phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-04 12:07:34 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

feat: introduce API leveling, post_training to v1alpha

Browse source 
Rather than have a single `LLAMA_STACK_VERSION`, we need to have a `_V1`, `_V1ALPHA`, and `_V1BETA` constant.

This also necessitated addition of `level` to the `WebMethod` so that routing can be handeled properly.

For backwards compat, the `v1` routes are being kept around and marked as `deprecated`. When used, the server will log a deprecation warning.

move:

post_training to v1alpha as it is under heavy development and not near its final state
eval: job scheduling is not implemented. Relies heavily on the datasetio API which is under development missing implementations of specific routes indicating the structure of those routes might change. Additionally eval depends on the inference API which is going to be deprecated, eval will likely need a major API surface change to conform to using completions properly

Signed-off-by: Charlie Doern <cdoern@redhat.com>
This commit is contained in:
Charlie Doern 2025-09-12 13:23:57 -04:00
parent a50b63906c
commit 03399cebf3
35 changed files with 1507 additions and 260 deletions
Download patch file Download diff file Expand all files Collapse all files

6
docs/openapi_generator/generate.py
View file

@ -16,7 +16,7 @@ import sys
import fire
import ruamel.yaml as yaml
from llama_stack.apis.version import LLAMA_STACK_API_VERSION # noqa: E402
from llama_stack.apis.version import LLAMA_STACK_API_V1 # noqa: E402
from llama_stack.core.stack import LlamaStack # noqa: E402
from .pyopenapi.options import Options # noqa: E402
@ -25,7 +25,7 @@ from .pyopenapi.utility import Specification, validate_api # noqa: E402
def str_presenter(dumper, data):
if data.startswith(f"/{LLAMA_STACK_API_VERSION}") or data.startswith(
if data.startswith(f"/{LLAMA_STACK_API_V1}") or data.startswith(
"#/components/schemas/"
):
style = None
@ -58,7 +58,7 @@ def main(output_dir: str):
server=Server(url="http://any-hosted-llama-stack.com"),
info=Info(
title="Llama Stack Specification",
version=LLAMA_STACK_API_VERSION,
version=LLAMA_STACK_API_V1,
description="""This is the specification of the Llama Stack that provides
a set of endpoints and their corresponding interfaces that are tailored to
best leverage Llama Models.""",

2
docs/openapi_generator/pyopenapi/generator.py
View file

@ -829,7 +829,7 @@ class Generator:
else:
raise NotImplementedError(f"unknown HTTP method: {op.http_method}")
route = op.get_route()
route = op.get_route(op.webmethod)
route = route.replace(":path", "")
print(f"route: {route}")
if route in paths:

233
docs/openapi_generator/pyopenapi/operations.py
View file

@ -11,7 +11,7 @@ import typing
from dataclasses import dataclass
from typing import Any, Callable, Dict, Iterable, Iterator, List, Optional, Tuple, Union
from llama_stack.apis.version import LLAMA_STACK_API_VERSION
from llama_stack.apis.version import LLAMA_STACK_API_V1, LLAMA_STACK_API_V1BETA, LLAMA_STACK_API_V1ALPHA
from termcolor import colored
@ -113,11 +113,13 @@ class EndpointOperation:
request_examples: Optional[List[Any]] = None
response_examples: Optional[List[Any]] = None
def get_route(self) -> str:
if self.route is not None:
return "/".join(["", LLAMA_STACK_API_VERSION, self.route.lstrip("/")])
def get_route(self, webmethod) -> str:
api_level = webmethod.level
route_parts = ["", LLAMA_STACK_API_VERSION, self.name]
if self.route is not None:
return "/".join(["", api_level, self.route.lstrip("/")])
route_parts = ["", api_level, self.name]
for param_name, _ in self.path_params:
route_parts.append("{" + param_name + "}")
return "/".join(route_parts)
@ -152,33 +154,39 @@ def _get_endpoint_functions(
functions = inspect.getmembers(endpoint, inspect.isfunction)
for func_name, func_ref in functions:
webmethod = getattr(func_ref, "__webmethod__", None)
if not webmethod:
webmethods = []
# Check for multiple webmethods (stacked decorators)
if hasattr(func_ref, "__webmethods__"):
webmethods = func_ref.__webmethods__
if not webmethods:
continue
print(f"Processing {colored(func_name, 'white')}...")
operation_name = func_name
if webmethod.method == "GET":
prefix = "get"
elif webmethod.method == "DELETE":
prefix = "delete"
elif webmethod.method == "POST":
prefix = "post"
elif operation_name.startswith("get_") or operation_name.endswith("/get"):
prefix = "get"
elif (
operation_name.startswith("delete_")
or operation_name.startswith("remove_")
or operation_name.endswith("/delete")
or operation_name.endswith("/remove")
):
prefix = "delete"
else:
# by default everything else is a POST
prefix = "post"
for webmethod in webmethods:
print(f"Processing {colored(func_name, 'white')}...")
operation_name = func_name
if webmethod.method == "GET":
prefix = "get"
elif webmethod.method == "DELETE":
prefix = "delete"
elif webmethod.method == "POST":
prefix = "post"
elif operation_name.startswith("get_") or operation_name.endswith("/get"):
prefix = "get"
elif (
operation_name.startswith("delete_")
or operation_name.startswith("remove_")
or operation_name.endswith("/delete")
or operation_name.endswith("/remove")
):
prefix = "delete"
else:
# by default everything else is a POST
prefix = "post"
yield prefix, operation_name, func_name, func_ref
yield prefix, operation_name, func_name, func_ref
def _get_defining_class(member_fn: str, derived_cls: type) -> type:
@ -239,105 +247,101 @@ def get_endpoint_operations(
"update",
],
):
# extract routing information from function metadata
webmethod = getattr(func_ref, "__webmethod__", None)
if webmethod is not None:
# Get all webmethods for this function
webmethods = getattr(func_ref, "__webmethods__", [])
# Create one EndpointOperation for each webmethod
for webmethod in webmethods:
route = webmethod.route
route_params = _get_route_parameters(route) if route is not None else None
public = webmethod.public
request_examples = webmethod.request_examples
response_examples = webmethod.response_examples
else:
route = None
route_params = None
public = False
request_examples = None
response_examples = None
# inspect function signature for path and query parameters, and request/response payload type
signature = get_signature(func_ref)
# inspect function signature for path and query parameters, and request/response payload type
signature = get_signature(func_ref)
path_params = []
query_params = []
request_params = []
multipart_params = []
path_params = []
query_params = []
request_params = []
multipart_params = []
for param_name, parameter in signature.parameters.items():
param_type = _get_annotation_type(parameter.annotation, func_ref)
for param_name, parameter in signature.parameters.items():
param_type = _get_annotation_type(parameter.annotation, func_ref)
# omit "self" for instance methods
if param_name == "self" and param_type is inspect.Parameter.empty:
continue
# omit "self" for instance methods
if param_name == "self" and param_type is inspect.Parameter.empty:
continue
# check if all parameters have explicit type
if parameter.annotation is inspect.Parameter.empty:
# check if all parameters have explicit type
if parameter.annotation is inspect.Parameter.empty:
raise ValidationError(
f"parameter '{param_name}' in function '{func_name}' has no type annotation"
)
is_multipart = _is_multipart_param(param_type)
if prefix in ["get", "delete"]:
if route_params is not None and param_name in route_params:
path_params.append((param_name, param_type))
else:
query_params.append((param_name, param_type))
else:
if route_params is not None and param_name in route_params:
path_params.append((param_name, param_type))
elif is_multipart:
multipart_params.append((param_name, param_type))
else:
request_params.append((param_name, param_type))
# check if function has explicit return type
if signature.return_annotation is inspect.Signature.empty:
raise ValidationError(
f"parameter '{param_name}' in function '{func_name}' has no type annotation"
f"function '{func_name}' has no return type annotation"
)
is_multipart = _is_multipart_param(param_type)
if prefix in ["get", "delete"]:
if route_params is not None and param_name in route_params:
path_params.append((param_name, param_type))
else:
query_params.append((param_name, param_type))
return_type = _get_annotation_type(signature.return_annotation, func_ref)
# operations that produce events are labeled as Generator[YieldType, SendType, ReturnType]
# where YieldType is the event type, SendType is None, and ReturnType is the immediate response type to the request
if typing.get_origin(return_type) is collections.abc.Generator:
event_type, send_type, response_type = typing.get_args(return_type)
if send_type is not type(None):
raise ValidationError(
f"function '{func_name}' has a return type Generator[Y,S,R] and therefore looks like an event but has an explicit send type"
)
else:
if route_params is not None and param_name in route_params:
path_params.append((param_name, param_type))
elif is_multipart:
multipart_params.append((param_name, param_type))
event_type = None
def process_type(t):
if typing.get_origin(t) is collections.abc.AsyncIterator:
# NOTE(ashwin): this is SSE and there is no way to represent it. either we make it a List
# or the item type. I am choosing it to be the latter
args = typing.get_args(t)
return args[0]
elif typing.get_origin(t) is typing.Union:
types = [process_type(a) for a in typing.get_args(t)]
return typing._UnionGenericAlias(typing.Union, tuple(types))
else:
return t
response_type = process_type(return_type)
if prefix in ["delete", "remove"]:
http_method = HTTPMethod.DELETE
elif prefix == "post":
http_method = HTTPMethod.POST
elif prefix == "get":
http_method = HTTPMethod.GET
elif prefix == "set":
http_method = HTTPMethod.PUT
elif prefix == "update":
http_method = HTTPMethod.PATCH
else:
request_params.append((param_name, param_type))
raise ValidationError(f"unknown prefix {prefix}")
# check if function has explicit return type
if signature.return_annotation is inspect.Signature.empty:
raise ValidationError(
f"function '{func_name}' has no return type annotation"
)
return_type = _get_annotation_type(signature.return_annotation, func_ref)
# operations that produce events are labeled as Generator[YieldType, SendType, ReturnType]
# where YieldType is the event type, SendType is None, and ReturnType is the immediate response type to the request
if typing.get_origin(return_type) is collections.abc.Generator:
event_type, send_type, response_type = typing.get_args(return_type)
if send_type is not type(None):
raise ValidationError(
f"function '{func_name}' has a return type Generator[Y,S,R] and therefore looks like an event but has an explicit send type"
)
else:
event_type = None
def process_type(t):
if typing.get_origin(t) is collections.abc.AsyncIterator:
# NOTE(ashwin): this is SSE and there is no way to represent it. either we make it a List
# or the item type. I am choosing it to be the latter
args = typing.get_args(t)
return args[0]
elif typing.get_origin(t) is typing.Union:
types = [process_type(a) for a in typing.get_args(t)]
return typing._UnionGenericAlias(typing.Union, tuple(types))
else:
return t
response_type = process_type(return_type)
if prefix in ["delete", "remove"]:
http_method = HTTPMethod.DELETE
elif prefix == "post":
http_method = HTTPMethod.POST
elif prefix == "get":
http_method = HTTPMethod.GET
elif prefix == "set":
http_method = HTTPMethod.PUT
elif prefix == "update":
http_method = HTTPMethod.PATCH
else:
raise ValidationError(f"unknown prefix {prefix}")
result.append(
EndpointOperation(
# Create an EndpointOperation for this specific webmethod
operation = EndpointOperation(
defining_class=_get_defining_class(func_name, endpoint),
name=operation_name,
func_name=func_name,
@ -354,7 +358,10 @@ def get_endpoint_operations(
request_examples=request_examples if use_examples else None,
response_examples=response_examples if use_examples else None,
)
)
# Store the specific webmethod with this operation
operation.webmethod = webmethod
result.append(operation)
if not result:
raise ValidationError(f"no eligible endpoint operations in type {endpoint}")