litellm/tests/code_coverage_tests/router_surface_check.py
Nate Armstrong 9752869d84
feat(router): add the Rust router core with the host as its only boundary (#49513)
Adds the litellm-router-core crate, its step-driven PyO3 bridge
(RouterCoreHandle / RouterCall) and RustCore (litellm/router_rust/),
served behind LITELLM_RUST through the ROUTER route rule. The host
call-out is the only boundary: every attempt goes through Host::invoke,
there is no native executor path, and RustCore never builds, wraps or
forwards to PythonCore, so router_python_core.py stays deletable.
Attempt preparation lives in litellm/router_attempt.py so it outlives
router_python_core.py.

RustCore matches PythonCore's behavior, including its known bugs (fixed
in follow-up PRs). Routing decisions (strategy picks, routing groups,
PTU, pre-routing, pre-call checks, fallback walk, health routing, access
groups, weights, cooldowns, allowed fails, usage) live in the crate;
work that needs live Python objects (request shaping, endpoints,
streaming, model info facts, callbacks, alerting, prompt management)
lives in the host-side litellm/router_rust/ package (steps/, invoke/,
surface/) driven by Rust steps.

The crate owns the registry. It runs the deployment build loop (secret
references, ids, organization fan-out, install order) and asks the host
for what needs Python, one deployment at a time so errors surface where
Python raised them: secrets, validate, provider and added steps. It
holds the model list, the index maps and the pattern routers with the
Python core's history rules (host_index.rs), and answers the registry
lookups (lookup.rs). The Python views of that state in
litellm/router_rust/surface/ (model_list, index(), names(), patterns())
read the handle, and every write to them becomes a changeset. Model
info, deployment model info, configured limits and model group info are
merged in the crate (model_info.rs, group_info.rs) from catalog rows and
facts the host answers on a model_info step asked at query time, so
price reloads and register_model stay visible. The usage cache key
format is defined once, in the crate.

The crate records each attempt's outcome itself rather than through
litellm callbacks. A streamed attempt is deferred: it counts its
request when the stream opens, and the host's stream wrappers report
the success with the final usage, or a failure after the first chunk,
when the stream ends, as the Python core's callbacks do.

tests/unit/test_router/core characterizes PythonCore's public behavior
and runs every test on both cores, with nothing pending; only the names
of the Python core's own outcome callbacks are asserted on Python alone.

litellm.Router is now final: subclassing it raises TypeError at the class statement, and the
machinery that kept subclass overrides working is gone. The Rust core never honoured subclass
overrides, so they would have silently stopped running there. Custom routing goes through
register_plugin(), settings through apply() and outcomes through observe(). This is a breaking
change for anyone subclassing Router.

Fuzz parity fixes on top of the rebase: the Rust core keeps num_retries,
max_fallbacks, allowed_fails, retry_after, timeout, the tag settings,
routing_strategy_args and the fallback lists as given (the crate reads
them as numbers, update_settings casts as Python does); a deprecated
`model_list = [...]` write reaches the Rust core; a failed set_model_list
is all or nothing on both cores (the deployments served before the call
are put back and the error raised); config_deployments() is a generator and `del` of a plain
attribute works on the Rust core. Least-busy pick reservations from
main now work on the Rust core: its selectors share one reservation
store, routed calls reserve and free their unstarted picks, and
select() and batch cancel do not reserve. On the Python core a dropped
upsert takes its name back off deployment_names, and
get_deployment_order reads every numeric spelling as one int so mixed
orders no longer raise TypeError.

Rulings on the remaining differences, the Python core changed where it
was the odd one out: sync picks and picks by deployment id apply the
team filter as async picks do, so another team's deployment is refused;
the fallback lists and default_litellm_params in RouterSettings are
copies, so an in-place edit is ignored and settings change only through
apply(); one health record with an unreadable timestamp drops out alone;
an empty fallback entry is skipped; a bool cooldown_time is refused at
construction, a bool or string cooldown_time on a deployment or a call
falls back to the router default, and a failure status that no HTTP
status can be is recorded as 500; an infinite or NaN cost and a whole
number wider than 64 bits (ptu_shares) are refused on both cores with
the same error (litellm/router_utils/deployment_numbers.py); and a
/v1/messages stream whose provider iterator raises cools the deployment
and retries like a non-stream call, with the Rust core waiting for that
cooldown before its retry picks.

Differences the proxy parity fuzz and its regression tests found on the
Rust core: a file upload or a batch on an uploaded file falls back only
within its own model group, as on the Python core, since the file lives
with one provider (the host sends same_group_fallbacks_only and the walk
skips targets in other groups); Decisions calls drop deployments whose
provider has no Decisions config, and deployments that refuse the audio
or videos a call carries, unless that would leave none, as the Python
core does, while the retry count still reads the whole group; a file
upload that fails on one order level is created on the next level's
deployments rather than the first level again; and a response that
cannot be converted (a text_completion body without a message) reports
only its own traceback, so a rate limit handled by the attempt before it
no longer turns the error into a 429, and both cores return 500.
2026-11-04 21:47:58 -08:00

246 lines
9.5 KiB
Python

"""
Fence for the Router's public surface.
Code outside the router may use only the Router members listed in
router_surface_allowlist.json, and may import only the litellm.router_utils
modules listed there. The lists hold what is used today and may only shrink:
- a member or module used outside the router that is not listed fails the
check (use an existing entry point instead, or discuss widening the surface);
- a listed member or module that nothing outside the router uses any more
also fails, so the entry is removed in the same change.
"Outside" means every non-test file under litellm/ and enterprise/ except
litellm/router.py and the router's own modules (including every
module under litellm/router_rust/, the Rust core's host-side glue). Uses from the router family
(litellm/router_utils, litellm/router_strategy) are listed separately, because
those modules become private parts of the router core in a later step.
Member uses are found the same way the boundary survey found them: attribute
access on a receiver whose name looks like a router (llm_router,
litellm_router_instance, router, ...) or on the Router class, plus
getattr/setattr/hasattr with a constant name on such a receiver. Members
include the container file endpoints Router.__init__ registers by name.
Run: python tests/code_coverage_tests/router_surface_check.py [--write]
--write regenerates the allowlist from the current tree (for shrinking it).
"""
import ast
import json
import os
import re
import sys
from typing import Dict, List, Set
ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", ".."))
ALLOWLIST = os.path.join(os.path.dirname(__file__), "router_surface_allowlist.json")
ROUTER_FILES = {
"litellm/router.py",
"litellm/router_api.py",
"litellm/router_config.py",
"litellm/router_core.py",
"litellm/router_python_core.py",
"litellm/router_python_host.py",
"litellm/router_rust/core.py",
"litellm/router_deprecated.py",
"litellm/router_host.py",
}
# The Rust core's host-side glue modules are part of the router.
ROUTER_FILE_PREFIXES = ("litellm/router_rust/",)
ROUTER_CLASS_FILES = ["litellm/router.py", "litellm/router_python_core.py", "litellm/router_deprecated.py"]
ROUTER_CLASS_NAMES = {"Router", "PythonCore", "DeprecatedRouterMembers"}
SCAN_ROOTS = ["litellm", "enterprise"]
SKIP_DIRS = {
".git",
"node_modules",
".venv",
"__pycache__",
"litellm-rust",
"ui",
"docs",
"cookbook",
"tests",
"test",
}
RECEIVER = re.compile(
r"(^|[._])(llm_router|router|litellm_router|litellm_router_instance|router_instance"
r"|_router|proxy_router|llm_router_instance|router_obj|mock_router|Router)$"
)
# Router.__init__ registers these with setattr(self, name, ...), so they never appear as self.<name>.
CONTAINER_FILE_ENDPOINTS = "litellm/containers/endpoints.json"
DYNAMIC_FUNCS = {"getattr", "setattr", "hasattr"}
def _area(rel: str) -> str:
if rel.startswith(("litellm/router_utils/", "litellm/router_strategy/")):
return "router_family"
return "outside"
def _is_test(rel: str) -> bool:
name = os.path.basename(rel)
return "/tests/" in rel or name.startswith("test_") or name.endswith("_test.py")
def router_members() -> Set[str]:
members: Set[str] = set()
for rel in ROUTER_CLASS_FILES:
with open(os.path.join(ROOT, rel), encoding="utf-8") as f:
tree = ast.parse(f.read())
for cls in tree.body:
if not isinstance(cls, ast.ClassDef) or cls.name not in ROUTER_CLASS_NAMES:
continue
for node in cls.body:
if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
members.add(node.name)
elif isinstance(node, ast.AnnAssign) and isinstance(node.target, ast.Name):
members.add(node.target.id)
elif isinstance(node, ast.Assign):
members.update(t.id for t in node.targets if isinstance(t, ast.Name))
for node in ast.walk(cls):
if (
isinstance(node, ast.Attribute)
and isinstance(node.value, ast.Name)
and node.value.id == "self"
and isinstance(node.ctx, ast.Store)
):
members.add(node.attr)
with open(os.path.join(ROOT, CONTAINER_FILE_ENDPOINTS), encoding="utf-8") as f:
for endpoint in json.load(f)["endpoints"]:
members.update((endpoint["name"], endpoint["async_name"]))
return {m for m in members if not (m.startswith("__") and m.endswith("__"))}
def _python_files():
for top in SCAN_ROOTS:
base = os.path.join(ROOT, top)
for dirpath, dirnames, filenames in os.walk(base):
dirnames[:] = [d for d in dirnames if d not in SKIP_DIRS]
for fn in filenames:
if fn.endswith(".py"):
path = os.path.join(dirpath, fn)
yield os.path.relpath(path, ROOT).replace(os.sep, "/"), path
def scan() -> Dict[str, Dict[str, Set[str]]]:
"""Return {"members": {area: {member: files}}, "router_utils_imports": {module: files}}."""
members = router_members()
uses: Dict[str, Dict[str, Set[str]]] = {"outside": {}, "router_family": {}}
imports: Dict[str, Set[str]] = {}
for rel, path in _python_files():
if rel in ROUTER_FILES or rel.startswith(ROUTER_FILE_PREFIXES) or _is_test(rel):
continue
with open(path, encoding="utf-8") as f:
src = f.read()
has_router = "router" in src.lower()
if not has_router:
continue
try:
tree = ast.parse(src)
except SyntaxError:
continue
area = _area(rel)
for name in member_uses(tree, members):
uses[area].setdefault(name, set()).add(rel)
if area == "outside":
for node in ast.walk(tree):
for mod in _router_utils_modules(node):
imports.setdefault(mod, set()).add(rel)
return {"members": uses, "router_utils_imports": imports}
def member_uses(tree: ast.AST, members: Set[str]) -> Set[str]:
"""The Router members a module uses, on an instance or on the Router class itself."""
used: Set[str] = set()
for node in ast.walk(tree):
if isinstance(node, ast.Attribute) and node.attr in members:
if RECEIVER.search(ast.unparse(node.value)):
used.add(node.attr)
elif (
isinstance(node, ast.Call)
and isinstance(node.func, ast.Name)
and node.func.id in DYNAMIC_FUNCS
and len(node.args) >= 2
and isinstance(node.args[1], ast.Constant)
and node.args[1].value in members
and RECEIVER.search(ast.unparse(node.args[0]))
):
used.add(node.args[1].value)
return used
def _router_utils_modules(node: ast.AST) -> List[str]:
if isinstance(node, ast.ImportFrom) and node.module:
if node.module == "litellm.router_utils":
return ["litellm.router_utils." + a.name for a in node.names]
if node.module.startswith("litellm.router_utils."):
return [node.module]
if isinstance(node, ast.Import):
return [a.name for a in node.names if a.name.startswith("litellm.router_utils.")]
return []
def build_allowlist(found) -> dict:
return {
"_doc": (
"Router members and litellm.router_utils modules used outside the router. "
"Shrink-only: see tests/code_coverage_tests/router_surface_check.py."
),
"members_outside": sorted(found["members"]["outside"]),
"members_router_family": sorted(found["members"]["router_family"]),
"router_utils_imports_outside": sorted(found["router_utils_imports"]),
}
def compare(allow: dict, found) -> List[str]:
errors: List[str] = []
sections = [
("members_outside", found["members"]["outside"], "Router member"),
("members_router_family", found["members"]["router_family"], "Router member (router family)"),
("router_utils_imports_outside", found["router_utils_imports"], "litellm.router_utils import"),
]
for key, used, what in sections:
listed = set(allow.get(key, []))
for name in sorted(set(used) - listed):
files = ", ".join(sorted(used[name]))
errors.append(
f"new {what} outside the router: {name!r} in {files}. The router surface may only "
f"shrink: use an entry point instead."
)
for name in sorted(listed - set(used)):
errors.append(
f"{what} {name!r} is listed in {os.path.basename(ALLOWLIST)} [{key}] but nothing uses it "
f"any more: remove it from the list."
)
return errors
def main(argv: List[str]) -> int:
found = scan()
if "--write" in argv:
with open(ALLOWLIST, "w", encoding="utf-8") as f:
json.dump(build_allowlist(found), f, indent=2)
f.write("\n")
print(f"wrote {ALLOWLIST}")
return 0
with open(ALLOWLIST, encoding="utf-8") as f:
allow = json.load(f)
errors = compare(allow, found)
for e in errors:
print(e)
if errors:
return 1
print(
f"router surface ok: {len(allow['members_outside'])} members outside, "
f"{len(allow['members_router_family'])} from the router family, "
f"{len(allow['router_utils_imports_outside'])} router_utils modules imported outside"
)
return 0
if __name__ == "__main__":
sys.exit(main(sys.argv[1:]))