llama-stack 0.0.42__py3-none-any.whl → 0.3.4__py3-none-any.whl

llama_stack/core/client.py

@@ -0,0 +1,205 @@
+# 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.
+
+import inspect
+import json
+import sys
+from collections.abc import AsyncIterator
+from enum import Enum
+from typing import Any, Union, get_args, get_origin
+
+import httpx
+from pydantic import BaseModel, parse_obj_as
+from termcolor import cprint
+
+from llama_stack.providers.datatypes import RemoteProviderConfig
+
+_CLIENT_CLASSES = {}
+
+
+async def get_client_impl(protocol, config: RemoteProviderConfig, _deps: Any):
+    client_class = create_api_client_class(protocol)
+    impl = client_class(config.url)
+    await impl.initialize()
+    return impl
+
+
+def create_api_client_class(protocol) -> type:
+    if protocol in _CLIENT_CLASSES:
+        return _CLIENT_CLASSES[protocol]
+
+    class APIClient:
+        def __init__(self, base_url: str):
+            print(f"({protocol.__name__}) Connecting to {base_url}")
+            self.base_url = base_url.rstrip("/")
+            self.routes = {}
+
+            # Store routes for this protocol
+            for name, method in inspect.getmembers(protocol):
+                if hasattr(method, "__webmethod__"):
+                    sig = inspect.signature(method)
+                    self.routes[name] = (method.__webmethod__, sig)
+
+        async def initialize(self):
+            pass
+
+        async def shutdown(self):
+            pass
+
+        async def __acall__(self, method_name: str, *args, **kwargs) -> Any:
+            assert method_name in self.routes, f"Unknown endpoint: {method_name}"
+
+            # TODO: make this more precise, same thing needs to happen in server.py
+            is_streaming = kwargs.get("stream", False)
+            if is_streaming:
+                return self._call_streaming(method_name, *args, **kwargs)
+            else:
+                return await self._call_non_streaming(method_name, *args, **kwargs)
+
+        async def _call_non_streaming(self, method_name: str, *args, **kwargs) -> Any:
+            _, sig = self.routes[method_name]
+
+            if sig.return_annotation is None:
+                return_type = None
+            else:
+                return_type = extract_non_async_iterator_type(sig.return_annotation)
+                assert return_type, f"Could not extract return type for {sig.return_annotation}"
+
+            async with httpx.AsyncClient() as client:
+                params = self.httpx_request_params(method_name, *args, **kwargs)
+                response = await client.request(**params)
+                response.raise_for_status()
+
+                j = response.json()
+                if j is None:
+                    return None
+                # print(f"({protocol.__name__}) Returning {j}, type {return_type}")
+                return parse_obj_as(return_type, j)
+
+        async def _call_streaming(self, method_name: str, *args, **kwargs) -> Any:
+            webmethod, sig = self.routes[method_name]
+
+            return_type = extract_async_iterator_type(sig.return_annotation)
+            assert return_type, f"Could not extract return type for {sig.return_annotation}"
+
+            async with httpx.AsyncClient() as client:
+                params = self.httpx_request_params(method_name, *args, **kwargs)
+                async with client.stream(**params) as response:
+                    response.raise_for_status()
+
+                    async for line in response.aiter_lines():
+                        if line.startswith("data:"):
+                            data = line[len("data: ") :]
+                            try:
+                                data = json.loads(data)
+                                if "error" in data:
+                                    cprint(data, color="red", file=sys.stderr)
+                                    continue
+
+                                yield parse_obj_as(return_type, data)
+                            except Exception as e:
+                                cprint(f"Error with parsing or validation: {e}", color="red", file=sys.stderr)
+                                cprint(data, color="red", file=sys.stderr)
+
+        def httpx_request_params(self, method_name: str, *args, **kwargs) -> dict:
+            webmethod, sig = self.routes[method_name]
+
+            parameters = list(sig.parameters.values())[1:]  # skip `self`
+            for i, param in enumerate(parameters):
+                if i >= len(args):
+                    break
+                kwargs[param.name] = args[i]
+
+            # Get all webmethods for this method (supports multiple decorators)
+            webmethods = getattr(method, "__webmethods__", [])
+
+            if not webmethods:
+                raise RuntimeError(f"Method {method} has no webmethod decorators")
+
+            # Choose the preferred webmethod (non-deprecated if available)
+            preferred_webmethod = None
+            for wm in webmethods:
+                if not getattr(wm, "deprecated", False):
+                    preferred_webmethod = wm
+                    break
+
+            # If no non-deprecated found, use the first one
+            if preferred_webmethod is None:
+                preferred_webmethod = webmethods[0]
+
+            url = f"{self.base_url}/{preferred_webmethod.level}/{preferred_webmethod.route.lstrip('/')}"
+
+            def convert(value):
+                if isinstance(value, list):
+                    return [convert(v) for v in value]
+                elif isinstance(value, dict):
+                    return {k: convert(v) for k, v in value.items()}
+                elif isinstance(value, BaseModel):
+                    return json.loads(value.model_dump_json())
+                elif isinstance(value, Enum):
+                    return value.value
+                else:
+                    return value
+
+            params = {}
+            data = {}
+            if webmethod.method == "GET":
+                params.update(kwargs)
+            else:
+                data.update(convert(kwargs))
+
+            ret = dict(
+                method=webmethod.method or "POST",
+                url=url,
+                headers={
+                    "Accept": "application/json",
+                    "Content-Type": "application/json",
+                },
+                timeout=30,
+            )
+            if params:
+                ret["params"] = params
+            if data:
+                ret["json"] = data
+
+            return ret
+
+    # Add protocol methods to the wrapper
+    for name, method in inspect.getmembers(protocol):
+        if hasattr(method, "__webmethod__"):
+
+            async def method_impl(self, *args, method_name=name, **kwargs):
+                return await self.__acall__(method_name, *args, **kwargs)
+
+            method_impl.__name__ = name
+            method_impl.__qualname__ = f"APIClient.{name}"
+            method_impl.__signature__ = inspect.signature(method)
+            setattr(APIClient, name, method_impl)
+
+    # Name the class after the protocol
+    APIClient.__name__ = f"{protocol.__name__}Client"
+    _CLIENT_CLASSES[protocol] = APIClient
+    return APIClient
+
+
+# not quite general these methods are
+def extract_non_async_iterator_type(type_hint):
+    if get_origin(type_hint) is Union:
+        args = get_args(type_hint)
+        for arg in args:
+            if not issubclass(get_origin(arg) or arg, AsyncIterator):
+                return arg
+    return type_hint
+
+
+def extract_async_iterator_type(type_hint):
+    if get_origin(type_hint) is Union:
+        args = get_args(type_hint)
+        for arg in args:
+            if issubclass(get_origin(arg) or arg, AsyncIterator):
+                inner_args = get_args(arg)
+                return inner_args[0]
+    return None

llama_stack/core/common.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+
+# 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.
+
+cleanup() {
+  # For venv environments, no special cleanup is needed
+  # This function exists to avoid "function not found" errors
+  local env_name="$1"
+  echo "Cleanup called for environment: $env_name"
+}
+
+handle_int() {
+  if [ -n "$ENVNAME" ]; then
+    cleanup "$ENVNAME"
+  fi
+  exit 1
+}
+
+handle_exit() {
+  if [ $? -ne 0 ]; then
+    echo -e "\033[1;31mABORTING.\033[0m"
+    if [ -n "$ENVNAME" ]; then
+      cleanup "$ENVNAME"
+    fi
+  fi
+}
+
+
+
+# check if a command is present
+is_command_available() {
+  command -v "$1" &>/dev/null
+}

llama_stack/{distribution → core}/configure.py

@@ -4,28 +4,28 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 import textwrap
-
 from typing import Any
 
-from llama_stack.distribution.datatypes import *  # noqa: F403
-from termcolor import cprint
-
-from llama_stack.distribution.distribution import (
+from llama_stack.core.datatypes import (
+    LLAMA_STACK_RUN_CONFIG_VERSION,
+    DistributionSpec,
+    Provider,
+    StackRunConfig,
+)
+from llama_stack.core.distribution import (
     builtin_automatically_routed_apis,
     get_provider_registry,
 )
-from llama_stack.distribution.utils.dynamic import instantiate_class_type
-from llama_stack.distribution.utils.prompt_for_config import prompt_for_config
+from llama_stack.core.stack import cast_image_name_to_string, replace_env_vars
+from llama_stack.core.utils.dynamic import instantiate_class_type
+from llama_stack.core.utils.prompt_for_config import prompt_for_config
+from llama_stack.log import get_logger
+from llama_stack.providers.datatypes import Api, ProviderSpec
 
+logger = get_logger(name=__name__, category="core")
 
-from llama_stack.apis.models import *  # noqa: F403
-from llama_stack.apis.shields import *  # noqa: F403
-from llama_stack.apis.memory_banks import *  # noqa: F403
 
-
-def configure_single_provider(
-    registry: Dict[str, ProviderSpec], provider: Provider
-) -> Provider:
+def configure_single_provider(registry: dict[str, ProviderSpec], provider: Provider) -> Provider:
     provider_spec = registry[provider.provider_type]
     config_type = instantiate_class_type(provider_spec.config_class)
     try:
@@ -40,17 +40,15 @@ def configure_single_provider(
     return Provider(
         provider_id=provider.provider_id,
         provider_type=provider.provider_type,
-        config=cfg.dict(),
+        config=cfg.model_dump(),
     )
 
 
-def configure_api_providers(
-    config: StackRunConfig, build_spec: DistributionSpec
-) -> StackRunConfig:
+def configure_api_providers(config: StackRunConfig, build_spec: DistributionSpec) -> StackRunConfig:
     is_nux = len(config.providers) == 0
 
     if is_nux:
-        print(
+        logger.info(
             textwrap.dedent(
                 """
         Llama Stack is composed of several APIs working together. For each API served by the Stack,
@@ -65,7 +63,7 @@ def configure_api_providers(
     if config.apis:
         apis_to_serve = config.apis
     else:
-        apis_to_serve = [a.value for a in Api if a not in (Api.telemetry, Api.inspect)]
+        apis_to_serve = [a.value for a in Api if a not in (Api.inspect, Api.providers)]
 
     for api_str in apis_to_serve:
         api = Api(api_str)
@@ -76,18 +74,12 @@ def configure_api_providers(
 
         existing_providers = config.providers.get(api_str, [])
         if existing_providers:
-            cprint(
-                f"Re-configuring existing providers for API `{api_str}`...",
-                "green",
-                attrs=["bold"],
-            )
+            logger.info(f"Re-configuring existing providers for API `{api_str}`...")
             updated_providers = []
             for p in existing_providers:
-                print(f"> Configuring provider `({p.provider_type})`")
-                updated_providers.append(
-                    configure_single_provider(provider_registry[api], p)
-                )
-                print("")
+                logger.info(f"> Configuring provider `({p.provider_type})`")
+                updated_providers.append(configure_single_provider(provider_registry[api], p))
+                logger.info("")
         else:
             # we are newly configuring this API
             plist = build_spec.providers.get(api_str, [])
@@ -96,25 +88,29 @@ def configure_api_providers(
             if not plist:
                 raise ValueError(f"No provider configured for API {api_str}?")
 
-            cprint(f"Configuring API `{api_str}`...", "green", attrs=["bold"])
+            logger.info(f"Configuring API `{api_str}`...")
             updated_providers = []
-            for i, provider_type in enumerate(plist):
-                print(f"> Configuring provider `({provider_type})`")
+            for i, provider in enumerate(plist):
+                if i >= 1:
+                    others = ", ".join(p.provider_type for p in plist[i:])
+                    logger.info(
+                        f"Not configuring other providers ({others}) interactively. Please edit the resulting YAML directly.\n"
+                    )
+                    break
+
+                logger.info(f"> Configuring provider `({provider.provider_type})`")
+                pid = provider.provider_type.split("::")[-1]
                 updated_providers.append(
                     configure_single_provider(
                         provider_registry[api],
                         Provider(
-                            provider_id=(
-                                f"{provider_type}-{i:02d}"
-                                if len(plist) > 1
-                                else provider_type
-                            ),
-                            provider_type=provider_type,
+                            provider_id=(f"{pid}-{i:02d}" if len(plist) > 1 else pid),
+                            provider_type=provider.provider_type,
                             config={},
                         ),
                     )
                 )
-                print("")
+                logger.info("")
 
         config.providers[api_str] = updated_providers
 
@@ -122,16 +118,12 @@ def configure_api_providers(
 
 
 def upgrade_from_routing_table(
-    config_dict: Dict[str, Any],
-) -> Dict[str, Any]:
+    config_dict: dict[str, Any],
+) -> dict[str, Any]:
     def get_providers(entries):
         return [
             Provider(
-                provider_id=(
-                    f"{entry['provider_type']}-{i:02d}"
-                    if len(entries) > 1
-                    else entry["provider_type"]
-                ),
+                provider_id=(f"{entry['provider_type']}-{i:02d}" if len(entries) > 1 else entry["provider_type"]),
                 provider_type=entry["provider_type"],
                 config=entry["config"],
             )
@@ -166,19 +158,46 @@ def upgrade_from_routing_table(
     config_dict["apis"] = config_dict["apis_to_serve"]
     config_dict.pop("apis_to_serve", None)
 
-    return config_dict
+    # Add default storage config if not present
+    if "storage" not in config_dict:
+        config_dict["storage"] = {
+            "backends": {
+                "kv_default": {
+                    "type": "kv_sqlite",
+                    "db_path": "~/.llama/kvstore.db",
+                },
+                "sql_default": {
+                    "type": "sql_sqlite",
+                    "db_path": "~/.llama/sql_store.db",
+                },
+            },
+            "stores": {
+                "metadata": {
+                    "namespace": "registry",
+                    "backend": "kv_default",
+                },
+                "inference": {
+                    "table_name": "inference_store",
+                    "backend": "sql_default",
+                    "max_write_queue_size": 10000,
+                    "num_writers": 4,
+                },
+                "conversations": {
+                    "table_name": "openai_conversations",
+                    "backend": "sql_default",
+                },
+            },
+        }
 
+    return config_dict
 
-def parse_and_maybe_upgrade_config(config_dict: Dict[str, Any]) -> StackRunConfig:
-    version = config_dict.get("version", None)
-    if version == LLAMA_STACK_RUN_CONFIG_VERSION:
-        return StackRunConfig(**config_dict)
 
+def parse_and_maybe_upgrade_config(config_dict: dict[str, Any]) -> StackRunConfig:
     if "routing_table" in config_dict:
-        print("Upgrading config...")
+        logger.info("Upgrading config...")
         config_dict = upgrade_from_routing_table(config_dict)
 
     config_dict["version"] = LLAMA_STACK_RUN_CONFIG_VERSION
-    config_dict["built_at"] = datetime.now().isoformat()
 
-    return StackRunConfig(**config_dict)
+    processed_config_dict = replace_env_vars(config_dict)
+    return StackRunConfig(**cast_image_name_to_string(processed_config_dict))