def delete_collection(collection_id=None):
# Deleting a collection affects many associated objects and requires
# checks, so this is done manually and in detail here.
q = db.session.query(Collection).filter(Collection.id == collection_id)
collection = q.first()
if collection is None:
log.error("No collection with ID: %r", collection_id)
log.info("Deleting collection [%r]: %r", collection.id, collection.label)
deleted_at = datetime.utcnow()
for entity in collection.entities:
entity.collections = [c for c in entity.collections
if c.id != collection.id]
db.session.add(entity)
if not len(entity.collections):
delete_entity(entity)
else:
update_entity(entity)
for document in collection.documents:
document.collections = [c for c in document.collections
if c.id != collection.id]
if not len(document.collections):
delete_document(document, deleted_at=deleted_at)
else:
if collection_id == document.source_collection_id:
document.source_collection_id = None
db.session.add(document)
update_document(document)
collection.delete(deleted_at=deleted_at)
db.session.commit()
index_delete(collection_id)
with graph.transaction() as tx:
graph.remove_collection(tx, collection_id)
def update_document(document):
# These are operations that should be executed after each
# write to a document or its metadata.
analyze_document_id.delay(document.id)
index_document(document, index_records=False)
with graph.transaction() as tx:
graph.load_document(tx, document)
def update_document(document):
# These are operations that should be executed after each
# write to a document or its metadata.
analyze_document_id.apply_async([document.id], queue=USER_QUEUE,
routing_key=USER_ROUTING_KEY)
index_document(document, index_records=False)
with graph.transaction() as tx:
graph.load_document(tx, document)
def update_document(document):
# These are operations that should be executed after each
# write to a document or its metadata.
analyze_document_id.apply_async([document.id],
queue=USER_QUEUE,
routing_key=USER_ROUTING_KEY)
index_document(document, index_records=False)
with graph.transaction() as tx:
graph.load_document(tx, document)
def update_entity_full(entity_id):
"""Perform update operations on entities."""
query = db.session.query(Entity).filter(Entity.id == entity_id)
entity = query.first()
generate_entity_references(entity)
with graph.transaction() as tx:
graph.load_entity(tx, entity)
graph.generate_paths(tx, entity)
reindex_entity(entity)
Alert.dedupe(entity.id)
def update_entity_full(entity_id):
"""Perform update operations on entities."""
query = db.session.query(Entity).filter(Entity.id == entity_id)
entity = query.first()
generate_entity_references(entity)
with graph.transaction() as tx:
graph.load_entity(tx, entity)
graph.generate_paths(tx, entity)
reindex_entity(entity)
Alert.dedupe(entity.id)
def analyze_document(document):
log.info("Analyze document: %r", document)
analyzers = []
meta = document.meta
for cls in get_analyzers():
analyzer = cls(document, meta)
analyzer.prepare()
analyzers.append(analyzer)
for text in document.text_parts():
for analyzer in analyzers:
analyzer.on_text(text)
for analyzer in analyzers:
analyzer.finalize()
document.meta = meta
db.session.add(document)
db.session.commit()
index_document(document)
with graph.transaction() as tx:
graph.load_document(tx, document)
def analyze_document(document):
log.info("Analyze document: %r", document)
analyzers = []
meta = document.meta
for cls in get_analyzers():
analyzer = cls(document, meta)
analyzer.prepare()
analyzers.append(analyzer)
for text in document.text_parts():
for analyzer in analyzers:
analyzer.on_text(text)
for analyzer in analyzers:
analyzer.finalize()
document.meta = meta
db.session.add(document)
db.session.commit()
index_document(document)
with graph.transaction() as tx:
graph.load_document(tx, document)
def delete_collection(collection_id=None):
# Deleting a collection affects many associated objects and requires
# checks, so this is done manually and in detail here.
q = db.session.query(Collection).filter(Collection.id == collection_id)
collection = q.first()
if collection is None:
log.error("No collection with ID: %r", collection_id)
log.info("Deleting collection [%r]: %r", collection.id, collection.label)
deleted_at = datetime.utcnow()
for entity in collection.entities:
entity.collections = [
c for c in entity.collections if c.id != collection.id
]
db.session.add(entity)
if not len(entity.collections):
delete_entity(entity)
else:
update_entity(entity)
for document in collection.documents:
document.collections = [
c for c in document.collections if c.id != collection.id
]
if not len(document.collections):
delete_document(document, deleted_at=deleted_at)
else:
if collection_id == document.source_collection_id:
document.source_collection_id = None
db.session.add(document)
update_document(document)
collection.delete(deleted_at=deleted_at)
db.session.commit()
index_delete(collection_id)
with graph.transaction() as tx:
graph.remove_collection(tx, collection_id)
def update_collection(collection):
"""Create or update a collection."""
with graph.transaction() as tx:
graph.load_collection(tx, collection)
def delete_entity(entity, deleted_at=None):
update_entity(entity)
with graph.transaction() as tx:
graph.remove_entity(tx, entity.id)
graph.delete_paths(entity.id)
entity.delete(deleted_at=deleted_at)
def delete_document(document, deleted_at=None):
with graph.transaction() as tx:
graph.remove_document(tx, document.id)
index_delete(document.id)
document.delete(deleted_at=deleted_at)
def delete_document(document, deleted_at=None):
with graph.transaction() as tx:
graph.remove_document(tx, document.id)
index_delete(document.id)
document.delete(deleted_at=deleted_at)
def delete_entity(entity, deleted_at=None):
update_entity(entity)
with graph.transaction() as tx:
graph.remove_entity(tx, entity.id)
graph.delete_paths(entity.id)
entity.delete(deleted_at=deleted_at)
def update_collection(collection):
"""Create or update a collection."""
with graph.transaction() as tx:
graph.load_collection(tx, collection)
def update_collection(collection):
with graph.transaction() as tx:
graph.load_collection(tx, collection)
def update_entity(entity):
reindex_entity(entity, references=False)
with graph.transaction() as tx:
graph.load_entity(tx, entity)
update_entity_full.delay(entity.id)
