# Copyright 2021 Google LLC All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """Classes for representing bundles for the Google Cloud Firestore API.""" import datetime import json from google.cloud.firestore_bundle.types.bundle import ( BundledDocumentMetadata, BundledQuery, BundleElement, BundleMetadata, NamedQuery, ) from google.cloud._helpers import _datetime_to_pb_timestamp, UTC # type: ignore from google.cloud.firestore_bundle._helpers import limit_type_of_query from google.cloud.firestore_v1.async_query import AsyncQuery from google.cloud.firestore_v1.base_client import BaseClient from google.cloud.firestore_v1.base_document import DocumentSnapshot from google.cloud.firestore_v1.base_query import BaseQuery from google.cloud.firestore_v1.document import DocumentReference from google.cloud.firestore_v1 import _helpers from google.protobuf.timestamp_pb2 import Timestamp # type: ignore from google.protobuf import json_format # type: ignore from typing import ( Dict, List, Optional, Union, ) class FirestoreBundle: """A group of serialized documents and queries, suitable for longterm storage or query resumption. If any queries are added to this bundle, all associated documents will be loaded and stored in memory for serialization. Usage: .. code-block:: python from google.cloud.firestore import Client, _helpers from google.cloud.firestore_bundle import FirestoreBundle db = Client() bundle = FirestoreBundle('my-bundle') bundle.add_named_query('all-users', db.collection('users')._query()) bundle.add_named_query( 'top-ten-hamburgers', db.collection('hamburgers').limit(limit=10), ) serialized: str = bundle.build() # Store somewhere like a Google Cloud Storage bucket for retrieval by # a client SDK. Args: name (str): The Id of the bundle. """ BUNDLE_SCHEMA_VERSION: int = 1 def __init__(self, name: str) -> None: self.name: str = name self.documents: Dict[str, "_BundledDocument"] = {} self.named_queries: Dict[str, NamedQuery] = {} self.latest_read_time: Timestamp = Timestamp(seconds=0, nanos=0) self._deserialized_metadata: Optional[BundledDocumentMetadata] = None def add_document(self, snapshot: DocumentSnapshot) -> "FirestoreBundle": """Adds a document to the bundle. Args: snapshot (DocumentSnapshot): The fully-loaded Firestore document to be preserved. Example: .. code-block:: python from google.cloud import firestore db = firestore.Client() collection_ref = db.collection(u'users') bundle = firestore.FirestoreBundle('my bundle') bundle.add_document(collection_ref.documents('some_id').get()) Returns: FirestoreBundle: self """ original_document: Optional[_BundledDocument] original_queries: Optional[List[str]] = [] full_document_path: str = snapshot.reference._document_path original_document = self.documents.get(full_document_path) if original_document: original_queries = original_document.metadata.queries # type: ignore should_use_snaphot: bool = ( original_document is None # equivalent to: # `if snapshot.read_time > original_document.snapshot.read_time` or _helpers.compare_timestamps( snapshot.read_time, original_document.snapshot.read_time, ) >= 0 ) if should_use_snaphot: self.documents[full_document_path] = _BundledDocument( snapshot=snapshot, metadata=BundledDocumentMetadata( name=full_document_path, read_time=snapshot.read_time, exists=snapshot.exists, queries=original_queries, ), ) self._update_last_read_time(snapshot.read_time) self._reset_metadata() return self def add_named_query(self, name: str, query: BaseQuery) -> "FirestoreBundle": """Adds a query to the bundle, referenced by the provided name. Args: name (str): The name by which the provided query should be referenced. query (Query): Query of documents to be fully loaded and stored in the bundle for future access. Example: .. code-block:: python from google.cloud import firestore db = firestore.Client() collection_ref = db.collection(u'users') bundle = firestore.FirestoreBundle('my bundle') bundle.add_named_query('all the users', collection_ref._query()) Returns: FirestoreBundle: self Raises: ValueError: If anything other than a BaseQuery (e.g., a Collection) is supplied. If you have a Collection, call its `_query()` method to get what this method expects. ValueError: If the supplied name has already been added. """ if not isinstance(query, BaseQuery): raise ValueError( "Attempted to add named query of type: " f"{type(query).__name__}. Expected BaseQuery.", ) if name in self.named_queries: raise ValueError(f"Query name conflict: {name} has already been added.") # Execute the query and save each resulting document _read_time = self._save_documents_from_query(query, query_name=name) # Actually save the query to our local object cache self._save_named_query(name, query, _read_time) self._reset_metadata() return self def _save_documents_from_query( self, query: BaseQuery, query_name: str ) -> datetime.datetime: _read_time = datetime.datetime.min.replace(tzinfo=UTC) if isinstance(query, AsyncQuery): import asyncio loop = asyncio.get_event_loop() return loop.run_until_complete(self._process_async_query(query, query_name)) # `query` is now known to be a non-async `BaseQuery` doc: DocumentSnapshot for doc in query.stream(): # type: ignore self.add_document(doc) bundled_document = self.documents.get(doc.reference._document_path) bundled_document.metadata.queries.append(query_name) # type: ignore _read_time = doc.read_time return _read_time def _save_named_query( self, name: str, query: BaseQuery, read_time: datetime.datetime, ) -> None: self.named_queries[name] = self._build_named_query( name=name, snapshot=query, read_time=read_time, ) self._update_last_read_time(read_time) async def _process_async_query( self, snapshot: AsyncQuery, query_name: str, ) -> datetime.datetime: doc: DocumentSnapshot _read_time = datetime.datetime.min.replace(tzinfo=UTC) async for doc in snapshot.stream(): self.add_document(doc) bundled_document = self.documents.get(doc.reference._document_path) bundled_document.metadata.queries.append(query_name) # type: ignore _read_time = doc.read_time return _read_time def _build_named_query( self, name: str, snapshot: BaseQuery, read_time: datetime.datetime, ) -> NamedQuery: return NamedQuery( name=name, bundled_query=BundledQuery( parent=name, structured_query=snapshot._to_protobuf()._pb, limit_type=limit_type_of_query(snapshot), ), read_time=_helpers.build_timestamp(read_time), ) def _update_last_read_time( self, read_time: Union[datetime.datetime, Timestamp] ) -> None: _ts: Timestamp = ( read_time if isinstance(read_time, Timestamp) else _datetime_to_pb_timestamp(read_time) ) # if `_ts` is greater than `self.latest_read_time` if _helpers.compare_timestamps(_ts, self.latest_read_time) == 1: self.latest_read_time = _ts def _add_bundle_element(self, bundle_element: BundleElement, *, client: BaseClient, type: str): # type: ignore """Applies BundleElements to this FirestoreBundle instance as a part of deserializing a FirestoreBundle string. """ from google.cloud.firestore_v1.types.document import Document if getattr(self, "_doc_metadata_map", None) is None: self._doc_metadata_map = {} if type == "metadata": self._deserialized_metadata = bundle_element.metadata # type: ignore elif type == "namedQuery": self.named_queries[bundle_element.named_query.name] = bundle_element.named_query # type: ignore elif type == "documentMetadata": self._doc_metadata_map[ bundle_element.document_metadata.name ] = bundle_element.document_metadata elif type == "document": doc_ref_value = _helpers.DocumentReferenceValue( bundle_element.document.name ) snapshot = DocumentSnapshot( data=_helpers.decode_dict( Document(mapping=bundle_element.document).fields, client ), exists=True, reference=DocumentReference( doc_ref_value.collection_name, doc_ref_value.document_id, client=client, ), read_time=self._doc_metadata_map[ bundle_element.document.name ].read_time, create_time=bundle_element.document.create_time, # type: ignore update_time=bundle_element.document.update_time, # type: ignore ) self.add_document(snapshot) bundled_document = self.documents.get(snapshot.reference._document_path) for query_name in self._doc_metadata_map[ bundle_element.document.name ].queries: bundled_document.metadata.queries.append(query_name) # type: ignore else: raise ValueError(f"Unexpected type of BundleElement: {type}") def build(self) -> str: """Iterates over the bundle's stored documents and queries and produces a single length-prefixed json string suitable for long-term storage. Example: .. code-block:: python from google.cloud import firestore db = firestore.Client() collection_ref = db.collection(u'users') bundle = firestore.FirestoreBundle('my bundle') bundle.add_named_query('app-users', collection_ref._query()) serialized_bundle: str = bundle.build() # Now upload `serialized_bundle` to Google Cloud Storage, store it # in Memorystore, or any other storage solution. Returns: str: The length-prefixed string representation of this bundle' contents. """ buffer: str = "" named_query: NamedQuery for named_query in self.named_queries.values(): buffer += self._compile_bundle_element( BundleElement(named_query=named_query) ) bundled_document: "_BundledDocument" # type: ignore document_count: int = 0 for bundled_document in self.documents.values(): buffer += self._compile_bundle_element( BundleElement(document_metadata=bundled_document.metadata) ) document_count += 1 buffer += self._compile_bundle_element( BundleElement( document=bundled_document.snapshot._to_protobuf()._pb, ) ) metadata: BundleElement = BundleElement( metadata=self._deserialized_metadata or BundleMetadata( id=self.name, create_time=_helpers.build_timestamp(), version=FirestoreBundle.BUNDLE_SCHEMA_VERSION, total_documents=document_count, total_bytes=len(buffer.encode("utf-8")), ) ) return f"{self._compile_bundle_element(metadata)}{buffer}" def _compile_bundle_element(self, bundle_element: BundleElement) -> str: serialized_be = json.dumps(json_format.MessageToDict(bundle_element._pb)) return f"{len(serialized_be)}{serialized_be}" def _reset_metadata(self): """Hydrating bundles stores cached data we must reset anytime new queries or documents are added""" self._deserialized_metadata = None class _BundledDocument: """Convenience class to hold both the metadata and the actual content of a document to be bundled.""" def __init__( self, snapshot: DocumentSnapshot, metadata: BundledDocumentMetadata, ) -> None: self.snapshot = snapshot self.metadata = metadata