def record_insert_or_replace(json, skip_files=False):
"""Insert or replace a record."""
pid_type = get_pid_type_from_schema(json['$schema'])
control_number = json['control_number']
try:
pid = PersistentIdentifier.get(pid_type, control_number)
record = InspireRecord.get_record(pid.object_uuid)
record.clear()
record.update(json, skip_files=skip_files)
if json.get('legacy_creation_date'):
record.model.created = datetime.strptime(
json['legacy_creation_date'], '%Y-%m-%d')
record.commit()
except PIDDoesNotExistError:
record = InspireRecord.create(json, id_=None, skip_files=skip_files)
if json.get('legacy_creation_date'):
record.model.created = datetime.strptime(
json['legacy_creation_date'], '%Y-%m-%d')
inspire_recid_minter(str(record.id), json)
if json.get('deleted'):
new_recid = get_recid_from_ref(json.get('new_record'))
if not new_recid:
record.delete()
return record
def _create_and_index_record(record):
record = InspireRecord.create(record)
inspire_recid_minter(record.id, record)
db.session.commit()
es.indices.refresh('records-hep')
return record
def record_upsert(json):
"""Insert or update a record."""
control_number = json.get('control_number', json.get('recid'))
if control_number:
control_number = int(control_number)
pid_type = InspireRecordIdProvider.schema_to_pid_type(json['$schema'])
try:
pid = PersistentIdentifier.get(pid_type, control_number)
record = Record.get_record(pid.object_uuid)
record.update(json)
record.commit()
except PIDDoesNotExistError:
record = Record.create(json, id_=None)
# Create persistent identifier.
inspire_recid_minter(str(record.id), json)
if json.get('deleted'):
new_recid = get_recid_from_ref(json.get('new_record'))
if new_recid:
merged_record = get_db_record(pid_type, new_recid)
merge_pidstores_of_two_merged_records(merged_record.id, record.id)
else:
soft_delete_pidstore_for_record(record.id)
return record
def _create_and_index_record(record):
record = InspireRecord.create(record)
inspire_recid_minter(record.id, record)
db.session.commit()
es.indices.refresh('records-hep')
return record
def record_insert_or_replace(json, skip_files=False):
"""Insert or replace a record."""
pid_type = get_pid_type_from_schema(json['$schema'])
control_number = json['control_number']
try:
pid = PersistentIdentifier.get(pid_type, control_number)
record = InspireRecord.get_record(pid.object_uuid)
record.clear()
record.update(json, skip_files=skip_files)
if json.get('legacy_creation_date'):
record.model.created = datetime.strptime(json['legacy_creation_date'], '%Y-%m-%d')
record.commit()
except PIDDoesNotExistError:
record = InspireRecord.create(json, id_=None, skip_files=skip_files)
if json.get('legacy_creation_date'):
record.model.created = datetime.strptime(json['legacy_creation_date'], '%Y-%m-%d')
inspire_recid_minter(str(record.id), json)
if json.get('deleted'):
new_recid = get_recid_from_ref(json.get('new_record'))
if not new_recid:
record.delete()
return record
def _create_and_index_record(record):
record = InspireRecord.create(record)
inspire_recid_minter(record.id, record)
# invenio-collections will populate _collections field in record upon
# commit
db.session.commit()
es.indices.refresh('records-hep')
return record
def _create_and_index_record(record):
record = Record.create(record)
inspire_recid_minter(record.id, record)
# invenio-collections will populate _collections field in record upon
# commit
db.session.commit()
# Record needs to be indexed since views fetch records from ES
r = RecordIndexer()
r.index(record)
es.indices.refresh('records-hep')
return record
def store_record(obj, *args, **kwargs):
"""Create and index new record in main record space."""
assert "$schema" in obj.data, "No $schema attribute found!"
# Create record
# FIXME: Do some preprocessing of obj.data before creating a record so that
# we're sure that the schema will be validated without touching the full
# holdingpen stack.
record = Record.create(obj.data, id_=None)
# Create persistent identifier.
pid = inspire_recid_minter(str(record.id), record)
# Commit any changes to record
record.commit()
# Dump any changes to record
obj.data = record.dumps()
# Commit to DB before indexing
db.session.commit()
# Index record
indexer = RecordIndexer()
indexer.index_by_id(pid.object_uuid)
def store_record(obj, *args, **kwargs):
"""Create and index new record in main record space."""
assert "$schema" in obj.data, "No $schema attribute found!"
# Create record
# FIXME: Do some preprocessing of obj.data before creating a record so that
# we're sure that the schema will be validated without touching the full
# holdingpen stack.
record = Record.create(obj.data, id_=None)
# Create persistent identifier.
pid = inspire_recid_minter(str(record.id), record)
# Commit any changes to record
record.commit()
# Dump any changes to record
obj.data = record.dumps()
# Commit to DB before indexing
db.session.commit()
# Index record
indexer = RecordIndexer()
indexer.index_by_id(pid.object_uuid)
def record_insert_or_replace(json):
"""Insert or replace a record."""
control_number = json.get('control_number', json.get('recid'))
if control_number:
pid_type = get_pid_type_from_schema(json['$schema'])
try:
pid = PersistentIdentifier.get(pid_type, control_number)
record = InspireRecord.get_record(pid.object_uuid)
record.clear()
record.update(json)
record.commit()
except PIDDoesNotExistError:
record = InspireRecord.create(json, id_=None)
# Create persistent identifier.
inspire_recid_minter(str(record.id), json)
if json.get('deleted'):
new_recid = get_recid_from_ref(json.get('new_record'))
if not new_recid:
record.delete()
return record
def store_record(obj, *args, **kwargs):
"""Create and index new record in main record space."""
if '$schema' in obj.data:
obj.data['$schema'] = url_for(
'invenio_jsonschemas.get_schema',
schema_path="records/{0}".format(obj.data['$schema'])
)
# Create record
rec_uuid = str(Record.create(obj.data, id_=None).id)
# Create persistent identifier.
pid = inspire_recid_minter(rec_uuid, obj.data)
db.session.commit()
# Index record
indexer = RecordIndexer()
indexer.index_by_id(pid.object_uuid)
def store_record(obj, eng):
"""Insert or replace a record."""
is_update = obj.extra_data.get('is-update')
if is_update:
record = InspireRecord.get_record(obj.extra_data['head_uuid'])
record.clear()
record.update(obj.data, files_src_records=[obj])
else:
record = InspireRecord.create(obj.data, id_=None)
# Create persistent identifier.
created_pid = inspire_recid_minter(str(record.id), record).pid_value
# Now that we have a recid, we can properly download the documents
record.download_documents_and_figures(src_records=[obj])
obj.data['control_number'] = created_pid
# store head_uuid to store the root later
obj.extra_data['head_uuid'] = str(record.id)
record.commit()
obj.save()
db.session.commit()
def store_record(obj, *args, **kwargs):
"""Create and index new record in main record space."""
assert "$schema" in obj.data, "No $schema attribute found!"
# Create record
record = Record.create(obj.data, id_=None)
# Create persistent identifier.
pid = inspire_recid_minter(str(record.id), record)
# Commit any changes to record
record.commit()
# Dump any changes to record
obj.data = record.dumps()
# Commit to DB before indexing
db.session.commit()
# Index record
indexer = RecordIndexer()
indexer.index_by_id(pid.object_uuid)
def store_record(obj, eng):
"""Insert or replace a record."""
def _get_updated_record(obj):
"""TODO: use only head_uuid once we have them merger."""
if 'head_uuid' in obj.extra_data:
updated_record = InspireRecord.get_record(
obj.extra_data['head_uuid'],
)
else:
pid_type = get_pid_type_from_schema(obj.data['$schema'])
updated_record_id = obj.extra_data['record_matches'][0]
updated_record = get_db_record(pid_type, updated_record_id)
return updated_record
is_update = obj.extra_data.get('is-update')
if is_update:
record = _get_updated_record(obj)
obj.data['control_number'] = record['control_number']
record.clear()
record.update(obj.data, files_src_records=[obj])
else:
# Skip the files to avoid issues in case the record has already pid
# TODO: remove the skip files once labs becomes master
record = InspireRecord.create(obj.data, id_=None, skip_files=True)
# Create persistent identifier.
created_pid = inspire_recid_minter(str(record.id), record).pid_value
# Now that we have a recid, we can properly download the documents
record.download_documents_and_figures(src_records=[obj])
obj.data['control_number'] = created_pid
# store head_uuid to store the root later
obj.extra_data['head_uuid'] = str(record.id)
record.commit()
obj.save()
db.session.commit()
def store_record(obj, eng):
"""Insert or replace a record."""
def _get_updated_record(obj):
"""TODO: use only head_uuid once we have them merger."""
if 'head_uuid' in obj.extra_data:
updated_record = InspireRecord.get_record(
obj.extra_data['head_uuid'], )
else:
pid_type = get_pid_type_from_schema(obj.data['$schema'])
updated_record_id = obj.extra_data['record_matches'][0]
updated_record = get_db_record(pid_type, updated_record_id)
return updated_record
is_update = obj.extra_data.get('is-update')
if is_update:
record = _get_updated_record(obj)
obj.data['control_number'] = record['control_number']
record.clear()
record.update(obj.data, files_src_records=[obj])
else:
# Skip the files to avoid issues in case the record has already pid
# TODO: remove the skip files once labs becomes master
record = InspireRecord.create(obj.data, id_=None, skip_files=True)
# Create persistent identifier.
created_pid = inspire_recid_minter(str(record.id), record).pid_value
# Now that we have a recid, we can properly download the documents
record.download_documents_and_figures(src_records=[obj])
obj.data['control_number'] = created_pid
# store head_uuid to store the root later
obj.extra_data['head_uuid'] = str(record.id)
record.commit()
obj.save()
db.session.commit()
def mint(id_, data):
"""Mint the record."""
return inspire_recid_minter(id_, data)
def create_author(profile):
"""Create a new author profile based on a given signature.
The method receives a dictionary representing an author.
Based on the values, it creates a dictionary in the invenio_records format.
After all the fields are processed, the method calls create_record
from invenio_records.api to put the new record.
:param profile:
A signature representing an author's to be created as a profile.
Example:
profile = {u'affiliations': [{u'value': u'Yerevan Phys. Inst.'}],
u'alternative_name': None,
u'curated_relation': False,
u'email': None,
u'full_name': u'Chatrchyan, Serguei',
u'inspire_id': None,
u'orcid': None,
u'profile': u'',
u'recid': None,
u'role': None,
u'uuid': u'd63537a8-1df4-4436-b5ed-224da5b5028c'}
:return:
A recid, where the new profile can be accessed.
Example:
"1234"
"""
name = profile.get('full_name')
# Template of an initial record.
record = {'collections': [{'primary': 'HEPNAMES'}],
'name': {'value': name},
'$schema': _get_author_schema()}
# The author's email address.
# Unfortunately the method will not correlate a given e-mail address
# with an affiliation.
if 'email' in profile:
email = profile.get('email')
record['positions'] = []
record['positions'].append({'email': email})
# The author can be a member of more than one affiliation.
if 'affiliations' in profile:
affiliations = profile.get('affiliations')
if 'positions' not in record:
record['positions'] = []
for affiliation in affiliations:
name = affiliation.get('value')
recid = affiliation.get('recid', None)
if recid:
record['positions'].append(
{'institution': {'name': name, 'recid': recid}})
else:
record['positions'].append(
{'institution': {'name': name}})
# FIXME: The method should also collect the useful data
# from the publication, like category field, subject,
# etc.
# Disconnect the signal on insert of a new record.
after_record_insert.disconnect(append_new_record_to_queue)
# Create a new author profile.
record = InspireRecord.create(record, id_=None)
# Create Inspire recid.
record_pid = inspire_recid_minter(record.id, record)
# Extend the new record with Inspire recid and self key.
record['control_number'] = record_pid.pid_value
record['self'] = inspire_dojson_utils.get_record_ref(
record_pid.pid_value, 'authors')
# Apply the changes.
record.commit()
db.session.commit()
# Reconnect the disconnected signal.
after_record_insert.connect(append_new_record_to_queue)
# Report.
logger.info("Created profile: %s", record_pid.pid_value)
# Return the recid of new profile to which signatures will point to.
return record_pid.pid_value
def mint(id_, data):
"""Mint the record."""
return inspire_recid_minter(id_, data)
def create_author(profile):
"""Create a new author profile based on a given signature.
The method receives a dictionary representing an author.
Based on the values, it creates a dictionary in the invenio_records format.
After all the fields are processed, the method calls create_record
from invenio_records.api to put the new record.
:param profile:
A signature representing an author's to be created as a profile.
Example:
profile = {u'affiliations': [{u'value': u'Yerevan Phys. Inst.'}],
u'alternative_name': None,
u'curated_relation': False,
u'email': None,
u'full_name': u'Chatrchyan, Serguei',
u'inspire_id': None,
u'orcid': None,
u'profile': u'',
u'recid': None,
u'role': None,
u'uuid': u'd63537a8-1df4-4436-b5ed-224da5b5028c'}
:return:
A recid, where the new profile can be accessed.
Example:
"1234"
"""
name = profile.get('full_name')
# Template of an initial record.
record = {'collections': [{'primary': 'HEPNAMES'}],
'name': {'value': name},
'$schema': _get_author_schema()}
# The author's email address.
# Unfortunately the method will not correlate a given e-mail address
# with an affiliation.
if 'email' in profile:
email = profile.get('email')
record['positions'] = []
record['positions'].append({'email': email})
# The author can be a member of more than one affiliation.
if 'affiliations' in profile:
affiliations = profile.get('affiliations')
if 'positions' not in record:
record['positions'] = []
for affiliation in affiliations:
name = affiliation.get('value')
recid = affiliation.get('recid', None)
if recid:
record['positions'].append(
{'institution': {'name': name, 'recid': recid}})
else:
record['positions'].append(
{'institution': {'name': name}})
# FIXME: The method should also collect the useful data
# from the publication, like category field, subject,
# etc.
# Disconnect the signal on insert of a new record.
after_record_insert.disconnect(append_new_record_to_queue)
# Create a new author profile.
record = Record.create(record, id_=None)
# Create Inspire recid.
record_pid = inspire_recid_minter(record.id, record)
# Extend the new record with Inspire recid and self key.
record['control_number'] = record_pid.pid_value
record['self'] = inspire_dojson_utils.get_record_ref(
record_pid.pid_value, 'authors')
# Apply the changes.
record.commit()
db.session.commit()
# Add the record to Elasticsearch.
indexer = RecordIndexer()
indexer.index_by_id(record_pid.object_uuid)
# Reconnect the disconnected signal.
after_record_insert.connect(append_new_record_to_queue)
# Report.
logger.info("Created profile: %s", record_pid.pid_value)
# Return the recid of new profile to which signatures will point to.
return record_pid.pid_value