llama-stack 0.4.1__tar.gz → 0.4.2__tar.gz

{llama_stack-0.4.1/src/llama_stack.egg-info → llama_stack-0.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llama_stack
-Version: 0.4.1
+Version: 0.4.2
 Summary: Llama Stack
 Author-email: Meta Llama <llama-oss@meta.com>
 License: MIT
@@ -46,7 +46,7 @@ Requires-Dist: psycopg2-binary
 Requires-Dist: tornado>=6.5.3
 Requires-Dist: urllib3>=2.6.3
 Provides-Extra: client
-Requires-Dist: llama-stack-client==0.4.1; extra == "client"
+Requires-Dist: llama-stack-client==0.4.2; extra == "client"
 Dynamic: license-file
 
 # Llama Stack

{llama_stack-0.4.1 → llama_stack-0.4.2}/pyproject.toml

@@ -7,7 +7,7 @@ required-version = ">=0.7.0"
 
 [project]
 name = "llama_stack"
-version = "0.4.1"
+version = "0.4.2"
 authors = [{ name = "Meta Llama", email = "llama-oss@meta.com" }]
 description = "Llama Stack"
 readme = "README.md"
@@ -57,7 +57,7 @@ dependencies = [
 
 [project.optional-dependencies]
 client = [
-    "llama-stack-client==0.4.1",
+    "llama-stack-client==0.4.2",
 ]
 
 [dependency-groups]
@@ -106,7 +106,7 @@ type_checking = [
     "lm-format-enforcer",
     "mcp",
     "ollama",
-    "llama-stack-client==0.4.1",
+    "llama-stack-client==0.4.2",
 ]
 # These are the dependencies required for running unit tests.
 unit = [

{llama_stack-0.4.1 → llama_stack-0.4.2}/src/llama_stack/cli/stack/run.py

@@ -202,6 +202,9 @@ class StackRun(Subcommand):
         # Set the config file in environment so create_app can find it
         os.environ["LLAMA_STACK_CONFIG"] = str(config_file)
 
+        # disable together banner that spams llama stack run every time
+        os.environ["TOGETHER_NO_BANNER"] = "1"
+
         uvicorn_config = {
             "factory": True,
             "host": host,

{llama_stack-0.4.1 → llama_stack-0.4.2}/src/llama_stack/core/stack.py

@@ -53,6 +53,7 @@ from llama_stack_api import (
     PostTraining,
     Prompts,
     Providers,
+    RegisterBenchmarkRequest,
     Safety,
     Scoring,
     ScoringFunctions,
@@ -61,6 +62,7 @@ from llama_stack_api import (
     ToolRuntime,
     VectorIO,
 )
+from llama_stack_api.datasets import RegisterDatasetRequest
 
 logger = get_logger(name=__name__, category="core")
 
@@ -91,18 +93,21 @@ class LlamaStack(
     pass
 
 
+# Resources to register based on configuration.
+# If a request class is specified, the configuration object will be converted to this class before invoking the registration method.
 RESOURCES = [
-    ("models", Api.models, "register_model", "list_models"),
-    ("shields", Api.shields, "register_shield", "list_shields"),
-    ("datasets", Api.datasets, "register_dataset", "list_datasets"),
+    ("models", Api.models, "register_model", "list_models", None),
+    ("shields", Api.shields, "register_shield", "list_shields", None),
+    ("datasets", Api.datasets, "register_dataset", "list_datasets", RegisterDatasetRequest),
     (
         "scoring_fns",
         Api.scoring_functions,
         "register_scoring_function",
         "list_scoring_functions",
+        None,
     ),
-    ("benchmarks", Api.benchmarks, "register_benchmark", "list_benchmarks"),
-    ("tool_groups", Api.tool_groups, "register_tool_group", "list_tool_groups"),
+    ("benchmarks", Api.benchmarks, "register_benchmark", "list_benchmarks", RegisterBenchmarkRequest),
+    ("tool_groups", Api.tool_groups, "register_tool_group", "list_tool_groups", None),
 ]
 
 
@@ -199,7 +204,7 @@ async def invoke_with_optional_request(method: Any) -> Any:
 
 
 async def register_resources(run_config: StackConfig, impls: dict[Api, Any]):
-    for rsrc, api, register_method, list_method in RESOURCES:
+    for rsrc, api, register_method, list_method, request_class in RESOURCES:
         objects = getattr(run_config.registered_resources, rsrc)
         if api not in impls:
             continue
@@ -213,10 +218,17 @@ async def register_resources(run_config: StackConfig, impls: dict[Api, Any]):
                     continue
                 logger.debug(f"registering {rsrc.capitalize()} {obj} for provider {obj.provider_id}")
 
-            # we want to maintain the type information in arguments to method.
-            # instead of method(**obj.model_dump()), which may convert a typed attr to a dict,
-            # we use model_dump() to find all the attrs and then getattr to get the still typed value.
-            await method(**{k: getattr(obj, k) for k in obj.model_dump().keys()})
+            # TODO: Once all register methods are migrated to accept request objects,
+            # remove this conditional and always use the request_class pattern.
+            if request_class is not None:
+                request = request_class(**obj.model_dump())
+                await method(request)
+            else:
+                # we want to maintain the type information in arguments to method.
+                # instead of method(**obj.model_dump()), which may convert a typed attr to a dict,
+                # we use model_dump() to find all the attrs and then getattr to get the still typed
+                # value.
+                await method(**{k: getattr(obj, k) for k in obj.model_dump().keys()})
 
         method = getattr(impls[api], list_method)
         response = await invoke_with_optional_request(method)

{llama_stack-0.4.1 → llama_stack-0.4.2}/src/llama_stack/providers/inline/tool_runtime/rag/memory.py

@@ -50,8 +50,11 @@ log = get_logger(name=__name__, category="tool_runtime")
 async def raw_data_from_doc(doc: RAGDocument) -> tuple[bytes, str]:
     """Get raw binary data and mime type from a RAGDocument for file upload."""
     if isinstance(doc.content, URL):
-        if doc.content.uri.startswith("data:"):
-            parts = parse_data_url(doc.content.uri)
+        uri = doc.content.uri
+        if uri.startswith("file://"):
+            raise ValueError("file:// URIs are not supported. Please use the Files API (/v1/files) to upload files.")
+        if uri.startswith("data:"):
+            parts = parse_data_url(uri)
             mime_type = parts["mimetype"]
             data = parts["data"]
 
@@ -63,7 +66,7 @@ async def raw_data_from_doc(doc: RAGDocument) -> tuple[bytes, str]:
             return file_data, mime_type
         else:
             async with httpx.AsyncClient() as client:
-                r = await client.get(doc.content.uri)
+                r = await client.get(uri)
                 r.raise_for_status()
                 mime_type = r.headers.get("content-type", "application/octet-stream")
                 return r.content, mime_type
@@ -73,6 +76,8 @@ async def raw_data_from_doc(doc: RAGDocument) -> tuple[bytes, str]:
         else:
             content_str = interleaved_content_as_str(doc.content)
 
+        if content_str.startswith("file://"):
+            raise ValueError("file:// URIs are not supported. Please use the Files API (/v1/files) to upload files.")
         if content_str.startswith("data:"):
             parts = parse_data_url(content_str)
             mime_type = parts["mimetype"]

{llama_stack-0.4.1 → llama_stack-0.4.2}/src/llama_stack/providers/utils/memory/vector_store.py

@@ -135,15 +135,20 @@ def content_from_data_and_mime_type(data: bytes | str, mime_type: str | None, en
 
 async def content_from_doc(doc: RAGDocument) -> str:
     if isinstance(doc.content, URL):
-        if doc.content.uri.startswith("data:"):
-            return content_from_data(doc.content.uri)
+        uri = doc.content.uri
+        if uri.startswith("file://"):
+            raise ValueError("file:// URIs are not supported. Please use the Files API (/v1/files) to upload files.")
+        if uri.startswith("data:"):
+            return content_from_data(uri)
         async with httpx.AsyncClient() as client:
-            r = await client.get(doc.content.uri)
+            r = await client.get(uri)
         if doc.mime_type == "application/pdf":
             return parse_pdf(r.content)
         return r.text
     elif isinstance(doc.content, str):
-        pattern = re.compile("^(https?://|file://|data:)")
+        if doc.content.startswith("file://"):
+            raise ValueError("file:// URIs are not supported. Please use the Files API (/v1/files) to upload files.")
+        pattern = re.compile("^(https?://|data:)")
         if pattern.match(doc.content):
             if doc.content.startswith("data:"):
                 return content_from_data(doc.content)

{llama_stack-0.4.1 → llama_stack-0.4.2/src/llama_stack.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llama_stack
-Version: 0.4.1
+Version: 0.4.2
 Summary: Llama Stack
 Author-email: Meta Llama <llama-oss@meta.com>
 License: MIT
@@ -46,7 +46,7 @@ Requires-Dist: psycopg2-binary
 Requires-Dist: tornado>=6.5.3
 Requires-Dist: urllib3>=2.6.3
 Provides-Extra: client
-Requires-Dist: llama-stack-client==0.4.1; extra == "client"
+Requires-Dist: llama-stack-client==0.4.2; extra == "client"
 Dynamic: license-file
 
 # Llama Stack

{llama_stack-0.4.1 → llama_stack-0.4.2}/src/llama_stack.egg-info/SOURCES.txt

@@ -586,6 +586,73 @@ src/llama_stack_api/inspect_api/models.py
 src/llama_stack_api/internal/__init__.py
 src/llama_stack_api/internal/kvstore.py
 src/llama_stack_api/internal/sqlstore.py
+src/llama_stack_api/llama_stack_api/__init__.py
+src/llama_stack_api/llama_stack_api/agents.py
+src/llama_stack_api/llama_stack_api/connectors.py
+src/llama_stack_api/llama_stack_api/conversations.py
+src/llama_stack_api/llama_stack_api/datasetio.py
+src/llama_stack_api/llama_stack_api/datatypes.py
+src/llama_stack_api/llama_stack_api/eval.py
+src/llama_stack_api/llama_stack_api/inference.py
+src/llama_stack_api/llama_stack_api/models.py
+src/llama_stack_api/llama_stack_api/openai_responses.py
+src/llama_stack_api/llama_stack_api/post_training.py
+src/llama_stack_api/llama_stack_api/prompts.py
+src/llama_stack_api/llama_stack_api/py.typed
+src/llama_stack_api/llama_stack_api/rag_tool.py
+src/llama_stack_api/llama_stack_api/resource.py
+src/llama_stack_api/llama_stack_api/router_utils.py
+src/llama_stack_api/llama_stack_api/safety.py
+src/llama_stack_api/llama_stack_api/schema_utils.py
+src/llama_stack_api/llama_stack_api/scoring.py
+src/llama_stack_api/llama_stack_api/scoring_functions.py
+src/llama_stack_api/llama_stack_api/shields.py
+src/llama_stack_api/llama_stack_api/tools.py
+src/llama_stack_api/llama_stack_api/vector_io.py
+src/llama_stack_api/llama_stack_api/vector_stores.py
+src/llama_stack_api/llama_stack_api/version.py
+src/llama_stack_api/llama_stack_api/admin/__init__.py
+src/llama_stack_api/llama_stack_api/admin/api.py
+src/llama_stack_api/llama_stack_api/admin/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/admin/models.py
+src/llama_stack_api/llama_stack_api/batches/__init__.py
+src/llama_stack_api/llama_stack_api/batches/api.py
+src/llama_stack_api/llama_stack_api/batches/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/batches/models.py
+src/llama_stack_api/llama_stack_api/benchmarks/__init__.py
+src/llama_stack_api/llama_stack_api/benchmarks/api.py
+src/llama_stack_api/llama_stack_api/benchmarks/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/benchmarks/models.py
+src/llama_stack_api/llama_stack_api/common/__init__.py
+src/llama_stack_api/llama_stack_api/common/content_types.py
+src/llama_stack_api/llama_stack_api/common/errors.py
+src/llama_stack_api/llama_stack_api/common/job_types.py
+src/llama_stack_api/llama_stack_api/common/responses.py
+src/llama_stack_api/llama_stack_api/common/training_types.py
+src/llama_stack_api/llama_stack_api/common/type_system.py
+src/llama_stack_api/llama_stack_api/datasets/__init__.py
+src/llama_stack_api/llama_stack_api/datasets/api.py
+src/llama_stack_api/llama_stack_api/datasets/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/datasets/models.py
+src/llama_stack_api/llama_stack_api/file_processors/__init__.py
+src/llama_stack_api/llama_stack_api/file_processors/api.py
+src/llama_stack_api/llama_stack_api/file_processors/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/file_processors/models.py
+src/llama_stack_api/llama_stack_api/files/__init__.py
+src/llama_stack_api/llama_stack_api/files/api.py
+src/llama_stack_api/llama_stack_api/files/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/files/models.py
+src/llama_stack_api/llama_stack_api/inspect_api/__init__.py
+src/llama_stack_api/llama_stack_api/inspect_api/api.py
+src/llama_stack_api/llama_stack_api/inspect_api/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/inspect_api/models.py
+src/llama_stack_api/llama_stack_api/internal/__init__.py
+src/llama_stack_api/llama_stack_api/internal/kvstore.py
+src/llama_stack_api/llama_stack_api/internal/sqlstore.py
+src/llama_stack_api/llama_stack_api/providers/__init__.py
+src/llama_stack_api/llama_stack_api/providers/api.py
+src/llama_stack_api/llama_stack_api/providers/fastapi_routes.py
+src/llama_stack_api/llama_stack_api/providers/models.py
 src/llama_stack_api/providers/__init__.py
 src/llama_stack_api/providers/api.py
 src/llama_stack_api/providers/fastapi_routes.py

{llama_stack-0.4.1 → llama_stack-0.4.2}/src/llama_stack.egg-info/requires.txt

@@ -29,4 +29,4 @@ tornado>=6.5.3
 urllib3>=2.6.3
 
 [client]
-llama-stack-client==0.4.1
+llama-stack-client==0.4.2