scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-11-12 08:42:18 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 207

test(count_tokens): check a Converse Bedrock deployment's full count beta set on Mantle and the runtime
Some checks failed
Cost map pins recheck / rerun-cost-map-guard (push) Has been cancelled
Details
Terraform Modules / fmt, validate, test (aws) (push) Has been cancelled
Details
Terraform Modules / fmt, validate, test (gcp) (push) Has been cancelled
Details
Terraform Provider / Provider endpoints vs proxy OpenAPI schema (push) Has been cancelled
Details
LiteLLM Rust DB / db-queries (push) Has been cancelled
Details
LiteLLM Rust / rust-lint (push) Has been cancelled
Details
LiteLLM Rust / rust-test (push) Has been cancelled
Details
LiteLLM Rust / rust-wheel (push) Has been cancelled
Details
Terraform Provider / gofmt, vet, build, test (push) Has been cancelled
Details

Browse source 
The Mantle case now sends several caller betas and expects exactly the set a
bedrock_mantle deployment's count sends, and the runtime case for a Converse
deployment expects the betas invoke inference sends, so a count that drops a
kept beta or filters through Converse's config fails
This commit is contained in:
mateo-berri 2026-11-07 06:35:35 -08:00
parent 6ccfda3618
commit 80efb123d9
1 changed files with 24 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

35
tests/unit/llms/bedrock/count_tokens/test_bedrock_token_counter.py
View file

@ -771,14 +771,20 @@ async def test_mantle_count_for_a_bedrock_deployment_keeps_and_drops_the_caller_
_INTERLEAVED_THINKING: Final = "interleaved-thinking-2025-05-14"
_CONTEXT_1M: Final = "context-1m-2025-08-07"
_MANTLE_COUNT_MODEL: Final = "bedrock_mantle/anthropic.claude-opus-5-5"
_CONVERSE_DEPLOYMENT: Final[Mapping[str, JsonValue]] = {
"litellm_params": {**DEPLOYMENT["litellm_params"], "model": "bedrock/converse/us.anthropic.claude-opus-5-5"}
}
@pytest.mark.asyncio
@pytest.mark.usefixtures("local_beta_headers_config", "aws_credentials_in_the_environment")
async def test_runtime_count_without_a_deployment_sends_the_caller_betas_invoke_inference_sends() -> None:
caller_beta: Final = f"{_INTERLEAVED_THINKING},{_CONTEXT_1M}"
@pytest.mark.parametrize("deployment", (None, _CONVERSE_DEPLOYMENT), ids=("no-deployment", "converse-deployment"))
async def test_runtime_count_without_a_deployment_or_for_a_converse_one_sends_the_caller_betas_invoke_inference_sends(
deployment: Mapping[str, JsonValue] | None,
) -> None:
caller_beta: Final = f"{_INTERLEAVED_THINKING},{_CONTEXT_1M},{ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER}"
sent: Final = await _invoke_inference_betas(caller_beta, {})
bedrock: Final = await _count_with_caller_beta("us.anthropic.claude-opus-5-5", caller_beta, None)
bedrock: Final = await _count_with_caller_beta("us.anthropic.claude-opus-5-5", caller_beta, deployment)
assert bedrock.posted_hosts() == (RUNTIME_HOST,)
counted: Final = _betas_besides_effort(_invoke_body(bedrock.requests[0]))
@ -912,9 +918,6 @@ async def test_mantle_count_for_a_converse_deployment_sends_the_format_and_betas
assert _sent_betas(bedrock.requests[0]) == _sent_betas(inference.requests[0]) != ()
_CONVERSE_DEPLOYMENT: Final = {
"litellm_params": {**DEPLOYMENT["litellm_params"], "model": "bedrock/converse/us.anthropic.claude-opus-5-5"}
}
_SYSTEM_OUTPUT_CONFIG_AFTER_A_USER_TURN: Final[tuple[Mapping[str, JsonValue], ...]] = (
{"role": "user", "content": "Hi"},
{"role": "system", "content": "Answer briefly", "output_config": _EFFORT["output_config"]},
@ -940,15 +943,25 @@ _SYSTEM_OUTPUT_CONFIG_DIRECTIVE_ONLY: Final[tuple[Mapping[str, JsonValue], ...]]
),
ids=("tool-input-examples", "system-output-config-after-a-user-turn", "system-output-config-directive-only"),
)
async def test_mantle_count_for_a_converse_deployment_keeps_the_caller_beta_its_body_fields_need(
async def test_mantle_count_for_a_converse_deployment_sends_a_mantle_deployments_betas_with_the_one_its_body_needs(
messages: Sequence[Mapping[str, JsonValue]], tools: Sequence[Mapping[str, JsonValue]] | None, caller_beta: str
) -> None:
bedrock: Final = await _count_with_caller_beta(
_MANTLE_COUNT_MODEL, caller_beta, _CONVERSE_DEPLOYMENT, messages=messages, tools=tools
caller_betas: Final = f"{caller_beta},{_CONTEXT_1M},{_INTERLEAVED_THINKING}"
converse: Final = await _count_with_caller_beta(
_MANTLE_COUNT_MODEL, caller_betas, _CONVERSE_DEPLOYMENT, messages=messages, tools=tools
)
mantle: Final = await _count_with_caller_beta(
_MANTLE_COUNT_MODEL,
caller_betas,
{"litellm_params": {**DEPLOYMENT["litellm_params"], "model": _MANTLE_COUNT_MODEL}},
messages=messages,
tools=tools,
)
assert bedrock.posted_hosts() == (MANTLE_HOST,)
assert caller_beta in _sent_betas(bedrock.requests[0])
assert converse.posted_hosts() == mantle.posted_hosts() == (MANTLE_HOST,)
counted: Final = _sent_betas(converse.requests[0])
assert counted == _sent_betas(mantle.requests[0])
assert caller_beta in counted
@pytest.mark.asyncio