feat: add runtime property reindex API for Weaviate >= 1.39 (#2097)

Adds collection.config methods for the GA runtime property reindex API:
- update_property_index: declarative create-or-migrate upsert (PUT
  /schema/{class}/properties/{prop}/index/{indexType}) with optional
  wait_for_completion polling of the index status endpoint
- rebuild_property_index: rebuild an existing index from scratch, with
  tenant selection and optional wait_for_completion
- cancel_property_index_task: idempotent cancellation of a live task
- get_property_indexes: parse GET /schema/{class}/indexes into new
  CollectionPropertyIndexes/PropertyIndexStatus read-side types

All methods raise WeaviateUnsupportedFeatureError below server 1.39.0.

Author: g-despot

weaviate/collections/classes/config.py

@@ -2207,6 +2207,83 @@ class _ShardStatus:
 ShardStatus = _ShardStatus
 
 
+class PropertyIndexTaskStatus(str, BaseEnum):
+    """The status of a runtime property index task submission.
+
+    Attributes:
+        STARTED: A reindexing task was submitted and started.
+        CANCELLED: A live reindexing task was cancelled.
+        NO_OP: No work was necessary, e.g. the index configuration already matched the request
+            or there was no live task to cancel.
+    """
+
+    STARTED = "STARTED"
+    CANCELLED = "CANCELLED"
+    NO_OP = "NO_OP"
+
+
+class PropertyIndexState(str, BaseEnum):
+    """The state of a property index as reported by the index status endpoint.
+
+    Attributes:
+        READY: The index is ready to serve queries.
+        PENDING: A reindexing task for the index is queued but has not started yet.
+        INDEXING: A reindexing task for the index is in progress.
+        FAILED: The reindexing task for the index failed.
+        CANCELLED: The reindexing task for the index was cancelled.
+    """
+
+    READY = "ready"
+    PENDING = "pending"
+    INDEXING = "indexing"
+    FAILED = "failed"
+    CANCELLED = "cancelled"
+
+
+@dataclass
+class _PropertyIndexTask(_ConfigBase):
+    task_id: Optional[str]
+    status: PropertyIndexTaskStatus
+
+
+PropertyIndexTask = _PropertyIndexTask
+
+
+@dataclass
+class _PropertyIndexStatus(_ConfigBase):
+    type: IndexName  # noqa: A003
+    status: PropertyIndexState
+    progress: Optional[float]
+    task_id: Optional[str]
+    tokenization: Optional[Tokenization]
+    target_tokenization: Optional[Tokenization]
+    algorithm: Optional[str]
+    target_algorithm: Optional[str]
+
+
+PropertyIndexStatus = _PropertyIndexStatus
+
+
+@dataclass
+class _PropertyIndexes(_ConfigBase):
+    name: str
+    data_type: str
+    description: Optional[str]
+    indexes: List[PropertyIndexStatus]
+
+
+PropertyIndexes = _PropertyIndexes
+
+
+@dataclass
+class _CollectionPropertyIndexes(_ConfigBase):
+    collection: str
+    properties: List[PropertyIndexes]
+
+
+CollectionPropertyIndexes = _CollectionPropertyIndexes
+
+
 class _TextAnalyzerConfigCreate(_ConfigCreateModel):
     """Text analysis options for a property.
 

weaviate/collections/classes/config_methods.py

@@ -4,8 +4,11 @@
 from weaviate.collections.classes.config import (
     DataType,
     GenerativeSearches,
+    IndexName,
     PQEncoderDistribution,
     PQEncoderType,
+    PropertyIndexState,
+    PropertyIndexTaskStatus,
     ReplicationDeletionStrategy,
     Rerankers,
     StopwordsPreset,
@@ -19,6 +22,7 @@
     _BQConfig,
     _CollectionConfig,
     _CollectionConfigSimple,
+    _CollectionPropertyIndexes,
     _GenerativeConfig,
     _InvertedIndexConfig,
     _MultiTenancyConfig,
@@ -31,6 +35,9 @@
     _PQConfig,
     _PQEncoderConfig,
     _Property,
+    _PropertyIndexes,
+    _PropertyIndexStatus,
+    _PropertyIndexTask,
     _PropertyVectorizerConfig,
     _ReferenceProperty,
     _ReplicationConfig,
@@ -558,3 +565,49 @@ def _references_from_config(schema: Dict[str, Any]) -> List[_ReferenceProperty]:
         for prop in schema["properties"]
         if not _is_primitive(prop["dataType"])
     ]
+
+
+def _property_index_task_from_json(response: Dict[str, Any]) -> _PropertyIndexTask:
+    return _PropertyIndexTask(
+        task_id=response.get("taskId"),
+        status=PropertyIndexTaskStatus(response["status"]),
+    )
+
+
+def _property_index_status_from_json(index: Dict[str, Any]) -> _PropertyIndexStatus:
+    tokenization = index.get("tokenization")
+    target_tokenization = index.get("targetTokenization")
+    return _PropertyIndexStatus(
+        type=cast(IndexName, index["type"]),
+        status=PropertyIndexState(index["status"]),
+        progress=index.get("progress"),
+        task_id=index.get("taskId"),
+        tokenization=Tokenization(tokenization) if tokenization is not None else None,
+        target_tokenization=(
+            Tokenization(target_tokenization) if target_tokenization is not None else None
+        ),
+        algorithm=index.get("algorithm"),
+        target_algorithm=index.get("targetAlgorithm"),
+    )
+
+
+def _collection_property_indexes_from_json(response: Dict[str, Any]) -> _CollectionPropertyIndexes:
+    properties: List[_PropertyIndexes] = []
+    for prop in response.get("properties") or []:
+        data_type = prop.get("dataType")
+        if isinstance(data_type, list):
+            data_type = data_type[0] if len(data_type) > 0 else ""
+        properties.append(
+            _PropertyIndexes(
+                name=prop["name"],
+                data_type=cast(str, data_type),
+                description=prop.get("description"),
+                indexes=[
+                    _property_index_status_from_json(index) for index in prop.get("indexes") or []
+                ],
+            )
+        )
+    return _CollectionPropertyIndexes(
+        collection=response["collection"],
+        properties=properties,
+    )

weaviate/collections/config/async_.pyi

@@ -1,15 +1,19 @@
-from typing import Dict, List, Literal, Optional, Union, overload
+from typing import Dict, List, Literal, Optional, Sequence, Union, overload
 
 from typing_extensions import deprecated
 
 from weaviate.collections.classes.config import (
     CollectionConfig,
     CollectionConfigSimple,
+    CollectionPropertyIndexes,
     IndexName,
     Property,
+    PropertyIndexStatus,
+    PropertyIndexTask,
     ReferenceProperty,
     ShardStatus,
     ShardTypes,
+    Tokenization,
     _GenerativeProvider,
     _InvertedIndexConfigUpdate,
     _MultiTenancyConfigUpdate,
@@ -90,3 +94,47 @@ class _ConfigCollectionAsync(_ConfigCollectionExecutor[ConnectionAsync]):
         self, *, vector_config: Union[_VectorConfigCreate, List[_VectorConfigCreate]]
     ) -> None: ...
     async def delete_property_index(self, property_name: str, index_name: IndexName) -> bool: ...
+    @overload
+    async def update_property_index(
+        self,
+        property_name: str,
+        index_name: IndexName,
+        *,
+        tokenization: Optional[Tokenization] = None,
+        algorithm: Optional[Literal["blockmax"]] = None,
+        tenants: Optional[Sequence[str]] = None,
+        wait_for_completion: Literal[True],
+    ) -> PropertyIndexStatus: ...
+    @overload
+    async def update_property_index(
+        self,
+        property_name: str,
+        index_name: IndexName,
+        *,
+        tokenization: Optional[Tokenization] = None,
+        algorithm: Optional[Literal["blockmax"]] = None,
+        tenants: Optional[Sequence[str]] = None,
+        wait_for_completion: Literal[False] = False,
+    ) -> PropertyIndexTask: ...
+    @overload
+    async def rebuild_property_index(
+        self,
+        property_name: str,
+        index_name: IndexName,
+        *,
+        tenants: Optional[Sequence[str]] = None,
+        wait_for_completion: Literal[True],
+    ) -> PropertyIndexStatus: ...
+    @overload
+    async def rebuild_property_index(
+        self,
+        property_name: str,
+        index_name: IndexName,
+        *,
+        tenants: Optional[Sequence[str]] = None,
+        wait_for_completion: Literal[False] = False,
+    ) -> PropertyIndexTask: ...
+    async def cancel_property_index_task(
+        self, property_name: str, index_name: IndexName
+    ) -> PropertyIndexTask: ...
+    async def get_property_indexes(self) -> CollectionPropertyIndexes: ...