feat(storage): add vector and analytics buckets support

Makefile

@@ -32,7 +32,7 @@ help::
 
 mypy: $(call FORALL_PKGS,mypy)
 help::
-	@echo "  mypy          -- Run mypy on all files"
+	@echo "  mypy           -- Run mypy on all files"
 
 ruff:
 	@uv run ruff check --fix

src/storage/run-unasync.py

@@ -5,7 +5,15 @@
 paths = Path("src/storage3").glob("**/*.py")
 tests = Path("tests").glob("**/*.py")
 
-rules = (unasync._DEFAULT_RULE,)
+rules = (
+    unasync.Rule(
+        fromdir="/_async/",
+        todir="/_sync/",
+        additional_replacements={"AsyncClient": "Client"},
+    ),
+    unasync._DEFAULT_RULE,
+)
+
 
 files = [str(p) for p in list(paths) + list(tests)]
 

src/storage/src/storage3/_async/analytics.py

@@ -0,0 +1,52 @@
+from typing import List, Optional
+
+from httpx import QueryParams
+
+from ..types import (
+    AnalyticsBucket,
+    AnalyticsBucketDeleteResponse,
+    AnalyticsBucketsParser,
+    SortColumn,
+    SortOrder,
+)
+from .request import AsyncRequestBuilder
+
+
+class AsyncStorageAnalyticsClient:
+    def __init__(self, request: AsyncRequestBuilder) -> None:
+        self._request = request
+
+    async def create(self, bucket_name: str) -> AnalyticsBucket:
+        body = {"name": bucket_name}
+        data = await self._request.send(http_method="POST", path=["bucket"], body=body)
+        return AnalyticsBucket.model_validate_json(data.content)
+
+    async def list(
+        self,
+        limit: Optional[int] = None,
+        offset: Optional[int] = None,
+        sort_column: Optional[SortColumn] = None,
+        sort_order: Optional[SortOrder] = None,
+        search: Optional[str] = None,
+    ) -> List[AnalyticsBucket]:
+        params = dict(
+            limit=limit,
+            offset=offset,
+            sort_column=sort_column,
+            sort_order=sort_order,
+            search=search,
+        )
+        filtered_params = QueryParams(
+            **{k: v for k, v in params.items() if v is not None}
+        )
+        data = await self._request.send(
+            http_method="GET", path=["bucket"], query_params=filtered_params
+        )
+        print(data.content)
+        return AnalyticsBucketsParser.validate_json(data.content)
+
+    async def delete(self, bucket_name: str) -> AnalyticsBucketDeleteResponse:
+        data = await self._request.send(
+            http_method="DELETE", path=["bucket", bucket_name]
+        )
+        return AnalyticsBucketDeleteResponse.model_validate_json(data.content)

src/storage/src/storage3/_async/client.py

@@ -8,8 +8,11 @@
 from storage3.constants import DEFAULT_TIMEOUT
 
 from ..version import __version__
+from .analytics import AsyncStorageAnalyticsClient
 from .bucket import AsyncStorageBucketAPI
 from .file_api import AsyncBucketProxy
+from .request import AsyncRequestBuilder
+from .vectors import AsyncStorageVectorsClient
 
 __all__ = [
     "AsyncStorageClient",
@@ -80,3 +83,18 @@ def from_(self, id: str) -> AsyncBucketProxy:
             The unique identifier of the bucket
         """
         return AsyncBucketProxy(id, self._base_url, self._headers, self._client)
+
+    def vectors(self) -> AsyncStorageVectorsClient:
+        return AsyncStorageVectorsClient(
+            url=self._base_url.joinpath("v1", "vector"),
+            headers=self._headers,
+            session=self.session,
+        )
+
+    def analytics(self) -> AsyncStorageAnalyticsClient:
+        request = AsyncRequestBuilder(
+            session=self.session,
+            headers=self._headers,
+            base_url=self._base_url.joinpath("v1", "iceberg"),
+        )
+        return AsyncStorageAnalyticsClient(request=request)

src/storage/src/storage3/_async/request.py

@@ -0,0 +1,47 @@
+from typing import Optional
+
+from httpx import AsyncClient, Headers, HTTPStatusError, QueryParams, Response
+from pydantic import ValidationError
+from yarl import URL
+
+from ..exceptions import StorageApiError, VectorBucketErrorMessage
+from ..types import JSON, RequestMethod
+
+
+class AsyncRequestBuilder:
+    def __init__(self, session: AsyncClient, base_url: URL, headers: Headers) -> None:
+        self._session = session
+        self._base_url = base_url
+        self.headers = headers
+
+    async def send(
+        self,
+        http_method: RequestMethod,
+        path: list[str],
+        body: JSON = None,
+        query_params: Optional[QueryParams] = None,
+    ) -> Response:
+        response = await self._session.request(
+            method=http_method,
+            json=body,
+            url=str(self._base_url.joinpath(*path)),
+            headers=self.headers,
+            params=query_params or QueryParams(),
+        )
+        try:
+            response.raise_for_status()
+            return response
+        except HTTPStatusError as exc:
+            try:
+                error = VectorBucketErrorMessage.model_validate_json(response.content)
+                raise StorageApiError(
+                    message=error.message,
+                    code=error.code or "400",
+                    status=error.statusCode,
+                ) from exc
+            except ValidationError as exc:
+                raise StorageApiError(
+                    message=f"The request failed, but could not parse error message response:'{response.text}'",
+                    code="LibraryError",
+                    status=response.status_code,
+                ) from exc

src/storage/src/storage3/_async/vectors.py

@@ -0,0 +1,213 @@
+from __future__ import annotations
+
+from typing import List, Optional
+
+from httpx import AsyncClient, Headers
+from yarl import URL
+
+from ..exceptions import StorageApiError, VectorBucketException
+from ..types import (
+    JSON,
+    DistanceMetric,
+    GetVectorBucketResponse,
+    GetVectorIndexResponse,
+    GetVectorsResponse,
+    ListVectorBucketsResponse,
+    ListVectorIndexesResponse,
+    ListVectorsResponse,
+    MetadataConfiguration,
+    QueryVectorsResponse,
+    VectorBucket,
+    VectorData,
+    VectorFilter,
+    VectorIndex,
+    VectorMatch,
+    VectorObject,
+)
+from .request import AsyncRequestBuilder
+
+
+# used to not send non-required values as `null`
+# for they cannot be null
+def remove_none(**kwargs: JSON) -> JSON:
+    return {key: val for key, val in kwargs.items() if val is not None}
+
+
+class AsyncVectorBucketScope:
+    def __init__(self, request: AsyncRequestBuilder, bucket_name: str) -> None:
+        self._request = request
+        self._bucket_name = bucket_name
+
+    def with_metadata(self, **data: JSON) -> JSON:
+        return remove_none(vectorBucketName=self._bucket_name, **data)
+
+    async def create_index(
+        self,
+        index_name: str,
+        dimension: int,
+        distance_metric: DistanceMetric,
+        data_type: str,
+        metadata: Optional[MetadataConfiguration] = None,
+    ) -> None:
+        body = self.with_metadata(
+            indexName=index_name,
+            dimension=dimension,
+            distanceMetric=distance_metric,
+            dataType=data_type,
+            metadataConfiguration=dict(metadata) if metadata else None,
+        )
+        await self._request.send(http_method="POST", path=["CreateIndex"], body=body)
+
+    async def get_index(self, index_name: str) -> Optional[VectorIndex]:
+        body = self.with_metadata(indexName=index_name)
+        try:
+            data = await self._request.send(
+                http_method="POST", path=["GetIndex"], body=body
+            )
+            return GetVectorIndexResponse.model_validate_json(data.content).index
+        except StorageApiError:
+            return None
+
+    async def list_indexes(
+        self,
+        next_token: Optional[str] = None,
+        max_results: Optional[int] = None,
+        prefix: Optional[str] = None,
+    ) -> ListVectorIndexesResponse:
+        body = self.with_metadata(
+            next_token=next_token, max_results=max_results, prefix=prefix
+        )
+        data = await self._request.send(
+            http_method="POST", path=["ListIndexes"], body=body
+        )
+        return ListVectorIndexesResponse.model_validate_json(data.content)
+
+    async def delete_index(self, index_name: str) -> None:
+        body = self.with_metadata(indexName=index_name)
+        await self._request.send(http_method="POST", path=["DeleteIndex"], body=body)
+
+    def index(self, index_name: str) -> AsyncVectorIndexScope:
+        return AsyncVectorIndexScope(self._request, self._bucket_name, index_name)
+
+
+class AsyncVectorIndexScope:
+    def __init__(
+        self, request: AsyncRequestBuilder, bucket_name: str, index_name: str
+    ) -> None:
+        self._request = request
+        self._bucket_name = bucket_name
+        self._index_name = index_name
+
+    def with_metadata(self, **data: JSON) -> JSON:
+        return remove_none(
+            vectorBucketName=self._bucket_name,
+            indexName=self._index_name,
+            **data,
+        )
+
+    async def put(self, vectors: List[VectorObject]) -> None:
+        body = self.with_metadata(vectors=[v.as_json() for v in vectors])
+        await self._request.send(http_method="POST", path=["PutVectors"], body=body)
+
+    async def get(
+        self, *keys: str, return_data: bool = True, return_metadata: bool = True
+    ) -> List[VectorMatch]:
+        body = self.with_metadata(
+            keys=keys, returnData=return_data, returnMetadata=return_metadata
+        )
+        data = await self._request.send(
+            http_method="POST", path=["GetVectors"], body=body
+        )
+        return GetVectorsResponse.model_validate_json(data.content).vectors
+
+    async def list(
+        self,
+        max_results: Optional[int] = None,
+        next_token: Optional[str] = None,
+        return_data: bool = True,
+        return_metadata: bool = True,
+        segment_count: Optional[int] = None,
+        segment_index: Optional[int] = None,
+    ) -> ListVectorsResponse:
+        body = self.with_metadata(
+            maxResults=max_results,
+            nextToken=next_token,
+            returnData=return_data,
+            returnMetadata=return_metadata,
+            segmentCount=segment_count,
+            segmentIndex=segment_index,
+        )
+        data = await self._request.send(
+            http_method="POST", path=["ListVectors"], body=body
+        )
+        return ListVectorsResponse.model_validate_json(data.content)
+
+    async def query(
+        self,
+        query_vector: VectorData,
+        topK: Optional[int] = None,
+        filter: Optional[VectorFilter] = None,
+        return_distance: bool = True,
+        return_metadata: bool = True,
+    ) -> QueryVectorsResponse:
+        body = self.with_metadata(
+            queryVector=dict(query_vector),
+            topK=topK,
+            filter=filter,
+            returnDistance=return_distance,
+            returnMetadata=return_metadata,
+        )
+        data = await self._request.send(
+            http_method="POST", path=["QueryVectors"], body=body
+        )
+        return QueryVectorsResponse.model_validate_json(data.content)
+
+    async def delete(self, keys: List[str]) -> None:
+        if 1 < len(keys) or len(keys) > 500:
+            raise VectorBucketException("Keys batch size must be between 1 and 500.")
+        body = self.with_metadata(keys=keys)
+        await self._request.send(http_method="POST", path=["DeleteVectors"], body=body)
+
+
+class AsyncStorageVectorsClient:
+    def __init__(self, url: URL, headers: Headers, session: AsyncClient) -> None:
+        self._request = AsyncRequestBuilder(session, base_url=URL(url), headers=headers)
+
+    def from_(self, bucket_name: str) -> AsyncVectorBucketScope:
+        return AsyncVectorBucketScope(self._request, bucket_name)
+
+    async def create_bucket(self, bucket_name: str) -> None:
+        body = {"vectorBucketName": bucket_name}
+        await self._request.send(
+            http_method="POST", path=["CreateVectorBucket"], body=body
+        )
+
+    async def get_bucket(self, bucket_name: str) -> Optional[VectorBucket]:
+        body = {"vectorBucketName": bucket_name}
+        try:
+            data = await self._request.send(
+                http_method="POST", path=["GetVectorBucket"], body=body
+            )
+            return GetVectorBucketResponse.model_validate_json(
+                data.content
+            ).vectorBucket
+        except StorageApiError:
+            return None
+
+    async def list_buckets(
+        self,
+        prefix: Optional[str] = None,
+        max_results: Optional[int] = None,
+        next_token: Optional[str] = None,
+    ) -> ListVectorBucketsResponse:
+        body = remove_none(prefix=prefix, maxResults=max_results, nextToken=next_token)
+        data = await self._request.send(
+            http_method="POST", path=["ListVectorBuckets"], body=body
+        )
+        return ListVectorBucketsResponse.model_validate_json(data.content)
+
+    async def delete_bucket(self, bucket_name: str) -> None:
+        body = {"vectorBucketName": bucket_name}
+        await self._request.send(
+            http_method="POST", path=["DeleteVectorBucket"], body=body
+        )