def _reindex_records(self):
"""Re-index all records in the collection.
- (leave records unchanged)
- wipe existing indexes, create new
- uses search_mapping associated with doc type
"""
search_indices = []
for doc_type in self.doc_types:
index_name = return_index_name(self._collection_id,
doc_type.__name__)
search_indices.append(index_name)
# overwrite temp indices of same name if they exist
if self.client.index_exists(index_name):
self.client.drop_index(index_name)
self.client.new_index(index_name=index_name,
mapping=doc_type.index_mapping)
records_dict = self.records_db.df.index
for document_id in records_dict:
record = self.return_record(document_id)
doc_type = record["document_type"]
link_content = self.gen_link_content(record["document_id"])
(document_id, record_index) = doc_type.gen_search_index(
record, link_content=link_content)
index_name = return_index_name(self._collection_id,
doc_type_str=doc_type.__name__)
self.client.index_record(
document_id=document_id,
record_index=record_index,
index_name=index_name,
)
def register(self, records_db_path=None, config_path=None, write=True):
"""Register collection/define asset paths.
Registration (and subsequent collection.write) are used to create a persistent collection
from an "in-memory" collection. Paths and search index references are stored in a record keyed
by collection name.
"""
if self._register.exists(self.collection_id):
raise KeyError(
f"collection name {self.collection_id} already taken.")
if records_db_path is None:
records_db_path = os.path.join(COLLECTIONS_FOLDER,
self.collection_id, "records.pkl")
if config_path is None:
config_path = os.path.join(COLLECTIONS_FOLDER, self.collection_id,
"config.json")
# rename temp indexes to collection
search_indices = []
for doc_type in self.doc_types:
tmp_index_name = return_index_name(self._collection_id,
doc_type.__name__)
index_name = return_index_name(self.collection_id,
doc_type.__name__)
self.client.rename_index(index_name=tmp_index_name,
new_index_name=index_name)
search_indices.append(index_name)
# register record, add attributes to collection
collection_record = {
"collection_id": self.collection_id,
"records_db_path": records_db_path,
"config_path": config_path,
"search_indices": search_indices,
}
self._register.add(collection_record)
for k, v in collection_record.items():
setattr(self, k, v)
self.registered = True
if write:
self.write()
def remove_record(self, document_id):
"""Remove (resolved) document_id record from collection."""
# get doc_type to remove from search index
record = self.return_record(document_id)
self.records_db.rm_record(document_id)
if self.registered:
self.records_db.write(self.records_db_path)
doc_type = record["document_type"]
index_name = return_index_name(self._collection_id,
doc_type_str=doc_type.__name__)
self.client.delete_record(document_id, index_name=index_name)
def add_record(self, record, index_linked_content, write=True):
"""Add record to collection, write to disk if registered."""
self.records_db.update_record(record)
if self.registered:
self.records_db.write(self.records_db_path)
doc_type = record["document_type"]
if index_linked_content:
link_content = self.gen_link_content(record["document_id"])
else:
link_content = None
(document_id,
record_index) = doc_type.gen_search_index(record,
link_content=link_content)
index_name = return_index_name(self._collection_id,
doc_type_str=doc_type.__name__)
self.client.index_record(document_id=document_id,
record_index=record_index,
index_name=index_name)
def new(cls, collection_id, doc_types):
"""Create a new collection.
Until registered, collection assets are in-memory/overwritable.
"""
if cls._register.exists(collection_id):
raise KeyError(f"collection name {collection_id} already taken.")
# create records db
collection_schema = {}
for doc_type in doc_types:
collection_schema = update_dict(collection_schema, doc_type.schema)
records_db = Data.new(collection_schema)
# create/overwrite temporary search indexes
_collection_id = f"tmp_{collection_id}"
search_indices = []
for doc_type in doc_types:
index_name = return_index_name(_collection_id, doc_type.__name__)
search_indices.append(index_name)
# overwrite temp indices of same name if they exist
if cls.client.index_exists(index_name):
cls.client.drop_index(index_name)
cls.client.new_index(index_name=index_name,
mapping=doc_type.index_mapping)
configd = {
"doc_types": [doc_type.__name__ for doc_type in doc_types],
}
coll = cls(
collection_id=collection_id,
doc_types=doc_types,
records_db=records_db,
configd=configd,
)
coll.search_indices = search_indices
return coll
