llama-stack 0.3.4__py3-none-any.whl → 0.4.0__py3-none-any.whl

llama_stack/providers/registry/agents.py

@@ -5,12 +5,12 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import (
+from llama_stack.core.storage.kvstore import kvstore_dependencies
+from llama_stack_api import (
     Api,
     InlineProviderSpec,
     ProviderSpec,
 )
-from llama_stack.providers.utils.kvstore import kvstore_dependencies
 
 
 def available_providers() -> list[ProviderSpec]:
@@ -30,11 +30,15 @@ def available_providers() -> list[ProviderSpec]:
             config_class="llama_stack.providers.inline.agents.meta_reference.MetaReferenceAgentsImplConfig",
             api_dependencies=[
                 Api.inference,
-                Api.safety,
                 Api.vector_io,
                 Api.tool_runtime,
                 Api.tool_groups,
                 Api.conversations,
+                Api.prompts,
+                Api.files,
+            ],
+            optional_api_dependencies=[
+                Api.safety,
             ],
             description="Meta's reference implementation of an agent system that can use tools, access vector databases, and perform complex reasoning tasks.",
         ),

llama_stack/providers/registry/batches.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec
+from llama_stack_api import Api, InlineProviderSpec, ProviderSpec
 
 
 def available_providers() -> list[ProviderSpec]:

llama_stack/providers/registry/datasetio.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import (
+from llama_stack_api import (
     Api,
     InlineProviderSpec,
     ProviderSpec,

llama_stack/providers/registry/eval.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
+from llama_stack_api import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
 
 
 def available_providers() -> list[ProviderSpec]:

llama_stack/{apis/datasets/__init__.py → providers/registry/file_processors.py}

@@ -4,4 +4,8 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from .datasets import *
+from llama_stack_api import ProviderSpec
+
+
+def available_providers() -> list[ProviderSpec]:
+    return []

llama_stack/providers/registry/files.py

@@ -4,8 +4,8 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
-from llama_stack.providers.utils.sqlstore.sqlstore import sql_store_pip_packages
+from llama_stack.core.storage.sqlstore.sqlstore import sql_store_pip_packages
+from llama_stack_api import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
 
 
 def available_providers() -> list[ProviderSpec]:
@@ -28,4 +28,13 @@ def available_providers() -> list[ProviderSpec]:
             config_class="llama_stack.providers.remote.files.s3.config.S3FilesImplConfig",
             description="AWS S3-based file storage provider for scalable cloud file management with metadata persistence.",
         ),
+        RemoteProviderSpec(
+            api=Api.files,
+            provider_type="remote::openai",
+            adapter_type="openai",
+            pip_packages=["openai"] + sql_store_pip_packages,
+            module="llama_stack.providers.remote.files.openai",
+            config_class="llama_stack.providers.remote.files.openai.config.OpenAIFilesImplConfig",
+            description="OpenAI Files API provider for managing files through OpenAI's native file storage service.",
+        ),
     ]

llama_stack/providers/registry/inference.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import (
+from llama_stack_api import (
     Api,
     InlineProviderSpec,
     ProviderSpec,
@@ -61,6 +61,7 @@ def available_providers() -> list[ProviderSpec]:
             pip_packages=[],
             module="llama_stack.providers.remote.inference.cerebras",
             config_class="llama_stack.providers.remote.inference.cerebras.CerebrasImplConfig",
+            provider_data_validator="llama_stack.providers.remote.inference.cerebras.config.CerebrasProviderDataValidator",
             description="Cerebras inference provider for running models on Cerebras Cloud platform.",
         ),
         RemoteProviderSpec(
@@ -137,10 +138,11 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter_type="bedrock",
             provider_type="remote::bedrock",
-            pip_packages=["boto3"],
+            pip_packages=[],
             module="llama_stack.providers.remote.inference.bedrock",
             config_class="llama_stack.providers.remote.inference.bedrock.BedrockConfig",
-            description="AWS Bedrock inference provider for accessing various AI models through AWS's managed service.",
+            provider_data_validator="llama_stack.providers.remote.inference.bedrock.config.BedrockProviderDataValidator",
+            description="AWS Bedrock inference provider using OpenAI compatible endpoint.",
         ),
         RemoteProviderSpec(
             api=Api.inference,
@@ -149,6 +151,7 @@ def available_providers() -> list[ProviderSpec]:
             pip_packages=["databricks-sdk"],
             module="llama_stack.providers.remote.inference.databricks",
             config_class="llama_stack.providers.remote.inference.databricks.DatabricksImplConfig",
+            provider_data_validator="llama_stack.providers.remote.inference.databricks.config.DatabricksProviderDataValidator",
             description="Databricks inference provider for running models on Databricks' unified analytics platform.",
         ),
         RemoteProviderSpec(
@@ -158,6 +161,7 @@ def available_providers() -> list[ProviderSpec]:
             pip_packages=[],
             module="llama_stack.providers.remote.inference.nvidia",
             config_class="llama_stack.providers.remote.inference.nvidia.NVIDIAConfig",
+            provider_data_validator="llama_stack.providers.remote.inference.nvidia.config.NVIDIAProviderDataValidator",
             description="NVIDIA inference provider for accessing NVIDIA NIM models and AI services.",
         ),
         RemoteProviderSpec(
@@ -167,6 +171,7 @@ def available_providers() -> list[ProviderSpec]:
             pip_packages=[],
             module="llama_stack.providers.remote.inference.runpod",
             config_class="llama_stack.providers.remote.inference.runpod.RunpodImplConfig",
+            provider_data_validator="llama_stack.providers.remote.inference.runpod.config.RunpodProviderDataValidator",
             description="RunPod inference provider for running models on RunPod's cloud GPU platform.",
         ),
         RemoteProviderSpec(
@@ -292,6 +297,20 @@ Available Models:
 Azure OpenAI inference provider for accessing GPT models and other Azure services.
 Provider documentation
 https://learn.microsoft.com/en-us/azure/ai-foundry/openai/overview
+""",
+        ),
+        RemoteProviderSpec(
+            api=Api.inference,
+            provider_type="remote::oci",
+            adapter_type="oci",
+            pip_packages=["oci"],
+            module="llama_stack.providers.remote.inference.oci",
+            config_class="llama_stack.providers.remote.inference.oci.config.OCIConfig",
+            provider_data_validator="llama_stack.providers.remote.inference.oci.config.OCIProviderDataValidator",
+            description="""
+Oracle Cloud Infrastructure (OCI) Generative AI inference provider for accessing OCI's Generative AI Platform-as-a-Service models.
+Provider documentation
+https://docs.oracle.com/en-us/iaas/Content/generative-ai/home.htm
 """,
         ),
     ]

llama_stack/providers/registry/post_training.py

@@ -7,7 +7,7 @@
 
 from typing import cast
 
-from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
+from llama_stack_api import Api, InlineProviderSpec, ProviderSpec, RemoteProviderSpec
 
 # We provide two versions of these providers so that distributions can package the appropriate version of torch.
 # The CPU version is used for distributions that don't have GPU support -- they result in smaller container images.

llama_stack/providers/registry/safety.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import (
+from llama_stack_api import (
     Api,
     InlineProviderSpec,
     ProviderSpec,

llama_stack/providers/registry/scoring.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec
+from llama_stack_api import Api, InlineProviderSpec, ProviderSpec
 
 
 def available_providers() -> list[ProviderSpec]:

llama_stack/providers/registry/tool_runtime.py

@@ -5,13 +5,13 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import (
+from llama_stack.providers.registry.vector_io import DEFAULT_VECTOR_IO_DEPS
+from llama_stack_api import (
     Api,
     InlineProviderSpec,
     ProviderSpec,
     RemoteProviderSpec,
 )
-from llama_stack.providers.registry.vector_io import DEFAULT_VECTOR_IO_DEPS
 
 
 def available_providers() -> list[ProviderSpec]:

llama_stack/providers/registry/vector_io.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from llama_stack.providers.datatypes import (
+from llama_stack_api import (
     Api,
     InlineProviderSpec,
     ProviderSpec,
@@ -163,14 +163,14 @@ The SQLite-vec provider supports three search modes:
 Example with hybrid search:
 ```python
 response = await vector_io.query_chunks(
-    vector_db_id="my_db",
+    vector_store_id="my_db",
     query="your query here",
     params={"mode": "hybrid", "max_chunks": 3, "score_threshold": 0.7},
 )
 
 # Using RRF ranker
 response = await vector_io.query_chunks(
-    vector_db_id="my_db",
+    vector_store_id="my_db",
     query="your query here",
     params={
         "mode": "hybrid",
@@ -182,7 +182,7 @@ response = await vector_io.query_chunks(
 
 # Using weighted ranker
 response = await vector_io.query_chunks(
-    vector_db_id="my_db",
+    vector_store_id="my_db",
     query="your query here",
     params={
         "mode": "hybrid",
@@ -196,7 +196,7 @@ response = await vector_io.query_chunks(
 Example with explicit vector search:
 ```python
 response = await vector_io.query_chunks(
-    vector_db_id="my_db",
+    vector_store_id="my_db",
     query="your query here",
     params={"mode": "vector", "max_chunks": 3, "score_threshold": 0.7},
 )
@@ -205,7 +205,7 @@ response = await vector_io.query_chunks(
 Example with keyword search:
 ```python
 response = await vector_io.query_chunks(
-    vector_db_id="my_db",
+    vector_store_id="my_db",
     query="your query here",
     params={"mode": "keyword", "max_chunks": 3, "score_threshold": 0.7},
 )
@@ -244,7 +244,7 @@ Two ranker types are supported:
 Example using RAGQueryConfig with different search modes:
 
 ```python
-from llama_stack.apis.tools import RAGQueryConfig, RRFRanker, WeightedRanker
+from llama_stack_api import RAGQueryConfig, RRFRanker, WeightedRanker
 
 # Vector search
 config = RAGQueryConfig(mode="vector", max_chunks=5)

llama_stack/providers/remote/datasetio/huggingface/huggingface.py

@@ -6,12 +6,9 @@
 from typing import Any
 from urllib.parse import parse_qs, urlparse
 
-from llama_stack.apis.common.responses import PaginatedResponse
-from llama_stack.apis.datasetio import DatasetIO
-from llama_stack.apis.datasets import Dataset
-from llama_stack.providers.datatypes import DatasetsProtocolPrivate
-from llama_stack.providers.utils.kvstore import kvstore_impl
+from llama_stack.core.storage.kvstore import kvstore_impl
 from llama_stack.providers.utils.pagination import paginate_records
+from llama_stack_api import Dataset, DatasetIO, DatasetsProtocolPrivate, PaginatedResponse
 
 from .config import HuggingfaceDatasetIOConfig
 

llama_stack/providers/remote/datasetio/nvidia/datasetio.py

@@ -8,10 +8,7 @@ from typing import Any
 
 import aiohttp
 
-from llama_stack.apis.common.content_types import URL
-from llama_stack.apis.common.responses import PaginatedResponse
-from llama_stack.apis.common.type_system import ParamType
-from llama_stack.apis.datasets import Dataset
+from llama_stack_api import URL, Dataset, PaginatedResponse, ParamType
 
 from .config import NvidiaDatasetIOConfig
 

llama_stack/providers/remote/eval/nvidia/eval.py

@@ -7,17 +7,23 @@ from typing import Any
 
 import requests
 
-from llama_stack.apis.agents import Agents
-from llama_stack.apis.benchmarks import Benchmark
-from llama_stack.apis.datasetio import DatasetIO
-from llama_stack.apis.datasets import Datasets
-from llama_stack.apis.inference import Inference
-from llama_stack.apis.scoring import Scoring, ScoringResult
-from llama_stack.providers.datatypes import BenchmarksProtocolPrivate
 from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper
+from llama_stack_api import (
+    Agents,
+    Benchmark,
+    BenchmarkConfig,
+    BenchmarksProtocolPrivate,
+    DatasetIO,
+    Datasets,
+    Eval,
+    EvaluateResponse,
+    Inference,
+    Job,
+    JobStatus,
+    Scoring,
+    ScoringResult,
+)
 
-from .....apis.common.job_types import Job, JobStatus
-from .....apis.eval.eval import BenchmarkConfig, Eval, EvaluateResponse
 from .config import NVIDIAEvalConfig
 
 DEFAULT_NAMESPACE = "nvidia"

llama_stack/providers/remote/files/openai/__init__.py

@@ -0,0 +1,19 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from typing import Any
+
+from llama_stack.core.datatypes import AccessRule, Api
+
+from .config import OpenAIFilesImplConfig
+
+
+async def get_adapter_impl(config: OpenAIFilesImplConfig, deps: dict[Api, Any], policy: list[AccessRule] | None = None):
+    from .files import OpenAIFilesImpl
+
+    impl = OpenAIFilesImpl(config, policy or [])
+    await impl.initialize()
+    return impl

llama_stack/providers/remote/files/openai/config.py

@@ -0,0 +1,28 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from typing import Any
+
+from pydantic import BaseModel, Field
+
+from llama_stack.core.storage.datatypes import SqlStoreReference
+
+
+class OpenAIFilesImplConfig(BaseModel):
+    """Configuration for OpenAI Files API provider."""
+
+    api_key: str = Field(description="OpenAI API key for authentication")
+    metadata_store: SqlStoreReference = Field(description="SQL store configuration for file metadata")
+
+    @classmethod
+    def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
+        return {
+            "api_key": "${env.OPENAI_API_KEY}",
+            "metadata_store": SqlStoreReference(
+                backend="sql_default",
+                table_name="openai_files_metadata",
+            ).model_dump(exclude_none=True),
+        }

llama_stack/providers/remote/files/openai/files.py

@@ -0,0 +1,253 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from datetime import UTC, datetime
+from typing import Any
+
+from fastapi import Response, UploadFile
+
+from llama_stack.core.access_control.datatypes import Action
+from llama_stack.core.datatypes import AccessRule
+from llama_stack.core.storage.sqlstore.authorized_sqlstore import AuthorizedSqlStore
+from llama_stack.core.storage.sqlstore.sqlstore import sqlstore_impl
+from llama_stack_api import (
+    DeleteFileRequest,
+    ExpiresAfter,
+    Files,
+    ListFilesRequest,
+    ListOpenAIFileResponse,
+    OpenAIFileDeleteResponse,
+    OpenAIFileObject,
+    OpenAIFilePurpose,
+    Order,
+    ResourceNotFoundError,
+    RetrieveFileContentRequest,
+    RetrieveFileRequest,
+    UploadFileRequest,
+)
+from llama_stack_api.internal.sqlstore import ColumnDefinition, ColumnType
+from openai import OpenAI
+
+from .config import OpenAIFilesImplConfig
+
+
+def _make_file_object(
+    *,
+    id: str,
+    filename: str,
+    purpose: str,
+    bytes: int,
+    created_at: int,
+    expires_at: int,
+    **kwargs: Any,
+) -> OpenAIFileObject:
+    """
+    Construct an OpenAIFileObject and normalize expires_at.
+
+    If expires_at is greater than the max we treat it as no-expiration and
+    return None for expires_at.
+    """
+    obj = OpenAIFileObject(
+        id=id,
+        filename=filename,
+        purpose=OpenAIFilePurpose(purpose),
+        bytes=bytes,
+        created_at=created_at,
+        expires_at=expires_at,
+    )
+
+    if obj.expires_at is not None and obj.expires_at > (obj.created_at + ExpiresAfter.MAX):
+        obj.expires_at = None  # type: ignore
+
+    return obj
+
+
+class OpenAIFilesImpl(Files):
+    """OpenAI Files API implementation."""
+
+    def __init__(self, config: OpenAIFilesImplConfig, policy: list[AccessRule]) -> None:
+        self._config = config
+        self.policy = policy
+        self._client: OpenAI | None = None
+        self._sql_store: AuthorizedSqlStore | None = None
+
+    def _now(self) -> int:
+        """Return current UTC timestamp as int seconds."""
+        return int(datetime.now(UTC).timestamp())
+
+    async def _get_file(
+        self, file_id: str, return_expired: bool = False, action: Action = Action.READ
+    ) -> dict[str, Any]:
+        where: dict[str, str | dict] = {"id": file_id}
+        if not return_expired:
+            where["expires_at"] = {">": self._now()}
+        if not (row := await self.sql_store.fetch_one("openai_files", where=where, action=action)):
+            raise ResourceNotFoundError(file_id, "File", "files.list()")
+        return row
+
+    async def _delete_file(self, file_id: str) -> None:
+        """Delete a file from OpenAI and the database."""
+        try:
+            self.client.files.delete(file_id)
+        except Exception as e:
+            # If file doesn't exist on OpenAI side, just remove from metadata store
+            if "not found" not in str(e).lower():
+                raise RuntimeError(f"Failed to delete file from OpenAI: {e}") from e
+
+        await self.sql_store.delete("openai_files", where={"id": file_id})
+
+    async def _delete_if_expired(self, file_id: str) -> None:
+        """If the file exists and is expired, delete it."""
+        if row := await self._get_file(file_id, return_expired=True):
+            if (expires_at := row.get("expires_at")) and expires_at <= self._now():
+                await self._delete_file(file_id)
+
+    async def initialize(self) -> None:
+        self._client = OpenAI(api_key=self._config.api_key)
+
+        self._sql_store = AuthorizedSqlStore(sqlstore_impl(self._config.metadata_store), self.policy)
+        await self._sql_store.create_table(
+            "openai_files",
+            {
+                "id": ColumnDefinition(type=ColumnType.STRING, primary_key=True),
+                "filename": ColumnType.STRING,
+                "purpose": ColumnType.STRING,
+                "bytes": ColumnType.INTEGER,
+                "created_at": ColumnType.INTEGER,
+                "expires_at": ColumnType.INTEGER,
+            },
+        )
+
+    async def shutdown(self) -> None:
+        pass
+
+    @property
+    def client(self) -> OpenAI:
+        assert self._client is not None, "Provider not initialized"
+        return self._client
+
+    @property
+    def sql_store(self) -> AuthorizedSqlStore:
+        assert self._sql_store is not None, "Provider not initialized"
+        return self._sql_store
+
+    async def openai_upload_file(
+        self,
+        request: UploadFileRequest,
+        file: UploadFile,
+    ) -> OpenAIFileObject:
+        purpose = request.purpose
+        expires_after = request.expires_after
+
+        filename = getattr(file, "filename", None) or "uploaded_file"
+        content = await file.read()
+        file_size = len(content)
+
+        created_at = self._now()
+
+        expires_at = created_at + ExpiresAfter.MAX * 42
+        if purpose == OpenAIFilePurpose.BATCH:
+            expires_at = created_at + ExpiresAfter.MAX
+
+        if expires_after is not None:
+            expires_at = created_at + expires_after.seconds
+
+        try:
+            from io import BytesIO
+
+            file_obj = BytesIO(content)
+            file_obj.name = filename
+
+            response = self.client.files.create(
+                file=file_obj,
+                purpose=purpose.value,
+            )
+
+            file_id = response.id
+
+            entry: dict[str, Any] = {
+                "id": file_id,
+                "filename": filename,
+                "purpose": purpose.value,
+                "bytes": file_size,
+                "created_at": created_at,
+                "expires_at": expires_at,
+            }
+
+            await self.sql_store.insert("openai_files", entry)
+
+            return _make_file_object(**entry)
+
+        except Exception as e:
+            raise RuntimeError(f"Failed to upload file to OpenAI: {e}") from e
+
+    async def openai_list_files(
+        self,
+        request: ListFilesRequest,
+    ) -> ListOpenAIFileResponse:
+        after = request.after
+        limit = request.limit
+        order = request.order
+        purpose = request.purpose
+
+        if not order:
+            order = Order.desc
+
+        where_conditions: dict[str, Any] = {"expires_at": {">": self._now()}}
+        if purpose:
+            where_conditions["purpose"] = purpose.value
+
+        paginated_result = await self.sql_store.fetch_all(
+            table="openai_files",
+            where=where_conditions,
+            order_by=[("created_at", order.value)],
+            cursor=("id", after) if after else None,
+            limit=limit,
+        )
+
+        files = [_make_file_object(**row) for row in paginated_result.data]
+
+        return ListOpenAIFileResponse(
+            data=files,
+            has_more=paginated_result.has_more,
+            first_id=files[0].id if files else "",
+            last_id=files[-1].id if files else "",
+        )
+
+    async def openai_retrieve_file(self, request: RetrieveFileRequest) -> OpenAIFileObject:
+        file_id = request.file_id
+        await self._delete_if_expired(file_id)
+        row = await self._get_file(file_id)
+        return _make_file_object(**row)
+
+    async def openai_delete_file(self, request: DeleteFileRequest) -> OpenAIFileDeleteResponse:
+        file_id = request.file_id
+        await self._delete_if_expired(file_id)
+        _ = await self._get_file(file_id, action=Action.DELETE)
+        await self._delete_file(file_id)
+        return OpenAIFileDeleteResponse(id=file_id, deleted=True)
+
+    async def openai_retrieve_file_content(self, request: RetrieveFileContentRequest) -> Response:
+        file_id = request.file_id
+        await self._delete_if_expired(file_id)
+
+        row = await self._get_file(file_id)
+
+        try:
+            response = self.client.files.content(file_id)
+            file_content = response.content
+
+        except Exception as e:
+            if "not found" in str(e).lower():
+                await self._delete_file(file_id)
+                raise ResourceNotFoundError(file_id, "File", "files.list()") from e
+            raise RuntimeError(f"Failed to download file from OpenAI: {e}") from e
+
+        return Response(
+            content=file_content,
+            media_type="application/octet-stream",
+            headers={"Content-Disposition": f'attachment; filename="{row["filename"]}"'},
+        )