mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-27 06:13:07 +00:00
* feat(router): expose supported_endpoints on /model_group/info * feat(router): route batch creates to mode batch deployments in mixed model groups * fix(router): prefer batch deployments after eligibility filters and in group-name credential lookups * fix(router): keep group-name credential lookups on the first deployment and prefer batch deployments before cooldown Model-encoded ids carry only the group name, so preferring the batch deployment there moved ids created before the upgrade onto its credentials. Batch creates now pick batch-capable deployments before health checks and cooldown, routing and group info share one batch-capable check, the cost map's /v1/batch is dropped only for Azure, and /model/info applies the same per-deployment rule * test(router): map _listed_endpoints to its /model/info tests in the router coverage check * fix(router): skip only Azure chat deployments for batch creates and list batch support per caller team Narrow the batch create filter to Azure and azure_ai deployments that are not mode batch, so OpenAI groups keep their chat deployments during cooldown and under tag filters. Mark batch deployments by mode batch only, since stored endpoint lists from older Admin UI saves copy the cost map. Leave /model/info unchanged so the Admin UI never saves a computed endpoint list, and pass the caller's team into /model_group/info. * test(proxy): let the model hub router fakes take the caller's team_id * fix(router): gate batch-create routing on deployments the caller's team may use and list only /v1/batch for Azure batch deployments * fix(router): skip a malformed model_info.supported_endpoints instead of failing group info and chat * fix(router): apply the batch create filter after cooldown, managed files, and tags The batch filter ran before health checks and cooldown, so a batch create on a mixed Azure group answered with a router 429 or 401 when cooldown, tags, or a managed file's targets left no batch deployment, where main sends the create on and returns Azure's 400. The filter now runs right before order filtering, after every eligibility filter, so those cases answer as main does and creates still skip Azure chat deployments whenever a batch deployment is left * test(router): annotate the batch routing test assignments with Final
195 lines
12 KiB
Python
195 lines
12 KiB
Python
import ast
|
|
import os
|
|
|
|
|
|
def get_function_names_from_file(file_path):
|
|
"""
|
|
Extracts all function names from a given Python file.
|
|
"""
|
|
with open(file_path, "r", encoding="utf-8") as file:
|
|
tree = ast.parse(file.read())
|
|
|
|
function_names = []
|
|
|
|
for node in tree.body:
|
|
if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
|
|
# Top-level functions
|
|
function_names.append(node.name)
|
|
elif isinstance(node, ast.ClassDef):
|
|
# Functions inside classes
|
|
for class_node in node.body:
|
|
if isinstance(class_node, (ast.FunctionDef, ast.AsyncFunctionDef)):
|
|
function_names.append(class_node.name)
|
|
|
|
return function_names
|
|
|
|
|
|
def get_all_functions_called_in_tests(base_dir):
|
|
"""
|
|
Returns a set of function names that are called in test functions
|
|
inside the unit test directories,
|
|
specifically in files containing the word 'router'.
|
|
"""
|
|
called_functions = set()
|
|
test_dirs = ["unit"]
|
|
|
|
for test_dir in test_dirs:
|
|
dir_path = os.path.join(base_dir, test_dir)
|
|
if not os.path.exists(dir_path):
|
|
print(f"Warning: Directory {dir_path} does not exist.")
|
|
continue
|
|
|
|
print("dir_path: ", dir_path)
|
|
for root, _, files in os.walk(dir_path):
|
|
for file in files:
|
|
file_path = os.path.join(root, file)
|
|
if file.endswith(".py") and (
|
|
"router" in os.path.relpath(file_path, dir_path).lower() or test_dir == "unit"
|
|
):
|
|
print("file: ", file)
|
|
with open(file_path, "r", encoding="utf-8") as f:
|
|
try:
|
|
tree = ast.parse(f.read())
|
|
except SyntaxError:
|
|
print(f"Warning: Syntax error in file {file_path}")
|
|
continue
|
|
if file == "test_router_validate_fallbacks.py":
|
|
print(f"tree: {tree}")
|
|
for node in ast.walk(tree):
|
|
if isinstance(node, ast.Call) and isinstance(node.func, ast.Name):
|
|
called_functions.add(node.func.id)
|
|
elif isinstance(node, ast.Call) and isinstance(node.func, ast.Attribute):
|
|
called_functions.add(node.func.attr)
|
|
|
|
return called_functions
|
|
|
|
|
|
def get_functions_from_router(file_path):
|
|
"""
|
|
Extracts all functions defined in router.py.
|
|
"""
|
|
return get_function_names_from_file(file_path)
|
|
|
|
|
|
ignored_function_names = [
|
|
"_acancel_batch",
|
|
"_acreate_batch",
|
|
"_acreate_file",
|
|
"__init__",
|
|
"avector_store_create", # Tested via proxy vector_store_endpoints (files lack "router" in name)
|
|
"_override_vector_store_methods_for_router", # No-op placeholder, called during Router init
|
|
"_merge_tools_from_deployment", # Tested indirectly via _update_kwargs_with_deployment (test files lack "router" in name)
|
|
"_invalidate_access_groups_cache", # Tested indirectly via set_model_list, upsert_model etc. (test files lack "router" in name)
|
|
"has_buffered_provider_output", # Property, so its reads in test_router.py are never an ast.Call
|
|
"chunks", # Property on FallbackAwareAnthropicMessagesStream, so its reads in tests are never an ast.Call
|
|
"messages", # Property on FallbackAwareAnthropicMessagesStream, so its reads in tests are never an ast.Call
|
|
"model", # Property on FallbackAwareAnthropicMessagesStream, so its reads in tests are never an ast.Call
|
|
"_routing_groups", # Property getter and setter reads are never ast.Call nodes
|
|
"_request_header", # Tested through Claude Code session routing in test_router.py
|
|
"_claude_code_session_router_cache_key", # Tested through Claude Code session routing in test_router.py
|
|
"_delete_claude_code_session_router_binding", # Tested through Redis cleanup failure in test_router.py
|
|
"_resolve_claude_code_session_router", # Tested through Claude Code session routing in test_router.py
|
|
"_get_claude_code_session_router_binding", # Tested through the two-worker session routing test in test_router.py
|
|
"_apply_updated_routing_strategy_args", # Tested via update_settings in test_lowest_latency.py (file lacks "router" in name)
|
|
"_wait_for_scheduler_turn", # Tested through prioritized acompletion and atext_completion in test_router.py
|
|
"arm_routing_read_prefetch", # Tested in tests/unit/caching/test_request_redis_batch_pre_call.py (file lacks "router" in name)
|
|
"_configured_model_info", # Tested through get_configured_service_tiers in test_router.py
|
|
"_routable_deployments", # Tested through get_configured_service_tiers and get_routable_upstream_model in test_router.py
|
|
"_async_get_available_deployment", # Body of the `route {model}` phase wrapper, exercised through async_get_available_deployment in test_router.py
|
|
"_async_get_available_deployment_for_pass_through", # Same, through async_get_available_deployment_for_pass_through in test_router.py
|
|
"_embedding",
|
|
"_aembedding",
|
|
"_stream_pre_content_error", # Tested through the non-retriable retry error tests in test_router.py
|
|
"_narrow_to_video_deployment", # Tested through video_status and avideo_status polls in test_router_provider_endpoints.py
|
|
"_video_call_pin", # Tested through the video_status pin tests and the chat video_id test in test_router_provider_endpoints.py
|
|
"_deployment_num_retries", # Tested through the deployment num_retries mid-stream budget test in test_router.py
|
|
"_request_fallback_list", # Tested through every mid-stream retry test in test_router.py
|
|
"_request_model_group", # Tested through test_anthropic_messages_retry_budget_precedence_direct_call
|
|
"_mid_stream_retry_trigger", # Tested through the retry policy mid-stream budget test in test_router.py
|
|
"_anthropic_messages_group_retry_policy", # Tested through the retry budget precedence test in test_router.py
|
|
"_anthropic_messages_resolved_retry_policy", # Tested through the malformed retry policy tests in test_router.py
|
|
"_anthropic_messages_plain_retry_budget", # Tested through the retry budget precedence test in test_router.py
|
|
"_stream_should_retry", # Tested through every mid-stream retry test in test_router.py
|
|
"_astream_retry_same_group", # Tested through the dropped-before-content retry tests in test_router.py
|
|
"_aanthropic_messages_yield_recovered", # Tested through every mid-stream retry and fallback test in test_router.py
|
|
"_anthropic_messages_policy_retries", # Tested through the retry budget precedence test in test_router.py
|
|
"_get_wildcard_deployments", # Tested through the get_model_list_of_routed_group wildcard test in test_router.py
|
|
"_is_fallback_hop", # Tested through the order fallback hop tests in test_router_order_fallback.py
|
|
"_deployment_that_just_failed", # Tested through the same-boundary hop test in test_router_order_fallback.py
|
|
"_merged_tools", # Tested through the _update_kwargs_with_deployment tools merge tests in test_router.py
|
|
"_drops_malformed_request_tools", # Tested through the _update_kwargs_with_deployment tools merge tests in test_router.py
|
|
"_collect_ptu_share_teams", # Cached behind team_ptu_ceiling, tested through the live model list test in test_router.py
|
|
"_resolve_team_ptu_ceiling", # Cached behind team_ptu_ceiling, tested through the live model list test in test_router.py
|
|
"_register_continuation_prefill_check", # Tested through the continuation tests in test_router.py
|
|
"_set_mid_stream_fallback_continuation", # Tested through the update_settings continuation test in test_router.py
|
|
"_is_flag_enabled", # Tested through the string-flag continuation tests in test_router.py
|
|
"_params_sent_to_deployment", # Tested through the configured-constraint continuation tests in test_router.py
|
|
"_names_deployment_that_cannot_continue", # Tested through the deployment-id continuation tests in test_router.py
|
|
"_set_retry_policy", # Tested through update_settings in test_router_retry_policy_update.py
|
|
"_resolve_pre_call_check_input_tokens", # Tested through the token-count failure pre-call check tests in test_router.py
|
|
"_params_neither_dropped_nor_allowed", # Tested through the allowed-or-dropped param pre-call check tests in test_router.py
|
|
"_deployment_rejects_params", # Tested through the unsupported-param pre-call check tests in test_router.py
|
|
"_deployment_lacks_params", # Tested through the unsupported-param pre-call check tests in test_router.py
|
|
"_pre_call_check_requested_chat_params", # Tested through the unsupported-param pre-call check tests in test_router.py
|
|
"_deployment_supports_param", # Tested through the deployment-capability pre-call check tests in test_router.py
|
|
"_capability_model_and_provider", # Tested through the authenticating-provider pre-call check test in test_router.py
|
|
"_exact_base_model_max_input_tokens", # Tested through get_router_model_info in test_router_model_cost_isolation.py
|
|
"_base_model_widened_window", # Tested through get_router_model_info in test_router_model_cost_isolation.py
|
|
"_provider_model_id", # Tested through get_router_model_info in test_router_model_cost_isolation.py
|
|
"_names_route_model", # Tested through get_router_model_info in test_router_model_cost_isolation.py
|
|
"_declared_base_model", # Tested through get_router_model_info in test_router_model_cost_isolation.py
|
|
"async_admit_deployment_turn", # Bound as the WebSocket turn_admission partial, tested through it in test_router_helper_utils.py
|
|
"_deployment_as_configured_now", # Read by async_admit_deployment_turn, tested through turn_admission in test_router_helper_utils.py
|
|
"_admit_turn_through_routing_strategy", # Run by async_admit_deployment_turn, tested through turn_admission in test_router_helper_utils.py
|
|
"without_deployments_that_cannot_batch", # Tested through the Azure and OpenAI mixed-group acreate_batch tests in test_router.py
|
|
"batch_create_routing", # Tested through the mixed-group acreate_batch tests in test_router.py
|
|
"is_batch_create_routing", # Tested through the chat-after-batch-create test in test_router.py
|
|
"is_batch_capable", # Tested through the mixed-group acreate_batch and model group info tests in test_router.py
|
|
"deployments_the_caller_may_use", # Tested through the usage-based-routing team batch create test in test_router.py
|
|
"deployment_supported_endpoints", # Tested through the model group info batch endpoint tests in test_router.py
|
|
"_endpoint_paths", # Read by deployment_supported_endpoints, tested through the malformed endpoint list group info test in test_router.py
|
|
"_deployment_field", # Read by is_batch_capable, tested through the mixed-group tests in test_router.py
|
|
"_deployment_provider", # Read by without_deployments_that_cannot_batch, tested through the mixed-group tests in test_router.py
|
|
"_cannot_run_batches", # Tested through the Azure and OpenAI mixed-group acreate_batch tests in test_router.py
|
|
]
|
|
|
|
|
|
def main():
|
|
router_file = [
|
|
"./litellm/router.py",
|
|
"./litellm/router_utils/batch_utils.py",
|
|
"./litellm/router_utils/pattern_match_deployments.py",
|
|
]
|
|
# router_file = [
|
|
# "../../litellm/router.py",
|
|
# "../../litellm/router_utils/pattern_match_deployments.py",
|
|
# "../../litellm/router_utils/batch_utils.py",
|
|
# ] ## LOCAL TESTING
|
|
tests_dir = "./tests/" # Update this path if your tests directory is located elsewhere
|
|
# tests_dir = "../../tests/" # LOCAL TESTING
|
|
|
|
router_functions = []
|
|
for file in router_file:
|
|
router_functions.extend(get_functions_from_router(file))
|
|
print("router_functions: ", router_functions)
|
|
called_functions_in_tests = get_all_functions_called_in_tests(tests_dir)
|
|
untested_functions = [fn for fn in router_functions if fn not in called_functions_in_tests]
|
|
|
|
if untested_functions:
|
|
all_untested_functions = []
|
|
for func in untested_functions:
|
|
if func not in ignored_function_names:
|
|
all_untested_functions.append(func)
|
|
untested_perc = (len(all_untested_functions)) / len(router_functions)
|
|
print("untested_perc: ", untested_perc)
|
|
if untested_perc > 0:
|
|
print("The following functions in router.py are not tested:")
|
|
raise Exception(
|
|
f"{untested_perc * 100:.2f}% of functions in router.py are not tested: {all_untested_functions}"
|
|
)
|
|
else:
|
|
print("All functions in router.py are covered by tests.")
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|