid)
check_for_special_series(record)
for serial in get_serials_by_child_recid(hit.legacy_recid):
volume = get_migrated_volume_by_serial_title(
record, serial["title"])
create_parent_child_relation(serial, record,
SERIAL_RELATION, volume)
RecordRelationIndexer().index(record, serial)
# mark done
record["_migration"]["has_serial"] = False
record.commit()
db.session.commit()
except Exception as exc:
handler = relation_exception_handlers.get(exc.__class__)
if handler:
legacy_recid = None
if hasattr(hit, "legacy_recid"):
legacy_recid = hit.legacy_recid
handler(exc, new_pid=hit.pid, legacy_id=legacy_recid)
else:
raise exc
def link_record_and_journal(record_cls, search):
click.echo(f"FOUND {search.count()} journal related records.")
for hit in search.params(scroll='1h').scan():
click.echo(f"Processing record {hit.pid}.")
try:
if "legacy_recid" not in hit:
continue
record = record_cls.get_record_by_pid(hit.pid)
for journal in \
hit["_migration"]["journal_record_legacy_recids"]:
serial = get_record_by_legacy_recid(
series_class, journal_legacy_pid_type,
journal["recid"])
create_parent_child_relation(serial, record,
SERIAL_RELATION,
journal["volume"])
# mark done
record["_migration"]["has_journal"] = False
record.commit()
db.session.commit()
except Exception as exc:
handler = relation_exception_handlers.get(exc.__class__)
if handler:
legacy_recid = None
if hasattr(hit, "legacy_recid"):
legacy_recid = hit.legacy_recid
handler(exc, new_pid=hit.pid, legacy_id=legacy_recid)
else:
raise exc
click.echo("Creating serial relations...")
link_records_and_serial(
document_class,
document_search.filter("term", _migration__has_serial=True))
link_records_and_serial(
series_class,
series_search.filter(
"bool",
filter=[
Q("term", mode_of_issuance="MULTIPART_MONOGRAPH"),
Q("term", _migration__has_serial=True),
],
),
)
link_record_and_journal(
document_class,
document_search.filter("term", _migration__has_journal=True),
)
